ST/ESA/STAT/SER.F/88
(Final Draft)
25 February 2002
Handbook
of Statistical Organization, Third Edition: The Operation
and Organization
of
a Statistical Agency
United
Nations Statistics Division
December
2001
Final Manuscript of a United Nations Publication now being edited and printed.
Contents
Foreword.
Introduction.
Chapter 1 FOUNDATION OF A STATISTICAL AGENCY
1.1. Introduction
1.2. Independence
1.3. Relevance
1.4. Credibility
1.5. Respondent policy
Chapter 2 THE STATISTICAL SYSTEM
2.1. Introduction.
2.2. Structure of
the statistical system
2.2.1. Single institutions.
2.2.2. More than
one institution.
2.2.3. Importance
of a brand name recognition.
2.3. Coordination
tools.
2.3.1. Incentives.
2.3.2. Some useful
mechanisms.
2.3.3. National Statistics
Council
2.3.4. Coordinating
budgets and standards.
2.3.5. Placement
of staff
2.3.6. Coordination
tools for a decentralized organization.
2.3.7. The role of
the Central Bank.
2.4. The chief statistician.
2.4.1. Introduction.
2.4.2. The chief
statistician’s superiors.
2.4.3. The chief
statistician’s colleagues.
2.4.4. Qualifications
of the chief statistician.
2.4.5. Term of office
of the chief statistician.
2.4.6. Demanding
resignation and threatening resignation.
2.4.7. Authority
of the chief statistician.
2.4.8. Competence
of the chief statistician.
2.5. The National
Statistics Council
2.5.1. Oversight
or advice.
2.5.2. Membership.
2.5.3. Chairmanship.
2.5.4. Secretariat
2.5.5. Agenda.
2.5.6. Frequency
of meetings.
2.5.7. Advisory committees.
2.5.8. Ad hoc bodies.
2.6. The law
2.6.1. Introduction.
2.6.2. Main Actors.
2.6.3. The law: short
or long.
2.6.4. The law: deterrence
and enforcement
2.6.5. Access to
information protected by other laws.
2.6.6. Legal advice.
2.6.7 Special legal
arrangements in decentralized systems.
2.7. Financing the
statistical system
2.7.1. Sources of
finance.
2.7.2. Financing
through government budgets.
2.7.3. Market Pricing
of goods and services.
Chapter 3 USERS AND
THEIR NEEDS.
3.1. The needs of
government
3.1.1. Introduction.
3.1.2. Ministries
of Finance.
3.1.3. Other Ministries.
3.1.4. Organizing
and establishing contacts.
3.1.5. Regional and
local government
3. 2. The needs of
the public.
3.2.1. The community
at large.
3.2.2. Schools and
high schools.
3.2.3. The press.
3. 3. The needs of
business.
3.3.1. Large business.
3.3.2. Small business.
3. 4. Research and
other needs.
3.4.1. Academic world.
3.4.2. International
institutions.
Chapter 4 SETTING
PRIORITIES.
4.1. Analysis of
user requirement
4.2. Developing a
progress plan.
4.2.1. Introduction.
4.2.2. The annual
plan and its constituents.
4.2.3. Making allowances
for size and uncertainties.
4.2.4. Ceremony.
4.2.5. Planning and
budgeting.
4.2.6. Planning and
accountability.
4.2.7. Evaluation
of the plan.
Chapter 5 PRINCIPLES
OF ORGANIZATION AND REORGANIZATION
5.1. Introduction.
5.2. Theories and
trends.
5.3. Organization
by subject or by function.
5.4. Organization
and reorganization.
5.5. Coordination
mechanisms and corporate culture.
Chapter 6 MANAGING
STAFF
6.1. Introduction.
6.2. Staff composition.
6.3. A human resources
policy.
6.4. Recruitment
6.5. The first day.
6.6. Training.
6.7. Job rotation.
6.8. Staff retention.
6.9. Options for
a new chief statistician.
6.10. Human resource
management: a package.
Chapter 7 MANAGING
INFORMATION TECHNOLOGY
7.1. Introduction.
7.2. IT management
models.
7.3. Review of some
concerns prevalent at the end of the 70’s.
7.4. Working with
the commercial sector – outsourcing.
7.5. Sound IT project
management practices.
Chapter 8 INTANGIBLES ASSETS.
8.1. Definition.
8.2. The stamp of
quality.
8.3. The force of
law
8.4. Policies.
8.5. Registers.
8.6. System of National
Accounts.
8.7. Central review
and control of questionnaires.
8.8. Nomenclatures
and classifications.
8.9. Finance.
Chapter 9 MANAGING
THE CORE FUNCTIONS.
9.1. Essential Functions.
9.1.1. Planning.
9.1.2. Executing
and Processing.
9.1.3. Analysis.
9.1.4. Dissemination.
9.1.5. Coordination
and Standardization.
9.2. Survey and census.
9.2.1. Introduction.
9.2.2. Project management
9.2.3. Working with
project staff
9.2.4. Working with
respondents.
9.2.5. Repository
of statistical expertise.
9.2.6. Flexible survey
taking capability.
9.2.7. Administrative
records.
Chapter 10 FIELD
ORGANIZATION
10.1. Interacting
with respondents.
10.1.1. Introduction.
10.1.2. Households.
10.1.3. Enterprises.
10.1.4. The consumer
price index: a special case.
10.1.5. Government
10.1.6. Special case
of provincial and local government
10.1.7. Non-profit
institutions.
10.2. Interacting
with a subject matter organization.
10.2.1. Perspectives.
10.2.2. Censuses
and surveys.
10.2.3. Managing
interactions.
Chapter 11 GETTING
INFORMATION TO THE USERS.
11.1. General dissemination
issues.
11.1.1. Introduction.
11.1.2. Providing
users information on the properties of statistical data.
11.1.3. Dissemination
policy.
11.2. Different forms
of dissemination.
11.2.1. Traditional
means.
11.2.2. Multimedia.
11.2.3. Internet
11.3. Recovering
the cost of publication.
11.3.1. Introduction.
11.3.2. Commercial
policies.
11.3.3. Using third
parties for dissemination.
11.3.4. Copyright
and royalties.
11.4. Supplementing
information with analysis.
11.4.1. Should an
NSO do analysis?.
11.4.2. Facts and
interpretation.
11.4.3. Analytical
functions and information.
11.4.4. Review of
publications.
11.5.The Statistical
Yearbook.
Chapter 12 RESPECTING
PRIVACY AND PRESERVING CONFIDENTIALITY: HONOURING THE CONTRACT
12.1. Respondent
policies.
12.1.1. Introduction.
12.1.2. Principles
of respondent relations.
12.1.3 Compulsion
and voluntary response.
12.1.4. Incentives
and assurances.
12.1.5. Sharing Administrative
Information.
12.2. Data protection.
12.2.1 Physical and
Electronic Security.
12.2.2. Sanctions.
12.3. Confidentiality
and disclosure.
12.3.1. Principles.
12.3.2 Dominance
and residual disclosure.
12.3.3 Confidentiality
and household statistics.
12.3.4 Disclosure
with consent
12.3.5 Forced disclosure.
12.3.6 Passive confidentiality.
12.3.7 Arrangements
for Research.
Chapter 13 PUTTING
IT ALL TOGETHER
13.1 Introduction.
13.2. Principles
of organization.
13.3. Internal Functions.
13.3.1. Office of
the chief statistician.
13.3.2. Dissemination.
13.3.3. The field
organization.
13.3.4. Subject matter
departments.
13.3.5. Statistical
infrastructure.
13.3.6. Technical
infrastructure.
13.3.7. The analytical
function.
13.3.8. Planning.
13.3.9. Horizontal
and vertical mechanisms.
13.4. Building external
support
APPENDIX
In 1954,
two years after sponsoring an interregional seminar to discuss the matter,
the United Nations Statistical “Handbook of Statistical Organization” in the series
Studies in Methods, Series F, No. 6 was published. In 1980, seven years after convening
a second interregional seminar, the Statistical Office published the first
revision to the Handbook as Series F No. 28. In 1999, at a data quality seminar
cosponsored by the International Monetary Fund and the United Nations, several
countries requested that the Handbook be updated. The United Nations Statistical
Commission confirmed this request.
This
version of the Handbook is largely based on a document prepared by Jacob Ryten
under a consulting arrangement with the United Nations Statistics Division.
The initial review of the document was made by staff in the Statistics Division
and an editorial team representing countries from each of the United Nations
Regional Commissions and the International Monetary Fund. Subsequently individual
Chapters were used at various seminars conducted by the Statistics Division.
Finally,
one must allow that implementation of this Handbook can proceed in one of
two quite different directions. On the one hand, the measures advocated in
this Handbook should not be adopted one at a time but rather as a package
since they act in concert with one another. However, it must also be recognized
that most agencies desiring to implement the principles in this Handbook will
not be in a position to implement that all at once. In fact, the most sensible
course of action for a country may be to concentrate on just one area, say,
for example, managing the core functions of a statistical office. In any event,
each statistical office must decide for itself – given its situation and that
of its country – which, if any, of the principles advocated in this Handbook
are applicable to it.
Without
the work of Jacob Ryten this Handbook would not have been written. He wrote
a complete first draft, which turned out to be a Herculean effort. Only those
who have attempted projects of this complexity can truly appreciate his efforts.
The draft was reviewed by an editorial board of: Paul Cheung, Miguel Cevera,
Svein Longva, Guest Charumbira, Hasan Abu Libdeh and Carol Carson who made
numerous suggestions for improvements. Jason Brody made many helpful suggestions
on how the text could be made more easily understandable. Several United Nations
Statistics Division staff contributed to the final draft but three deserve
special mention: Angela Me, who organized the production of early drafts and
in addition made many valuable substantive suggestions at all stages of the
process; Sabine Warschburger and Isabela Heng, who organized and produced
the final draft. Chapter 7 on Managing Information Technology is substantially
the work of Jack Arthur. However, responsibility for the final version of
the Handbook and any weaknesses rests with Hermann Habermann and Willem de
Vries, who made the final decisions on the content and structure of the Handbook.
Nitin Desai
Under-Secretary-General
Department of Economic and Social Affairs
Introduction
There
are two main intended audiences for this Handbook: (i) the Chief statistician
(or soon to be Chief statistician of a statistical agency) and his/her colleagues;
and (ii) those charged with oversight of the official statistics function.
As the
heads of agencies, they are both formally and emotionally committed to continuity
of a tradition embodied by the Fundamental Principles of Official Statistics. They establish public faith in the
facts turned out by the agency; drive innovation in their agency; foster an
ethos of professionalism without which quality standards cannot be maintained;
communicate with both government and public at large regarding the condition
of their country on the basis of those aspects amenable to quantitative measurement.
The Handbook then is designed for those who are responsible for providing
leadership to a statistical agency and for ensuring that the public trust
is not misplaced.
It would
be welcome indeed, if in addition, the Handbook were also useful to those
who simply want to understand why a statistical agency operates in the manner
that it does.
The Handbook
is about national statistical offices – government departments of varying
status, created in order to compile, interpret and make public official statistics.
Such offices come in all sizes and many different shapes. Some are general
in the sense that they compile statistics on all that must be known about
economic and social processes in a country. Others are very specialized: they
only compile statistics on transport, banking or agriculture. Some have high
visibility and have their identity sharply defined by statute. Others are
nested within other organizations and were created chiefly to inform the governing
body of their organization. However, except in size, scale of operations and
scope statistical offices do not differ that much from each other in terms
of their main function, their expected behaviour, and in terms of the rules
that apply to them. They all process raw data, convert them into statistics,
apply objective standards to their operations and make it a condition of survival
to be impartial, neutral, and objective.
The Handbook
is organized into the following areas: general principles for a statistical
agency and system, data collection and respondent policies, principles and
examples of organization and management, and dissemination guidelines.
The Handbook
is not supposed to be read as a story. Rather, it is designed to serve as
a checklist of items a successful statistical agency must take into consideration;
a set of examples that illustrates how principles can be put into practice
in certain typical contexts; a list of factors that may persuade an agency’s
staff to favour one application over another; and finally as a caution that
while certain solutions maybe be preferred, none are universal. In particular
it must be understood that although the Handbook provides clear and firm guidance
in these areas, the situation in each country is unique and only those in
the country can determine the usefulness of the guidance in this Handbook.
For this reason, the Handbook cannot be prescriptive in a generic form. But
the Handbook can be illustrative by mentioning examples and suggesting ways
of learning from them and it can lay out a small number of general principles
which appear to have stood the tests of time and location, tradition and legal
context, politics and economic development, needs and expectations. If it
succeeds in doing that much, its objectives will be met.
Notwithstanding
its flexibility, the Handbook, like its predecessors, is uncompromising in certain respects.
It enjoins the chief statistician and his/her staff to allow no interference
by interested parties in the way facts are assembled and combined into statistics,
nor in the method and timing of their release to the public. And it underscores
repeatedly that the information that statistical agencies collect is to remain
confidential and inviolate. The failure to treat individual information as
a trust would prevent the statistical agency from functioning effectively.
A new
generation, with capabilities and resources heretofore unmatched, finds itself
with the responsibility of heading today’s statistical offices. In fulfilling
its responsibility, it may be aided by records of past practices, especially
those that yielded favourable results. However, the environment in which official
statistics are created and released has changed, and continues to change,
so profoundly that past experience is only partly applicable. Some of the
changes are technology driven and proceed at an incredible pace. Others are
the consequence of social, legal, and institutional innovations. Others still
are the fruit of greater affluence, better education, and more sophisticated
use of knowledge than ever before.
Never
has so much been expected from statistics, never have statisticians had such
means at their disposal, and never has there been so much willingness to learn
from each other and standardize internationally the fruit of that learning.
This does not imply that the distribution of these gains is even or best or
that there are no tensions or even crises. There undoubtedly are. In some
instances, the strong nexus that used to bind governments to their statistical
agencies has been shaken and is more fragile today than it was in the past.
Some agencies are bound to look back on the individual performance of their
particular agency and feel that it has not reflected the technological and
economic changes referred to above.
Previously,
official statistics catered mainly to Ministries of Industry and Finance,
to Planning Commissions and to Ministries of Trade and of course to its most
traditional supporters: the Ministries of Agriculture, Transportation and
Labor. Today the body of users has expended to include departments that look
after the natural environment, housing, welfare, education and health, justice
and energy. In addition, official statistics are now used by large businesses
on a scale commensurate with that of government, and by the general public
(for example in measuring government accountability).
The drive
to make inter-country comparisons is far more widespread than it was when
the first of the Handbooks was drafted. Whereas standards taken for granted
today such as the activity, commodity and occupation classifications were
still in their infancy and in any case were literally handed down to national
statistical offices by a tiny handful of international experts, the participation
of agencies in the drafting of today’s successors to these standards has become
universal. Fifty years ago the domain of international standard setting was
left to international statistical offices. Today, it is a matter that engages
all statistics agencies.
These
then are the features, largely prompted by developments outside the province
of statisticians that necessitate updating old standards while at the same
time paying attention to past experiences of statistical offices. Indeed,
among the many difficulties that statisticians must take into account, perhaps
the most noteworthy is managing an organization that thrives on continuity
but must develop in an environment marked by constant change.
In this
revision of the Handbook, the following are principal innovations:
v
More attention
is paid to the statistical process as a process and as a system, rather than
as various institutions linked by coordinating devices. Accordingly, the debate
between centralization and decentralization that was heavily featured in Handbook
Rev. 1 is now incorporated into a discussion of methods available to strengthen
coordination.
v
It is recognized
that while the accessibility of up-to-date computer equipment and computer
knowledgeable staff has increased greatly in the last twenty years, the management
of information technology is a critical issue deserving special attention.
v
Planning systems
for large statistical undertakings such as censuses are now better understood
and more easily put into practice, and are therefore no longer in need of
detailed examination.
v
The emphasis
on the relationships between statistical offices, respondents and users that
emerged in Handbook Rev. 1 now occupies
a prominent place in this version. There is also greater emphasis placed on
the coordinating power of such elements of the statistical infrastructure
as common classifications, the system of economic accounts and other integrating
tools.
v
This version of the Handbook, unlike its predecessors, attempts
to deal with the concerns of the different classes of users and the statistics
they will require as a result of their differences.
Ultimately,
the successful statistical agency recognizes that ideal solutions are elusive
and that practical, robust, and workable solutions are created when the advantages
and disadvantages of each approach are balanced successfully. Virtually each
chapter that follows illustrates such balances. Of course, these cannot be
universal, because the way in which various factors are weighed against one
another toward workable ends differs according to location and time. It is
up to each agency to gauge its particular set of circumstances sensibly.
The successful
statistical agency also recognizes that measures are not to be adopted one
at a time but rather in concert. Even if initiatives are first evaluated on
their individual merits, they eventually must be judged in relation to other
organizational and procedural measures.
Chapter 1
FOUNDATION OF A STATISTICAL AGENCY
1. National statistical offices exist to
provide information to the general public, government and the business community
in the economic, demographic, social and environmental fields. This information
is essential for development in these areas and for mutual knowledge and trade
among the States and peoples of the world.
2. The quality of official statistics depends
largely on the cooperation of citizens, enterprises, and other respondents
in providing appropriate and reliable data to the statistics agencies.
3. In order that the public trust official
statistics, a statistical agency must have a set of fundamental values and
principles that earn the respect of the public. These include, independence,
relevance, credibility and treating respondents as means rather than as ends.
4. These principles have been codified in
the “Fundamental Principles of Official Statistics”
5. A widely acknowledged position of independence
is necessary for a statistical agency to have credibility and to carry out
its function to provide an unhindered flow of useful, high-quality information
for the public and policy makers. Without the credibility that comes from
a strong degree of independence, users may lose trust in the accuracy and
objectivity of the agency’s data, and data providers may become less willing
to cooperate with agency requests.
6. In essence, a statistical agency should
be distinct from those parts of the department government that carry
out enforcement and policy-making activities. It should be impartial and avoid
even the appearance that its collection, analysis, and reporting process might
be manipulated for political purposes or that individually identifiable data
might be turned over for administrative, regulatory, or enforcement purposes.
7. Characteristics related to independence
are:
v Authority
for professional decisions over the scope, content, and frequency of data
compiled, analysed, or published.
v Authority
for selection and promotion of professional, technical, and operational staff.
v Recognition
by policy officials outside the statistical agency of its authority to release
statistical information without prior clearance.
v Authority
for the chief statistician and qualified staff to speak about the agency’s
statistics before the Government and the public bodies.
v Adherence
to predetermined schedules in public release of important economic or other
indicator data to prevent even the appearance of manipulation of release dates
for political purposes.
v Maintenance
of a clear distinction between the release of statistical information and
policy interpretations of such information by the senior members of the Government.
v Dissemination
policies that foster regular, frequent release of major findings from and
agency’s statistical programmes to the public via the media, the Internet,
and other means.
8. In 2000 the United States National Research
Council in writing the Principles and Practices for a Federal Statistical
Agency,
recognized as fundamental goals of a statistical system the following:
v to
protect confidentiality of responses
v to
minimize the burden of the people who provide the responses
v to
assure accuracy which requires proper concern for consistency across geographic
areas and across the time, as well as statistical measures of errors in the
data;
v to
assure timeliness which requires concern for issuing data as frequently as
is needed to reflect important changes in what is being studied, as well as
disseminating data as soon as practicable after they are collected;
v to
assure relevance which requires concern for improving data that help users
meet their current needs for decision making and analysis, as well as anticipating
future data needs;
v to
establish credibility which requires concern for both the reality and appearance
of impartiality, and of independence from political control.
9. If a statistical agency operates from a
strong position of independence, then how are its objectives and priorities
fixed? The answer is that its objectives are fixed by law, and its priorities must be decided by the
chief statistician. The objectives are often seemingly very simple. For example,
the law governing the Canadian statistical agency states: “…there shall be
a bureau…the duties of which shall be to collect, abstract, compile, and publish
statistical information relative to the commercial, industrial, social, economic
and general activities and condition of the people…”.
However, in its Strategic Overview,
the Canadian chief statistician states that the agency’s medium-term priorities
are provincial statistics, the service sector, science and technology, etc.
The former describes the agency’s accountability; the latter is the chief
statistician’s best interpretation of what the agency should do in the medium-term
in light of perceived demand and the conditions necessary to meet it.
10. A statistical agency is a
service agency, so its independence is related to methods and results, not
to ultimate objectives. For this reason, we begin the discussion of organizational
matters with the topic of relevance. There is no question that the products
of a statistical agency must be national in scope— that is to say, they must
apply to all sectors of a nation’s society and economy. But what does relevance
mean? And what are the constraints, both physical and psychological, that
limit any attempt to be “relevant”?
11. Statistical
agencies should continually look to improve their data systems to provide
information that is accurate, timely, and relevant for changing public policy
needs. A problem is, however, that policy interests may change at a faster
pace than a statistical system can adapt. It takes little time for a concern
to emerge; first as curiosity, next as a subject of discussion, and lastly
as a matter of substantial importance to policy makers. The question of whether
there was a “new economy,” not accounted for by conventional statisticians,
first surfaced in the early nineties in the press and in popular literature.
Within two or three years, this issue became a political priority in a number
of advanced countries, and eventually called into question whether statisticians
had correctly measured the GDP of their respective countries. If this concern
had justified the creation of a new research programme, leading to the possible
replacement of the current system of economic accounts and supporting basic
data, it would have taken years if not decades for such a programme to yield
useful measurements.
12. Another example concerns
the service sector. It took two to three years for the issue of the service
sector-its configuration, productivity, and quality of jobs it offered to
become a serious political concern. But it took over a decade to formulate,
accept, and institute the basic international classifications required to
collect service sector data. Indeed, as late as the end of the nineties, most
statistical agencies were still experimenting with operational frameworks
that would enable them to deal in a meaningful way with the service sector.
13. From these two examples it
follows that recognizing a problem takes far less time than deploying the
necessary means to measure its extent, and making the measurement internationally
comparable. Given this disparity, in a world of rapidly shifting priorities,
a statistical agency striving to be instantly relevant could become systematically
irrelevant in the sense of dealing with matters that once were at the top
of the policy agenda but are no longer of such great importance.
14. There is little point in
attempting to deal with concerns perceived as transient. By the time a programme
devised to deal with them is implemented, the policy agenda will have changed
several times over. In fact, when examining priority options, the statistical
agency will have to sort out transient from more permanent concerns.
15. Once a priority is determined,
it is difficult for a statistical agency to modify it as fast as policy agenda
concerns appear to change. This is why it is crucial to exercise good judgment
in setting priorities and to accurately foresee changes in policy direction.
The chief statistician’s planning involves four important elements:
v devising
programmes that are sufficiently general to adapt easily (after a fashion)
to small changes in policy direction
v building
up a reserve and creating a state of preparedness such that unforeseen contingencies
can be addressed without disturbing the regular functioning of the statistical
agency
v developing
human resource policies designed to make the staff of statistical agencies
adaptable and redeployable so as to track effectively changes in the agency’s
programmes
v sharing
technical information and ideas with other statistical agencies. Such sharing
can stimulate the development of innovative data collection, analysis, and
dissemination methods
16. Ensuring these capabilities gives the statistical
agency a great amount of leverage in its attempt to adapt to problems arising
from shifts in priority.
17. In an environment of social
turbulence, answering to a government using statistical information for planning
and allocating resources, the chief statistician is advised to keep some flexible
and adoptive capability. Also, he/she should avoid overly detailed, very specific
surveys, keeping in mind that policies may change unexpectedly, rendering
such surveys of doubtful applicability. In addition, there is no substitute
for gaining advance information on issues troubling policy makers themselves,
so that the statistician is aware of impending changes in the priorities of
the policy agenda. No matter how small an office, the chief statistician must
spend a significant proportion of his/her time in the company of senior government
officials in order to gain the necessary awareness of impending changes.
18. But it is not sufficient
to engage only the head of the agency. Awareness has to extend to the entire
agency and this is why a considerable amount of space is devoted to this topic
particularly in Chapter 3.
19. A special circumstance affects
statistics: the results of the activities of statistical agencies must be
replicable to be believable, but realistically the user cannot replicate them.
This is why a statistical agency must work hard to bolster credibility, and
why there is such extreme sensitivity to any attack on credibility or to notions
of a loss of public faith in the reliability of a statistical agency’s output.
20. Statistical agencies must
be extremely rigorous with respect to the standards that inputs must meet,
methods of processing, and derivation of results, and must instill in their
staff an ethos of quality on par with such rigorous high standards. In this
way, the sense that what is produced is the result of quality inputs, as well
as quality methods of production and control, is constantly reinforced.
21. The need to inspire an ethos
of quality, and to convince all users of the quality of adopted production
processes, has a number of organizational consequences. For example, it is
reassuring to users if periodically the methods adopted by a statistical agency
are subject to an outside process of evaluation and the findings are made
public and open to discussion. It is difficult to convey the notion of limited
openness without giving the impression of patronizing the public. And yet,
uncritical openness might end up having contrary effects to those expected.
For example, no matter how high a quality involved in the compilation of the
national accounts, there is an inevitable residue of estimation based on assumptions
that may be plausible but are not necessarily backed by evidence. Conveying
this bald fact to the public may give an impression of arbitrariness that
in turn could bring the rest of the structure into disrepute. Any sophisticated
analyst would know the limits to the effects of these assumptions in the light
of the system of identities imposed by the accounting framework. How to convey
this to the public in a manner that is not harmful is a matter for careful
thought above all in an environment where open inspection of methods is actively
encouraged.
22. The underlying issue in the
discussion of credibility is how one part of the statistical system can obtain
information from one preceding it in the production chain with complete faith
that quality has not been compromised in the process. For this to be assured,
there must be a subtle combination of subjective elements in place. The spirit
of quality shared by the staff of the agency must never falter, and methods
of inspection and control, of checks and balances, powerful enough to detect,
correct, and prevent future avoidable errors must always be exercised.
23. In discussing credibility,
there is the matter of statisticians interacting with respondents and ensuring
that they provide the best possible answers to the questions that statistical
agencies put to them. “Best possible” means that the required information
should be made available to the official statistical agency without distortion
caused by fear of subsequent use or by failure to comprehend survey questions,
and without reluctance due to the agency’s perceived disrespect for the respondents’
time and privacy.
24. The objectives implied in
the previous paragraph are easier to list than to achieve, and there is no
single method to achieve them. All methods tried so far rely on a combination
of four basic elements: legal instruments to force compliance or to dissuade
disobedience; appeals to respondents’ sense of morality to encourage cooperation;
assurances that the information will not be misused; and increasingly incentives
are being used in some countries.
25. An international survey of
practices and opinions would likely reveal that the confidentiality of individual
information is the greatest concern among respondents. Agencies that have
not yet managed to persuade respondents that the information provided to a
statistical agency is absolutely confidential cannot rely on the quality of
the information they collect.
26. In all likelihood, most statistical
agencies would agree that the power given them by law to solicit information
is of little use unless all sectors of society are willing to cooperate. Those
offices that have made a strenuous effort to convince respondents that the
information they provide is valuable, and that the time taken to provide statistical
information is respected and appreciated, tend to be the ones with the highest
response rates. It should be clear that low response rates are as much of
a flaw in statistical work as is carelessness in the editing and dissemination
of numbers.
27. Persuading respondents to
part with information is a difficult task. (There is a detailed discussion
of this matter in chapter 12). However, it may have a high rate of return
in improving overall quality. In this connection, the national statistical
office should be aware of the relationship between the marginal additional
expense of improving cooperation and its impact on the overall quality of
the resulting statistics. However, other factors must also be taken into consideration.
For example, the rate of return on the marginal expense on improving editing
might be higher than that on improving response rates.
28. All offices must have a
unit dedicated to interacting with respondents. That unit may be part of the
office’s field organization, or the matter may be of sufficient importance
to justify a higher profile unit dealing exclusively with the matter of respondents
policies. The objective of a respondents policy unit is to help raise response
rates, and to ensure that respondents part with information willingly. The
unit’s staff must be equipped to answer questions about the use of the information,
the care with which it is handled, and the general attitude of their agency.
They must avoid the appearance of harassment and of heavy-handedness in quoting
from the law but must be fair and consistent in the way they treat businesses
and households. If there is a perceived crisis in respondents’ relations,
the chief statistician is advised to raise the matter to his level by placing
in charge someone who has his confidence and who reports to as high a level
in the organization as possible. Reporting directly to the chief statistician
may be a good way of showing reluctant respondents the seriousness with which
the agency views the matter.
Conclusions
Statistical agencies are service organizations. Their reason for existing,
growing, and making a visible contribution to the affairs of their government
and society is rooted in their capacity to provide information for the solution
of important issues. However, priorities can change more rapidly than the
agency’s capacity to modify its productive effort. For this reason, it is
important that its senior officers have the intuition and contacts that allow
them to detect serious problems and distinguish them from what may turn out
to be no more than fads.
A strong position of independence is essential for a statistical organization
to establish credibility among its users and to create a relationship of mutual
respect and trust. Collecting, analysing, and disseminating statistical information
should always be distinct from policy-making activities. The chief statistician
should commit to impartiality when dealing with collection and release of
information.
A statistical organization must assure the soundness of the statistical
collection and compilation process, and its inner workings. For this soundness
to be credible to the public, and inspiring to the staff of a statistical
agency, a number of conditions are required:
v the
process must be logically sound
v the
machinery that produces it must be robust
v the
descriptions of machinery and process must be open for inspection and the
result of inspection amenable to public debate
v both
process and machinery must have the capacity to grow and to adapt to new circumstances
and a new environment
Unless a
statistical agency is able to ensure that the information provided to it by
respondents is absolutely confidential it will not be able to rely on the
quality of information it collects and the credibility of the agency will
be in danger.
2.1. Introduction
1. What is the statistical system? Who is
at its head? What is the legal basis of the system? Who makes sure that the
head of the system does the right thing? Who pays for the operation of the
system? In this chapter we examine alternative institutional arrangements
as an indication of possibilities rather than as suggestions for what is recommended.
Clearly, institutional arrangements are divided into those that have worked
well and those that have not, but the relative importance of the factors are
country specific.
2.
While appropriate
organization of the national statistical service is necessary to achieve efficiency
and effectiveness, other factors also contribute to its performance. Often,
it is believed that problems of statistical policy can be solved by organizational
rearrangements, when, in reality, their solution requires other means, such
as capable and inspiring leadership, qualified staff, application of appropriate
methods and techniques, and common sense and hard work. The needs for reorganization
should be identified and satisfied; however, the effect of reorganization
on improvements in statistics should not be overestimated.
3. No where in the world is there a country
that has vested in a single institution the responsibility to collect all
the official statistics of a nation. Rather, statistical systems exist on
a continuum. At one end of the continuum stand those nations in which there
exists a single institution that is responsible for most of the official statistics.
Examples include Australia, Canada and Mexico. However, even on this end of
the continuum some official statistics are the responsibility of others. These
may be agricultural, labor or environmental statistics or statistical information
based on taxes or civil registration and vital statistics. The country most
representative of the other end of the continuum is probably the United States
of America, which has numerous statistical agencies that are, for the most
part, devoted to particular subject matters. When we speak of centralized
or decentralized systems we are generally referring to those that are nearer
the ends of this continuum.
4. We begin with a detailed analysis of the
different types of statistical systems, addressed primarily to those who are
still working through problems of status, organization, and relationship of
their agency to the rest of government.
5. The first version of the Handbook describes
the single institution as one type of statistical system:
“…A system of this nature is typified by the establishment of one department
within the government to organize and operate a scheme of co-ordinated social
and economic [today we would have added and environmental] statistics pertaining
to the whole country. This department collects, compiles and publishes statistical
information …and, in addition, collaborates with other departments of government
in the compilation of administrative and specialized statistics.”
Handbook
6. The advantages it cites for this form of organization are respectively:
v concentration
of specialized and scarce manpower in one centre
v
co-ordination of programmes within one office
v concentration
of scarce equipment and know how to operate it
v easy
recognition of the institution that guarantees quality, impartiality, and
freedom from political interference.
7.
There is a more general formulation given to this kind of system in
the first revision of the Handbook:
“…A national statistical service is centralized
if the management and operations of the statistical programmes are predominantly
the responsibility of a single autonomous government agency, headed by the
country’s chief statistician. Centralization can include outposting of staff
to other departments or the delegation of certain functions to geographically
separate units, which, however remain subordinate to the central authority.”
Handbook Rev.
1
8. The
same handbook adds to the list of advantages mentioned above “…it is furthermore
convenient and efficient for users to secure statistical materials in a variety
of fields from a single source…[and] respondents to censuses and surveys find
it convenient to deal with a single office, especially if they suspect duplication.”
The text goes on to argue that a centralized organizational arrangement makes
it easier to create customized tabulations, and cross cutting arrays of data
sometimes necessitated by user concerns. The text adds that protection of
“incoming” confidentiality as well as “outgoing” confidentiality is easier
and more credible if conducted by a single rather than by several agencies.
Finally the text concludes, “…it should be easier for such an office to maintain
balance in the priorities assigned to different statistical fields, that is,
to co-ordinate the entire service.” The subsequent discussion stresses the
need for an integrated system of social and economic statistics; assuming
that both systems have sufficient funding and capable leadership, this is
more easily achieved if the co-ordination problems are internal rather than
external.
9. Clearly, the circumstances that exist
today are not the ones that existed when the Handbook Rev. 1 was developed.
Some changes since then relate to:
v the
availability of computing and printing equipment at relatively low cost and
the wide diffusion of the necessary know how to operate it;
v the
easier access to mathematical/statistical techniques and their availability
in “canned” form for easier application to problems of estimation and formal
data analysis
v the
introduction of advanced quantitative methods in the teaching of the social
sciences
v the
globalisation of curricula
v the
advent of universal telecommunications, including worldwide access to and
use of the Internet.
v the
sophistication of use and of users has increased noticeably, implying that
their demand for integrated statistics has also increased
v the
pressure on public sector resources has increased greatly, despite the progressive
relative cheapening of computing and data storage facilities
v increasingly,
private sector enterprise carries out some functions once associated exclusively with statistical
agencies, centralized or otherwise
v the
pressures to produce internationally comparable data have multiplied not only
within new inter-country arrangements such as the European Union but also
because of the greater number of Free Trade and Customs Union entities
10. Despite these changes in the environment,
most of the advantages of centralization given in the Handbook Rev.1 are still
valid.
11. Whether or not these changes,
taken together in a national context, argue for a radical change in the institutional
arrangements prevalent to date is something for each government to answer,
fully aware of its own circumstances. However, today’s environment has probably
affirmed the following note of caution from Handbook Rev. 1:
“ …amendments in the scope and range of activities
within a national statistical service require specific high level policy decision
by the central government and may even involve special legislation. Such a
decision is not likely to be influenced entirely by the results of an objective
analysis of the alternatives. Inevitably, past practices, interdepartmental
rivalries, the structure and size of government, the impact of tradition and
personalities and so on come into play. Moreover, when possible changes of
the organization of a national statistical service are considered, short-run
disruptions in service that may be induced by the change must be weighed against
the long-run gains…”
12. There are two advantages
that are usually given for decentralized systems. The first is that the statistical
agency is close, both physically and intellectually to the subject matter
policy office. In this case it may be more likely that the statistical agency
will be responsive to policy needs and be aware of impending requirements.
The second has to do with the chief statistician. As this Handbook points
out the chief statistician is enormously important in defining the culture
of the institution, and in providing intellectual and ethical leadership.
In a decentralized system a poor choice for the head of one agency or institution
will not directly affect the culture in the other agencies.
13. The first edition of the
Handbook
distinguishes three types of decentralized systems:
v a
statistical system decentralized by subject with a co-ordinating authority.
v statistical
systems decentralized by subject with no central control or co-ordination;
v statistical
systems decentralized by subject with a minimum of control or co-ordination;
14. The first of these is to be avoided if at
all possible. It is the product of historical accident and rapid, disorderly
rise in demand for quantitative information by a handful of ministries and
other government agencies.
15. The second type of system,
in which there is effective coordination even though it is not mandated by
law, provides contrast to a situation in which there are de jure, but not de facto,
methods for bringing about coordination. This situation would arise in the
case of indifferent leadership or if the system has become ineffectual as
a result of inter-institutional jealousies and rivalries.
16. So far, the forms of decentralization that
were discussed had to do with institutions outside the central statistical
agency dealing with certain subject matter areas. An entirely different form
of decentralization that exists in many countries, in various forms, is regional
decentralization. This form of decentralization too, may imply certain coordination
problems. Whether such decentralization is, on balance, efficient and effective
depends very much on national circumstances. Often it is a function of either
the size or the country or of the governmental (federal) structure.
17 There are two basic forms of such decentralization,
but of course there are many national situations that are a mix of the two.
18 The first form is comprised of regional
statistical offices, sometimes in two hierarchical layers (e.g. regions and
districts), but totally controlled by the central statistical office. Such
systems are sometimes called ‘vertical’. In this form, the role of the regional
offices is usually exclusively data collection as prescribed by the central
office. Often, such a system has to do with the size of the country, in combination
with developing communication systems. Coordination in such systems does not
have to be a significant problem, provided the central office has sufficient
resources for training and instruction of regional staff.
19
In some cases, the regional offices are not only responsible for data
collection, but also have an important role to play as regional dissemination
centres, e.g. in France and Canada. In the case of Australia, the regional
offices act as national centres
for certain areas of statistics. e.g. the office in Victoria (Melbourne) is
responsible for the compilation and dissemination of statistics about the
services sector of the entire country.
20
In the other form of regional decentralization, the regional offices
are not directly or not exclusively
controlled by the central office, but are part of the regional administration
of the country. Often, this occurs in federal systems of government. A well-known
example is Germany, where cooperation between the Federal Statistical Office
and the statistical offices of the German State governments is largely based
on a complex system of agreements about data collection, statistical standards
etc. Switzerland is another example. Coordination in such systems need not
be a major problem, but may be rather more difficult to achieve and require
more effort than in ‘vertical’ systems. For example in Spain, the Autonomous
Regions have their own statistical offices that are part of the Regional administration,
but the statistical offices in the provinces that are part of the Autonomous
Regions belong to the central system.
“…A distinction must be made between legal status
and status in the sense of an agency’s professional and administrative standing
in the eyes of other government bodies and the public. It is in fact status
in the latter sense that matters most for the external capability of a statistical
agency or service.”
Handbook Rev.
1,
21.
Usually a stand alone statistical agency is recognized as an identifiable
agency within the central government, its director reports to a minister or
equivalent; generally it occupies its own building and in the eyes of the
public its activities are attributed to the agency itself, rather than associated
with the governing ministry or an equivalent governing body. Thus in countries
where an Institute of Statistics compiles the CPI, the public would recognize
that the CPI is compiled by the Institute of Statistics and not, for example,
by the Ministry of Economic Affairs.
22.
Being a recognizable agency within government goes with budgetary recognition. This means that in the government accounts
there will be one line mentioning the Institute and the corresponding appropriation.
Virtually in all countries where the statistical system is of some size and
there is a dominant agency, that agency stands alone whether or not it is
called an “Institute”. In smaller countries though (and during the period
preceding WWII) the examples of Directorates or Directorates General nested
in a ministry are numerous.
Brands and no brand statistics
Recognizing quality in statistics and using them
with trust are closely associated with the recognition of the agency that
has compiled them. The more instinctive the recognition, the greater the chance
of taking the information at face value because of the element of trust. But
for recognition to be instinctive, the agency must be visible and its visibility
increases if it stands on its own as part of central government.
2.3. Coordination tools
23. Regardless of where on the
continuum is the particular structure of a nation’s statistical system some
coordination will be necessary. Coordination is desirable for a variety of
reasons, including:
v To
create a national statistical system in which the outcomes of various data
collections are comparable or can at least meaningfully be related to each
other, harmonization of concepts, definitions, classifications and sampling
frames is necessary.
v To
avoid duplication of effort, as well as undue burdening of respondents, agreements
about efficient and effective data collection are required, including the
use of non-statistical government data files.
v To
strengthen the position of and enhance the image of official statistics, coordination
of dissemination methods and pricing is very useful.
v To
represent a country’s interests internationally, coordination of international
policies is needed.
This section
addresses effective coordination tools in both centralized and decentralized
systems. We first address incentives and deterrents for effective co-ordination;
issues on which co-ordination is required; tools that make co-ordination possible;
and mechanisms that ensure the spread of information and monitor compliance.
These mechanisms are applicable more or less to both systems. We then discuss
some tools specific to decentralized systems.
24. If the various agencies in the system have
the distinct view that they are better off by not being part of a co-coordinated
system, it is unlikely that that co-ordination will succeed no matter how
well legislated. Governments are wary of enforcing coordination with severe
sanctions, and without these, only incentives are likely to work. With respect
to data collection activities, there are, in fact objective reasons why a
ministry may decide to continue operating its own statistical cell rather
than assign the work that it has carried out in the past to the central statistical
agency. Those reasons include:
v the
view that users’ demands can only be met accurately and on time if the work
is handled within the ministry
v the
impression that the central statistical agency does not possess the skills
and knowledge required to do a good job, largely because the ministry is in
charge of a very specialized area or because its infrastructure is better
suited for the particular statistical activity
v the
view that the statistical work of the ministry is an offshoot of its administrative
procedures and is therefore handled more efficiently than if taken over by
third parties
v the
acknowledgement by the ministry that in any case its legislation prevents
the sharing of information with other statistical agencies and that the effort
to change the law would exceed the benefits derived from transferring from
such sharing the responsibility
The incentives
involved in taking part in a well co-ordinated system could include any of
the following:
v
sharing of information that would improve the ministry’s capacity
of statistical response, assuming it has the same attributes (classification,
standards, coverage etc.) as the statistics produced centrally.
v
access to statistical expertise which the statistical cell in
the ministry might lack
v
influence in determining priorities in all-purpose statistical
initiatives such as the Census of Population or in relevant classification
systems
v
engaging in joint statistical activities so as to provide leverage
to its own budget
25. Of
course, the relative weight of any of these will vary over time, and according
to the individual agency’s stance on whether or not the prospects of coordination
might be beneficial, given its programme, users, and budget. Chances are,
it will not be moved by such abstractions as the benefit of an integrated
system of social and economic statistics. It may not even be moved by the
argument that its credibility with users will rise, as will the receptivity
of respondents, once they are aware of the efforts made at avoiding duplication
and simplifying requests. In the majority of situations, such gains will not
be immediately evident. But a clear demonstration that the ministry’s users
feel better off if they are provided with a broader range of data may be the
decisive factor.
26.
Co-ordination
will invariably result in the loss – real and perceived – of independence
by those coordinated. For example, the avoidance of duplication could mean
that a particular statistical cell may have to rely on the statistical or
administrative work of others to find its information. It may find that it
is no longer free to adapt standards and nomenclatures to its convenience
but will have to settle for those adopted by the majority or imposed by the
co-ordinating agency. Sooner or later, the proximity of the cell’s practices
and results to those of other statistical agencies will lead users to question
their consistency, and eventually coordinated answers will be called for.
27.
The most frequently
seen coordination tools are nomenclatures and some combination of questionnaire
and budgetary control. The application of these controls can be formal or
informal, depending on tradition and an assessment of the most efficient way
to ensure compliance. There may even be a law concerning the details of coordination.
Matters are helped if there is a political decision or at least a decision
made at the highest level of the civil service that the only legitimate standards
of classification are those promulgated by the central statistical agency.
In many countries, the statistical agency will act as a mediator between international
organizations and domestic statistical agencies. The former are perceived
to have the legitimacy and knowledge required to determine standard classifications.
28.
In the case
of questionnaires, and even administrative forms that request information
to be used for statistical purposes, one way to ensure co-ordination is to
create a logo or a visa indicating that the form has been inspected and approved
by the relevant central body. It is important that in exercising control,
the central statistical agency be seen as a facilitator, not as an impediment
to initiatives. This is not easy. An inter-agency commission that approves
data collection instruments; controls respondents’ burden; and is not associated
with any particular agency but is provided by the central statistical agency,
might allow for the perception of the central statistical agency as a coordinating
friend rather than as a controlling enemy.
29.
In general,
co-ordinating mechanisms tend to be in the form of committees, formal or informal,
but in any case sufficiently manageable to meet regularly and reach decisions
that the represented parties can put into practice.
30.
For the work
of these committees to be fruitful, they must have the power to approve forms
certifying the use of certain nomenclatures used for statistical purposes;
they require support from higher levels in the political hierarchy; and officials
in each ministry or agency with a stake in the statistical system must demonstrate
that they regard taking part in collective initiatives as a necessary duty.
31.
The central
statistical agency should remember that its concerns will seldom be the highest
priorities for other departments and agencies. Accordingly, it should shoulder
the burden of maintaining coordination. So long as it perseveres and takes
its responsibilities as co-ordinator of the system seriously, thoughtfully,
and in a spirit of co-operation, the central statistical agency can give cohesion
to the system. If the statistical agency were to show the least bit of impatience
or lack of interest, the other participants would take it as a signal that
the coordination effort should be of little consequence to them.
32. Co-ordination may be considerably helped
where there is a National Statistics Council. Countries with no previous experience
with such councils but convinced that their existence may be of help, should
take advantage of any relevant opportunity that offers itself. For example,
having a nationally visible, prestigious figure serve as chair on a sustained
basis provides an opportunity to launch the council and give it substantive
powers. The central statistical agency or the chief statistician should find
an ex-officio role in the council,
thereby strengthening the office’s influence in discussions about co-ordination.
The National Statistics Council will be discussed in detail in Session 2.5.
33.
A powerful
budgetary tool to ensure co-ordination is to estimate a budget for total statistical
activities and to give the co-ordinators the power to allocate it, with fairness
regarding quality and response burden in mind. A co-ordinating agency armed
with the power to make budgetary allocations will find it a good deal easier
to impose standards, although it is clearly in the best interest of users
and of the statistical system to integrate, rather than to maintain sets of
non-comparable statistics
34.
At any time,
a legislator might ask the question, “How much does it cost to produce official
statistics in my country?” This can be an ominous question for statisticians.
Ignorant of the answer, a government in the process of making serious cutbacks
may opt for a disproportionate share to be allocated to the central statistical
agency, oblivious to the fact that it may only carry a small portion of total
statistical activity.
35.
In certain countries, the answer is relatively straightforward. The
cost for a few specialized activities must be added to the initial budget
of the central statistical agency, and the result is the total amount committed
to statistics. However, in many other countries the answer is difficult to
provide with any degree of certainty. The majority of statistical activities
are conducted in other Ministries and agencies, and in those cases where they
are not explicitly noted in their department’s financial statement, there
is no reliable way of estimating their expenditures. Often a literal “guess”
is required, particularly in cases where statistics are a by-product rather
than the focus of a department’s work. Since staff members, and particularly
professional staff, will be engaged in a variety of overlapping activities,
they certainly would not be able to accurately gauge the resources invested
in statistics per se. While estimating expenditures gives a rough idea of
the relative amount committed to statistics, it is unreliable in discerning
year-to-year changes.
36.
The tools of
co-ordination mentioned so far exclude co-ordination through the placement
and intermingling of staff. Managing government statisticians is an issue
that is relevant in only a few countries. Essentially it entails placing in
the hands of a capable authority, the chief statistician being the obvious
choice, the power to certify that one is allowed to work in the public sector
as a statistician. The system can have additional attributes. For example,
the school or institute that trains statisticians may be attached to the statistical
agency. If it is highly respected, it may find itself catering to private
as well as to public sector needs. In those circumstances, the chief statistician
becomes not only the effective head of public sector units conducting statistical
work, but head of the statistical profession irrespective of where its members
exercise their skills.
37.
Where co-ordination is exercised through staff members that share the
same training and professional leadership, many of the coordination tools
mentioned may not be required. If the need for such devices does arise, they
will be created more or less spontaneously, without the need of any great
formality.
38.
A number of agencies have implemented the regular exchange of staff
as a means to expose staff members to a broader range of ideas and practices.
If the central agency is the interlocutor with other countries’ agencies,
it will enhance its prestige and status and serve to foster the ability to
coordinate. Even though this is practiced on a small scale, and involves relatively
few countries, the idea is praiseworthy, as it strengthens the sense of international
community. It can also be used as an incentive by a central statistical agency,
particularly in the case of developing countries, if the exchange takes place
between the agency and its peers abroad.
39. Previous paragraphs discussed
different types of coordination tools that should be considered when the statistical
system is a more centralized one. Clearly the more decentralized a system
the more important coordination becomes. Many of the tools discussed in the
last Section are also applicable to a decentralized one. However, experience
shown that the most important tools that should be considered are the following.
These coordinating functions may be executed by a separate body and not by
one of the several statistics agencies.
v the
ability to control or at least to significantly influence the budgets of the
statistics agencies.
v the
ability to control whether or not an agency is allowed to carry out a data
collection activity.
v the
ability to decide on nomenclatures for the statistical system.
40.
Even if these three tools are employed there is one other important
consideration. This has to do with the manner of the application of the tools.
More specifically the application can be done in an informal manner or there
may be a formal process. In the first place there may be an informal agreement
between the several statistical agencies and the agency that is doing the
coordinating (or a separate coordinating body) that the coordinating agency
will have the authority to apply the tools above. In well-established statistical
systems this informal application can indeed be very powerful. If there is
a lengthy tradition behind the agreement the force to continue their use will
be very powerful. In new or reasonably new systems, however, this may not
be the case and a more powerful application mechanism may be needed. This
is the formal mechanism. The weaker form of the formal approach is through
the order of the Prime Minister or President and the stronger is through legislative
action. While it may be difficult to affect either of these formal mechanisms
once they are in place it will be clear what are the responsibilities of the
members of the statistical system. The coordinating agency will also be in
a much stronger position to apply these tools.
Centralization vs. Decentralization
The
subject of centralization and decentralization is subject of intensive discussion
in the statistical community. This box summarizes some of the most important
issues related to this subject.
- the stronger the powers of co-ordination
at the centre of the system the greater the chances of integrating statistics
effectively
- integrated statistics (definitionally, conceptually,
through the use of harmonized nomenclatures and classification devices)
are immensely more powerful than statistics collected without harmonization.
- If the office in charge of co-ordination is backed by
numbers, legal power to apply the tools of coordination, and a healthy budget
its capacity to co-ordinate is correspondingly greater than that of an office
bereft of these assets. See Section 2.3.6 above.
- in a system where the tools discussed in 2.3 are lacking,
it is vital that the chief co-ordinator have status, access, and the capacity
to offer career opportunities to intelligent, competent and motivated staff.
- physical proximity to policy analysts
can be important to statisticians in determining the policy needs. However,
examples exist of centralized systems, which have found ways to be responsive
to the requirements of policy officials.
- the ability of centralized systems to create convincing
career opportunities tends to increase with size. Small statistical institutes
with a limited number of professional posts give the impression of limiting
career prospects
- economies of scale give the chief
statistician more flexibility in efficient management of his/her budget
- whatever the institutional arrangements,
policy analysts and decision makers particularly in the more specialized
activities must feel they can involve statisticians directly in their quest
for more relevant and reliable quantitative information
Notwithstanding these arguments,
Handbook Rev. 1 recognizes that the practical issue to address is that of
“degree of centralization” about which it says:
There seems to be a consensus
that one of the statistical institutes among the several responsible for
the collection and dissemination of statistics, should “…be responsible
for population censuses, household surveys, demographic statistics and a
wide range of statistics including establishment and enterprise censuses
and surveys, as well as prices and international trade, the national accounts
and other across-the-board activities. There seems to be less of a consensus
in regard to the social sphere – education, health, crime etc. – where statistics
are based to a substantial degree on administrative records collected by
other government departments…” Handbook Rev. 1, p.14.
41. The Statistical Commission noted at its Nineteenth
Session in 1976:
“…many developing countries have experienced
severe losses of professional staff in their central statistical organizations,
thus reducing their contributions to planning and policy formulation processes.
Those losses and the generally severe shortages of trained statistical personnel
in those countries reflected in large part the disadvantaged position of the
statistical services with respect to pay scales and related working conditions…”
42.
This situation
has not changed. In fact the disparity between the working conditions of statistical
agencies and Central Banks, for example, may be increasing. This Handbook
is not the place to lobby for improvement in working conditions for the staff
of statistical agencies. But many of the remarks about organization included
in this book may seem superfluous so long as such serious imbalances persist.
43.
Central banks have taken over a number of key statistical activities
in countries where in spite of a longstanding tradition of statistical production
the budget of the central government is no longer thought sufficient to support
it.
44.
When the macroeconomic statistics and the key economic indicators are
in the hands of the Central Bank, the latter commissions basic statistics
from the statistical agency. In addition, a number of Ministries have their
own statistical budget and use it to compile special performance statistics.
45.
Often there is no co-ordinating mechanism in this type of system, only
the standards required by the system of national accounts and by the needs
of the economic policy makers. While this situation may not be ideal with
respect to the long-term development of the statistical system, it is nonetheless
a workable model.
46.
Existing imbalances in salary and working conditions between the Central
Bank and the statistical agency may significantly limit the agency’s capacity.
In light of such difficulties, it is prudent to recognize the reality of the
situation. Even if there are no legal provisions on which to base co-ordination,
the statistical agency should work out a de facto arrangement with the head of the
Central Bank’s research department.
“…The external capability [of a statistical agency]
is also much influenced by the status of the head of the statistical service
in relation to colleagues in the government hierarchy. There can be no doubt
that the building of a robust statistical service is contingent on assigning
top people to it. In view of the great national importance of many decisions
which may be affected by statistics, it would appear to be in the national
interest for the statistical service to have an equal opportunity to compete
with other government departments for an outstanding individual of proven
capability to be its leader and manager…He or she should be in a position
to develop rapport with policy advisers of other government departments and
to negotiate various proposals on statistics with authority and insight…He
should have the same status as the top civil servant of a ministry.”
Handbook Rev. 1
47. If there is an agency such
that it can be called the central statistical agency, it is the head of that
agency that is referred to here as the chief statistician. The chief statistician
should be the highest authority in the government’s statistical system with
respect to substantive statistical matters.
48. Not all national statistical
systems have an obvious chief statistician.
In this section we pay attention to the matter of status and to the practices
governing appointment and resignation; to matters of authority and competence;
and to the underlying issue of qualifications.
49.
The terms of reporting vary a great deal. By and large, in situations
where the statistical agency is an independent agency the head reports to
a minister. However, the terms of reporting vary a great deal among countries.
In some cases, the minister is responsible for appointment and dismissal or
requesting resignation of the chief statistician. In other cases, the chief
statistician is appointed by Parliament and it is Parliament that delegates
its supervisory functions to one of its members, naturally the minister to
whom the responsibility for the central statistical agency is assigned.
50.
More crucial than the chief statistician’s formal superior is the official
with whom the chief statistician interacts on a day-to-day basis or on key
occasions. For example, there are statistical agencies that are located in
the Prime Minister’s Office, but it does not follow that the chief statistician
interacts regularly with the Prime Minister. In fact, because of the Prime
Minister’s other concerns, the chief statistician may find that his/her regular
liaison is of much lower rank than if the statistical agency were placed under
the supervision of another minister.
51.
There is a question relating to which minister should be placed in
charge of the statistical agency (and if so with what power). The trade off
is easy enough to understand. The more powerful the cabinet position of the
minister in charge, the more the status of that position reflects in principle
on the chief statistician. Other things being equal, more status accrues to
the chief statistician if he reports directly to the Minister of Finance than
to, say, the Minister of Communications.
52.
The matter of reporting to a minister and the coherence between the
minister’s portfolio and the service provided by the statistical agency have
to be understood with a proper perspective. A relationship between chief statistician
and an elected politician is required because the former must be accountable
and take political guidance. Neither of these constraints implies a loss of
independence in the sense that in the end, it is the chief statistician who
decides on methods, results, and their presentation.
53.
Different benefits and drawbacks arise depending on the amount of influence
(not personal, but rather formal) a minister possesses. A highly influential
minister would have more of a hand in determining budgetary allocations. However,
public credibility might be challenged (how would an important minister such
as the Minister of Economic Affairs resist the temptation of putting pressure
on the chief statistician to get more acceptable results for the Government
or for the Party?) and the many claims on the minister’s time would heavily
affect his/her sense of priorities so that his/her interest in the statistical
budget might be lukewarm. Conversely, a less influential minister in charge
of a more neutral and less demanding portfolio – or even a Cabinet member
without portfolio – might preserve credibility and prove to have more time
and interest, but have less influence on the outcome of budgetary allocations.
54.
In many developing countries, a relationship was cemented between the
planning body and the statistical agency. The former, because of its nature
extended its interest to virtually the entire output of the statistical agency
and therefore was in a position to articulate its priorities in a balanced
way. Where Ministries of Planning changed functions, were retired, or simply
renamed, the links forged tended to be transferred to the economic portfolios
that took over from the planning body.
55.
Perhaps the most important advantage of high status is that it affords
the chief statistician greater access to key people in government, and others
who influence public opinion, the economy, and so on. The critical definition
is that of “user”, which extends all the way from the senior official who
proposes quantitative criteria for an assessment of the performance of the
country’s health system, to the junior professional who writes a paragraph
on what the latest indicator of industrial production appears to suggest.
In the structure of the statistical system, access to the latter may be useful,
but access to the former is essential for the preservation of the agency’s
claim to relevance.
56.
Accordingly, the chief statistician must have easy access to the highest
official – the minister or, in many instances, someone just below the minister
– in each of the major user departments Such access, provided it is frequent
and sustained, allows a chief statistician to make informed decisions about
priorities and internal allocations of resources. Without it, the decision-making
process could lack valuable insight, particularly during times of shifting
priorities.
57.
The chief statistician must preserve a delicate relationship, not only
with potential users, but also with the suppliers of information contained
in administrative records, for their degree of responsiveness has substantial
effects on the success of a statistical program. Interaction with senior officials
responsible for taxation, customs administration, social security and the
like is fundamentally different from interaction with potential users. With
users, the chief statistician inquires into concerns, priorities, and trends,
with the distinct objective of providing a service. With suppliers of information,
the chief statistician is vitally dependent on the goodwill of his/her counterparts,
but can give little in return; hence the importance of high status with strong
political support.
58.
If the chief statistician is also the chief co-ordinator of the statistical
system, his/her status must give him/her access to the principals of departments
containing statistical cells and have regular meetings with managers of those
cells.
59.
In the case of federations, if the statistical system is to have a
truly national character the chief statistician must have frequent, unfettered
access to the relevant authorities in the federated entities and be recognized
as the chief statistician or at least the primus
inter pares by his/her regional peers.
60.
Finally the chief statistician must also maintain a rapport with his/her
peers abroad. Heads of statistical agencies, particularly the smaller ones,
who cannot maintain a body of expert economists, sociologists, demographers
and the like, should consult their peers frequently and informally about problems
they share, and judge whether tested solutions abroad could be implemented
at home. Once a fraternal atmosphere is created and maintained, these judgements
can be made more easily. Maintaining these relationships has become incomparably
easier with the relative cheapening of telecommunications and, above all,
with the advent of the Internet. If new chief statisticians do not inherit
this invaluable resource from their predecessors, it should be acquired immediately
and fostered.
61.
There are three aspects to this issue: Who should the chief statistician
be? Who should do the appointment? How should the appointment process be conducted?
62.
There are several attributes a chief statistician should possess. First,
he/she should be a statistician, someone proficient in statistics, or someone
with a profound understanding of statistics. Second, the chief statistician
should be capable of running a large professional organization; and third,
the chief statistician should understand and be sensitive to the wishes of
users whose needs will likely have profound influence on the activities of
his/her agency.
63.
Usually, these attributes are seldom to be found with equal weight
in one single person. Some chief statisticians are distinguished professionals
(statisticians, quantitative economists, demographers, etc.) who bring to
their post the knowledge, prestige, and wisdom they have acquired in the course
of their careers. Others have demonstrated excellent managerial instincts
by successfully directing public programmes, agencies, or private institutions,
preferably with a strong professional bent. Still others bring the experience
and the contacts that only long years in government can produce.
64.
Obviously, the nominating authority will seek to maximize all three
attributes, but it may have to make a choice at a very early stage. It may
decide to select a chief statistician from within the statistical agency,
or prefer to scout the outside world—it may seek the most suitable candidate
in academic or research organizations in an attempt to emphasize the professional
component of the ideal string of attributes, or else it may consider successful
programme managers with the right disposition and academic background to assume
leadership of a statistical agency.
Needed attributes
change over time
As the needs of
the statistical agency evolve, the weights of the various attributes desired
in a chief statistician will vary. Clearly an agency that has run into trouble
over personality conflicts or budgetary difficulties; appears to be chronically
late with its outputs; or has an amorphous programme of future activities;
requires a leader with high-level managerial capacity. An agency that has
allowed itself to be placed at the margin of governmental concerns or appears
to be insensitive to the immediate needs of decision makers, might require
a leader who knows the ins and outs of government. But once the perceived
crisis has subsided, the requirements may change again.
65.
A wise nominating authority demands that the chief statistician be
highly independent. If it appears that the chief statistician has no mind
of his/her own but rather is overly influenced by an interested party in government,
the credibility of key economic indicators, and eventually of the entire statistical
programme, is jeopardized.
66.
The perception of personal and institutional independence is so important
that the process of selection and appointment is given special attention,
at least in some countries. For example, in countries that have a thriving
professional society, government often seeks its advice before making the
final choice.
67. The
term of office of the chief statistician is just as crucial as that of nomination.
There are three standard situations:
v the
chief statistician’s term of office is the same as the Executive’s (this is
the case with the chief statisticians in a number of Latin American countries)
v the
chief statistician is appointed for a term of office that can be renewed
(this is the case with the New Zealand Statistician; the Commissioner, United
States Bureau of Labor Statistics; and the Director, Office for National Statistics,
UK; and appears to be an emerging trend)
v the
chief statistician serves an unlimited term, which ends either with his/her
resignation or removal from office (this is the case with the chief statistician
of Canada)
68
The considerations
at play are fairly obvious. Short spells of office interfere with continuity;
statistical programmes tend to extend over significant periods of time. A
series of chief statisticians would not share a common vision and programme
commitment, and would therefore be unlikely to produce a consistent approach
to statistical policy or predictably adapt to new circumstances. On the other
hand, excessively long stays in office may produce stagnant programmes, lacking
energy and innovation. While such a situation may keep the statistical agency
out of potential conflicts, it may also create a public sector backwater with
no appeal, an inert staff, and eventually reduced budgetary support.
69
These situations represent the two extremes. In designing laws or regulations
and establishing standard practices, one must guard against the greater of
the two evils. One would imagine that in all or at least most cases, there
is a provision that authorizes the political official responsible for the
statistical agency
to request the chief statistician’s resignation.
70
It stands to reason that requesting a resignation is less likely to
be necessary if the chief statistician serves a limited, rather than an open-ended,
term.
71
The existence of an effective National Statistics Council is bound
to have an effect, particularly in preventing arbitrary appointments and dismissals.
72.
Responsible ministers will be very reluctant
to demand the resignations of chief statisticians for reasons other than clear
incompetence. Refusal to disclose confidential information or to obscure or
delay the presentation of a key result should never be a reason. Most governments
will understand that credibility is an essential element of a functioning
system of official statistics. If, for example, a government is elected on
the basis of commitment to full employment, and in fact statistics show strongly
decreasing unemployment numbers during its term, these statistics will only
be important if there is no doubt about the trustworthiness of the statistical
system and indeed the chief statistician. Moreover, the prospect that a chief
statistician will publicly resign rather than compromise his/her integrity
is a strong deterrent against such a request. Identifying confidential data
and modifying data or its presentation are, as mentioned before, examples
of such breaches of integrity.
73.
Conversely, a chief statistician should not
use his letter of resignation as a bargaining tool, except when the integrity
of the office is threatened
74.
So far we have examined the chief statistician’s surroundings – who
appoints, from whom he/she takes orders, and those with whom he/she associates.
We now turn to the authority of the chief statistician.
75.
There are operative failures that the user community should not accept
from any chief statistician. For example, it would be unacceptable if the
CPI was published two months after the period of reference or if prescriptive
comments were made about the adequacy of a particular Government policy. But
it is acceptable for the chief statistician to advise that the extent of revision
of quarterly or annual GDP cannot be cut down substantially with the current
budget constraint. Likewise, the chief statistician can claim that if questions
relating to ethnicity must be the object of a dedicated survey, the budget
required must be increased. If the government lacks confidence in the chief
statistician’s professional assessment, it should request a resignation.
76.
Assuming that the chief statistician’s overall performance is acceptable,
the government should not become overly involved in the internal affairs of
the statistical office—for example, it is unacceptable for the government
to suggest to the chief statistician that in presenting quarterly GDP only
seasonally adjusted numbers should be shown, or that the measure of inflation
should exclude food (this is quite different from asking for a CPI sub-aggregate
that excludes food; the former suppresses information, whereas the latter
is a modification of a standard presentation for the convenience of the analyst).
77.
It follows that the chief statistician must have the legal basis, and
would be expected, to put forward authoritatively his/her best interpretation
of users’ combined wishes; the characteristics (frequency, accuracy, timeliness,
detail) with which he/she believes he/she can meet those wishes; the methods
used to estimate; the burden imposed on the community; and the overall cost
involved. He/she should not be second-guessed or hindered unless it is to
show that the government has lost confidence. It is also understood that when
the chief statistician announces his/her programme, all available resources
will have been consulted.
78.
In situations where in addition to being the head of the official statistical
agency the chief statistician is also the chief co-ordinator of the statistical
system, he/she should be in a position to speak as a co-ordinator (as primus
inter pares). In this capacity the chief statistician
should have the authority to commit the system to the use of certain standards
and quality characteristics, as well as to address the social burden on behalf
of all members of the system.
79.
Finally, the chief statistician should be the spokesman for his/her
country in matters involving international co-ordination and exchanges;
its representative in international meetings; the person who determines standards
where they require revision or updating.
80.
The detailed knowledge required to choose among different systems options
for a Census of Population or for a survey of large and complex enterprises
is what one might commonly associate with a specialist, not with a senior
executive officer. Nevertheless, such knowledge is expected from the head
of a statistical agency. In addition, the chief statistician will be confronted
with matters dealing with health, education, ethnicity, aggregation bias,
treatment of outliers, leads and lags in the foreign exchange market, assets
held by private non banking residents and so on. On each of these subjects
he/she will be expected to provide an authoritative opinion. Gaps in the chief
statistician’s expertise will diminish the professionalism that is one of
his/her claims to independence.
81.
For these reasons, the chief statistician requires above all the prudence
to surround himself/herself with specialists who can propose comprehensible
options and recommend those believed to be in line with the overall framework
for which the chief statistician is responsible. In his/her debriefing on
various options, the chief statistician must be a quick study at times, patient
at others, and must always demonstrate an enormous capacity to listen.
82.
In the last few years, considerable attention has been devoted to increasing
the capabilities of chief statisticians through various forms of international
co-operation. Without creating very intensive co-operative arrangements such
as those that exist in the framework of the European Union, United Nations
regional conferences appear to be growing in stature and a considerable investment
has been made in informal meetings on specialized subjects, albeit at levels
lower than that of chief statistician. The primary impetus for these conferences
is to facilitate access and promote the exchange of opinions on potential
solutions to common problems.
“…Irrespective of the degree of centralization
of the national statistical service a national statistical council or commission,
composed of representatives of the private sector, the universities and government,
may be established either at the top of the external committee structure or
independently of it…, a top-level co-ordinating council may serve as a guiding
and protective device – but mainly as regards the solution of problems of
a general nature. Whether the benefits derived justify the cost is likely
to depend, to a large degree, on national circumstances, including the political
situation…”
83. This
constitutes the only reference made in Handbook
Rev. 1 to councils as advisory or governing institutions. Since these
words were written, the growing importance of such bodies and their introduction
in areas where they did not already exist, has become a visible trend. Their
roles vary
but it is possible to state the following, taking into consideration their
formal mission and the responsibilities they have assumed as their relationship
with the statistical agency and its senior officers has flourished:
v A
national statistical council can be used for the defence of the statistical
agency. It exists to protect the statistical agency from attacks to which
it cannot properly reply due to restraints on public servants.
v The
council can assume the role of guardian of fundamental values such as the
protection of privacy.
v The
council is the ultimate guarantor that within existing resources the statistical
programme as defined by the chief statistician and instituted by his/her agency
preserves the best possible balance among contending claimants for statistical
attention – economic, environmental and social statistics; national and regional
details; reliability and timeliness and so on;
v The
council is the interlocutor a minister might choose if he/she wishes to have
the professional opinion of the chief statistician validated by a group of
impartial experts.
v The
council is the body a minister could turn to for advice and succession management
in the case of a disagreement with the chief statistician.
v The
council’s proceedings would constitute a venue for registering opinions about
the output of the statistical agency.
84.
The combination of these roles is much too important for the study
of a council to be left out of the Handbook. Since the various roles imply different relationships: to the chief
statistician, to the minister and to the public – we must also examine how
the council may or indeed should be inserted in the top-level communications
circuit.
85.
The following schematic presentation summarizes a few of the possible
basic relationships between a statistical agency and what is referred to in
Handbook Rev. 1 in a generic way
as a “National Statistics Council”.
Fig.
1
Illustration of three different sets
of relationships between minister, National Statistics Council and chief
statistician
|
Case 1
|
Case
2
|
Case
3
|
|
|
Legend
|
NSC
|
National
Statistics Council
|
|
CS
|
Chief
statistician
|
|
Minister
|
Elected
officer (generally cabinet officer)
responsible
for statistical agency
|
Case 1
86. In this case there is no
formal contact between the head of the agency and his/her minister. The chief
statistician answers to, and receives broad guidance from, the council, whereas
the minister instructs, and is advised and reassured, by the council.
Case 2
87. While the
chief statistician takes direction from the minister,
the latter communicates with both the chief statistician and the council.
It is also the minister who appoints the members of the council, and it is
to him/her that they are ultimately accountable.
Case 3
88. The chief statistician appoints
the council, and while he/she is free to share the advice received with the
minister, the chief statistician alone is ultimately accountable to the minister.
89. The texts of statistics acts
of many countries are crowded with references to a statistical council or
commission with real and imaginary powers. Often it is said that the council
has either never met or has only had its initial ceremonial meeting. It is
actually a loss of credibility to specify in detail what a council should
do but not to be able to follow through with constituting it and seeing that
it has documented meetings out of which is generated a flow of advice. It
follows that in approaching this matter, the chief statistician (or his/her
minister or both) should:
v choose
from the available list which of the roles they feel are the most important
in the medium term and which will therefore establish the powers assigned
to the Council;
v identify
those individuals who possess the knowledge and prestige to sit on the council;
ensure their availability; and assess the likelihood of their being active
and interested members;
93. The
personality, range of acquaintances, interest, and availability of the council’s
chairman are key to the success of the enterprise. If no one is qualified
to fill the position as the council is being launched, perhaps it is best
to postpone the enterprise until someone with the right characteristics is
found.
94.
There are various schools of thought regarding the secretariat. According
to one, the chief statistician is also the Secretary ex officio of the Council. This view fits
with the provisions of Case 1. While the chief statistician would call meetings,
propose the agenda (under the guidance of the Chairman), and draft minutes,
he/she would be in a subordinate position, surrendering some of the powers
normally associated with being chief executive officer to the chairman. This
course of action is most appropriate to situations where there is a chairman
of great renown and seniority willing to maintain a high level relationship
with the statistical agency but not to work in it as a regular officer. But
these are exceptional circumstances, and do not dictate a general rule.
95.
The chief statistician may act as an ex officio member, taking active part in the discussions of the Council
usually at the invitation of the Chairman, or he/she may act as any other
member would, although the chief statistician would naturally possess more
knowledge of the day-to-day activities of the statistical agency. This is
a common situation, corresponding to the relationships illustrated in Case
3. This situation has the advantage that is more acceptable because the chief
statistician does not surrender any of his/her prerogatives while at the same
time being engaged with other members of the council in a continuing discussion
on the agency’s programme.
96.
Another arrangement for the secretariat is that the secretary of the
Council is a designated individual from the central statistical agency, who
– as a rule- does this work on a part-time basis. In such a construction it
is important that it is somehow (e.g. in legislation) made clear to whom the
secretary reports; either to the (chair) of the Council or to the chief statistician).
97.
In yet another arrangement, the chief statistician may only show up
when invited, and then only to answer questions from members of the council.
This is the obvious offshoot of Case 2, which is sometimes designed deliberately
so that an actively involved minister would be presented with two distinct
opinions on any one subject.
98.
There are no assumptions about political interference with any of these
systems. In all cases, things can go wrong, or alternatively everyone can
behave according to the best of expectations. In some instances, the existence
of a body placed in between the chief statistician and his/her minister may
prove to be an essential insulator for the statistical agency. In others,
ministers may try to politicise the council and use it as an additional body
of influence on the behaviour of the statistical agency. In some instances,
of the case 2 variety, a minister may try to dilute a hard view expressed
by the chief statistician with a more nuance opinion arising from the deliberations
of the council. In still other cases, a minister may find that his/her efforts
are impeded by both the chief statistician and the Council. In summary, one
cannot predict the action of these systems without placing them in their proper
context.
99.
To avoid a dangerous confusion of roles and responsibilities, the chief
statistician should exercise careful judgement when proposing an agenda for
the council’s discussions. For example, he/she should not involve the council
in the micro-management of the statistical agency. Such matters as detailed
costs of surveys, or improvement of the efficiency of projects at the operational
level, should be handled within the statistical agency, without external interference.
100.
But the agenda should include such matters as the council’s view of
the definition of unemployment; the worth of extending activities to the field
of the environment at the expense of other statistical fields; or opinions
on the most acceptable balance between quality and detail. If the chief statistician
keeps on insisting that these are the issues on which he/she needs outside
help, then he/she will not be overrun with superfluous advice, and over time,
will see an improvement in the quality and value of the advice that either
he/she or his/her minister receives from the council.
101.
The nature of the agenda will tend to dictate the frequency of meetings
but a few words of caution are in order. If meetings are too frequent, most
members will find them too onerous, and participation will probably lessen.
If meetings occur too infrequently (for example, once every two years), the
members will tend not to know each other and to have forgotten whatever they
learned about the statistical agency during their induction. If the group
of participants is relatively small – ten to fifteen – frequent meetings are
easier to schedule, whereas with a larger body of members logistics get more
complicated. Also, it is best to work with fixed schedules (for example, setting
a fixed day of the week) so that meeting dates are predictable.
102.
The reports should be publicly available, even if the subjects discussed
are not the most engaging. Technology now makes it possible to publicize such
reports at low cost. Placing on the agency’s web site a copy of the agenda,
a summary of the decisions made, and a list of actual participants, will help
to demystify the work conducted by the statistical agency and the National
Statistical Council.
103.
In addition to the National Statistical Committee, it has been found
in many countries to be useful to form advisory committees.
In theory, these advisory committees could meet regarding a multitude of matters,
but in practice they are most often adept at dealing with technical issues.
Thus, such committees tend to be specifically oriented toward key problems
of measurement (e.g., replacing an outmoded way of measuring the flows into
the labor market; estimating the productivity of sectors which have no priced
output; dealing with the production of the informal sector; determining whether
the CPI has a bias and so on). The difference between these committees and
the ones described in the paragraphs below is that while their opinion is
authoritative they seldom deal with issues that are key to the survival of
the statistical agency. Whereas certain ad
hoc committees or commissions are convened in response to a crisis, advisory
committees of the nature described here are designed mostly to avoid crises
rather than to manage them.
104.
Many countries have more than one such committee. The number of advisory
committees depends much on each country’s capacity to support them; on the
size of the community of experts sufficiently interested to take part in such
discussions; and on the capacity of the agency to take notice of expert opinion
and to institute recommendations where these are found to be appropriate.
One benefit of such committees is that they help narrow the distance between
academic and research statisticians on the one hand and practising government
statisticians on the other, since many of the participants in such committees
come from academic circles. Academic interest and participation in solving
key problems of measurement narrows the gap between the two communities, and
keeps official statisticians in close contact with intellectual advances that
might contribute to their various branches of specialization.
105.
Advisory committees have other advantages. If their membership includes
interested government officials, they improve the sense of policy relevance
of current statistics. If official data are criticized because of their conceptual
or methodological inadequacy, advisory committees serve as sources of respected
advice, as well as protection in the face of unfair criticism. Moreover, as
additional friends of the statistical agency, they contribute to its reputation
as well as to its capacity to reach out.
106. The creation of an ad hoc body should be reserved for crisis
situations. It would be pointless to convene an authoritative body chaired
by someone with a great deal of prestige, whose opinion would have to be taken
into consideration by government, if the problem under review were not of
appropriate importance. For example, if users were concerned that calculation
of the GDP accounts is biased or plainly wrong, the very essence of economic
measurement would be affected. If not promptly addressed by a dispassionate
group of experts at the highest level, such doubts could profoundly threaten
the credibility of all economic figures published by the statistical agency.
There are two general points to mention in relation to these advisory bodies:
v the
membership needs to come from outside the central statistical agency
v the
report relationship of advisory and ad hoc committees varies considerably
among countries and depends upon the circumstances of each individual country
“…The subject of statistical legislation can be
reduced to two major issues: the compulsory aspect, that is, the power the
government asserts through the statistical agency to collect data; and the
guarantees it provides for safeguarding the confidentiality of the information
collected from individual respondents…”
107.
Laws regarding statistical agencies have largely
similar connotations, though the wording may differ: The state (or the government,
or the people’s assembly etc.) grants certain rights to a body hereinafter
designated as a statistical agency. This body’s organic structure is explicitly
laid out, including the requirements for the person at its head; the constraints
under which it is supposed to operate; and the accountability that prevents
it from abusing its rights or acting arbitrarily. The law dictates what the
statistical agency is expected to do with the information respondents submit
to it, and for which it is accountable. The community of respondents is asked
to comply with the statistical agency’s demands for information so long as
they can be justified in the name of the objectives set by the law. In exchange
for intrusion upon privacy rights, the statistical agency is required to safeguard
respondents’ information. If the agency breaks this commitment, its officers
are subject to certain sanctions. If respondents do not comply, they too are
subject to certain sanctions. While laws differ from each other in length,
style, detail, and scope, if they do not cover the fundamental points outlined
above, they are incomplete.
108.
The Handbook includes an exhaustive list of subjects that may be covered
by the law. Consulting the Handbook in this regard is worthwhile, partly as
a matter of historical interest and partly because some heads of agency may
still find it useful as a checklist if they entertain any wishes to get their
own law changed. Appendix II contains a “model” statistical law.
109. Usually the law defines the main actors and
their rights and accountabilities. In the case of the legal provisions for
an official statistical activity, to be referred to as a Statistics Act, those actors
are:
v
the minister;
v
the chief statistician;
v
the statistical agency and its staff; and
v
the agency responsible for the coordination of the statistical
system
v
the respondents.
110. The Statistics Act will also define a set of
relationships with other bodies that could include:
117.
It is best if the Statistics Act makes clear provision for the statistical
agency’s right of access to other government data holdings. This should be
done partly in the interest of streamlining government operations but, more
importantly, to alleviate excessive paperwork on the part of respondents.
118.
The right of access by the statistical agency to administrative holdings
of information useful for statistical purposes should be explicitly recognized
as an exception in the legislation that protects such holdings or in general,
administrative registers. An ideal state of affairs is one of reciprocity,
where the statistical legislation lays down the rights and conditions of access
ands the specific legislation that protects administrative holdings wherever
they may be within government recognizes as an exception the right of access
by the statistical agency for statistical purposes.
119.
Even though the structure and content of a
Statistics Act should be straightforward, a chief statistician should have
access to legal advice, preferably a specialized legal adviser, expert in
the interpretation of the Statistics Act and in the treatment of its various
exceptions. The legal adviser will be of inordinate value in cases of conflict
between the Statistics Act’s provisions for rights of access and the formal
restrictions embodied in other Acts.
120.
In the case of decentralized systems, there
is a need for the statistical act to apply to all the members of the statistical
system. For example, if there is a statistical agency or a research department
within the Central Bank, is its legal authority to collect data from other
banks determined by the general authority of the Central Bank or by a specific
law or regulation authorizing the responsible department to collect supplementary
information for statistical purposes? If there are discrete statistical agencies
in the Ministries of Transportation, Agriculture, Public Works, Interior and
so on, what are their legal rights and restrictions in terms of collection,
access to micro-data, and form in which individual records are stored and
accessed? How does the central statistical agency decide whether the particular
cell is or should be a bona fide member of the statistical system? In an ideal
state of affairs the following are minimum legal provisions:
v all
members of a statistical system should have a legal basis for their collection
operations;
v all
members should have provisions defining their legitimacy, accountability,
obligation to hold individual information in trust, and sanctions if those
obligations are not heeded;
v
all members should recognize the form in which individual information
can move from one to the other be shared for purposes of statistical integration
and generally for effective analytical work;
v
provisions acknowledging the need for, and definition of, statistical
coordination, as well as guidelines on how it is carried out.
121.
In this chapter we will look at finance from
a general perspective: who should bear the financial burden of producing statistics?
The subject was glossed over in the two previous versions of the Handbook,
probably because it was felt that the issue was no major concern to the international
statistical community. However, things have changed. It is now generally felt
that adequate funding of statistics is a key issue in sustained statistical
capacity building across the globe. Of course, there will always be debate
on what ‘adequate’ means and even in developed countries there are ongoing
budgetary pressures that make priority setting in statistical programmes an
everyday fact of life. Nevertheless, few will argue that, in general terms,
the funding of statistics in developed countries is inadequate. The situation
is entirely different, on the other hand, in developing and so-called ‘transition’
countries.
122.
On a philosophical level, there has been some
discussion on the financial burden and proper allocation of the cost of official
statistics. Basically, there are two polar views (and of course a wide range
of practices that mix the two extremes):
v Government
pays for information which is either needed for decision making, is a public
good or is needed to inform the electorate and provides this information free
of charge to the public (or at most at the marginal cost of dissemination),
or
v Information
should be collected (and paid for by the government) primarily because government
needs it for its own business and the costs of collecting, processing and
disseminating any other information should be borne by the user.
123. Something of a watershed in this discussion
has been the Rainer Report in the United Kingdom, but it should be
noted that the latter view (which is the gist of this report) has since fallen
out of favour, because it is incompatible with the principles of political
transparency and accountability.
However, some other views of the Rainer Report have since been adopted by
some offices, for example that the introduction of payments between departments
is one mechanism for improving the allocation of resources. It should be recognized
that these discussions are mostly relevant for countries where the use of
statistics is firmly anchored in tradition, where the community of quantitative
analysis both in and outside the government is considerable, where political
decisions are largely evidence based and statistics are an integral part of
this evidence and where allocation of funds and transfers among parts of the
community are driven by statistical measures.
124. On a more practical level it is useful, first
of all, to look at the two main sources for the financing of official statistics,
which are:
v Appropriations
through the government budget, both for the central statistical agency (if
there is any) and for ‘statistical cells’ in ministries.
v Revenue
that statistical offices generate by selling products and services at market
prices.
125.
Overwhelmingly, the first source is far more
important than the second. Few statistical offices generate more than 10-20%
of their ‘income’ from sales. In addition, in quite a few countries, the revenue
that statistical offices generate, cannot be used for their own operations,
but goes directly to the Treasury.
126.
Some of the main financing issues that are
on today’s agenda are:
v
What information is placed in the public domain free of charge
or at the marginal cost of dissemination and what information will be provide
at a cost?
v
If information is provided at a cost, what should the charges
be based on? How does an agency guard against abusing its monopoly powers?
127.
These questions have been complicated by the fact that conventional
publications are no longer the main vehicle for the dissemination of statistics,
but rather the CD-ROM or indeed the statistical agency’s web site on the Internet.
128. The process of securing sufficient government
funding for statistical organizations may on the one hand be very country
specific, on the other hand there are many similarities in the procedures
that are actually followed. First of all, it is important that funding is
based on clear, systematic, transparent multi-annual and annual work programs.
In some countries, particularly in developing and transition countries, it
is also important that a long term ‘masterplan’
is developed and promulgated in the appropriate government circles. Secondly,
it is of course vital that the statistical agency (or agencies) mobilises
sufficient political support from their user community. A Statistics Council
may also be an important instrument to achieve such support. Thirdly, it is
very helpful to secure funding if the statistical agency is seen as a well-managed
organization that is proficient in planning, cost-accounting and in producing
clear management reports that show progress, income and expenditure, under-
and overspending etc.
129. There are several considerations in favour
of allowing a statistical agency to supplement its budget by the sale of goods
and services at market prices.
Two key definitions:
v
Goods, or “information products” are self-contained arrays of quantitative
information, with or without interpretation, which can be stored for future
retrieval. The medium in which these arrays are recorded is immaterial. Thus,
examples of “goods” include a Yearbook of National Accounts; a CD-ROM with
the Standard Industrial Classification; and tables on exports and imports
by commodity groupings downloadable from a web site.
v
Services are activities
carried out by the statistical agency to create a statistical information
product. Examples of “services” include providing an algorithm for the selection
of a sample of small businesses from a shared register; testing whether a
particular table contains residual disclosure; and testing whether a time
series meets a set of conditions that makes it eligible for seasonal adjustment.
130.
First of all, this gives some assurance that the entire community is
not funding a specialized commodity that is only of interest to a select number
of users. Secondly, by allowing statistical agencies to keep the proceeds
of its sales of services, one ensures that unused capacity is taken advantage
of which is preferable to leaving it unused and having the user organization
develop its own survey capacity, with all the duplication of efforts this
could represent. Thirdly, it may promote a user-oriented culture in statistical
organizations.
Conclusions
An examination of the classification of statistical
systems favours the creation of a stand-alone Institute or Bureau. It is better
to attach a recognizable name to the production of official statistics than
to disseminate them anonymously or too discreetly. In recent years, budgetary
problems affecting central governments have necessitated structures in which
the effective co-ordinating power lies in the hands of research departments
of Central Banks.
There is no question that concentration and critical mass provide opportunities
and means of action that dispersion or fragmentation hinder. Even so, the
expense and legal impediments related to the process (as distinct from the
state) of centralization may be such that its feasibility is improbable. In
those cases where the head of a central statistical agency is thwarted in
his/her attempts to reduce the fragmentation of the system, but wants to get
the most from co-ordination, there are several tools, all of which can be
tried in some degree. They
include in addition to mobilizing political support:
v
the creation of a National
Statistical Council
v
the co-ordination of
budgetary allocations for statistical activities
v
the management of the
corps of statisticians in the public sector
v
the international exchange
of staff
v
data collection approval
v
international standards
Strong leadership is
key to the effective performance of a modern statistical agency. In order
to attract good leadership, the job must have the right status. chief statisticians
must demonstrate their objectivity and impartiality by acting independently
of political controversy, but at the same time must maintain close contact
with their peers in other Ministries. Chief statisticians must display a rare
combination of professional and managerial talents, although the proportions
of each will vary according to the objective situation. Governments must not
appear to be frivolous or arbitrary in demanding that the chief statistician
resign if need arises, but they must not allow the continuation of a state
of affairs in which energy and inventiveness have long been exhausted. Lastly,
the capacity of a chief statistician can be augmented through intensive contacts
with his/her peers abroad.
Securing stable
(and in the case of developing countries: increased) financing for statistics
is an important responsibility for the chief statistician. To obtain political
support for stable finances, setting up sound planning instruments (such as
annual and long-term work programmes) are helpful, as well as promoting the
image of statistical offices as well-managed organizations.
Chapter 34
“We need statistics not only for
explaining things but also in order to know precisely what there is to
explain.”
“…Statistics is the area of political science
dealing with the collection, classification, and discussion of numerical facts
relating to the condition of a State or community. “
1. Statistics are compiled to answer questions
and, at times, to allow questions to be formulated with sufficient precision.
Some inquiries are developed once and made continually (e.g., how many people
who sought work this month were unsuccessful?). Other questions are asked less
frequently (e.g., how many ton/miles can the railways system deliver per
week?). Some questions are modified as the underlying reasons for asking them
are modified by technological advances or the shifting relationship between the
public and private sectors. Some questions still have not been answered (e.g.,
is the apparent slowdown in productivity increases merely a consequence of
measurement problems or is it real?). By and large, statistical agencies are
accustomed to converting a general question into one to which practicable
answer can be found.
5. Changes in environment, equipment, and
expertise create new needs, which dictate changes in the agenda, policies, and
organizational structures of statistical agencies. In the following paragraphs
we examine some of the typical features of the demand for statistical products
by various constituencies.
6. By
its very nature, statistics is of special interest to governments. In addition,
official statistical agencies are largely financed by central government,
though there are exceptions. For example, there are instances in which the
government’s contribution to the statistical agency’s budget is supplemented by
the proceeds of sales of goods and services to all users, including the private
sector. There are also instances in which statistics compiled by private sector
research agencies over a long period of time acquire a semi-official status.
But in the majority of countries, the largest proportion of the budget is
supplied by the national Treasury, aided at times by grants from international
agencies.
7. As the source of financing, the
government may try to exert leverage on the activities of the statistical
agency. Where it helps determine priority activities, this leverage can be
constructive. Conversely, where it attempts to delay the release of a vital
number, tries to modify a number, or defies a change in definitions or methods,
the leverage may be pernicious. We have already seen why the latter activities
are harmful. In this chapter, we will examine constructive activities.
8. The needs of government, like most
needs of potential users, are usually not expressed in terms that lend
themselves to instant data collection. They are divided into two categories:
needs that call for regular observations of the same variable and needs that
require information collected on an ad
hoc basis. Examples of the former include the need to know how average
consumer prices change from one week, month, or quarter to the next, so as to
maintain a current perspective on inflationary pressures; the need to determine
the relationship between receipts and payments from and to abroad in case the Treasury
runs out of foreign assets; and the need to know the expected annual school
enrolment by small area within particular regions, in order to ensure that
class space and teacher supply meet the demand. These needs are met by the
regular publication of statistics with known and describable properties: the
Consumer Price Index, the balance with the rest of the world on goods and
services, and the population projections by gender, age, and area. In the
majority of countries, there are intermediaries who
interpret the current statistics in light of those compiled earlier and inform
decision-makers about changes that may require attention.
9. In cases in which the government
requires ad hoc statistical
information to explain the roots of a particular problem, the situation becomes
complicated when there are no measurement conventions geared to the nature of
the problem and the expected answers. Examples of such problems include whether
the scale of evasion of value added tax collected appears to be growing; the
size of the underground economy; whether the current policy on incarceration
makes best use of resources to meet stated objectives; and whether current pay
scales discriminate against women. In such cases, the very nature of the
evidence may be open to interpretation and discussion. Clearly, translating
users’ information needs into the data collection or rearrangement that will
eventually take place is a complicated process.
10. Some
problems persist irrespective of time, place, and method of government or
structure of political system. “How many people are there of military age?” was
a popular question even two thousand years ago. “What gross tonnage is
available to feed the inhabitants of the British Isles?” has had a fairly long
tradition and was until recently a matter of concern to many cabinet ministers
in the UK. Other problems are entirely new and are the result of technological
innovation and of the changes in the way economic agents organize themselves in
response to new techniques. They too exert pressures on statistical agencies,
not only for the compilation of new information but eventually for the
reorganization of basic data as well.
19. Because
government is responsible for an overwhelming share of the provision of health
and education services in many countries, questions arise regarding efficiency and
public accountability. Examples of such questions include: How much does it
cost society to provide free schooling for all children up to the age of 15?
How are the costs recouped? Are the physical facilities adequate or is there
inequity in the provision of the service? Do the teacher-student ratios ensure
effective use of the resources assigned to education? Are they likely to
continue doing so in light of the age of the workforce and the growth in
population? Are there enough teacher training facilities to guarantee the
maintenance of the current teacher-student ratios? And the most important
question to which decisive answers have not yet been found: Is the marginal
public investment in health (education, science, etcetera), in light of the
observed outcomes, commensurate with the expected return?
20. Similar
questions arise in the health sector. Are marginal expenditures on additional
hospital beds more effective than expenditures on improved home care if the
objective is to minimize pain? In general, statistical agencies must resist the
temptation to compile information that is understandable and amenable to a
count but uninteresting from the point of view of the policy- and
decision-makers. Rather, they must devise methods and techniques that will allow
them to tackle the more difficult problem of measuring outcomes (in the case of
health, education, and science, there is no designated length for the period of
observation before an outcome can be assessed).
21. If
it becomes clear to all that the question of effectiveness drives the need for
information—and taking into account the difficulties in measuring effectiveness
without making very controversial assumptions—the national statistical office
should establish a close partnership with health economists, school
administrators, criminologists, and natural scientists. Such a partnership
should be designed to help clarify the limits of the information available in
these domains and the possible progress that a statistical agency can make.
38. The
public keeps a watchful eye on regular statistics such as the CPI, the
unemployment rate, and on overall economic performance. The statistical agency
should be able to address these concerns. At the same time, it should
capitalize on the public’s ongoing interest and ensure that its name is
automatically associated with the publication of data with which the public is
most concerned. The statistical agency should also strive to inspire confidence
in its efforts to maximize the quality of these key measurements.
39. The public also has
requirements that are not so substantial, permanent, or well defined. The
difficulty lies in their inchoate condition and in the fact that a particular
interest may flare up at any moment in response to a particular situation, most
often an unforeseen event of a threatening nature (e.g., What was the value of
the housing destroyed by last week’s hurricane? How much in savings has been wiped
out by the latest bout of inflation? How many illegal immigrants are there in a
particular city? How large is the underground economy?). One way to address
these concerns directly is to maintain a small ad hoc survey capacity characterized by very quick turnaround.
40. The
introduction of statistics in school curricula is a means of instilling respect
for quantitative information and analysis. Both primary schools and high
schools offer a variety of courses in which statistics play a visible role. The
following are examples of basic questions that appear in the curricula of many
schools at both the primary and secondary level: How many people live within
the country’s borders? What is the population density? How wealthy are they?
What do they do in their leisure time? How are they employed? How many of them
are poor and need help? Do they have access to educational and health
facilities?
41. The names of the relevant
courses vary from country to country but the subject matter is more consistent
than the terminology suggests. Several statistical agencies have a special
programme to support initiatives by the educational system to promote the use
of quantitative data at an early age. In
Poland, there is an active high school competition for the best essay in which
extensive use is made of official statistics. In other countries, schools are
granted premium access to the national database of the statistical agency.
42. The
press and other means of mass news dissemination play a special intermediary
role in the relay of statistical information, whether it is the subject of
headline or local news interest. For this reason, the statistical agency is
responsible for ensuring not only that the right information gets into the
right hands at the right time, but also that it is properly described and
imparted in the appropriate form to those who most need it.
43. No
statistical agency has the power to ensure, for example, that all those whose
lives are affected by a change in the measured level of consumer prices become
aware of the fact exclusively by consulting a statistical bulletin. Even in
countries where the population is measured in tens of millions statistical
bulletins are seldom published in quantities of more than a few thousand (with
the possible exception of the occasional prestige publication, which may be
issued in quantities of two or three times the average circulation).
44. For
these reasons, the agency will have to rely on the press to ensure that statistical
news reaches all interested parties. This creates several challenges, which
will have to be dealt with through a liaison organization:
v
ensuring that newspapers, television, radio, and
magazines all become aware of the latest statistical information at the right
time;
v
ensuring that the statistical office is responsive,
even during non normal office hours, to the deadlines of the press;
v
providing, along with the latest numbers, the necessary
amount of explanation so that the press does not put an unbalanced spin on the
news;
v
periodically reviewing with the press the arrangements
made to ensure efficient dissemination;
v
making all possible efforts to present statistical news
in manner that minimizes misinterpretation and enhances clarity;
v
holding seminars for members of the press to highlight
simple analytical techniques for interpreting and assessing current data.
A number of
offices have made permanent arrangements to ensure that these challenges are
met and that the value of the press to the statistical agency is maximized.
Colouring
statistical news
This heading is not meant to suggest impropriety. Ensuring that the
news gets interpreted properly does not imply manipulation of the subjective views
of the press. Rather, it implies that everyone should be aware of the pitfalls
of interpretation. For example, if the imports of toys escalate significantly
in the month of November (in a country where a large proportion of the
population celebrates Christmas), this is not necessarily a harbinger of
disastrous news for the national balance of payments. Rather, it may mean that
the domestic toyshops are preparing for Christmas, and the relevant comparison
is not with the previous month but with the same month one year earlier. Such
is the situation with unique events—e.g., the leasing of three jumbo jets for
the national airline or the effects of a strike or a natural disaster—which may
distort the expected figures and play a major role in their interpretation.
Recognition
The statistical agency should get recognition when statistical
headlines are broadcast. Often, mass media make references such as this:
“Government officials claim that this month’s production of cement is well
above expectations, leading to an optimistic outlook for the construction
industry….” Correcting the failure to single out the statistical agency
responsible for the estimation should be a matter of priority. Giving the
agency credit not only reminds the population of its existence, role, and
responsibilities, but also reinforces its imprimatur on the release of key
figures and impresses upon users that those figures have a special character of
reliability.
45. In
most countries and in most political systems, governments can be elected or
their tenure of power can be ratified on the basis of promises made to the
electorate. Promises range from the general (“we shall make this country secure
for our children”) to the very specific (“we shall keep the rate of
unemployment below 5 percent”). Political accountability in the latter example
can only take place in light of an objective measure (official statistics) that
informs public opinion about whether or not the government has kept its
promise. In this instance, opinion is formed in two ways: the public may decide
that the government was or was not faithful to its electoral platform or it may
decide that the standard of measurement—the official statistic—was unreliable
and therefore of no consequence in its ultimate judgement of the government’s
performance. The press can play a constructive role in ensuring that the
judgement of performance does not get mixed up with the reliability of the
statistics.
46. A
number of agencies have achieved public confidence in their word and institutional
guarantee. Clearly, this situation is desirable for every statistical
institution. Since the press can be a valuable ally in this task, investing in
a press liaison is worthwhile.
47. Ultimately, all business users’
interests in quantitative information are similar. The fundamental questions
are: How many businesses like ours exist? How do they compare to us? What are
the prospects for our business and for those with similar attributes? Other
interests are merely variations on these fundamental questions.
48. To
provide an answer, the statistical agency must first define “us.” From the
statistical agency’s point of view, the definition should be as wide as
possible, so that sampling techniques can be used effectively and the seemingly
insoluble problems of accurate coding can be overcome. From the point of view
of the potential user, “us” should be as narrowly defined as possible, to
account for the large number of idiosyncrasies of what is being compared.
49. Once
the scope of “us” has been defined, the next challenge is to select the
attributes that must be compared to satisfy the interests of business. In
principle, all economic variables are potential subjects of comparison. Thus,
the number of employed per unit of revenue and per unit of profit; the
structure of costs; the size of the market measured in number of customers and
in gross revenue; the composition of the market in terms of purchasers at home
and abroad; the return on capital invested; the rate of product innovation; and
the prices quoted are all examples of important attributes.
50. Of all these variables,
and in most lines of business, the two that appear to be of greatest interest
are the rate at which prices are adjusted to conditions of overall and specific
demand and the rate at which the products or services of a business are
influenced by technical and organizational innovation.
51. At
the limit, the interests of large businesses are very similar to those of the
Ministries of Finance, and the respective specialized sectors. Obviously, if
businesses cater to national markets and possess multiple lines of
specialization, the only useful point of comparison is the rest of the economy.
Their questions will resemble those of the Ministry of Finance, with an
emphasis on cost structures for activities of special interest; cost structures
of competitors abroad; information on barriers to their activities abroad (both
tariff and non-tariff); and financial information on investment at home and
abroad.
52. In many countries, large
business represents a disproportionately large share of the GDP, and therefore
neither its requirements for information nor the accuracy and the promptness of
its responses can be ignored. For this reason, a number of statistical agencies
have instituted a special unit with the exclusive function of managing
relations with large business. Not surprisingly, the first agencies to
institute such units were in those countries with the highest concentration of
large businesses. However, the payoff has been so significant that others have
followed suit. Today, there are examples of large business units in large and
small statistical offices, in industrially advanced and in emerging
industrialized countries.
53. The
role of a large business unit is manifold. Therefore, it stands out as an
exception to the simple rules of organization along the statistical production
process. Its responsibilities include:
v
keeping a current record of the boundaries and
structure of large businesses (these businesses change configuration very
quickly, so what is reported in one period may not be significantly comparable
with the previous period)
v
keeping track of the businesses’ accounting practices
to distinguish between internal transfer pricing and arm’s length market
pricing
v
deciding, in communication with designated business
contacts, what statistical questions can be answered directly, using account
information, and what questions require estimation
v
making sure that each business is required to answer
each question only once and that the answer is used in all databases for which
that information is required
v
communicating with the business’ research department regarding
its needs for information and the most convenient medium in which to provide
the data
v
gathering, from discussions with the research
department, information as to the intentions of the business regarding
innovation; investment; acquisitions abroad; and general confidence in the
domestic market.
The larger
the business, the more explicit and articulated each of these responsibilities.
54. Satisfying the information
needs of small business is the greatest challenge for business statistics for a
number of reasons: The ratio of effort deployed in answering requests for
information – regulatory and statistical --from the government to the economic
size of businesses is much higher for small than for medium-size and large
businesses. The ratio is almost as high as that for households. In addition,
even when the need for information is well
formulated, the specifications tend
to be so detailed—in terms of what the business does, where it is located, and
with whom it wishes to be compared—that the statistical agency finds it
difficult to comply with any professionally acceptable degree of reliability.
The third reason has to do with the nature of small business needs for
information—even though they tend to require highly specific statistics, small
businesses may be reluctant to comply with all of government’s demands for
detailed information because devoting the necessary time and resources would
significantly add to overall costs. The statistical agency will find it very
difficult to accommodate small business needs if it does not have access to key
administrative records and the ability to use them for statistical purposes.
Examples of administrative records of use for small business statistics
Under most circumstances, the
high level of corporate income tax information limits its usefulness in
deriving small business statistics. In addition, the apparatus required for
collecting direct taxes is unevenly developed, so income tax records are of
little help in many of the countries where the activities of small business and
their relationships to the informal sector are of special interest, even where
access is granted. Conversely, indirect taxation is a virtually universal
source of information . In recent years, the governments in many countries have
made special efforts to ensure that such taxes, particularly value added taxes,
are levied and faithfully reported to the proper authorities. They have also
made special efforts to ensure that social security records are complete and
accessible. While access to and use of these records is only useful to measure
and describe the organized sectors of the economy—at times leaving a large
swathe of informal economic activities nearly untouched—nevertheless, the
information that can be gathered from the records is more substantial than any
information the statistical agency could collect unaided. If it relies on
indirect tax records, the agency is free to concentrate its efforts on
household, and related, activities that may for a time resist the effects of
taxation and social security reforms.
56. Despite all these difficulties, small
businesses are of undeniable importance. They are numerous; play a key role in the service
sector; are given a large share of credit for innovation, particularly in the
dynamic activities that characterize a modern economy; and, according to some,
play a key role in job creation, particularly in the early phases of a cyclical
upswing. For these reasons, their wishes must factor into the dissemination
programme of a statistical agency.
57. There are two aspects to meeting the
information needs of small businesses, both of which affect organization—the
information itself and training small-business employees to use the information
to their advantage. One of the most significant
barriers to the survival of small businesses is gaining access to credit on
affordable terms, and to do so, they must demonstrate how their costs and gross
profits compare with those of their competitors. Thus, what is of prime
interest to them is not averages and totals but rather the distribution of key
ratios and where they fit in. Examples of such ratios are advertising costs to
total cost; equity to credit; and accounts receivable to total revenue.
58. Using a table with distributions of key
ratios may be intimidating at first, particularly for some entrepreneurs who
are apt to trust their intuition or salesmanship more than unfamiliar arrays of
numbers. For this reason, the statistical agency may wish to offer seminars for
small-business managers and take advantage of trade fairs and similar events to
display its relevant products.
59. The demographic characteristics of
markets are also of interest to businesses of any size, particularly those who
cater to the final consumers or to freelancers who work from home. The age,
gender, family, and income attributes of the population in very small areas are
of the utmost importance to them, but they may not have the knowledge or the
confidence to approach a statistical agency and request cross-tabulations
derived from the census of population. In order to cater to those wishes,
statistical agencies may have to utilize specialized consultants who can
estimate potential sales of a wide range of articles on the basis of a
combination of small-area census variables and the results of the latest survey
of family expenditure.
60. In any case, maintaining a unit within
the statistical agency that specializes in small-business concerns is valuable.
Its responsibilities should include the following:
v
maintaining expertise in the handling of administrative
records so as to satisfy current demands either exclusively or primarily with
the information they contain
v
organizing
events of interest to small business and using them as opportunities to
disseminate extracts of existing databases and to assist small businesses in
the effective use of statistical publications
v
championing small-business claims for limitations on
the paperwork burden generated by government in general
v
serving as permanent liaison between the statistical
agency and such agencies as social security, indirect taxation, and customs.
61. The academic world can place burdensome
demands on a statistical agency but it can also be the most important ally in
defining research and action programmes, in evaluating the quality of the
statistics produced, and in helping explain to laymen the underlying strengths
and weaknesses of official statistics. So, contacts with members of academic
institutions are a strategic resource and should be as extensive as possible.
62. The needs of the academic world cannot be
summarized as easily as those of other sectors. In theory, they encompass all activities
of a statistical agency. Any research in the social sciences that is conducted in an academic
institution and requires testing places demands on statistical agencies.
Academic needs may differ from the needs of other sectors in at least one fundamental
respect. All other sectors require statistics for formulating policy, examining
options, and ultimately making decisions, and are therefore less prone to worry
about how the data are generated. The reverse is true for the academic
community. For example, in testing a particular economic theorem, it may be
crucial to distinguish between data as they are observed and data altered by
standard editing processes. In addition, academic researchers are far more
likely to dispense with the statistical agencies’ analyses, in favour of their
own. For these reasons, the statistical agency should establish a liaison with
academic researchers, but a distinct unit is not required.
63. Agencies that have a special
methodological unit in charge of survey design and estimation may be tempted to
locate the liaison with universities in such a unit, at least for researchers
in applied sociology, demography, criminology, and microeconomics. This should
not preclude contacts with other parts of the agency.
64. Naturally, macroeconomists prefer to
establish their liaison with the agency’s national accountants and are less
interested in the sampling aspects of the underlying statistics. However, much
of the communication between the statistical agency and academia will be the
result of individual circumstances and some trial and error. The essential
point is to capitalize on the contribution to the statistical process that can
be made by academic institutions. Many statistical agencies have programmes
that allow for an interchange between university researchers and the more
research-inclined members of the statistical agency. To carry out such an
interchange, the university and statistical agency may each assign researchers
to the other’s institution, or they may collaborate on projects of mutual
interest.
65. In many countries with advanced
statistical systems, the programme requirements of international agencies represent no more than a marginal
addition to the statistical agencies’ domestic programmes and can be treated as
a by-product of current activities, but this is not always the case. Indeed, in
some statistical agencies, apart from special large-scale initiatives such as
the census of population, the requirements of international organizations
dominate the current programme.
66. For many statistical agencies,
international agencies remain the most important source of conceptual and
methodological guidance, not only in improving their capacity for inter-country
comparisons but also in structuring domestic programmes. The demands of
international organizations impose a discipline without which inter-country
comparisons would quickly become either unachievable or meaningless.
Comparisons are possible largely due to the use of international
classifications and standard accounting systems such as the SNA.
67. In addition to the benefits the
international agencies bring their intervention also often brings significant
problems. Since an international agency brings with it extremely needed
resources it is difficult, if not impossible, for many developing countries to
approach the offer from an international agency as an arms-length transaction.
Moreover, the intervention of the agency may cause significant disruptions in
the program of the national statistical office. Situations have also arisen
where the data collected by the agency is published without any review by the
county. Finally, too often, international agency programs have not been
accompanied by a transfer of technology from the agency to the country. Thus,
while, valuable information has been collected there has not often been a
corresponding improvement in the statistical capacity of the country.
Fortunately, this situation has been changing and international agencies are
increasingly committed to capacity building. One of the important
responsibilities of the chief statistician is to ensure the maximum amount of
capacity building while continuing the cooperation with the international
agencies.
68.
Throughout
this chapter, the need to create a specialized unit charged with particular
functions is discussed. This should not be taken as a panacea. A great deal
depends on the size of the office, its stage of development, and the weight of
the problems it faces. In many instances, it is sufficient to assign one
person. In other cases, it is best to distribute responsibility throughout the
agency. Where the specialized position can broaden agency employees’ experience
by exposing them to keen, analytically minded users, it is favourable to
distribute the responsibility
Conclusions
Statistical agencies should create a
progamme that is as versatile as possible, in order to serve users with a wide
range of questions and concerns. The
discussion above illustrates the variety of those concerned and questions. But a statistical agency cannot confidently
predict the nature of future concerns and the types of questions that might be
answered through the use of official statistics. Accordingly, it must invest in the ability to produce general purpose,
versatile tabulations. A statistical
agency can, and should, also invest in ensuring user access to techniques that
allow for rearranging basic data to conform better to specific objectives. But this is only possible if there is a
clear under understanding of the nature of the questions.
However, an agency is organized,
consideration should be given to the assignment of specialized functions (in a
distinct unit or not) with a dual mission:
v
Detecting, through close contact with all sectors,
upcoming matters of concern that may require statistical support
v
Following up with representatives of the various
sectors to determine the adequacy of the service provided by the statistical
agency.
Some functions would benefit from the
cooperation of the relevant subject-matter experts. In a few special cases, specialized units may be created to
address each function. Such cases
include relations with large business; small business and its associations; and
the press. In all instances, it is
imperative that the findings of these units be incorporated into the
formulation of priorities and the corresponding plan of action..
Chapter 4
SETTING PRIORITIES
1. The previous chapter underscored the
fact that meeting the demand from some users requires a specialized
understanding of the corresponding sectors (for example, health and education).
It also showed the need to maintain a special relationship with businesses
(multi-location, multi-activity, multi-national) that in many instances account
for a disproportionate share of a country’s GDP. These considerations have a
number of consequences for statistical agency. In the sections that follow, we
examine how various requirements give rise to a plan of action to ensure a
balance between the supply of and the demand for information.
“Ideally, priorities should be determined on
the basis of analyses of the costs and benefits of various alternative ways of
using scarce resources.…Although, with the exception noted, the cost of
statistical projects can, in principle, be identified and measured, this is
rarely the case with benefits…[because] statistics are intermediate products,
not end products. The benefit of a statistical series is a function of its
impact on policy decisions and their importance; it is generally very difficult
to pin it down and attach a meaningful value to it. Secondly, the process of
identifying benefits is all the more difficult both to the extent that the
series are part of a system in which the diverse elements are interdependent
and the value of each is enhanced by the availability of the others.…Thus
priority setting and allocation of resources among competing possibilities in
statistics cannot be calculated precisely but must be determined…on the basis
of judgements and insights based on past experience…in order to render
priorities and related resource allocation as rational and balanced as
possible.”
Handbook Rev.
1
2. This quote describes a
circular process. While there have been several attempts to establish a “calculus
of priorities,” none of them has gained wide acceptance. The idea behind some
of these attempts is to weigh several attributes—e.g., importance of the
request (using some arbitrary unit of measurement); cost of answering it fully;
cost of a partial, short-term answer; effects it might have on other activities
within the statistical agency; consequences of denying a request for
statistical information on the relevant decision or policy (again using some
arbitrary unit of measurement); and so on. The scores derived from these
attributes would be used to compile a table ranking each priority. But that is
not the way things are done, and while such a method introduces a consistent
and explicit set of factors, the conclusions are no less arbitrary.
3. A chief statistician cannot avoid
prioritising the activities to which an agency is committed.
4. Fortunately, priorities do not change
from one day to the next. The activities of a statistical agency possess a
constancy. Indeed, from one year to the next, a statistical agency will not
change its agenda by more than a small fraction of its total resources (5
percent is probably a generous estimate), with the exception of large-scale
initiatives such as the censuses of population, agriculture, and economic activities.
In addition to these endeavours, the statistical agency is likely to continue
measuring from one period to the next the variation in consumer prices, the
rate of industrial production, the amount of international trade in goods, the
size and condition of the labor force, and so on. Once these activities are
taken into account, the allocation of additional resources should be at the
discretion of the chief statistician, who can initiate subtle changes in
direction by placing a somewhat heavier accent on one activity or another.
5. That is not say that every single one
of the activities carried out regularly by a statistical agency is static. In
fact, as part of the standard work of an agency, products are added or removed
from the CPI, new nomenclatures replace old ones in the compilation of trade
statistics, and changes in the population are ultimately reflected in the
sample on which a labor force survey, for example, is based. Moreover, in many
offices there is a systematic drive to carry out habitual tasks in a more
productive and effective manner. The success of such drives frees up resources,
which in turn can be allocated to alternative activities. Accordingly, setting
priorities turns out to be not so much an exercise in the evaluation of discrete
subjects but rather a series of marginal changes in an ongoing programme to
which additions and subtractions are made at regular, and sometimes even
predictable, intervals.
6. In order to preside over this process,
the chief statistician, to a greater or lesser degree, requires the following:
v
authoritative information about what various
constituencies perceive as gaps in the range of statistical products and an
understanding of why those gaps are awkward for policy formulation
v
information about emergent issues of public concern
that may require statistical information
v
information about the cost of alterations to the
existing program, including possible additions in light of the two elements
noted above
v
a mechanism to sort out possible requirements.
7. These four elements underlie the
planning process, which can be handled formally or informally. The advantage of
formality is not so much the intrinsic value of the ceremony but rather that it
impresses on all participants that the planning of a statistical program is one
of the most important, if not the most important, statistical activity that an
agency’s management can undertake.
8. Priorities should be determined within
a framework. Nowadays, frameworks are referred to as strategic overviews,
strategic frameworks, or corporate strategies. Mostly, these headings apply to
the same thing: a medium-term plan that allows the agency to acquire, maintain,
or change existing capabilities while continuing to meet new requests.
9. Priorities ought to be selected from
within a framework not because it is fashionable to do so, but rather because
it is the most effective way to ensure that the priorities are clearly
articulated and that the priorities are considered as a whole rather than in
piecemeal fashion, which could result in inconsistent and duplicative
activities. The alternative is distinct descriptions of activities with no
coherence.
Balancing of requirements
“…[1]a balance must be
achieved between servicing users needing detailed data in specific fields and
those needing across-the-board aggregative data for macro-economic and social
analysis…[2] there is the balance between different subject matter fields [this
balance should ensure] that some resources are assigned (a) to projects that
are in their early stages of development, (b) to strengthening weak areas and
(c) to carrying out at least a minimum of research and analysis…[3] within each
subject matter field there is a balance between different projects both across
the board and in time. The importance of covering the entire field must be
balanced against the need for detailed tables in parts of the field…[4] there
is the balance between timeliness, accuracy, and publication of detail…[5]
there is the balance between satisfying user needs and avoiding undue burden on
the suppliers of primary data…” Handbook Rev. 1
10. It is generally accepted that one can precisely define the costs but not
the benefits of statistical undertakings. There is no way to devise a program
to answer the question of how many people will benefit directly or indirectly
from the publication of a particular number and to what extent. Nevertheless, a
chief statistician’s decisions regarding the allocation of resources implies
that the balance of costs relative to benefits is more favourable in one
direction than another.
11. The issue is to assess needs and
concerns—as a source of information on how a statistic can be beneficial to its
potential users. And since its usefulness cannot be quantified, one should have
at least a general impression of what that use might be.
12. Clearly, the formulation described above
is rather abstract. In practice, there is no substitute for an inquiring
attitude, one that seeks to relate systematically proposed new statistics to
the questions they are deemed to be helpful in answering. Furthermore, those
questions should relate to the concerns of the general public or the
government.
13. On the supply side it is important to
track costs and to keep a history of costs of different kinds of statistical
activity. A chief statistician should at least be able to argue in relative
terms about costs and benefits—for example, if x were spent on activity A, y
should (or not) be spent on activity B given the apparent (albeit non-quantifiable)
benefits of A relative to B. The tracking of costs for planning purposes should
be the object of considerable attention and respect within a statistical agency
and should include the following:
v
the ability to assess the value of a statistical
activity (e.g., a survey, the compilation of an account, the editing of an
administrative file, the publication of a statistical bulletin) with a degree
of accuracy adequate for planning purposes
v
the ability to archive and retrieve the results so that
they can be used to price future activities
v
the ability to reveal alternative uses for the same
resources
v
a mechanism to record decisions and assign
accountability.
14. Notwithstanding the limitations of
planning, experience shows that embarking on any complex initiative (conducting
a family income and expenditure survey, launching a quarterly labor force
survey, compiling an input-output table) without a plan is a prescription for
unmanageable turbulence and very uncertain outcomes. This uncertainty is
worsened by the interdependency of a statistical agency’s undertakings. For
example, the input-output table cannot be compiled without the survey of
inter-industry purchases and sales; the latter cannot be conducted without the
new commodity nomenclature, which in turn requires a production survey; and so
on. Planning for all the interdependencies and expecting them to function
precisely, though, is not realistic. Rather, in addition to careful planning, considerable
flexibility, local discretion, and fallback positions are necessary. The art of
planning statistical operations is finding a way to manage all of this in a
timely and realistic fashion.
15. In the next section we look at the
various purposes and elements of a plan; the difference between a multi-year
exercise and a single-year plan (which is more akin to a traditional budget);
the significance of “accountability”; and lastly, the plan as a teaching device
once its results are evaluated.
16. Essentially a plan consists of the
following elements:
v
a clear set of objectives to be met by the end of the
planning period
v
a description of the steps to be taken
v
a list of the requirements—environment and resources—to
ensure the practicality of the objectives
v
a discussion of fallback options should the outcomes
not meet the objectives
Additional
elements include the mechanisms for monitoring and evaluation.
17. Handbook Rev. 1 devotes a great deal of
space to describing in detail a three-tiered planning structure:
is
described, based on a long-term plan of a“…a
system of overall planning “strategic” nature, …a medium-term plan of a more
“tactical” nature and a short-term “operative” plan.”
Handbook Rev.
1
In this edition, we focus primarily
on “operative” plans.
18. The reason why an operative plan must be
annual—in addition to the fact that the standard budgetary cycle is annual—is
that if it were devised less frequently, it would be forgotten, and if it were
devised more frequently, the activity of planning would consume available
resources and impart a sense of instability. The annual plan must also fit into
a longer-term framework, thereby preventing the creation of a series of
disjointed annual plans and ensuring that the statistical system, while
maintaining its flexibility, will not be faced with changes in direction that
it cannot accommodate.
19. Many offices are small—100 staff members
or fewer—and have unpredictable budgets, either because of political
circumstances or because much of their resources come from international
initiatives over which the office limited influence. Thus, if there is a concerted
push to measure poverty on an international scale or to examine the factors
that affect fertility, the scale of operations required might account for a
very large component of the office’s total activity. Conversely the existence
of a multi-year plan can help ensure that international programmes reflect
national needs. Notwithstanding, the principle of having a five-year plan and
an annual set of objectives firmly anchored in that plan is as applicable to
small offices as to much larger ones, which tend to have a more predictable
budget. The difference is that in smaller offices, the actual implementation of
an undertaking can be much more informal, allowing for greater flexibility.
20. A plan of this kind has
multiple purposes, involving perceptions as well as organizational realities:
v
impressing upon the staff that their work is part of a
larger scheme and is not a routine continuation of some arbitrary disposition
creating a basis for measuring the performance of the office and of its
organizational units
v
recording experience in order to improve future
allocation of resources, reliability in carrying out tasks, etc.
v
increasing the efficient use of available resources
v
documenting inadequacy of resources relative to imposed
tasks
v
improving the perception that the statistical agency
uses rational criteria to organize its activities
21. These objectives are somewhat easier to
attain if planning is open; if a significant proportion of the staff takes part
in the establishment of the plan; and if there is a collective effort
to learn from experience by reviewing systematically the differences between
activities planned and activities carried out. But there is no avoiding the
cost involved in documentation, tracking performance, recording costs, and
organizing a structure that is cohesive and achievable.
Should there be a planning unit?
This has not been a recent matter of discussion
or systematic exchange of experiences in international venues. It is
nonetheless an important question of consequence to the larger and more complex
statistical agencies. The general advantages and disadvantages are easily
described. Having a specialized planning unit reduces the time spent planning
by the substantive experts of the office and allows for more control,
consistency in presentation, and ease in verifying internal consistency in
execution. But it increases very substantially the risk that the planning
activity will be treated as “someone else’s responsibility” by the substantive
part of the office and never be as internalised as it should. As a solution,
some offices have adopted a mixed regime, in which the planning unit is
responsible for the formal part of planning and for acting as a secretariat to the
planning effort (or committee); the substantive part of planning rests squarely
with those whose activities are to be planned. Naturally, in mixed
organizations of this type the cost of planning increases. But so do the
chances that it will be taken seriously.
22. On the one hand, the expectation that
detailed plans can be fulfilled with total precision is likely to cause
frustration with the entire planning exercise and may lead to a premature
abandonment. On the other hand, introducing the plan in a manner that is too
lax may lead to its never being addressed with the respect that is required for
it to function effectively. This is why in a number of offices where the annual
planning activity has become formalized, the discussion and promulgation of the
plan are subject to some ceremony.
23. Contrary to a natural first reaction, the
institution of sensible planning is a measure that usually leads to less,
rather than more, detailed control of the various parts of an office attaining
high-level objectives is only possible if local managers have the discretion to
modify their approaches in light of those unforeseeable circumstances. This
means that effective planning requires the establishment of a sound budgeting
approach, complete with a statement of how authority is to be delegated.
24. Conceptually, the relationship between
the two activities is as follows. A plan is created, taking into account all
the interdependencies and outlining a set of objectives, a schedule, and steps
to be taken to ensure that the objectives are met.
25. The plan will appoint agents to head its
various activities.. Each agent will be given a discretionary budget and the
responsibility for achieving a balance between income and expenditures. Handling
matters in this fashion has the added virtue that each agent will become more
conscious of costs. If the performance in meeting objectives is judged from the
point of view of efficiency, the various agents will have the incentive of
finding methods and techniques that will allow them to meet objectives with the
resources allocated (rather than the somewhat perverse situation that has been
known to occur, wherein managers spend as much as possible to demonstrate their
authority). Essentially, planning involves controlling the resources allocated
to agents, thereby requiring that they achieve their goals within the given
resource constraints. This arrangement applies to offices of all sizes.
26. The greater the discretionary authority
given to planning agents, the greater their accountability. In the framework of
an open plan with a great deal of devolution of authority and responsibility,
the chief statistician will ask for accountability from the staff. If the management
establishment is too inexperienced to carry out its part of the agency-wide
plan, the chief statistician will not be able to comfortably delegate budgetary
authority. An organization can be considered efficient if the head of the
system can rely on subordinates to help formulate and carry out both a
medium-term and an annual plan.
27. At the end of a planning exercise
(annually and, on a more strategic basis, every five years) the chief
statistician should receive a formal evaluation. This evaluation may be
conducted by an outside party or by a unit within the statistical agency,
depending on the political, legal, and regulatory circumstances. The evaluation
serves several purposes:
v
accounting to the government for the work of the agency
v
conveying to the staff the seriousness of the process
v
determining the degree of error in the process –
exogenous (beyond control) and endogenous (subject to control and future
improvement) – and attempting to explain the causes
v
validating or changing the agency’s strategy
v
using the results to engage the organization’s
management in a dialogue regarding accountability.
28. For the process to be credible, the
evaluators should not also be the planners; in fact, the greater the organizational
distance between the two, the better. It is also worth considering the
possibility of using the evaluation as a training opportunity. For example,
recently recruited professionals can serve as assistants to an evaluation
secretariat, thereby receiving their introduction to the planning exercise. It
is not critical to constitute an evaluation secretariat as a permanent
organization. Indeed, it may be rather difficult to justify it, as it is
unlikely that it will have work year-round. One option is to have a standing
committee headed up by someone who is indisputably knowledgeable and objective.
If circumstances allow, evaluation (and similar tasks, such as audit) might
consistently be handled by retired senior officials of the agency.
Conclusions
There is no objective “calculus of
priorities” in a statistical organization. The chief statistician must combine
intuition with the experience and current to make decisions with confidence.
Once this becomes routine, the agency will be able to formulate, albeit in a
qualitative way, some of the benefits of a planned course of action. It is
important to implement an effective mechanism to track, measure, and record
costs. For a planning exercise to be effective, it must reflect the right
balance of decisiveness and staff participation. It should be part of a
medium-term framework and treated as the annual definition of activities, their
cost, and the corresponding resource requirements. The staff, having taken part
in the creation of the plan, should be held accountable for its execution. The
practice of evaluating the plan ex post should be
strictly observed, as a means of enforcing accountability and learning from
experience.
Assuming that the annual plan is part of a five-year medium-term
programme, that resources are constant, and that the maximum permissible annual
shift in staff is only five percent gross, at the end of the term, one quarter
of the staff might be doing very different things from what they were doing at
the beginning.
Chapter 5
Chapter 5
PRINCIPLES OF ORGANIZATON AND REORGANIZATION
1. The
literature about organizational theory demonstrates that organization is not a
trivial matter and, most importantly that, while there is no single, ideal way
to structure an organization there are some commonly accepted general
principles.
2. This last point applies to statistical
organizations as well. There are in theory at least a dozen different, sensible
ways to organize a statistical office. Remarkably though, in spite of clear
differences in detail and even in organizational philosophy and culture, there
is a great deal of similarity in the organization of statistical offices across
the world. For example: a director of economic statistics from the statistical
office of country A is likely to find a counterpart with very similar
responsibilities in country B. The same applies to a director of social
statistics or national accounts, or the director (or head of department) of
price statistics, the business register, environment statistics or
dissemination. No doubt, there are common underlying principles at work when it
comes to organizing statistical agencies.
3. There are some widely accepted,
alternative principles for structuring organizations:
- Knowledge or skill (e.g. departments of cardiology,
internal medicine and neurology in hospitals)
ii.
Work process, often based on the technology used (e.g.
letterpress and offset departments in a print shop)
- Business function ( e.g. manufacturing, research,
engineering, accounting.)
- Output (different product lines or services in
different divisions, e.g. household appliances, machinery, maintenance)
- Client (different organization units for retail to
households, large business clients etc.)
vi.
Time (shifts in factories and
hospitals)
vii.
Place (different geographical areas
that are served)
4. An entirely different distinction, which
cuts across the principles mentioned above, is that between hierarchical
organization structures and network organizations.
5. Principles 1 and 2 are particularly
relevant for statistical offices. Principle 7 is also relevant, but was already
discussed in Chapter 3 (regional decentralization). Some of the other
alternatives (3, 4, 5) may be relevant for certain parts of statistical
organizations, but can hardly be seen as major guiding principles. While the
distinction between ‘hierarchy’ and ‘network’, is also relevant, government
rules and bureaucratic traditions in many countries may not be
‘network-friendly’.
6. Since the seventies, when Handbook Rev. 1 was written, new trends
emerged that affected organizations in general. Some of these trends apply more
particularly to statistical offices. Among the general trends, those of particular relevance are:
v
Need for greater flexibility to respond differently to
different situations
v
Fewer detailed rules and procedures
v
Greater autonomy and encouragement of initiative
v
Fewer levels of management
v
Workers empowered to make decisions
v
Cross unit team structures, project teams, matrix organizations,
networks
v
Outsourcing and downsizing
v
Increased budgetary pressure
7. For statistical offices in
particular relevant trends include:
v
Need to improve timeliness
v
Need to reduce the reporting burden
8. In addition, dramatic increases in
decentralized computer power, as well as related developments in
telecommunication have also affected the organization of statistical offices.
Most of all these technical changes have affected the response to the trends
mentioned above.
9. The relevant sections in Handbook Rev. 1 focused on two
recognized ways of structuring a statistical office: by subject matter and by
function. The term ‘subject matter’ is ambiguous and has several connotations.
It may apply to the institutional source of the basic information that is
collected. Education statistics, for example, would be classified on the basis
of the source from which the statistics are collected: all statistics collected
from educational institutions, including say, their finance data, would be
assigned to the responsibility of the education division. The ‘knowledge base’
of that division would consist of all the ins
and outs of the education system. An
alternative interpretation of ‘subject matter’ centres on main categories of
data. This interpretation leads to groupings such as employment and
unemployment, finance, investment etc. Ignoring these subtleties, the most
common interpretation of ‘subject matter’ vs. ‘function’ leads to the creation
of divisions such as trade, industry, health, education etc. and the second to
divisions such as sample survey design, data entry and data editing, field
operations, analysis, dissemination etc.
10. For reasons, some of which will are
described below, it has always been difficult to make a clear cut choice
between ‘subject matter’ and ‘function’. Accordingly, the structure of most
statistical offices is a mix of the two. There are some functions that are
ideally suited to be grouped into agency wide functional units, e.g. sample
design and field work. Other functions, such as questionnaire and publication
content and analysis/interpretation generally require the direct involvement of
subject matter specialist and should therefore be assigned to subject matter
units.
11. There are two additional requirements that
have come to the fore more recently, namely:
v the
requirement that data collection should be organized as efficiently as possible
in order to minimize the burden for both the respondents and the collecting
agency and at the same time to speed up enhance timeliness
v the
requirement that statistical information be disseminated in a manner relevant
to the problems or questions that it addresses implying that dissemination
should be user-focused rather than based on statistical sources and processes.
12. These two requirements could give rise to
two different organizational structures within an agency: one for data collection
and one for data dissemination. In order to ‘connect’ the two structures, a
bridge is required, a unit in charge of re-sorting data after they have been
collected and edited, into new groupings that lend themselves better to
analysis and dissemination. This aspect of organizational design applies to a
statistical agency of any size. Naturally, the smaller agencies can adopt more
flexible solutions.
13. To summarize:
v
in theory, there are many organizational principles
that one could use to structure statistical offices
v in
practice, most statistical offices have been organized according to either
‘subject matter’ or ‘function’, but usually according to some combination of
these two principles
v new
organizational trends have affected all organizations and some of them are
specifically relevant for statistical offices
v
among new developments is the realization that
statistical offices may be organized in two ways: for data-input according to
data-source, for data-output according to user categories, with a bridge in between where the data are
re-sorted
14.
The
difficulties involved in finding the right organizational structure are illustrated by a brief case history.
5.4
Organization
and reorganization
Before
1974 the organization chart of the Dutch Central Bureau of Statistics (CBS, now
Statistics Netherlands) was simple: a
Director-General, a Deputy, a director for coordination and about 20
departments, most of them subject matter oriented but a few constituted along
functional lines (including a large department for computerized data
processing). Some departments were
subdivided into divisions, some not.
15. In
1973, the government announced a plan to relocate the Bureau to a city
(Heerlen) at 200 kilometres from The Hague (the administrative centre of the
Natherlands), in order to create government jobs in that part of the
country. This caused turmoil, not only
among CBS staff, but also among other stakeholders and in political
circles. The Director-General
resigned. After awhile, the compromise
was reached to relocate only half of the Bureau, as well as to substantially
increase its resources, partly to compensate for inefficiencies, partly to
tackle new statistical work, partly to create new employment. Also, a new organization structure was
developed.
16.
This
structure regrouped statistical departments into four directorates:
Economic Statistics (12 departments) Essentially: subject matter groupsing
Social Statistics (9 departments) Essentially:
subject matter grouping
Methods and development (4
departments, incl. the central computer department)
Functional
grouping
Office services (4 departments) Functional
grouping
All departments were subdivided into
divisions and subdivisions, and often two more layers.
17.
At the
same time, the staff employed was substantially increased, from about 1,500 to
well over 2,000. During the seventies
and early eighties, new increases in staff were allowed. In 1982, the (budgetary) number of posts in
the CBS reached the level of about 3,500 (the actual number went up to just
over 3,000). Then came the turning
point.
18.
From
1982 to 1992, the CBS (and most other government agencies) was forced to cut
down its budget and staff-numbers. Cuts
in the early nineties brought the staff down to around 2,500. Overall, the Bureau managed to keepmost of
its statistical output intact, thanks to effective computerisation. In the early nineties, it had one of the
largest and most advanced computer networks in the country: about 2,500 PCs in local area networks and
with high-speed connections between the two locations. Moreover, efficient software (including
Blaise) for data collection, data editing and data dissemination was developed.
19.
Regardless
of budgetary decreases, demand for statistics increased, largely propelled by
demand from the European Communities.
Around 1992, the point was reached where the increases in efficiency
could no longer offset budgetary cuts.
This created tension. The
Central Commission for Statistics, which decides on the CBS’ program of work,
had great difficulties to restablish priorities. Several important users (e.g. ministries) became dissatisfied.
20.
In
this situation, the CBS management tried a ‘pre-emptive strike’ and
launched the so-called Operation TEMPO,
which stood for:
TIMELY
– EFFICIENT – MODERN – PROFESSIONAL – INDEPENDENT
The main focus of TEMPO was
external: input-co-ordination to minimise response burden and
output-co-ordination to maximise user-satisfaction.
21. Analysis
made clear that the problems were the following:
22. Problems
created by external factors included:
increasing demand for statistics, decreasing budget, consequent need to
improve efficiency, tension between European and national requirements,
pressure to reduce the response burden, need to improve presentation and public
image, need to improve timeliness, need to improve public image.
24. The
organizational structure was completely transformed. Top management of the CBS
would consist of the Director-General and two policy directors (including the
Deputy). There would be 8 directorates: 4 for statistical production and 4 for
support. The first of these supporting directorates was for
input-co-ordination, responsible for the business register, data collection
from households, development of EDI (Electronic Data Interchange), etc. The
second was for output-co-ordination, including the development of a central
database for overall dissemination purposes, integration of information (incl.
national accounts), general publications and marketing and public relations
activities. The third supporting directorate was responsible for basic
statistical infrastructure (incl. methodology and computer services) and the
fourth for general office services. Financial and human resource management
were placed in a staff directorate, directly under the Director-General, who in
addition was assisted by a small cabinet, responsible for international
relations, legal and policy matters and for providing the secretariat of the
Central Commission for Statistics.
25. The
four directorates for statistical production were organised on the basis of a
combination of principles, of which subject matter coherence and relations with
‘market segments’ (both in terms of users of information and of suppliers of
basic information) were the most important. These four directorates were
responsible for:
i.
Agriculture, manufacturing, environment, energy and technology
ii.
Trade, transport and commercial services
iii.
Public sector and well-being of the population
iv.
Demography, labor, income and consumption
26.
In terms of management, statistical divisions would be fairly
autonomous; their performance was monitored by a system of so-called contract
management. Within each directorate,
there were 8-10 statistical departments.
Their internal organization was such that production teams were
responsible for the complete statistical process.
27.
In 1999, the government announced plans to close down the
office in Heerlen, as well as to once again drastically reduce budget and staff
numbers. Another turbulent period
followed. Again, a compromise was
reached. The Heerlen office remained
open, but the CBS was asked to implenent a plan to further reduce staff
numbers.
28.
This meant another fundamental reorganization, be it with a
quite different perspective: while TEMPO was induced by the need to strengthen
the orientation towards clients and respondents, the new reorganization was
pushed by the task to further reduce cost and improve efficiency. Hence, this time the focus was on the
streamlining and standardization of processes.
29.
The organization of statistical processes was, therefore,
restructured from a subject matter to a process-orientation: production units were integrated and
subsequently segregated according to stage in the statistical process. The movement towards integration led to the
reduction of the number of Statistical Directorates from four to two: one for business statistics and one for
social and spatial statistics. Within
each of these Directorates, separate Departments were established for data
collection and administrative editing on one hand, and for statistical editing
and analysis on the other. Because of
the geographical Voorburg/Heerlen separation, there is one Department of each
type in both locations.
30.
In this design, there was no longer need for a separate Input
Directorate, also considering that external data sources for persons and for
businesses have little in common.
Conversely, the Output Directorate(Macro-economic statistics and
Publication) was maintained, and its responsibility with respect to publication
and dissemination was even strengthened:
all data produced by the statistical divisions are published by the
Output Division, which also accommodates the Central Information Service.
31.
The Directorate Technology and Facilities completes the
Division structure. To ensure close
cooperation between technology/methodology and production, a substantial part
of its IT and methodology staff is operationally assigned to the statistical
Directorates.
32.
The supervision over the four Directorates and twenty
Departments is in the hands of the Executive Board, consisting of the Director
General and his Deputy, who is in charge of statistical policy and
processing. Together with the Directors
of the four Directorates, they form the Management Committee.
33.
The process-oriented new structure is heavily supported by
advanced IT-tools. Actually, large
databases account for a clear demarcation between the various stages in the
statistical process:
v
Data collection departments and data analysis
departments are delimited by a database containing consistent source data at
the micro-level; these data are the inputs for statistical editing and
analysis;
v
Data analysis departments on the one hand and the
publication and integration departments on the other are delimited by a
database containing consistent statistical data at the micro- and aggregate
level; these data are the inputs for the integration and tabulation processes.
34. Finally, it should be stressed that
inegration of processes not only leads to more efficiency. It is also a major step towards the long
cherished goal of presenting a coherent picture of society, in the form of
consistent figures.
35.
Postscript:
It
is difficult to assess the necessity or indeed the costs and benefits of the
organizational changes that SN (Statistics Netherlands) has implemented over
the years. Clearly, any major reorganization
may generate innovation (fresh ideas), gains in efficiency and better focus on
strategic objectives. It is, however,
als0o energy consuming and jeopardizes continuity. Many statistical agencies innovate, improve their efficiency and
rethink their strategies without major organizational changes.
The
last reorganization of SN, however, intends to fundamentally restructure the
primary process and thus to produce substantial efficiency gains as well as
integration, doing away with existing stovepipes for separate statistics. It is difficult to see how this can be
achieved without a major organizational change.
36. Newly appointed heads of
statistical agencies may wonder whether the organization that they have taken
over is the most efficient and effective one and whether changes in structure
ought to be made. Likewise, long serving chief statisticians may at some point
in time realize that the circumstances have changed so much over time that
there is reason to consider a possible reorganization.
37.
Before
any major reorganization is initiated, it is wise to think at least twice,
because of the risks and costs involved. There are also some general
considerations to be borne in mind. They are expressed below as maxims:
v
make no changes to accommodate every valuable new idea;
this creates confusion and uncertainty
v
organizations are not mechanical structures that can be
simply re-engineered to meet new sets of circumstances; instead, they are
living organisms and tend not to perform well when they are restructured along
totally unfamiliar lines
v
organizations are not ends in themselves; they serve to
facilitate and they must accommodate traditions and most of all talented people
v
undue changes in top management personnel may
antagonize staff and create factions
v
while efficiency is an important concern, some overlap
in responsibilities and even some redundancy cannot always be completely
avoided without running the risk of overlooking important matters; obviously
the overlap should be minimized
v
the organization chart should be simple and easy to
explain to insiders and outsiders; if it looks complicated on paper, it will
probably also be complicated in practice
v
it is advisable to limit the number of hierarchical
levels (four is widely seen as the maximum), because messages tend to get
distorted as they are transferred from one level to the next; at the same time,
however, the span of control of supervisors should remain manageable (ten
professionals is probably the maximum number to manage although for production
staff the number is, of course, much higher)
v
if a reorganization is unavoidable, keep the
reorganization process as brief as possible
v
make sure that the views of all staff members are heard
v
the reasons for any re-organization should be made
perfectly clear, because people resent working within a framework that is not
fully understood
38.
While
these considerations apply to most organizations, there are also some points
that should be borne in mind as they apply to statistical agencies in
particular:
v
Much of what statistical offices do is marked by sharp
fluctuations in the workflow. For example for annual business surveys most of
the basic data may be received in the second quarter of the year, most of the
data editing may happen in the third quarter, and the mail-out of
questionnaires for the next survey round may happen in the fourth quarter. To
even out the workflow it may be a good solution to combine this work with tasks
that have another ‘seasonal pattern’, e.g. short-term data collections, into
one organizational unit. The extreme approach is, of course, to look for an
agency-wide solution.
v
For certain activities, particularly those that affect
the corporate identity, it is important to aim at strict agency uniformity, and
these activities should be placed under unified control, in one organizational
unit. The agency ‘stamp’ is important because it assigns quality and conveys
integration. Therefore, products (including questionnaires and electronic
products on the Internet) should have a common look and feel; control of their
design should preferably be centralized. Other examples of the importance of
conveying a unified corporate identity are organizational units for
international relations and press offices.
v
In times of decreasing budgets, statistical offices are
sometimes confronted with the question whether or not it is financially
efficient to outsource certain functions, and which services can responsibly be
bought outside. Catering, janitor services, security, in many cases printing
and sometimes part of the dissemination activities are likely candidates for
successful outsourcing. Some more general (e.g. basic computer skills) or on
the other hand highly specialized types of training (e.g. management training)
can often be bought easily. Much more controversial is outsourcing of
interviewing; apart from quality issues that have to be considered, few
commercial interviewing agencies seem to be able to compete –price-wise- with
the agency’s own fieldwork staff. And finally, a very controversial issue is to
what extent certain computer services should be bought or self-produced.
39.
In all
organizations, a sufficient degree of coordination (and not only statistical
coordination in a narrow sense, as described in chapter 2) is a must. While in
smaller statistical offices, coordination may be achieved through ‘mutual
adjustment’, and coordination in units of larger offices through direct
supervision, there will also be a need for standardization –agency-wide- of
processes, outputs and skills.
40.
Even
more importantly, though this is a difficult topic, is the creation of a strong
corporate culture: shared beliefs, values and norms. A strong system of
internal communication (newsletters, Intranet, periodic meetings of the chief
statistician with senior and middle-level managers and indeed rotating
selections of all staff) may help to achieve this. A seemingly trivial, but in
principle very functional technique of creating a corporate culture is for
senior managers to regularly visit the work floor.
41.
Among
the more formal mechanisms to coordinate and to promote corporate culture, a
well functioning system of committees is perhaps the most important and
effective. In the following sections the role and workings of such committees
are discussed. A distinction must be made here between committees and task
forces or project teams. The latter serve to deal with specific, finite
problems. Committees usually have a more permanent character and serve to
address long term or recurring issues.
42.
Handbook Rev. 1 distinguishes the following as sufficient
justification for the creation of a committee:
v
Improving across-the-board communications and
increasing the sense of participation in collective initiatives
v
Mobilizing agency-wide support for new and
high-priority tasks
v
Ensuring multi-disciplinary contributions to a new and
complex undertaking
v
Maintaining a balance of different
interests and perspectives when new methods, standards or concepts are introduced.
40. On the subject of
structuring a committee required for effective management of a statistical
agency, Handbook Rev. 1 states:
“Indeed, it is, or it should be, one of the
constant challenges of top management in a statistical agency, irrespective of
its pattern or organization, to keep the horizontal dimension alive, visible,
and effective.”
41 We may safely assume that
a sizable agency (i.e., 500+ employees) will reflect in its organizational
structure both subject matter and functional units. As a result, there will be
points of internal tension for which some form of mediation is required. For
example, both the department in charge of industry statistics and that in
charge of health statistics have the required funds to pay for the computer
processing of their latest data. Monthly, conflict arises over which department
takes priority, as does the threat of one or the other missing scheduled
release dates. The chains of command of the two departments only meet at the
very top. A system that reduces tension, forecasts events, and standardizes the
manner in which future conflicts arising from this kind of situation are
resolved, becomes necessary. Handbook
Rev. 1 points out that horizontal mechanisms must be put in place to
resolve conflict and promote integration in any organization with separate
units that must agree on an integrated programme. Such mechanisms are known as
horizontal committees. Some statistical agencies have a complex network of
committees, some standing and some ad
hoc, all created to promote an overriding objective and to minimize
inevitable tensions.
42.
The
influence of a horizontal committee depends on whether the statistical agency
wants to maintain its established hierarchical flow of information, or wishes
to introduce a second source of advice and information specialized in corporate
rather than single organization problems. In the case of the latter, committees
may become as powerful as the regular units listed on an agency’s organization
chart.
43.
There
are essentially two kinds of internal committees,
those that deal with managerial and those that deal with technical issues.
Certain statistical problems exist irrespective of the good will of
participants, the soundness of the agreements reached by them, and the quality
of basic statistics. For instance, in those countries where there are estimates
of quarterly national accounts, issues inevitably arise from the nature of the
basic statistics and resulting inconsistencies when the accounts are first
compiled. Where a central register of businesses, classified according to their
economic activity exists, subject matter experts should take part in
agency-wide discussions to ensure that different surveys do not end up with
overlapping populations.
44.
Situations
requiring periodic discussions that cut across several organizations extend
beyond those concerning subject matter. Issues such as the recruitment of
professionals; the provision of general training for all classes of staff; and
the standards of dissemination where different parts of the organization
contribute to the same publication, all require open discussion and review of
internal protocols. These matters are of consequence to both large and small
organizations, but in the case of smaller organizations, the formality and size
of committees decreases substantially.
45.
One
important committee, essential in instances where the statistical agency’s
staff is unionised, deals with matters of labor-management relations. There are
many situations in which the chief statistician needs a direct conduit to the
staff, just as staff members need direct access to the agency’s executive. The
nature of such a committee’s agenda depends partly on which elements are
handled on a government-wide basis and which are left for each agency head to
solve.
46.
In the
following discussion we address those concerns requiring regular attention.
Issues resulting from a temporary increase in interest justify the creation of
a Task Force and are not seen as crucial to the agency’s organization. The
elements required for a Task Force to operate successfully, though, are much
the same as those necessary for the functioning of a standing committee.
47.
Typically,
committees do not have a budget. Their power is a result of the respect they
are given by the rest of the organization and the established practice of
resolving cross cutting issues by committee. Still, committees have minimum
requirements — a chairman, a secretary, an agenda, and rules on membership. The
chairman should be a capable and impartial leader, and his/her judgement should
be respected, even though it is not an official decree. The committee’s purpose
is to advise the chief statistician on what it believes. While the latter has
the power to ignore the committee’s advice, in practice, and possibly with
slight modifications, the committee’s voice will be heard.
48.
The
Committee’s recommendations should be recorded. These decisions, together with
a short account of the reasoning behind them, should be open to inspection by
the rest of the staff. For large offices with a comprehensive internal system
of communications, all of the committee’s agenda, deliberations, and records of
decision should be posted on the agency’s Intranet. As an organizational rule,
while the committee’s recommendations are not reached by consensus, staff
members should always be privy to the committee’s decisions.
49.
Committee
membership is an opportunity to contribute to agency-wide policies, as well as
a training ground for future senior managers in the sense that it provides
members with a broader perspective than they could acquire from their regular
jobs, and makes them aware of the wide range of considerations important to the
agency. Committees must not grow too large, or else they become cumbersome and
incapable of reaching closure on the issues they debate. Generally, committee
meetings lasting over two hours represent a heavy tax on their members’ time.
In order to maximize the training benefits that such committees confer on their
members, some system of membership rotation should be adopted. One possibility
is that members serve on a committee for a mandatory period of two years,
extensible by one year, after which they move on. Inevitably, an informal
league table ranking committees by their perceived influence or prestige will
be established. Determination of membership should acknowledge prestige and
ensure that those managers who show promise serve on committees with the most
critical agendas.
Conclusions
Organizationally, there is no ‘ideal’ model. In
practice most statistical offices have been organized according to either
subject matter or function, but usually according to some combination of these.
Moreover, there has been a realization that one might organize statistical
offices in two ways: according to data source or for data output according to
user categories.
Before any major reorganization is initiated it
is wise to think at least twice. Some of the general considerations to be borne
in mind include:
v
organizations are not ends in themselves; they serve to
facilitate and they must accommodate traditions and most of all talented
peoplewhile efficiency is an important concern, some overlap in responsibilities
and even some redundancy cannot always be completely avoided without running
the risk of overlooking important matters; obviously the overlap should be
minimized.
Chapter 6
MANAGING STAFF
1. Nothing is as important to a statistical
agency as its staff. For the difference between a well-organized, strongly
motivated and technically competent staff, and one that only displays these
attributes to a very modest extent is the difference between a good and
credible office and one which is second rate. Expenditure on staff accounts for
a dominant fraction of a statistical agency’s budget. In a sample of offices
representing a broad array of sizes and stages of development, salaries account
for approximately 70 per cent or more of the total budget. In addition, heads
of agency tend to devote more time to staff problems than to any other issue.
2. Improvements have been made in “people
management” in bureaucracies in general and in statistical agencies in
particular. People management is acknowledged in discussions in international
venues as having been taken for granted in earlier times, and is now perceived
as critical. Also addressed in such discussions are the various stages of
training and the need to retain and develop staff, as well as the
considerations that determine how specialized the agency’s staff should be.
Handbook Rev. 1 on the subject of
staff and skills
“ A decisive factor in the internal
capability of a statistical agency is the calibre of its staff. An agency can
only function well if good people are available to make it work. Organizational
arrangements may contribute to enabling good people to do their best; but it is
essential to give utmost attention to building up the right kind of staff in
organizing and managing a statistical agency.”
“To acquire the kind of skills needed, a
statistical agency must have a well-planned policy and active programme of
recruitment, career development, education and training.”
“In addition to the personnel unit, other
units of the administrative branch should be involved in carrying out the
personnel policy. The acquisition and maintenance of adequate office space and
equipment can contribute not only to efficiency but also to the well-being and
satisfaction of the staff. Moreover, suitable accommodation, facilitating ready
communication between people working in related fields, is conducive to the
integration of statistics because, in a real sense, the “successful integration
of statistical end-products is contingent upon the ‘integration of the
statisticians’ .”
“After some time, when a minimum of
experience has been gained, every employee should be obliged to attend an
orientation course where further explanation is provided.”
“Officials of the personnel unit should teach
the introductory course for employees at lower levels [management should teach
professionals]; the establishment of good personal relations between this unit
and the new employees is of great importance.”
“The in-service education and training described
above requires not only teaching but also quite extensive administrative
efforts. Therefore permanent machinery should be established in the form of an
in-service training school. In a small or medium-sized agency, a part time head
of the school may be selected from among the best qualified professionals of
the agency. In a large agency, a full time head may be needed. Such machinery
is necessary to ensure the proper design of courses, recruitment of teachers,
selection of students, etc.”
Handbook Rev. 1, Chap. VIII
6.2. Staff composition
3. The share of the professional and
technical components has increased substantially even in cases where actual
numbers have dropped and relative resource endowments have not kept up with
changes elsewhere in the civil service. It happens in some statistical agencies
(at least in countries in the OECD area) that professionals account for one in
two staff members. Statistical surveys have become more sophisticated, at least
in terms of planning, and general expectations regarding the quality of the
estimates produced by statistical agencies have risen. The computer and
communications revolutions and the resultant gains in labor productivity have
surely contributed to these trends. One of the many results of these changes is
the impact on the human resource policies adopted by, or imposed on,
statistical agencies. These policies, which were still in the developmental
stage 20 years ago, appear to have fully evolved in many agencies and should
therefore, be open for study by statistical offices that are still undergoing
structural changes.
4. The staff of a statistical agency is
not solely made up of statisticians. But it is a common practice—although by no
means a rule—to appoint someone with strong credentials in statistics.
5. Irrespective of the incumbent’s other
virtues, the demonstration to the public that the government’s chief
statistician has a profound understanding of statistics tends to inspire
confidence. Due to the range of subjects covered by a modern statistical
agency, the staff tends to possess a variety of skills and academic
backgrounds. Even agencies that only conduct moderately complex and specific
statistical undertaking find themselves in need of a wide range of talents.
Statistical operations require a mix of economists, sociologists, demographers,
econometricians, model builders, geographers, anthropologists, criminologists,
engineers, computer experts and so on.
6. Some of these skills are not required
for day-to-day statistical operations, but they prove essential when new
operations are being planned and launched. Naturally there is a certain amount
of learning and technology transfer that makes the staff of statistical
agencies more versatile but in the end, medium and large offices will find that
there are limits to what can be borrowed from other countries and other
institutions and will therefore choose to enhance their own capabilities.
7. There are important reasons to build a
multi talented staff with a good mix of academic and work experience
backgrounds. The following is a discussion of the factors that dictate the
balance of skills required from the staff in an effective statistical agency.
As in many other situations, the question arises whether certain capabilities
are used so often that they should be available to the agency on a permanent
basis or whether given the expected usage it is sufficient if they are acquired
if and when their use demands it. This holds for people as much as for
equipment. For example, when adapting an international classification to suit
national purposes, there will be an inevitable growth in demand for very
specialized knowledge, such as chemicals, electronics, pharmaceuticals,
computing, the audio-visual industry, and financial services. Likewise, international
trade statistics require specialized commodity knowledge as well as knowledge
of national customs administration processes. These are examples of
requirements that tend to be in demand when new methods are introduced or new
classifications are instituted. In such cases, an agency may choose to employ
the services of a specialist, through a consulting contract or some other
non-permanent association.
8. The more general the capabilities
sought, the easier it is to administer and maintain them. Moreover, general
capabilities have a versatility that agencies, particularly the smaller ones
and those that have not yet fixed a medium-term agenda, find attractive.
General training can be provided formally; creates the basis for a shared
culture and predictable performance; and can more easily be supplemented by
specialized training as staff members move from one job to another.
9. Professional staff can often be divided
into two categories:
v General
personnel capable of performing at lower and intermediate levels of
sophistication in the areas of computing, statistical design and analysis,
national, public sector or enterprise accounting, and economic, social or
demographic analysis
v Specialized
personnel capable of performing at a substantially higher level in an area such
as non-response analysis or in a filed such as analysis of geographic
information, criminology, or health statistics.
10. This structure can be augmented by the
occasional purchase of services provided by national or international
consultants who are highly specialized or concerned with matters of policy and
public perception to contribute to some major undertaking.
11. Once a determination has been made as to
what is acquired as permanent staff and what is acquired through short-term
contracts, the chief statistician will evaluate supply and demand and attempt
to correct any imbalances through recruitment policies.
12. Irrespective of its
composition, the work force of a statistical agency is its most precious
resource and as such it must be treasured. An effective statistical agency will
manage its staff with the help of an explicit set of policies. The most
important elements of a personnel (human resources) policy are:
v ability
to ascertain personnel needs
v standards
and techniques for recruitment
v deployment
of personnel
v introductory
training
v monitoring
equity in recruitment and in career advancement
v intermediate
training
v career
development and job rotation
v training
for managerial functions
v staff
motivation and retention
13. These elements are listed in the
approximate order in which they naturally arise. Certainly, other features
could be added, but these are sufficient to support a policy with common
objectives. The following is an example of a set of coherent objectives, which
the elements listed above are designed to help attain:
v increase
the proportions of professional staff by expanding the annual intake of young
professionally qualified staff
v ensure
that once recruited, qualified people are assigned suitable jobs, and everyone
is treated fairly
v devise
a proper mixture of deterrents and incentives, to be made explicit to the staff
v administer
sufficient training to staff members at key points in their careers, so as to
maximize versatility and motivation
v ensure
that key jobs are staffed and have a possible successor ready to step in to the
incumbent’s shoes should the need arise
14. To facilitate a discussion of ways in
which these policies can be administered, four internal institutions should be
examined:
v office
of personnel
v personnel
committee
v in-service
training facility
v evaluation
machinery
15. The office of personnel and the personnel
committee assess needs, recruit and deploy staff, formulate retention policies
and ensure equitable treatment. The in-service training facility, as its name
suggests, oversees training. Finally, the evaluation machinery fields questions
regarding the adequacy of these mechanisms.
16. Handbook
Rev. 1 suggests that when a statistical agency reaches a certain size, it
should invest in a professional Office of Personnel. But experience shows that
while such an office is essential in dealing with the administration of all or most personnel policies and with advising
on the development of personnel policies and instruments, it seldom plays a
decisive role in detection, evaluation, or persuasion, all of which help ensure
that talented people come to work for a statistical agency, that the agency’s
stock of people gets renewed as often as required, and that there are
incentives to retain the best employees. This finding, while seemingly
critical, merely recognizes a simple fact: a professional Office of Personnel
primarily will administer public service acts, enforced in most countries, that
regulate the conditions of recruitment and employment in the public sector.
17. Cases in several countries have also shown
that the most successful formula for the management of personnel issues is one
in which the subject matter staff is intimately involved with such personnel processes
as recruitment, training, and retention. Heads of personnel units are the
advisers best qualified to validate suggested personnel approaches and
policies. They are also the right choice for the execution of policies once the
choices are made and the policies in question defined.
18. Often Ministries such as Planning,
Treasury, Industry, Employment, Agriculture, and Transport as well as agencies
such as Central banks, Stock Exchange Commissions and Comptrollers Offices will
draw recruits from the same pool as the statistical agency. In most countries,
initial, and subsequent, salaries in the statistical agency are relatively low
compared to Central Banks. Moreover, in some developing countries this
inequality also holds between the statistical agency and other ministerial
agencies. For this reason the statistical agency should develop a recruitment
plan, including attractive items such as:
v impressing
upon them that professional work tends to be more rewarding in a non-political
environment such as that maintained by the statistical agency.
v appealing
to the competitive instinct of recruits by promising them that merit will be
used to militate against the more onerous aspects of seniority rules in their
promotion prospects; and
v giving
recruits intensive and marketable training of a practical nature either
directly or by agreement with a teaching institution;
v striving
to send young people to acquire training abroad as soon as they prove
themselves ready;
19. All this is subject to the constraint that
the statistical agency should rather not recruit than deliberately recruit the
mediocre. The personal involvement of the chief statistician in these matters
can help transform a mechanical and not very fruitful approach to young people into
one filled with prospects and enthusiasm.
20. Young people who decide to go into public
service are often seduced by the prospect of working on policy analysis. The
opportunity to be in the proximity of those whom they perceive as the makers of
the Nation’s destiny is an unrivalled attraction. This creates a problem for a
statistical agency. Not only does it add to the difficulty of enticing top
graduates, it also makes it more likely that if recruited and given specialized
training, these graduates will move on to more prestigious centres of
government. Furthermore, once economists and social scientists start working
for other government agencies, the possibility of getting the brightest back to
a statistical agency is not high. For this reason, statistical agencies must
accept the cost and the risk of recruiting young people with no experience,
forge suitable training policies, and stick to them in the knowledge that they
face inevitable high rates of attrition particularly among the non-statisticians.
21. There are a great number of careers open
to young people with ability for quantitative analysis that promise greater
incomes and a faster rise to even higher monetary rewards than anything the
chief statistician can hold out for them. It is very difficult for a
statistical agency to compete with the private sector on initial or prospective
salary. However, statistical agencies should be looking for who are searching
for more then salary. Those who may be more successful in the government in general
and in the statistical agency in particular, would join the organization
because they wish to have a job with higher moral purpose that is public
interest. Still, as we have mentioned previously, it is often difficult for the
statistical agency to compete with other governmental agencies on the lookout
for the same kinds of talents. Mitigating this situation is the fact that those
who have graduated in mathematics with special reference to statistics, and who
do not wish to teach or to be an actuary, but are interested in mathematical
applications to sampling, statistical design of experiments, hypothesis testing
and so on, the statistical system presents the most interesting challenges. The
statistical agency should make it a priority to find professionally interesting
work for such employees as soon as they join the agency. In addition it is
probably easier as a matter of practice to get young statisticians to broaden
their interest so that they extend to economics and social organization and
stay with the statistical agency rather than expect economists and sociologists
to extend their stay for the sake of interesting issues in the theory of
measurement.
22. In official texts on organization, little
if anything has been said about a new employee’s first day in the office. And
yet the first day (or the first few days or the first week) can have a
pronounced effect on the recruit’s attachment, respect, and dedication. If the
statistical agency is just another government agency, it will not be able to
compete with more prestigious areas within government. It will have neither the
cohesive and elite forces that bind together Foreign Service officers nor the
high profile work that characterizes the business of analysts working on macroeconomic
policy. The strengths of a statistical agency are that it will allow long term
pursuits to be entertained in an atmosphere relatively free from political
turmoil and totally free, one hopes, from successful political pressures. Those
strengths can be cemented in a recruit’s first few days and converted into
loyalty to the office.
23. An effective statistical agency will
ensure that the following takes place on the first day:
v
the new recruit finds an agreeable place in which to
sit and something definable to do
v an
accessible person is identified to whom to address questions and get doubts
resolved
v
the organization conveys its willingness to consider
the individual’s career in addition to the immediate job
v
an explanation is given of what the individual’s
activities will be in the immediate and the near future, how they fit in with
the organization of the office and what specific objectives they are designed
to serve
v
a training plan is presented
v
orientation. Some(larger) offices have developed the
tradition of organizing, at regular intervals, e.g. every month or every second
month, an orientation day, which is meant to give new recruits a quick overview
of how the organization is structured and operates.
The first day has
become an incomparably easier affair to manage with the institution of the
Intranet. In several offices, recruits are equipped with a PC and a modem and
find a personal greeting from the chief statistician; the text of the law that
rules their agency; the structure of the organization, complete with the names
and telephone numbers of key people; extracts from major publications; and
usually a thriving exchange of messages among staff members that range from
private sales of cars or audio equipment to serious discussions of
methodological problems. Even if a PC and a Modem are not available, a printed
guide with the above material would still be of value.
“The
supply of professionals with the skills required in a statistical agency is as
a rule scarce….Frequently the progress made by new staff members is slow
because they do not know the characteristics of the agency sufficiently
well….After some time, when a minimum of experience has been gained, every
employee should be obliged to attend an orientation course where further
explanation is provided…”
“In many
countries, the knowledge acquired in universities, both in subject-matter field
and in statistics, is often too general and abstract to be immediately well
adapted to the needs of statistical agencies….Developing countries have access
to regional institutes of statistics, sponsored by several countries, which a
statistical agency should take advantage of.”
25. There is no denying that the demand for
training has increased. Generally speaking, the statistical process has become
increasingly complex. Traditional on-the-job training methods, while
indispensable, are not the most effective way to ensure versatility of skills
and standardization of methods. And on the other hand, training facilities for
developing countries seem to have diminished by insufficient budgets and lack
of available trainers. In addition, the demand on statistics in terms of
accuracy and timeliness has increased. This makes the choice of training
methods, particularly for more experienced, specialized statisticians a crucial
issue.
26. While there are different ways to provide
career-long training, one way that has proved to work in many countries is to
consider general-purpose training as having three distinct cycles:
27. Moreover, one should make the
corresponding cost part of the agency’s regular budget; and administer training
to all targeted staff members as a matter of course. Because such training is
regarded as general purpose, it is not sufficient and should be complemented by
more narrowly defined courses to meet specific short-term needs. Many offices
are not in a position to provide courses at all or at any of these levels. This
is why alternatives and special arrangements are so important. In general,
training can range from the very formal, given in a separate school by
qualified trainers and lasting for several months at the end of which course
graduates are awarded a diploma, to an informal series of discussions with
senior officers, possibly including the head of the agency.
28. For those offices, which are very small,
newly formed and where tradition is still being created, the alternative
solution may be to benefit from what is available in the offices of
better-endowed neighbours; or in more advanced statistical agencies in which
training modules for visitors from abroad are featured regularly. There are
numerous variations on this practice, and the possibility of offering
additional courses largely depends on the overall resources of the statistical
agency as well as on the time and money it wishes to devote to initial
training. Generally, the size of the agency and its endowment will determine
the formality and specialization of the courses it offers.
29. The idea of an in-service training
institute is almost as old as that of creating a stand-alone statistical agency
dedicated to the improvement of statistical methods and compilation and
dissemination of social and economic statistics.
“…[The]
education and training described above requires not only teaching but also quite
extensive administrative efforts. Therefore permanent machinery should be
established in the form of an in-service training school. In a small or
medium-sized agency, a part time head of the school may be selected from among
the best qualified professionals of the agency. In a large agency, a full time
head may be needed. Such machinery is necessary to ensure the proper design of
course, recruitment of teachers, selection of students, etc.”
30. Some countries have such a facility and run
it very much in accordance with these recommendations. Other countries have
different arrangements with local universities, in which they share the burden
of administering at least the more formal aspects of training. Others still
adopt ad hoc solutions and ensure
that their junior and intermediate staff are trained abroad, preferably in
neighbouring countries or else in a regional training facility.
31. An agency’s “staff model” may fall
anywhere between two extremes. At one end of the spectrum is the “no one moves”
model. Its objective is to maximize specialized human capital by allowing staff
members to learn more and more about their respective areas of responsibility
(e.g., industry; health; education; retailing; balance of payments). Staff
members would be expected to have a purely vertical career and would only leave
their areas when called upon to discharge agency-wide responsibilities.
32. At the other extreme, the driving
principle is versatility. Staff members are encouraged (or required) to acquire
the widest possible experience in the shortest possible time by moving from one
job to another. Where this movement is directed by the office of personnel, it
would likely be in keeping with an optimal pattern so that the acquired experience
would provide the most versatility.
33. It is virtually impossible to quantify the
many factors that affect how an agency will balance the need for specialized
knowledge and versatility. For example, staff morale is an important
consideration, and if the idea of permanent rotation goes against habits and
expectations, its introduction in an extreme form may create negative reactions
that outweigh its benefits. Conversely, in a culture where the staff are used
to and expect to change jobs every so many years, the absence of opportunities
to try different assignments might well lead to frustration and atrophy.
34. Somewhere in-between is a point that makes
possible to find a balance, capitalizing on the inherent strengths of the two
extremes. For example, an institution may require that at a certain level no
one is allowed to remain on the same job for more than five years; if one opts
to keep the same job for an indefinite period of time, it would be with the
knowledge that one’s chances of advancement are correspondingly reduced, even
if job performance is entirely satisfactory.
“Despite
the contributions of national, regional and international institutes to the
trained statistical work force, critical shortages exist in developing
countries. In a more fundamental sense, the shortages reflect the disadvantaged
position of the statistical service in many developing countries with respect
to pay scales and related working conditions as well as the low or moderate
status of the national statistical services within the government hierarchy.
Thus the most promising young people do not enter the statistical service as a
first choice…”
Handbook Rev.
1
35. Staff retention is one of
the major challenges for every office, large or small, well or poorly endowed.
The smaller and less well endowed the office, the greater the loss sustained
when a talented individual leaves. Several methods of retaining staff are worth
exploring but none is foolproof. Ultimately, losses of employees to other
offices and to the private sector must be included as part of the expected cost
of running a statistical agency.
36. One way to deal with
staff retention is simply to accept that no statistical agency can keep its
most talented people forever. Accordingly, the best it can do is to procure the
services of capable staff for a limited period. It is better to be explicit
about it for in this way there are no surprises at the end of the period for
which these services are contracted. One tactic is to convince departments of
government that attract talented professionals with abilities in quantitative
analysis that it is in their interest to place such people in the statistical
agency for a while, so that they can learn all about the limitations of data
and the possibilities of statistical inference. Moreover, such sharing of staff
will help the statistical agency sort out its priorities, connect the agencies,
and creating a better basis for ongoing dialogue. Since these features are
always valuable, the initial stationing of people in the statistical agency or
the ongoing exchange policy could become permanent features of a
government-wide personnel management program.
37. Another way to improve
retention is to provide positions of increased responsibility to young and
talented people as soon as they have demonstrated their capability. There is a
risk, but it is preferable to retaining staff members of mediocre talent while
losing those whose talents are exceptional.
38. Another option is to form a contractual
relationship with recruits that will deter them from leaving the statistical
agency for frivolous reasons. For example, in offices with structured
introductory training programmes, recruits could spend their first year on a
particular assignment, after which they would participate in the training
programme, provided they have signed up for a stay of at least three to five
years in the statistical agency.
39. There are also special programmes that can
be developed to assure that qualified persons are recruited such as the ‘cadet
programme’ of the Australian Bureau of Statistics (where the statistical office
pays for the education of talented future statisticians), and internships.
40. Traditionally the problem of staff
retention has been sharpest for experts in information technology (IT). As the
speed of deployment of IT has increased so did the demand for trained
personnel; however, the supply of qualified people did not increase as quickly.
In such circumstances, it became virtually impossible for a government
institution to compete with the private sector, the banking sector, or
companies owned and operated from abroad. Government institutions could not
offer competitive salaries, and what they could offer—job security—was not an
overriding concern for the young and mobile professional with those skills. The
standard response to this situation, which has shown no great variation in the
recent past, was to recruit ever younger, less-experienced technicians and
administer training. But this policy, in addition to consuming resources,
converted statistical agencies into an unrecognised training centre for the
private sector. As soon as recruits were trained, some other enterprise stepped
in, offering to double or triple their salaries.
41. Next chapter will consider in more detail
how to manage IT in a statistical agency.
42. No recently appointed chief
statistician—with the possible exception of one in a newly created office—can
make more than minor staff changes. Annual recruitment represents only a small
share of the total number of employees and is unlikely to make a profound
difference in the short term. Most agency heads will inherit a complete team;
some will inherit a ready-made team from which the top layer has been removed,
as when there is a change of government. Whether or not the existing team
is ideal, it is prudent to keep it intact (to do so may also be the only option
under the law), and to make improvements through gradual additions and by
taking advantage of voluntary departures and retirements.
43. Occasionally, total resources are
increased appreciably because of a special project (e.g., a Census or a very
complex and costly survey) for which supplementary staff members must be
recruited. In such cases, significant staff changes may be made; this is not
the rule but when it does happen experience shows that the risks involved in
hurried recruitment outweigh the bad impression created by a sluggish response
to an urgent need.
44. A chief statistician newly appointed to
run an existing agency may worry about being overly dependent on the existing
staff for advice, and may be concerned that the staff will take advantage of
the situation, perhaps to carry out a pet project, exaggerate a technical
difficulty, or add to resources in a less than unbiased fashion. Accordingly, a
chief statistician joining an agency may wish to bring in someone he has worked
with, whose judgement he trusts and with whom he can discuss options as an
alternative to his own immediate subalterns. While this practice may provide a
certain cushioning, it may well have the unintended effect of widening the gap
between the head of the agency and the regular staff. In fact, the greater the
access and the level of the newcomer the greater the danger of internal
rivalry. One device, which has been used to signal the transient nature of the
appointment, is to bring in the newcomers but not as part of the established
hierarchy.
45. So far we have described the elements of a
human resource policy and the factors that affect them. But makers of
policies—Chief statisticians, Heads of Personnel, Committees on Human
Resources—are not simply free to pick and choose those measures they believe
are required to solve a current problem without concern for the coherence of
the resulting set. Policies, in this domain perhaps more than in any other,
must reinforce each other in order to be effective. For example, a strong
impetus assigned to job rotation can only make sense if versatility is favoured
over specialization and general training over on-the-job training. The list
below details a possible set of measures and policies applicable to small to medium
sized offices:
v
a
steady influx of recruits, weighted in favour of those with post-secondary
training, with insistence on computer proficiency and flair for quantitative
analysis
v
as
heavy an investment in the beginner’s initial training as resources permit,
laced if at all possible with out of the institution or even out of the country
training opportunities for the most talented
v
an
agreement with other public sector institutions in which talented young people
remain in the statistical agency for a reasonable period of time
v
promotions
based on merit as opposed to seniority, in order to minimize the loss of able
staff members
v
semi-compulsory
job rotation in order to avoid excessive specialization, supported by ongoing
training to ensure versatility and acquired familiarity with a variety of jobs
within the statistical agency
Conclusions
Staff are
the most important asset of a statistical agency. It therefore deserves a
policy especially designed to ensure that:
v the
most talented and promising people are recruited
v the
selection process is as thorough and as professional as possible
v newly
recruited staff are properly assimilated into the statistical agency
v working
conditions are satisfactory and designed to facilitate communications between newcomers
and the rest of the organization
v personnel
administration is a necessary activity within a statistical agency and is
better handled in a centralized mode by professionals but not exclusively by
them
v training
needs may justify a special training centre with its own staff and a dedicated
(part time or full time) professional director
human
resource policies should constitute a coherent set. (small and medium sized
offices should be encouraged to choose policies that favour versatility and
maximize career opportunities within the office)
Chapter 7
MANAGING INFORMATION TECHNOLOGY
7.1. Introduction
1. Much has changed in the twenty years
since the matter of capital assets was addressed in Handbook Rev. 1. Four advances, familiar to all, have greatly expanded
the capabilities of a statistical agency:
i.
Powerful, relatively
inexpensive computer equipment has been made available to all staff members in
many statistical offices.
ii.
User-friendly applications
software has given staff members control over a number of key statistical
functions ranging from questionnaire design to collection, editing, tabulating,
mapping and publishing. Readily available off-the-shelf components have made
in-house application programming easier and component reuse within the organization
has become more common.
iii.
Computer networking has
facilitated internal access to data and metadata through established tiered
client/server environments.
iv.
Internet technology has made
timely access to the outside world feasible for staff members, thereby making
it possible to conduct research, gather general information, etc., at all
levels of an organization, rather than solely through top management.
2. These and related changes have given rise
to new concerns regarding how to manage the information technology (IT)
environment. We include in the IT framework hardware, software, staff
resources, and commercial services.
3. Our focus, here, is not with the
management of particular technologies but rather to investigate if there are
broader themes that will withstand the inevitable – and fast paced – changes in
technology to come. Thus, for example, we will not try to examine the pros and
cons of a client server model or optimal network structure within an agency.
Even if such competency were within the province of this handbook the opinions
would be out of date even before the handbook was published. Rather we will
first consider two organizational models which we believe have a reasonable
chance of being valid over the next ten or so years and then consider some
general prescriptions which should be valid independent of particular
technological winners and losers.
3. Model
1. In this model the agency makes detailed and comprehensive decisions
about its informatics infrastructure. This goes far beyond hardware decisions
and also includes a common software architecture. By this we mean not only the
underlying data base management software and consequent standardized definition
throughout the agency but also standard user interface, group ware (e.g. Lotus
Notes type products and again standard naming conventions and archiving
conventions), personal software such as spreadsheets and the integrating tools
to “glue” the software together. Some agencies have gone so far as to prohibit
private personal file storage to be attached to workstations to ensure that no
other software -- except that approved by the central informatics area – is
found on individual machines. Maintenance, updates and decisions on new
software are all determined by the central organization.
4. Since there is one common
infrastructure in the organization this approach serves to bind together the
organization and prevent information anarchy. As well it makes it much easier
to integrate different functions of the agency such as registers and cognitive
research. On the other hand it may stifle creativity and parts of the
organization may not have the tools they need to perform their functions
properly. The success of this model is clearly dependent on the effectiveness
of policies and personnel in the central informatics division.
5. Model
2. This model also recognizes the need to avoid information anarchy. This
approach relies on standards – hardware, software, and telecommunications – to
provide for a coherent informatics environment in the organization rather than
a standard and unyielding infrastructure. At first glance it may seem that this
is easier to implement. Certainly most staff outside the informatics
organization will embrace it because it provides for more local flexibility and
decision-making. From a central management view it is harder to implement. In
practice, it has proved exceedingly difficult to ensure that new software --
even if it meets the standards – will transparently integrate with existing
software and changes to the existing software may cause unforeseen problems in
the local new software products. This approach does allow some flexibility and
potential adaptation of new software even if the central informatics staff
makes decisions, which turn out to be short-sighted.
6. In either of these two models there is
general agreement today that software applications should be developed as
closely as possible to the substantive user. In the first model all the
software tools would be defined, provided and maintained by the central
informatics organization. In the second model the basic tools would be defined
and a set of standards that would have to be met are also defined by the
central organization. Both models have shown that they can be successful. The
choice is often made on the culture of the organization as much as on technical
grounds.
7. In the 1970’s and early 1980’s, as statistical
agencies tried to integrate mainframe computers into their operations, concerns
were very different. A number of prescriptions emerged, regarding not only the
way the systems analysis workforce should be managed, but also relations with
other government computing establishments, as well as what software to
implement and what to avoid. These are summarized below:
v
Do not
squander specialized human resources; they are scarce and must be kept together
under central supervision.
v
Document
all systems extensively.
v
Organize
the work force creatively so as to provide both functional and subject matter
experience i.e. balance versatility and specialization.
v
Promote
training activities where on-the-job training is built in.
8. With respect to the first recommendation,
it is still true that specialized informatics expertise is scarce. However,
depending on the size of the organization and its levels of decentralization it
may be more sensible to locate some of these resources with the functional
areas of the statistical office. Training and staff development should be
elaborated and updated on a regular basis. The importance of documentation
cannot be underestimated although the difficulties in achieving this objective
are as great today as they ever were. Finally, exchange of experience and
know-how between area specialists and application developers should be
encouraged.
9. While these recommendations generally
still hold we submit that several of the recommendations from the Handbook Rev.1 have been bypassed by the
changes in the IT industry:
v Do
not subcontract any systems work and programming, other than certain one-time
applications, to outside agencies.
v Do
not assign systems analysts and programmers to new projects until they have
completed clear documentation of data and programs for ongoing projects.
v Do
not leave administration of hardware to other parties unless it is guaranteed
that statistical processing will get the highest priority.
10. With respect to the first two
recommendations, instead of asking how it should be developing software
applications the correct question for the statistical agency might better be:
“Should a statistical agency be engaged in any customized software development
– regardless if it is developed by its own staff or subcontracted?” The
evidence to date says that an agency does this at its peril. The history of
customized software projects is replete with projects that were never
completed, or partially completed and over budget. The comprehensive
off-the-shelf software products now available reinforce this admonition. A rule
that some have used is that if seventy to eighty percent of an agency’s
requirements can be met by a product that actually works then the agency should
consider changing or dropping the remaining twenty percent of its requirements.
With respect to the last recommendation it is worth noting that many companies
are increasingly looking to find others to operate their hardware
infrastructure. This practice, of concentrating on the core functions of an organization
and contracting out – or outsourcing – others that can be better done by
specialized organizations is one of the greatest opportunities, and challenges,
to statistical agencies in the IT area.
11. Organizations generally receive best
value using commercial products and services, if they can be adapted to the
mission requirements. The common reasons for considering outsourcing are:
v
Proven, commercially viable
technologies are available only in the marketplace,
v
“Standard” or “commodity” products are
less expensive due to competition,
v
Vendors often have specialized skills
that are difficult to maintain “in-house”,
v
The agency can concentrate on the core
mission (which is not IT project development),
v
Competition can bring innovation to
bear on a mission requirement,
v
Risks can be shared with a vendor if
the rewards are great enough, and
v
Proprietary, or unique software or
systems may only be available from a vendor.
12. Standard IT products are those having
high market penetration, such as the personal computer, which have industry
standard characteristics, and are available as commodities from a variety of
sources. This category also includes some classes of software and network
service products such as computer operating systems, office software suites,
some statistical packages, network portal services, and web hosting. These are
the most likely candidates for outsourcing not only the product itself, but
often the maintenance, update, and replacement.
13. Commercial sources are often used to do
IT project development and implementation, or provide specialized skills (such
as network management, proprietary database knowledge, etc.). This is done
where cost-effective and the designated tasks fit an organizational strategy
for in-house versus out-house skills and responsibilities. The requirements for
this category are the most diverse, and vary not only from country to country,
but also from organization to organization. Past performance and the experience
of other similar organizations can be the most effective aid in deciding the
level, source and scope of outsourcing IT project development and support.
14. The accomplishment of a mission
requirement itself can often be competed through outsourcing, rather than
simply competing various pieces. With careful construction of “proper
government function” and protection of the information being handled, vendors
can be given the incentive to accomplish an end result by innovative methods.
Recent examples include the collection of delinquent taxes by commercial
vendors, using a for profit business model to govern the contract. It is
becoming more feasible to outsource or contract larger individual segments of
government functions. See also Section 9.1.8.
15. Outsourcing can often allow the risks to
be shared with a vendor, along with the rewards. Often a vendor can share in
the benefits from implementing a project, either through rights to data and/or
products developed, or receiving revenues from services provided, such as
dissemination. This arrangement can give a significant incentive to a vendor to
be successful and receive a return on his investment in the project.
16. A vendor can offer a unique product, such
as market leading software that is effective enough to outstrip the
competition. Careful market analysis is needed to be sure the acquisition of
the product is justified and does not present excessive risks in its’
application. In more specialized fields, such as Geographic Information
Systems, this is often a requirement because of the dominance of single or a
few vendors. This type of situation requires care not to invent something that
is much more cost-effectively available in the market, recognizing that the
commercial products may be unique and/or proprietary. Attempts at government
specification of formats and standards that these products must meet, has often
met with failure. Care must be taken not to create a sub-optimal solution to a
problem best left to the creativity and competition of the private sector.
17. Finally, one of the most vexing problems
in outsourcing is maintaining a healthy, productive, and ethical business
relationship with a commercial supplier. Too often, government contracts
provide significant penalties and “protections” for the government to impose,
but fail to recognize how to provide significant incentives for the vendor.
Good practice is to decide in advance what success should be for a vendor, and
then include that consideration in the design of the contract. Managing the
business relationship in an adversarial way almost always leads to project
failure at some point in larger IT projects. Conversely, most reputable vendors
are in business for the long run, and understand that the overall success of a
project is the most important objective. Finding or recruiting managers in the
statistical agency with the skill and experience to effectively manage an
outsourced function can be almost as difficult as finding specialized technical
skills to carry out the work internally.
18. Regardless of whether outsourcing is used
the likelihood of success of IT projects is greatly increased if it follows
sound management practices. In the next section we will examine the extent to
which such practices have been identified.
19. Because of the importance of IT and
because the results of too many IT projects have been so poor, a great deal of
effort has been applied lately in identifying good project management practices. These can be summarized as:
v
Involving top-level management,
v
Avoiding leading edge technologies, opting for small
projects, and
v
Involving end users in project formulation and
management.
20. The preparation of a detailed IT project
plan is now common practice for invoking the necessary discipline to ensure
that these practices are followed. The project plan is designed to analyse
return on investment, identify risks and mitigation strategies, ensure modular
development to avoid large-scale failures, and provide for oversight and review
at crucial steps along the way. No sizeable IT project should be undertaken
without this management process being utilized – in the United States for
example, the process is specified in detail in several laws governing IT
budgeting and procurement.
This direction describes three major phases of the investment and control
process for IT projects:
v
Selecting (screening, evaluating risks and return, and
assessing how the project will help accomplish agency mission),
v
Controlling (monitoring against costs, schedule and
performance), and
v
Evaluating (post implementation reviews, adjustments,
and lessons).
21. This management discipline has, for
example, been largely responsible for improving the success of larger IT
projects in the United States Government over the past five years.
While
this management discipline has been found by others to be useful the only
“certainty” that the chief statistician can rely on is the changing IT
environment characterized by:
v
Continual availability of new and
different IT capabilities and products,
v
Claims by marketing organizations of
amazing results in applying their products,
v
Lobbying by internal staff who have
become advocates of a product or specific technology in which they have become
skilled and knowledgeable, and
v
Continual pressure to reduce costs and
improve delivery of products.
22. Keeping an eye on how well commercial
products are faring in the marketplace is critical to avoid ending up in a
“blind alley” with an orphaned product. The larger vendors generally provide
more stability and support of products over the long term, while many of the
smaller vendors, who rely on a narrow product line, generally have more
volatile business prospects.
When you
have a solid technology management process in place in your organization, it is
still a challenge to choose appropriate technologies to leverage the mission.
Generally, it is best to:
v
Avoid new or “immature” technologies,
v
Assess results of others, who have
already applied the product,
v
Decide if the product is compatible
with the existing information/technology architecture of the organization, and
v
Assess the risks and price/performance
of the product.
23. New and emerging technologies are
notoriously difficult to assess in terms of when, or if, they will mature and
become mainstream products. In the 1980’s it appeared that imaging technology
for document storage and retrieval was just around the corner, and several
large IT companies bet their business future on their emergence. They all lost,
and that market still struggles below most of the market projections for the
technology. On the other hand, the worldwide computer access provided by the
Internet, and the associated explosion of the technology beyond any projections
made before 1995, is potent testimony to the power of a technology whose time
has come.
24. Learning from the experience of others
with similar business issues is also a powerful tool – visiting other countries
that have adapted technology for similar functions, and comparing their results
can be very helpful. It is often difficult, however, to account for differences
in application and operating environment from organization to another. It is
critical to assure that when applied successfully in one organization, the
technology can be similarly applied in another. Carefully inspecting the
adaptations necessary to operate in another environment can make the difference
between success and failure during implementation.
25. Finally, before choosing to adopt a new
technology, it is critical to assess the risks and determine, conservatively,
the projected price/performance. Remembering that the track record for such implementations
is notoriously poor, it is crucial that managers be convinced that the IT
project plan is complete and solidly demonstrates a successful implementation
strategy.
Conclusions
v Involving
top-level management,
v Employing
effective risk analysis in guiding direction,
v Avoiding
leading edge technologies, opting for small projects, and
Additionally, a deliberate
organizational strategy and management model for managing in-house versus
out-house technical skills can best guide the long-term direction for
accomplishing the Statistical Office mission.
Chapter
8
INTANGIBLES ASSETS
8.1.
Definition
1. In this section we look at an agency’s
non-staff assets -- its laws, policies, traditions, prestige, credibility in
the eyes of the public as well as its policies, registers, frameworks, and
nomenclatures.
2. A statistical agency must attach the
stamp of quality to what it does. For this to happen, its stamp must be
recognized. See also 2.2.3. The situation is similar to that faced by the
producer of a marketable product that wishes to differentiate itself from
competitors by instant recognition, assuming that the instant recognition is
associated with sound construction, attentive service, and durability. The
difference is that with few exceptions statistical agencies are monopolistic
producers of figures; the figures (at least the general aggregates) are freely
accessible to the public and the statistical production process cannot be
replicated by the majority of users.
3. A statistical agency’s stock in trade
is the recognition of its name and the unimpeachability of the attributes
associated with that name. Thus, the agency must be perceived to have the legal
basis to intrude into the privacy of individuals, enterprises, and
institutions; the integrity not to part under any circumstances with the
individual information it collects in the course of its activities; the
competence to use that information in order to derive the best possible
estimates of aggregate measures; and the sense of responsibility to ensure that
those measures are useful to the government and to the public at large. These
attributes stem from pride and respect for the agency’s image, and sufficient
legal competence to ensure that the agency works well within lawful boundaries.
4. In its administration, the statistical
agency should strive to make this posture universal among its staff. Larger
agencies may even wish to devote a department to liaising with media and
reviewing its reports, to ensure the maintenance of its public image.
5. A favourable image is necessary to the
functioning of a statistical agency, but insufficient by itself. Oftentimes, a
statistical agency demands, not merely once, but on an ongoing basis, that
respondents tolerate intrusions into their privacy and comply with its requests
for information. When the same is desired by private agencies (e.g., those
engaged in marketing), the cooperation of respondents is purchased or otherwise
remunerated, and in any case is typically inferior to what statistical agencies
judge to be acceptable. Government agencies tend to operate invoking the efficient
working of government and of the community as reasons to elicit information,
but this presupposes legal basis.
6. The law has to define clearly why
certain classes of information are required, what guarantees are offered by the
statistical agency regarding how individual records are protected, and what
penalties might be incurred by respondents who fail to provide truthful and
timely information (see chapter 12 for a broader discussion on what protection
is typically afforded to respondents).
7. A statistical agency
will need to ensure that it has both a sound legal basis and the right image
–with respondents and users alike.
8. Who in a statistical
agency answers a telephone call from the Prime Minister’s Office? Who is
responsible for meeting the press the day the CPI is released? Should an
academic doing research on the premises be allowed access to microdata,
provided it is anonymous?
9. Such questions, while
frequently posed, are not always answered consistently, or else consistency is
more a matter of habit than any actual policy. There have been instances,
however, where various practices and traditions are codified and brought
together in one policy manual. Usually such an exercise has two objectives: to
provide a standard of practices, thereby increasing consistency, and to
question the wisdom and propriety of such practices in the process.
10. Maintaining policies,
making staff aware of their existence, and ensuring compliance is a thankless
job, which usually earns the incumbent (if there is one) frosty relations with
the rest of the organization. However, if policies are not enforced, individual
agencies will revert to old habits, weakening any possibility of continuity or
consistency. It follows that this responsibility should be assigned to a
trusted official who ensures that at appropriate intervals the matter is
discussed with senior officers so that they are kept abreast of goings-on
within the agency and policies are revised if necessary.
Effects
of codified policies
The existence of a code of policies has several
beneficial effects. In situations where there is a high turnover of staff,
there will be ongoing pressure from outside organizations for special
concessions, such as waiving of fees for services, early access to new
statistical information, or access to material that under normal circumstances
would be regarded as confidential. If there is no relevant documentation to
bolster the refusal of such requests, and there exists a precedent of
inconsistent reactions, institutional resistance to improper requests is
weakened overall. Furthermore, the discomfort involved in appearing to be
negative is considerably lessened if the agency can show that it reviewed a
particular situation, decided on a course of action, and stood firm on the
matter by ensuring that all its staff are aware of the recommended guidelines.
The position of an agency is strengthened even further if it can show not only
that the stated policy is adhered to but also that it happens to be a matter on
which other agencies worldwide have come to similar conclusions. Lastly, public
image is considerably strengthened if a spokesperson for the agency can show
that there are written policies covering the particular matter being
challenged.
11. Because interests and concerns evolve in
an unpredictable fashion, a statistical agency should maintain a store of
available information on a wide variety of topics. It is incomparably better to
ensure that information is interrelated and, where it encompasses different
attributes of what is, hypothetically, the same population, the population in
question is defined and measured in the same way. For example, it is far better
to analyse net business formation with the knowledge that industry statistics
apply to the same population than to make tentative adjustments for a different
scope. Whereas in one case, energies are dissipated in trying to explain what
the data show or fail to show in spite of statistical discrepancies, in the
other case, analysis may proceed at once. However, it should be kept in mind
that the design and upkeep of the special tools required for this kind of
integration constitute yet another element in the statistical production
process, with its own set of organizational implications. In the remainder of
this section we consider a class of tools that serve to as integrating tools.
That is, they serve either to integrate information and/or to integrate the
functions of the agency itself.
12. Handbook
Rev. 1 has a special section on registers,
with particular emphasis on business registers, but there is little discussion
about the appropriate location for these in the organization of the statistical
agency.
13. The reasons are not difficult to fathom;
at the time, the introduction of business registers in the production process
was a novelty. It was made possible by two sets of circumstances: (a) some
statistical agencies were granted access to a computerized store of
administrative information (in this case, information derived from tax
records), and (b) largely as a result of the introduction of database
techniques, statistical agencies discovered that the computer could be used for
massive processing jobs other than international trade statistics or the census
of population.
14. Today, the notion that a statistical
agency must have a business register, a register of household addresses, a
catalogue of location codes, and possibly a number of specialized registers
(e.g., a farm register or a register of ocean-going vessels) is commonplace.
Therefore, there is little need to describe the contents of such registers or
why they are required. It is, however, necessary to discuss how they should be
managed and where they should be located.
15. There are various points of view. One is
that the business register should be located with the division of the agency in
charge of economic statistics, and the household address register should rest
with the part of the agency in charge of social statistics. The specialized
registers should be managed by those who make the most use of them (the farm
register with agricultural statistics, vessel register with maritime transport
and foreign trade statistics, and so on). Another view places registers in a
central organization within the statistical agency, where it can be shown that
their maintenance is an agency-wide service. Each method has its advantages and
disadvantages, and these should be known in advance.
16. Assume that the organization in charge of
business or economic statistics consists of half a dozen units, each in charge
of either an industry (manufacturing, mining, construction, business services,
etc.) or a type of expenditure (exports, capital formation, consumer outlays,
etc.). Whether the register is placed at the very centre of the organization as
a corporate asset, or within the economic statistics organization, will depend
a great deal on personalities and is therefore outside the scope of this book.
If they are to serve as integrating devices, it is important to keep in mind
that:
v registers must be placed in a service unit;
v no single user organization is entitled to modify the contents of the
register; and
v the custodian of the register is ultimately responsible for its
integrity, the classifications assignments, and the definition of the
statistical unit.
17. Where
registers are looked upon as agency-wide assets, it usually best if their
respective custodians are dependent on an internal board, rather than
exclusively on their hierarchical superiors. The composition of the board would
include supplier organizations (e.g., those responsible for the acquisition and
deployment of tax and social security information) and user organizations (all
whose surveys are based on the agency register files).
“The system of national accounts and balances
represents an [integrated set] of standard concepts and classifications
applicable to the most important macro-data of economic statistics and
characterized by the fact that all concepts are inter-related within a system
of definitional relations....They represent a useful frame for developing
statistics, facilitating the identification of gaps, and the setting of
priorities in quantitative perspective….It is a great advantage for these
integrating frameworks to be developed in the central statistical agency of a
centralized national statistical service or in the co-ordinating body of a
decentralized statistical service….The unit responsible for elaborating the
national accounts and balances and similar integrating frameworks should be
assigned a particularly important duty in regard to data improvement and
integration.”
Handbook Rev.
1
18. The central role of the
national accounts has become even more critical in today’s statistical agencies,
and its use in identifying weaknesses and gaps in basic statistics has grown
clearer with the publication of the 1993 United Nations version of the system.
There is no such widely accepted and powerful integrating tool for social and
demographic statistics.
19. The management of the
national accounts requires specialized skills and the closest of liaisons with
its key users. Moreover, the role of the economic accounts extends beyond the
production of data on the condition of the economy at regular intervals. Part
and parcel of the task of estimating the broad macroeconomic aggregates is the
related task of taking inventory and organizing the corresponding basic
economic statistics. In those cases where a specialized agency is exclusively
responsible for the economic accounts—the Research Department of the Central
Bank, for example—it will have a central role in guiding the statistical
system.
20. In some organizational
models the unit responsible for the compilation of the national accounts is separate
from its counterpart in charge of basic economic statistics. In this situation,
it can be far more difficult to signal gaps, point out weaknesses, and
generally exercise leadership when reporting to different supervisors.
“[These two
functions are] essential to integration. The subject matter divisions of a
statistical agency should as a rule be responsible for the content of
questionnaires and should work in co-operation with the functional units, especially
those responsible for methodology, survey design, and data processing…. Before
a questionnaire is put to use, a central authority should ensure that certain
requirements are satisfied as regards integration, efficient data processing,
editing requirements, and obligations towards the data suppliers. First and
foremost, it must be determined that the concepts and classifications applied
conform with the standards established so that integration is facilitated…”
Handbook Rev.
1
21. These words are as
relevant today as when they were first written. Moreover, to facilitate
integration, the statistical agency should ensure that questionnaires are
written in a language that respondents can easily understand, and which
accordingly minimizes response error. In larger agencies a specialized unit may
be established as an agency-wide service, devoted to cognitive research and to
efficient questionnaire layout. Moreover, a unit devoted to these matters is
ideally suited to conduct studies of equitable response burden.
“Unless concepts and classifications
applying to several subject-matter areas of statistics are standardized within
a coherent logical system, statistical data cannot be structured to meet the
needs of users for aggregated data or for compatible information from various
fields….Standard definitions are needed for various items, for example, urban
population, illiteracy, wages received, industry, agricultural holdings,
etc.…The general use of standard definitions is essential for compatibility of
different data referring to the same subject….Standards that are to enjoy
general acceptance should be formulated in consultation with specialists in the
respective fields. Technical committees are the most appropriate instruments
for this work. [However,] the responsibility for implementation should rest
with the chiefs of the subject matter divisions; …a separate central unit …must
monitor the implementation. [The] uniform implementation... and the maintenance
of coherence among them, require continuous attention and negotiation with the
suppliers and users of the data. This is particularly important in countries
undergoing rapid changes and economic development.”
Handbook Rev. 1
22. The organizational
elements specified in the earlier version of the Handbook are still required for effective coordination today. They
consist of a central unit charged with the duty of promulgating standards and
monitoring their implementation to ensure coherence within the system; a technical
committee to iron out difficulties and ambiguities in implementation; and a
negotiating mechanism to ensure that users and suppliers of data fully
comprehend the standard. In light of today’s experience, the requirement for
international comparability requires further discussion. Political unions,
customs unions, Free Trade areas, and the global character of foreign
investment all require that statistics (and particularly socio-economic
statistics) be expressed in the same language. The failure to use
internationally comparable standards should be regarded as a deficiency in the
organization of a statistical service. By the same token, the organizational
elements required for the effective coordination of classifications,
nomenclatures, and concepts must include a division in charge of ensuring that
what is adopted nationally is comparable with that adopted by other countries.
23. A unit in charge of
finance is necessary to the proper functioning of a statistical agency, for
obvious reasons. It should not be set up merely as a unit in charge of
bookkeeping, but rather as an analytical unit (because upper management
requires financial advice several times over the course of a year) and as a
legal unit (there must be an in-house legislator capable of determining the
permissibility of expenditures).
24. The administrative unit in
charge of finance has as its natural functions the keeping of a logbook into
which it enters cash and credit receipts and expenses. Registration and
annotation of transactions will be conducted in compliance with local
legislation, and in a method such that the results can satisfactorily pass
inspection by the government auditing services.
25. One obvious function of
the head of financial services is that of making recommendations as to what are
legally mandated expenses. In addition, the director of finance should be the
most significant source of advice with respect to proposed expenses that, while
they appear to be reasonable, may not meet all legal restrictions.
26. Statistical agencies
commonly run into situations where, through lack of financial foresight,
expenses must be prematurely curtailed, without regard for the effects of such
abrupt reductions. This is generally the case when, just prior to the end of a
fiscal review, the head of financial administration realizes that certain
obligations will not be met unless some sacrifice is made at once.
27. Sometimes, it also
happens that financial administrators find that at the end of a fiscal review,
the agency possesses a considerable portion of its resources. In such a case,
one might impulsively look for a way to spend them, so as not to arouse
suspicion. However, arbitrary spending can be more pernicious than hurried
measures to curtail expenditure, lead to poor managerial decisions and to waste
of the public treasury.
28. It follows that the chief
statistician will require advance warning of expected gaps between expenditures
and resources unless measures are taken well in advance and on the basis of
solid preparation. To fulfill this requirement, the financial administration
unit must have the information base and the analytical capacity to provide such
advance warning.
29. No matter how small the
agency, it is seldom the case that the information base required to provide
reliable estimates can be organized without help from all those with managerial
responsibilities. Accordingly, the custom of taking part in financial
forecasting and calculating rough but serviceable ideas on expenditures is one
that should permeate the agency. In a number of offices it has been observed
that that the quality of financial management goes up substantially if
responsibilities and accountability are devolved.
Conclusions
A statistical agency depends on a number of capital
assets, some of which are intangible: classifications, lists of names and
addresses, methodologies, policies determining proper procedure. Some assets
are designed to achieve efficiency of the agency’s operations, such as
computers, others are created to promote the integration of the agency’s
products, such as the national accounts and other databases. However, the most
important assets of a statistical agency are its people and the knowledge base
that they represent.
It is important that all the agency’s assets
are kept in the best possible condition. Therefore, successful human resource
management is probably the key factor for any statistical agency’s performance.
However, other tangible and intangible assets also need continuous conceptual,
technical and operational maintenance and when the time comes, replacement. To
achieve this, good management and effective organization are needed to take the
necessary precautions, financial and intellectual, in order to renew assets
when their condition demands it.
Even so, there are
well-established statistical operations that do require the almost daily
intervention of very specialized knowledge. Nothing could be better established
in the majority of Government statistical agencies than the Consumer Price
Index (CPI). And yet maintaining detailed specifications of goods and services
priced every month or every quarter in the face of technical and institutional
changes demands a knowledge usually possessed by a wide variety of specialists.
Thus, offices may require at least the frequent advice of civil engineers,
telecommunications, radio, and television specialists, banking and insurance
operators, etc.
There is little advice that
can be given to Chief statisticians in a generic way on what to do with the
people who are generally perceived to be a liability to the organization. Of
course, if they are very disruptive there may be no choice but to engage all
the means available by law, regulation, and tradition to free the organization
from their services. Such cases, however, tends to be the exception. More
often, people are on the margin and the Chief statistician receives conflicting
advice about their value to the organization.
Chapter 9
Chapter 9
MANAGING THE CORE FUNCTIONS
9.1. Essential Functions
1. An
important question to ask is whether it is possible to define the essence of a
national statistical office. Is there an irreducible set of functions that --
if not performed by a statistical office -- would invalidate its claim to the
title of a national statistical office? Which, if any, of the functions of a
national statistical office can be performed by the private sector without
compromising the credibility and integrity of government statistic? With little
fear of overstatement it can be said that these questions are fundamental to
any considerations of organization and management. For these reasons we shall,
in this section, consider these questions and provide some possible answers.
2. We
begin by listing those functions that are commonly attributed to a statistical
office. These include:
v
Planning
v
Executing and Processing
v
Analysis
v
Dissemination
v
Coordination and
Standardization
Let us consider for each of these functions the extent,
if any, to which they can be outsourced.
3. Planning
is the transformation of a policy question or other type of request for
information into a series of steps that will result in the desired information
being made available within the specified parameters of time, quality and
budget. It may be that the actual plan itself is developed outside the
government. This would include decisions on models to be employed, use of over
sampling, stratification and coverage. However, there is an inherent role of
the statistics agency here. It is the national statistics agency, which must
have credibility and an unbiased professional reputation so that they can
ensure that the plan is developed without favour to any politician party or
special interest group, and is only subject to choices based on the highest
professional standards. In fact, in many countries the details of, for example,
a survey plan is carried out by the private sector. The expertise and standing
of a national statistical agency is necessary to provide public confidence in
the data collection.
4. The execution of data collection
projects is increasingly being carried out by the private sector in developed
and developing countries. Of course, in many developing countries the execution
and processing of data collection activities – census, survey and/or
administrative records – was and is often carried out by international
agencies. As with the planning function there is nothing inherent in the execution
function that would require that it be carried out by the national statistical
agency. As with planning, however, there is a key role of the statistical
office, which cannot be given to those outside the government. The national
statistical office must ensure that the data collection is carried out in an
unbiased way, that confidentiality is assured and that decisions are made only
on professional grounds. This is not to denigrate those outside government but
rather to reinforce that a specific role of those within the government is to
uphold public trust.
5. At this point it is important to
consider for a moment the question of confidentiality and how this can be
maintained if the government statistics office is not carrying out the
execution and processing of the data collection. One of the most important
principles of official statistics is that the privacy must be respected and
confidentiality of identifiable data maintained. It is the function of the
statistical office to ensure this. However, this can be accomplished even if
the statistical office does not control the data. For example, when a contract
is written for an organization in the private sector to collect and process
data it can be made part of the contract that the confidentiality of the data
must be protected. These provisions may be very explicit and can incur severe
penalties if they are violated. There are cases, in the United States for
example, where the legal provisions protecting the confidentiality of data are
stronger than some of those for government statistical offices.
6. Analysis
is one of those functions that should be carried out by the national
statistical office and by those outside the government. It is important to the
intellectual vitality of a statistical office that members of its staff
critique the process and models used to produce the data as well as discuss the
strengths and weaknesses of the data. This is important for the growth of
individual staff members and to the office in its quest to improve the quality
of its data. However, it is also important that those outside the government do
their own analysis. This is vital to informed public policy debates and to
provide critiques that the statistical office can use to make improvements.
7. As with the other items discussed above
this is an area where, in many countries, the private sector is already engaged
in the dissemination of statistical data and information. In some cases the
products that are disseminated are basic statistical information and in others
the vendor provides a value added service. This may be in further analysis of
the data or in integrating the data with other information. What then is the
role of the statistical office? While a statistical office may not disseminate
all statistical information (it may not even be the primary source) it does
have an obligation to ensure that fundamental statistical information is
provided to all segments of the society on an equal basis. Depending on
national practices this may result in the statistical office providing the
information itself and in other cases assisting the private sector to provide
the information. This principle holds despite differences of opinion (discussed
in section 11.3) on whether statistical information is to be provided at the
marginal cost of dissemination or whether it is permissible to provide some
statistical products at market rates. While the private sector may be used as a
vehicle for the dissemination of statistical information the national statistical
office has an obligation to ensure – either directly or through others – that
statistical information is disseminated.
8. This last area is one in which there is
general agreement that this is uniquely a government function. This is true
regardless of whether coordination is among other functional ministries or
decentralized statistical agencies. It would simply not be possible for a
non-governmental agency or the private sector to accomplish this kind of
coordination. To do so would unalterably blur the lines between governmental
and non-governmental activity and change the very definition of government. For
similar reasons it would not be possible for anyone but the national
statistical office to be able to participate with others in the international
statistical system in the development of global standards. No one outside the
statistical agency could lay claim to an unbiased position or would have its
credibility.
10. While statistically not a ‘survey’, the
Census is obviously the flagship product of statistical offices when it comes
to their capabilities in ‘survey taking’: a Census is complex, costly, visible,
politically significant and/or sensitive and has to be completed within strict
time-frames. Therefore, good management of the Census is critically important.
External perspectives: In this example, a
Ministry of Tourism requests that the statistical office conduct a survey on
the number of arrivals and departures and on the related expenses during the
tour. The Ministry of Finance, responsible for the balance of payments
statistics and conscious of the fragility of the travel account in the balance
of payments and of its importance in explaining fluctuations in the current
balance,
supports the demand and offers to co-finance it. The project manager in this
hypothetical situation loses many nights’ sleep because the two sponsoring
agencies cannot agree on the possible interpretations of the priorities listed
as part of their contract with the statistical agency. The statistical agency,
anxious not to forego an important opportunity, agreed to formulations that
turned out to be too vague and did not specify a protocol for how sponsors
should behave during the execution of the project. As a result, and in spite of
the fact that the statistics on tourism improved, both the Ministry of Tourism
and the Ministry of Finance feel that their goals have not been met; neither
has much faith in the technical competence of the statistical agency although
both agree that it treated the results with moral integrity; and no party has
the energy or the resources to make another attempt, even though the survey
sample was limited in size and scope.
To avoid a conflict such as this one, the statistical agency should (a)
attempt to lay out all the “don’ts” it should have learned from the first
exercise (b) it should demand an agreed-upon list of objectives in line with
the budget; and (c) above all, it should insist that there is a commonly
agreed-upon definition of the goals of the project.
9.2.4. Working
with respondents
34. Part of a survey
capability is an established relationship with respondents which comprises a
way of: (i) finding them; (ii) explaining to them why they should be willing
and truthful respondents; (iii) persuading them to comply – without necessarily
threatening them with legal actions; (iv) dealing with complaints in cases
where surveys are too lengthy, unclear and continuous. Naturally, none of this
is attainable without a survey organization that includes a respondent
relations department, entrusted with the following duties:
v to
handle the public relations required for potential respondents to understand
why they have been selected, what is asked of them, and what is the public good
that is served as a result of their co-operation
v
to exercise special care and take all the required
precautions in cases where the announced survey is either unusually long (for
example, surveys of family expenditure) or unusually intrusive (for example
surveys of harmful drug consumption and surveys of fertility)
v
to keep a register of respondents contacted and survey
interviews completed so that recalcitrant respondents can be identified and
persuaded to participate
v
to share information with respondents, so that they
feel not only that they have made a contribution to the public good but that
there is some personal benefit as well
v
to have the resourcefulness, presence of mind, and
necessary information to find alternative respondents in cases where there is
strong resistance to the survey or where the original respondent can no longer
be located[4]
35. Accomplishing these tasks requires tact
and diplomacy, together with firmness and determination. There will always be
people in either the household or the business sector who will refuse to
comply, no matter how good a case for co-operation has been put forward. This
situation cannot be avoided and the officer in charge should not assume
personal responsibility for a small percentage of such cases. However, if the
rate of refusal appears to be increasing, procedures and methods should be
examined, for this may reveal a serious inadequacy in the methods adopted[5].
36. The capabilities that a
statistical agency must have if it wishes to be recognized as the repository of
statistical expertise and the rightful custodian of the official national statistical
infrastructure include:
v
the capability to design a comprehensive survey that
takes into account sample selection; survey design; estimator choice;
estimation method; and calculation of variances and sampling errors
v
the capability to design a questionnaire with
properties that minimize the respondent burden while recognizing the importance
of minimizing cognitive errors that could arise from lack of clarity in the
questions
v
the capability to relate collected information to
concerns that drive users to solicit information from the statistical agency
v
the capability to analyse information that has been
collected (and is about to be disseminated) so as to maximize its informative
role
37. These capabilities must be held by some
part or parts of the organization. Logically, the more technical
capabilities—sample survey design, estimation techniques, and advice on
drafting and questionnaire layout—should be the responsibility of one
organizational unit, in order to promote coherence in the agency’s survey
activities. Questionnaire content and substantive processing of the information
collected should be the responsibility of another part of the organization.
Designating responsibility for analysis of survey results is largely dependent
upon the availability of analysts and on the way in which the statistical
agency cooperates with outside specialists (if at all). Conventionally, all
issues relating to content—interaction with users, determination of the
schedule for data collection, financial arrangements to support the initiative
(particularly if sponsors are involved), and tabulation and accompanying
analysis—have been the domain of the subject matter staff. Finally, the field
organization is charged with the responsibilities of establishing contacts with
respondents and ensuring that they react favourably to the initiative and
provide the right kind of information.
38. It is commonly alleged (and there is truth
behind the allegation) that statistical agencies are slow in reacting to
requests for vitally important information. In fact, the lags involved in
responding with hard information, particularly information that includes a
measure of the rate of change, are aggravated by the typically vague
formulation of the requests. Such undertakings do take considerable time, but
while the lags can be accounted for and seem natural to fellow statistical
practitioners, it can be difficult to get users to understand why it takes so
long to conduct a survey and disseminate the results. To respond to such ad hoc
challenges, a valuable survey taking capability of statistical offices is a
unit that is able to mount a quick survey, either as a first instalment on a
more permanent effort, or as a one-off exercise.
39. One way of deploying such
a unit is to assign to it the responsibility of all feasibility tests in a
statistical agency, so that its staff becomes accustomed to launching quick
efforts designed to settle basic questions, preceding what might be a more
substantial survey. The mission of this unit might be to survey a given number
of respondents (households, businesses, or people in government or in public
institutions) in a set amount of time (e.g., a maximum of sixty or ninety
days). By developing such a capability and periodically demonstrating its power
and scope, an office could greatly increase its credibility and gradually
establish pre-eminence in the field of statistics
40. A statistical agency
should not automatically initiate a new survey in response to every demand for
information. Rather, it should systematically attempt to react to new demands
by exploring how they might be satisfied using regularly collected data or,
failing that, by examining whether the administrative records already in the
hands of government can address the new request, at least to some degree.
Whether or not, or rather to what extent administrative records can be used to
replace or to supplement statistical survey information, is a very complex
issue and the answer also depends very much on specific national situations.
Statisticians tend to be wary of the quality of administrative information, in
terms of concepts and coverage.
41. Nevertheless, the
attractive features of administrative records are that they are to be collected
or have been collected anyway. It is probably true in many countries that some
administrative records, e.g. tax records, have a very good coverage of parts of
the population, and that the rate of response is substantially better than that
achieved by a statistical agency. Moreover, there is always the possibility of
improving on the information yielded by those records by supplementing them
with data obtained from a much smaller sample of respondents.
42. If these advantages are recognized,
it follows that in some part of the statistical agency, preferably one that is
set up alongside the field organization, there be staff charged with the
following responsibilities:
v
keeping abreast of holdings of administratively
collected data by other parts of government
v
evaluating each new request to determine the extent to
which it can be met without resorting to a new or expanded survey
v
negotiating with the
custodians of the relevant information to determine how it can be shared within
the legal framework imposed on government information activities
Conclusions
In addition to capacity for the execution of a regular survey programme,
it is advisable to develop a rapid response capability in order to show
sensitivity to emerging ad hoc requirements, while continuing to use discretion
before committing to costly and cumbersome surveys. Access to administrative
records at all times is desirable, not only to promote rapid response, but also
to preserve the
good will of potential respondents.
Chapter 10
FIELD ORGANIZATION
1. In
this chapter we consider how an agency should be organized in order to deal
efficiently with respondents to statistical inquiries.
2. One method
is to invest in the creation and maintenance of a working field organization.
By definition, a field organization interacts with respondents to communicate
the agency’s need for information; elicit relevant information in the correct
form and at the appropriate time; maximize respondents’ cooperativeness and
goodwill; and adequately explain the consequences of refusal.
3. As
far back as the 1952 Interregional Seminar,
participants representing statistical offices of all sizes in both developed
and developing countries reached the conclusion that if anything was worth
managing according to function, rather than subject matter, it was the field
organization. All participants shared a similar opinion regarding the census of
population—the census itself cannot be managed without an effective national
field organization, and once created, that organization can be put to use at
low marginal cost in undertaking other surveys.
4. Two
factors can alter this straightforward conclusion. Some countries have a geographically
decentralized statistical system, not out of choice but out of constitutional
imperative. If the central statistical office’s role is limited to assembling
results compiled elsewhere, issuing coordinating guidelines, and perhaps
surveying its immediate surroundings, the issue of a national field
organization does not apply.
On the other hand, if the individual states, provinces, or cantons that make up
the decentralized system conduct multiple surveys, some of which are carried
out on a continuous basis, they will tend to have a field capability organized
by function basis.
5. The
reasons are straightforward. Economies of scale exist in recruiting, training
interviewers, managing a large workforce, ensuring that embarrassing
duplications are avoided, and preventing under-utilization of the workforce by
drawing up permanent, long-term, and short-term contracts. These arguments were
recognized fifty years ago and continue to hold today.
6. Suppose that the field organization is
established on a permanent basis, in charge of the collection operations for
all surveys (except those carried out through direct contact between subject
experts and respondents). One of its chief assets is its location in different
parts of the national territory. In a sense, the field organization acts as the
local eyes and ears of the central office.
Many field organizations are therefore organized so as to strategically
distribute resources throughout a nation, while maintaining a centrally located
cell in charge of coordination and liaison. The central cell’s responsibilities
are critical. It serves as the day-to-day link between the decentralized
network and the management of the statistical agency; ensures that the
different offices of the field organization work as a system rather than as a
loose confederation; and serves as the point of contact for various project
managers to convey their wishes, doubts, and complaints to local offices.
7. The field organization should be a key
participant in a standing committee that settles differences of opinion on how
to approach respondents. If the field organization believes that a recommended
approach to a particular survey threatens respondents’ goodwill, it should be
able to express its concern to the agency’s upper management. If the issue is
to develop an approach to the census of population or any other major census,
the field organization will take part in discussions at the highest level.
8. Of
course, the existence of a field organization is predicated on the need to divide
the statistical process into discrete segments, and in particular, to separate
collecting respondents’ information from other stages such as editing,
estimating, or tabulating. As the use of the Internet in general and in
statistical activities becomes more widespread, the role of a special field
organization will likely change in the face of technological progress.
However, current circumstances are such that these organizations are bound to
continue with their traditional functions, at least over the next decade or
two.
9. Large
field organizations tend to be divided into two or three groups. One will be in
charge of interviewing households, and its talents are best put to use in
household surveys conducted on a continuous basis (e.g., the labor force survey).
There will also be a group in charge of interviewing businesses, and its
membership will range from those with formal or informal bookkeeping or
accounting skills to those without a great deal of experience, who will be
primarily responsible for the delivery and pickup of paperwork. A third group
may be in charge of pricing consumer goods, and their knowledge will encompass
a wide variety of products such as foodstuffs, clothing, medicine,
entertainment, and so on.
10. The
groups themselves may have a two-tiered composition. They may include a small
group of permanent employees who provide leadership, organization, and
continuity, and a larger group of employees contracted on a renewable,
short-term basis. These short-term employees afford the field organization a
great deal of flexibility in the face of fluctuations in workflow.
11. A
field organization deals with several classes of respondents: households,
enterprises, governments, non-commercial institutions, and at times foreign
institutions. Each possesses unique attributes and may be sensitive, and even
resistant, to different types of requests. In this section, we examine some of
these issues and what methods a statistical agency can deploy to maximize
respondent participation.
12. Above
all, interactions between households and the statistical agency concern the
census of population; surveys of household incomes and expenses; labor force
surveys; and special surveys that supplement the latter two.
Successful field organization managers dealing with households should possess:
v
a comprehensive list of the
respondents in scope;
v
thorough knowledge of the
ethnic and socio-economic characteristics of the population, so as to maximize
cooperation and minimize the friction caused by invasions of privacy; and
v
a method of training
interviewers that reflects the organization’s accumulated knowledge of how to
successfully approach the respondent community.
13. Interviewers
should not be deployed until they have thoroughly understood the difference
between coercion and cooperation. In addition, a few other matters require
understanding. For example, households seldom keep usable records. The easiest
questions are those based on knowledge of their status (demographic,
participation in the workforce, and so on). The most difficult are those that
require exact recollection or access to detailed records (e.g., when and at
what cost was a particular household item purchased).
14. Households
may react poorly to certain questions on the basis of their cultural and
demographic characteristics. In some households, respondents may prefer to keep
the nature of their expenses hidden, and certain expenses are considered to be
morally wrong (alcohol, tobacco, and so on). Controversial topics such as
abortion may elicit diverse, possibly even violent, reactions.
Two
communities
In one country, the population was almost evenly
divided between the coast and highlands. While there was no strife between the
communities, they were still far from integrated and showed pronounced
differences in their attitudes toward finances and family matters. Certain
types of statistical inquiry were easiest if the interviewer belonged to the
other community, since this was considered less intrusive than sharing answers
with peers. For other types of questions, the opposite was true. Successful
administration of surveys requires detailed knowledge of such attitudes.
15. The
effectiveness of a field organization depends on:
v
the depth of its knowledge
of what works and what does not in household interviews;
v
an interviewer training
program that foster respondent/interviewer relations; and
v
the capacity to support an
interviewer faced with overwhelming resistance to parting with information.
16. These capabilities
are more easily found in a functional organization than in an organization
fragmented by subject matter.
17 Whatever
the legal basis for dealing with households (see chapter 12), the field
organization will want to deal with them gently. In some countries, there are
legal and political measures in place that protect respondents’ privacy.
Because statistics are generally not as high a priority for government as more
pressing issues, when friction arises from the nature of household surveys the
government is more likely to side with the respondents, without taking the time
to fully appreciate the loss of information and its consequences. The field
organization should consider this possibility when dealing with households and
make sure interviewers do not overstep their bounds.
18. Dealing
with enterprises is quite different from dealing with households. Enterprises
find it much easier to respond to questions on the basis of records than from
memory or opinion. One obvious reason is that once the decision to part with
information has been made, a junior officer can be authorized to provide
record-based data; the same officer would not be authorized to express opinions
on behalf of the firm or to provide facts from memory. In short, record-based
responses are the most predictable and time-efficient from the point of view of
business management. In the case of households, the head of the household
normally speaks on behalf of all household members.
19. Note
that the distinction made above applies mostly to properly organized and
constituted enterprises (generally medium and large-sized enterprises). Smaller
and/or informal enterprises are best treated as households, and the emphasis on
records is considerably less, since answers may not be based on a proper
bookkeeping system. Also note that the latitude given to interviewers in the
survey process is quite different in each case. Whereas interviewers can be
safely trusted (once they have received proper training) to follow the procedures
to replace one household or small firm in the informal sector with another, the
same is not true for larger firms. For this reason, one way of deploying the
field force is to assign to the more experienced and more tactful interviewers
the task of dealing with households, and to deal with situations where tact and
persuasion are of paramount importance. Less experienced interviewers could
take on regular contacts with properly constituted enterprises. Whereas initial
contact with an enterprise should be made at the highest level that the agency
can afford, the follow-up should be left to someone whose functions are
principally those of a messenger. In situations where the same matter can be
handled by mail, telephone, or e-mail, these are preferable, of course. Where
e-mail (or fax) is chosen, the field organization should ensure that proper
technical support is provided.
20. Whether
the workforce charged with taking prices every month should be an integral part
of the field organization or should be placed directly under the control of the
CPI management is a matter of preference, efficiency, and public relations.
Clearly, the CPI management will want to be certain that its
instructions—particularly on quality and replacement rules—are followed
scrupulously, and may fear that the divided attention of the field organization
may dilute the seriousness of those instructions. Moreover, it may also fear
that the public’s confidence in the CPI could be shaken if it were under the
impression that insufficiently trained personnel took information on prices.
Because the CPI conforms to neither household nor enterprise interviewing
protocol (the bulk of it is carried out by requesting access to public
information displayed by retailers, or, in certain cases, in advertisements of
catalogues), CPI interviewers require special training. These arguments favour
placing the CPI field force directly under the CPI management, rather than
within the field organization. On the other hand, in countries where size and
geographic layout necessitate local administration, an independent CPI
interview force would likely be inefficient. In such cases, it would be preferable
to make use of the local field organizations, under the supervision of
experienced CPI interviewers.
21. Information
is collected from government in a form different from that applied to
households and enterprises:
v
The government agencies may
have field organizations of their own; therefore, a combined effort is
required.
v
With respect to disclosing
information, the law that rules those agencies may oppose that of the statistical
agency, requiring that a legal conflict be resolved.
v
The collection of
information may require government agencies to adhere to a coding standard
developed by the statistical agency, even though the agency has no jurisdiction
to enforce it.
22. The
first situation is often the case with Ministries of Agriculture in countries
where agriculture is dominant and where there is a tradition of intensive use
of quantitative information in the ministry.
Rather than debate issues of control, the statistical agency should make use of
the existing results and the existing workforce, and engage the ministry in
more productive discussions, such as whether or not the collection should be
modified (as may be required for an integrating framework such as the National
Accounts).
23. The
Ministry of Transport is another institution where there is a tradition of
using quantitative information and where the work of the ministry generates
administrative records that can be used for statistical purposes at a sufficiently
low marginal cost. Again, the wisest policy is to reach an agreement with the
ministry so as to make the best possible use of the existing information.
Sometimes the administrative information does not conform to international
standards, but usually it is easier to make the necessary adjustments than to
embark on an independent data collection effort.
24. The
main barriers in making use of data collected by other ministries (especially
the government agencies in charge of tax collection and administration of
social security) are legal and regulatory. Tax collectors tend to be as
protective of individual records as statisticians. The very notion that sharing
occurs could deeply offend taxpayers and tarnish the ministry’s reputation, no
matter how powerful its detection and collection capacity. Furthermore, if
knowledge that the statistical agency receives information from other
government agencies creates the perception (no matter how ill-founded) that it
provides complementary information, the agency’s reputation may be tarnished.
In section 12.1.5, there is an attempt to show that in spite of these
difficulties, one-way sharing of tax information is the most powerful method of
substantially reducing statistical paperwork for enterprises.
25. Generally,
it is advisable that the statistical agency hold off on creating permanent
arrangements in which other ministries collect the data, and that they use
existing resources and political pressure to reach a constructive and
collaborative agreement. Perhaps the best way to proceed is to establish a
tradition of informal inter-ministerial committees. It is possible that the
agency will encounter much frustration along the way, as non-statisticians
rarely perceive collaborating in data collection to be an urgent task. On the
other hand, non-statisticians may be seduced by the prospect of taking credit
for an important service rendered to respondents, especially where the
respondent community is broad.
26. Lastly,
there is the matter of enforcing a common standard in the collection of
statistical data. This often arises in the context of specialized public
finance compilations, particularly for the national accounts. Assuring that
ministries use the same definition of current and capital expenditure at the
micro level is an arduous and at times thankless task, but one that cannot be
avoided.
27. In many instances, provincial or state
governments have highly developed statistical offices of their own. This is not
only the case with federal countries such as Brazil or Germany; even in
non-federal countries such as Spain, so-called autonomous communities have
offices with the capacity to collect, compile, and disseminate data of special
interest to the public.
28. In these situations, it is important to
prevent different field organizations from overlapping and exhausting
respondents’ goodwill. Where there are legal devices in place that allow
information to be shared between statistical agencies, and where working arrangements
are sufficiently intimate to support such devices, a balance can be achieved.
29. Dealing with other government
institutions does not automatically involve the field organization. In
principle, these matters are worked out between the heads of statistical
agencies and their legal advisors, where possible. If different levels of
government are involved, the political heads responsible for inter-governmental
affairs on both sides may be brought in. Of course, if the agreement results in
a delegation of tasks to field organizations then they will play a role in data
collection. In federal countries or in situations where local government is
involved, cooperation will tend to be country-wide. In such cases, the field
force operates nationally and has an agreement with the statistical agency as a
whole, rather than a coordinated set of understandings with each subject matter
unit.
30. These types of institutions are becoming
more important in almost all countries of the world. In recognition of this,
guidelines have been developed explaining how to account for the economic
impact of non-profit institutions. Most
non-profit organizations and voluntary bodies that engage in any monetary
transaction with the public have some rudimentary bookkeeping system, and the
survey demands from them tend to be relatively limited. However, non-profit
health and education institutions may require more attention from the field
organization.
31. The cases of non-profit health and
education constitute a separate issue largely because their relative size
measured as a share of GDP and their social importance in the development of a
country demand special consideration. In many countries, there is a tradition
of specialized surveys of health and education. But there is also a trend to
break away from the traditional surveys of these two services and in addition
to measuring material and human inputs,
attempting to measure outcomes. One framework that has gained some acceptance
shifts the emphasis on to longitudinal surveys (“films rather than photos…”).
However, the statistical complexity of such attempts places them beyond what
the administrative staff of non-profit hospitals, clinics, schools and
universities can do as a by-product of their regular activities. To the extent
a statistical system is still involved with the use of traditional indicators
of health and education there has to be an understanding between the providers
of the registers and the data collectors.
32. As a final comment on this matter, the
specialized nature of these two fields – Health and Education – is such that in
the majority of those countries that compile regular statistics on them, the
statistical or analytical units in charge of compilation are located in the responsible
ministry itself. For small and even medium sized statistical agencies that are
still sorting out their mandate it may be prudent to leave matters as they are
rather than attempting full centralization with few chances of succeeding.
33. For offices that are better established,
they may wish to experiment with the creation of dedicated Institutes or
statistical satellites. Since bodies come in a variety of forms. In addition to
having their own chief executive officer, they can answer to a board made up of
representatives of the education (or health) sector and of the statistical
agency.
34. Each major organizational unit within a statistical
office should possess a perspective broad enough to see how its actions may
intersect with the rest of the office or, in a decentralized system, with the
rest of the system. Such broad views are mostly required when agencies reach
important crossroads, which fortunately does not happen too often. In everyday
matters, the scope of interaction is usually limited to those with whom
information is exchanged. Thus, the management of a field organization will
bear in mind first and foremost its respondent policies (see chapter 12) and
communication with that part of the agency in charge of processing currently
collected data.
35. Invariably, a narrow perspective will
produce tension. For example, when requesting detailed information on consumer
expenditure, those responsible for economic statistics may consider minimizing
errors in the CPI weights as their overriding objective, because of the harm
such errors are bound to cause. They will therefore wish to add all possible
precautions to the relevant inquiry. This stance may be countered by those in
charge of field operations, whose experience tells them that the shorter the
interview, the more attentive the respondent.
36. The tension itself is a healthy
manifestation of a commitment to quality motivating both organizations. If
neither organization expressed its point of view on how to avoid pitfalls,
tensions might escalate even further. But, given the possibility of conflict,
there should be an organizational device in place that forces agreement without
having to call upon senior officers. In the case of new initiatives, the
mechanism should lie with the project management, and ultimately the authority
of the project manager.
Where the point of conflict is the result of an ongoing activity, its resolution
requires the involvement of a standing committee. The reason for a standing,
rather than an ad hoc, arrangement is that problems concerning the interaction
between a field organization and subject matter units are bound to arise with
considerable frequency. The important thing is that differences of opinion be
channelled into cooperative dialogue that yields constructive and binding
resolutions for all parties.
37. The management of the field organization
cannot treat censuses and surveys as if they were interchangeable. Censuses,
particularly the censuses of population and housing require a major civilian
mobilization and therefore should be considered as a national event, whereas
surveys should not, because of their scale, impact, and potential repetition
(though such a thing should be avoided).
38. The census of population is vastly
important to any field organization for several reasons. In many agencies,
field organizations were in fact created in response to the needs of the
census, so accordingly, the census is their prime commitment. Second, the most
visible part of any census is the collection of data, since it requires a brief
period of intense activity on the part of all households.
39. In contributing to the preparation of a
census, the field organization will usually call on its detailed knowledge of
the country and assist in translating it into a data collection approach. It
will also contribute its knowledge of local conditions to identify and mobilize
local leaders.
40. Whatever
the arrangements, a field organization, like any other function-based
organization within a statistical office must look after its connections within
the rest of the office with great care. There are several points of tension
arising from interfaces that must be watched:
v
the challenge from the
pressure of subject matter specialists who are furthest from respondents –
analysts, national accountants, specialists in marketing and dissemination
providing the outstanding examples;
v
the ambiguities about the
quality with which a survey should be conducted. The field organization must be
mindful to come to a sensible compromise between its standards of quality and
the survey budget; and
v
the proper balance between
those survey operations that can be effectively conducted in regional offices
(in a decentralized mode) and those that must be conducted in a centralized
environment or else the standards of consistency cannot be applied with
guarantees of success (surveys involving complicated coding for example).
In addition the field organization should be watchful
of technological developments that may short circuit it before its management
has had a chance to re-train the staff.
Contacting enterprises (and in a few years this may
extend to households) via the Internet is undoubtedly going to get the subject
matter expert in much closer contact to the respondent. Field organizations
should find a niche in the new process where they can play a necessary and
cost-effective role.
There are a number of infrastructural services that
the chief statistician should make sure are available to the staff employed
outside the central office. They include:
v
access to statistical
frames;
v
access to equal training
opportunities; and
v
access to experts in the
central office in real time to make rulings on the application of standard
concepts.
Conclusions
There is general agreement that the field organization is a top
candidate for organization by function.
The effectiveness of a field organization depends on:
v
the depth of its knowledge
of what works and what does not in household interviews;
v
an interviewer training
program that foster respondent/interviewer relations; and
v
the capacity to support an
interviewer faced with overwhelming resistance to parting with information.
If the field office is dispersed throughout the country, then it is
critical that there be a central function responsible for liaison and
coordination.
Chapter 11
GETTING INFORMATION TO THE USERS
11.1. General dissemination issues
Definitions
The words dissemination and publication
play crucial roles and possess broad definitions in the text that follows. They
are taken to mean “making available to the public,” without restrictions and
without regard for the way in which the action is carried out. Thus,
publication involves the action of making public, in printed form or through
the Internet. Publication likewise
extends beyond the notion of hard-copy publication to include CD-ROM’s, tapes,
audiocassettes, radio and TV broadcasts, and any other media that can meet the
same objectives. At certain points in the text, publication is used in its traditional sense; this will be made
clear by the context.
“ It is probably
not an exaggeration to say that in most countries the statistics available, in
published and other forms, are by no means fully taken advantage of by users,
partly because the statistics are not enough known…the fostering of a more
extensive exploitation of the statistics by active promotion based on user
studies is an integral part of the dissemination effort…”
Handbook Rev.
1
1. In part because of the
condition identified in this quotation, there is a class of intermediaries in
the dissemination process whose task is to seek out the users and tailor the
information to suit their needs, while simultaneously providing useful
interpretation of the data.
10. A statistical agency that
lacks a well-defined dissemination policy risks losing its credibility as an
independent agent. The following elements are essential to an effective
dissemination policy:
v
a release schedule that treats all constituencies
equally, defined well in advance
v
a clear policy with respect to what information should
be made available to the press and what is supporting detail that can be disseminated
through statistical bulletins
v
a policy regarding the cost of accessing detailed
statistical information
11. Statistical agencies have become
increasingly aware of the benefits of a pre-defined schedule of statistical
releases. It is better to work according to a timetable with conservative
deadlines than to work too fast and be perceived as unpredictable. Or, if news
is released later than usual and is more favourable than expected, the
statistical agency could be perceived as under political influence.
12. For many users, the one
line that encapsulates the entire measurement process (e.g., “last month the
CPI continued to increase at the rate of ten per cent per annum”) is more
important than the factors that explain the news. Since it is usually the media
that provides such summaries, it must be made aware of the significance of the
data and of the relevant metadata. From the point of view of others, the
statistical intermediaries among them, the explanatory factors and supporting
data play an important role in their efforts to explain the underlying forces
of change and to foresee future changes. The statistical agency should have a
clear policy regarding what data can be distributed to the press for mass
consumption and the body of data that, because it is more detailed and results
from a finer cross classification, appears some time later in a statistical
bulletin (in electronic or conventional form).
13. The
growth of the Internet has blurred the edges of what used to be a fairly
straightforward policy. In theory, nothing should stop a statistical agency
from placing all available information on its Web site and managing access
through the Internet rather than in more traditional ways. By “drilling”, users
can get to the detail they find appropriate to their concerns, with the added
advantage that the technology allows them to leave a trail of what they have
accessed.
14. The unit in charge of
dissemination in a statistical office should be aware of the fact that
technology is rapidly loosening the constraints that used to affect its
activities. Tracking technological advances has become almost as important as
conducting studies of user needs and satisfaction.
15. Information
can be disseminated using standard forms—bulletins, digests, abstracts,
yearbooks, etc. There is a pattern to which most established statistical
offices have adhered throughout most of the last half century. It consisted in
producing a yearbook (see Section 5: The Statistical Yearbook), a monthly or
quarterly digest, and the occasional specialized publication if its readership
could be readily identified. Both the yearbook and the digest covered virtually
every activity in which the statistical agency was engaged. Where it was
obvious that the demand for detail was far more important than the demand for
selected aggregates—typically the case with foreign trade statistics—the latter
would be included in a specialized volume.
16. The organization required
to support this form of data dissemination was equally straightforward. Each
substantive unit looked after the compilation of a special branch of statistics
(e.g., price statistics), which would be featured in the “digest” and the
“yearbook,” respectively. Each unit was responsible for the accuracy of its
data and for the relevance, consistency, and form of the accompanying
footnotes. It would also be available to answer specialized queries that could
not be handled by the editors of the publication.
17. At the agency level there
would be a unit in charge of dissemination (identifying readers, defining the
form of the publication, providing timetables to contributors, ensuring that
the printing process worked according to schedule, handling distribution and subscriptions,
etc.). This unit was also responsible for interacting with the subject-matter
organizations to ensure that they complied with the overall schedule of
dissemination.
18. A number of large offices
had their own printing presses; in several countries, the facilities available
to the statistical office were among the largest and most sophisticated in the
public sector. This form of organization is still viable, although modern forms
of dissemination, and in particular the use of the Internet, have opened up new
possibilities.
19. Perhaps the most important
change brought about by technological advances is the significantly diminished
need to control large printing facilities. Statistical agencies now have the
means to lay out their own publications and, for small printing runs, the tools
to print everything in-house. In addition, technology has done away with the
need to print large volumes of statistical tables (typically those related to
the censuses of population and to foreign trade statistics) by making it
possible to provide the information in other ways.
20. Twenty
years ago, the alternative to paper publications was to disseminate information
using computer tapes but this was only open to the very small set of users who
had access to a mainframe computer. Over the last two decades, access became
far more widespread with the proliferation of PC’s and floppy disks. And in the
last five to ten years, the practice of disseminating massive bodies of data
via CD-ROM has been almost universally adopted. CD-ROM’s allow information to
be conveyed in a more imaginative way—mixed with sound, accompanied by
processing software, etc.—and actually invite a greater effort to analyse raw
data than any means of dissemination previously available.
21. The advent of the
Internet has opened up an increasingly large number of possibilities for both
the providers and the users of statistics. Its main advantages are that it
offers interactivity, versatility, speed and cost efficiencies therefore
enabling the statistical agency to greatly enhance the service provided to
users.
22. The organizational
implications of intensive use of the Internet for dissemination purposes are
substantial. For example, the traditional two-tiered approach to publishing
referred to above can be modified. Each substantive unit would be equipped to
set up its own Web page on the agency’s site and handle the interactive aspect
of dissemination. A central unit would be in charge of releasing information to
the press and also have the power and the expertise to define (subject to
higher approval) a “code of good behaviour” with regard to the contents and
appearance of material posted on the agency’s Web site. The central unit would
also be in charge of convening meetings to determine how users as a community
were reacting to the range of outputs produced by the agency.
23. Another important
organizational consequence is that the electronic file/database used to prepare
the print dissemination could be used for the Internet version of the same
publication. A close integration of content and presentation between electronic
and conventional publications can also bring about substantial economies in the
dissemination process. This does not mean that an office is tied to print
media, as it may be suitable for an office to only offer electronic versions of
its publications whether they are Internet or CD-ROM versions.
24. The delivery of
time-sensitive statistics to users has been greatly enhanced by the Internet.
The time lag between data collection and data dissemination by traditional
print media office is greatly reduced when these data are disseminated via
Internet. In practice, Internet dissemination is usually possible around the
same time as the print version has been finalised and sent off for reproduction
and distribution. This often will led to the user having access to the
statistics disseminated via Internet a number weeks before they would normally
receive the print version.
25. Many statistics agencies
are finding that Internet is ideal for releasing market-sensitive statistics.
The Internet allows the agency to release the statistics in an orderly fashion;
that is one that allows equal access to the statistics to all users at the same
time. In the past, facsimile transmissions have been used for this purpose but
users have been critical of this method.
26. The set-up cost of
establishing an Internet service in a statistics agency has reduced
dramatically in the past few years. This is due to both the reduction in cost
of the hardware and software as well as the availability of “off-the-shelf”
software that can be used for a basic but very satisfactorily performing
Internet site.
27. The cost efficiencies produced
by disseminating statistics and statistical metadata via Internet are
substantial when the Internet dissemination is replacing established print or
other physical products such as tapes or CD-ROMS.
28. The Internet enables
statistics agencies to provide greater access to statistics and statistical
metadata. In particular, the amount of detail that can be provided is not
limited as in print media by the size of the page. Extensive statistical
metadata can now also be provided as a matter of course to the user. Statistics
and statistical metadata that has been archived by the statistics agency can
now be made readily available to users. In addition, out-of-print statistics
publications can also be made available to users on Internet by providing scanned
or other formats that preserve the document structure and layout.
29. Web
browser technology is very sophisticated, which allows agencies to provide
users with browser ready access to databases and Internet based statistics
dissemination systems. Web browsers now come with the functionality that
enables the browser to recognise the file type and allow the user to download
statistics in comma separated variable format (CSV), spreadsheet or other
common formats directly to their computer.
30. The Internet has provided
statistics agencies with a very useful way to test various ways of presenting
statistics or tracking the popularity of its releases, to invite comments and
to get feedback on the adequacy of its data and metadata. In the development
stage of a new database for Internet, users can test the database interface for
the statistics agency and provide real-time response to the designers on such
items as layout of the interface, functionality, response time of the Internet
connections to the database and general usability. This is very valuable to the
statistics agency in pre-testing of the “product” before release. Once a
service has been released the statistics agency has the opportunity to keep in
close contact with the users by providing a “feed-back” email address. By
providing this function, the statistics agency can get very useful suggestions
on ways to better serve their clients.
31. The agency can also use
the Internet to collect statistics and related metadata via hyper text mark-up
language (html) forms, direct access to remote databases, file transfer
protocol (FTP) to remote servers or email submission. These methods are now
commonly used and the choice between these depends on the complexity of the
task, the security required, the frequency of submission among other
requirements. Most recently, statistics agencies have been investigating the
use of the Extensible Mark-up Language (XML) to transfer statistics and
metadata between statistics agencies and data suppliers and users. XML potentially
offers substantial gains in efficiencies to both the statistics provider and
user as it allows standard transfers of structured data and metadata. The
important point here is not to promote the use of XML – in fact by the time you
read this book XML might have been replaced -- but rather that agencies will
have new tools at their disposal to more easily collect and disseminate data
and metadata.
Consistent with their governments’ “user pay” policies, some offices
charge market rates for at least some of their products and services, whether the users are
in government or in the private sector.
“The aims of
this policy are to encourage users to identify and address their real needs for
statistics, to enable the demand for ABS products to be used as an indicator of
how ABS resources should be used, and to offset the cost of production of the
statistics. The ABS has ‘public interest’ obligations to ensure that at least
basic statistics are both readily available and affordable [as a result the
ABS’s Library Extension Program, 529 Libraries across Australia] are provided
with free ABS publications…to meet the needs of their communities…”
Australian
Bureau of Statistics
32. The issue is whether one
should treat statistical publications in much the same way as a general
government service (e.g., police force and national defence) or as a government
service for which users pay upon each utilization (e.g., toll roads, parks, and
museums). And, in the case of the latter, what is the appropriate price? The
arguments vary according to a government’s communications and social policy.
The positions in the debate over user pay policies include:
v If
a government needs information to pursue its own ends—make better decisions,
show the electorate whether or not it has delivered on its pre-electoral
promises—it will request the production of statistical information. The cost
will be borne by all taxpayers and is part of the burden of providing good
government.
v The
availability of information about the social and economic progress of a society
is something to which any citizen in an open society is entitled and a
responsibility that good government must honour. Since it is not possible to
legislate on how much information should be provided free and how much should
be charged for, the only practical policy is not to charge for any.
v Since
information has to be produced anyway to meet government needs, the most that
should be charged to the user is the marginal cost of delivering information in
a requested form and by a given date.
v It
is manifestly unfair to place the burden of financing the production of highly
specialized information, which gratifies the objectives of the few (e.g.,
researchers, historians, professional statisticians), on the shoulders of the
entire community of taxpayers.
v It
would be nearly impossible to base a fee structure on marginal cost.
Accordingly, the average costs of delivery should be charged to all users,
regardless of what sector they come from. There is no other equitable way to
manage a cost recovery policy.
33. These arguments have been bandied for
some time, but no international consensus has emerged. Perhaps what can be said
at this stage is that for those agencies in developing countries where the
major challenge is to get citizens to recognize the existence and usefulness of
statistical information, the most pressing issue is not whether and how to
obtain maximum revenue from statistical publications. In more advanced
statistical agencies, where certain users have unique (and costly) specialized
requests, there is an increasing trend in favour of charging more than nominal
costs. The implementation of a fee policy differs greatly from one agency to
another.
34. In agencies that have
embarked upon a dissemination programme with commercial overtones — either
heavily dominated by the sale of a particular service or relying on the
complete set of publications — the creation of a special unit involved with
marketing and related activities is warranted. However, in other agencies
research has shown that the commercial potential of statistical publications is
limited and, therefore, the creation of a special unit is not called for.
Of course, even if there is limited or no commercial potential and where there
is no special marketing unit, agencies still have the responsibility to
investigate how they can increase the audience for their products.
35. A commercial policy
requires several elements and, above all, a record of costs that could justify
the prices attached to each publication if they were subjected to scrutiny on
the presumption of monopolistic unfairness. The elements of such a policy are
as follows:
v
an algorithm to price special tabulations and a policy regarding
subsequent requests for the same information by another user
v
an algorithm to price custom-designed surveys or
additional questionnaires drafted to accompany existing survey questionnaires
v
an explicit and justifiable objective (e.g., recovering
all costs associated with the production of camera-ready copy, printing, and
distribution, in the case of hard copy products, and a comparable set of costs
for distribution via the Internet)
v
a marketing plan including customer identification and
needs
v
for larger offices a separate marketing unit may have
to be identified
v
a catalogue of all publications with their respective
prices.
36. The
unit in charge of marketing publications must fully comprehend these elements
and objectives and be ready to deal with criticism from the sectors of the
community that, though they may be intensive users, are the most affected by
any set of commercially calculated prices.
37. In a number of countries,
third parties (in some instances private sector firms) have been contracted to
take over the marketing of part or all of the statistical agency’s
dissemination programme. This measure upholds the objective of gauging signals
from the market for the agency’s publications, eliminates the need to staff a
marketing unit, and releases the statistical office from the obligation to
explain its pricing policies in the context of broad social goals. Naturally,
such a program only makes sense in an environment where there is an active market.
38. Before adopting solutions
of this kind, though, senior officers in statistical agencies should be mindful
that the best of statistical publications do not have mass appeal and that many
users will be quite content to share copies or to access them via libraries or
other public collections. Experience suggests that among the few publications
with less limited appeal are those on foreign trade statistics and those
derived from the census of population.
39. An important objective for
an effective statistical agency is to disseminate the information it compiles
as widely as possible and to make all possible efforts to get government,
business researchers, and the community at large to use that information. It
therefore seems irrelevant or even contradictory to raise the matter of
copyright and royalties in connection with official statistics. In fact, what
is envisaged is to ensure by moral persuasion that intermediary users cite the
source of the statistical information and include the caveats to which it is
subject. Moreover, for those offices that market their information
commercially, copyright is a measure designed to prevent users of statistical
information from deriving personal profit from an undertaking that was collectively
funded by the taxpayers.
40. A number of statistical
agencies make use of the copyright mark and instruct all users to properly
attribute statistical information. It
is not likely that any agency has taken legal measures to address improper use
or attribution. Moral suasion is usually effective enough.
41. In the cases where it is
obvious that intermediaries also frequently buy statistical compilations in
order to support studies and analytical exercises, attribution may be insufficient and the payment of royalties
may be in order.
42. There is no international consensus
on how far a statistical agency should go in commenting on what the current
figures mean and even less on how a set of figures recently compiled may relate
to another set of figures compiled independently. To a certain extent, there is
convergence of opinion in the social and demographic field, and several
agencies prepare monographs on demography and various social issues. But there
is considerable diffidence toward doing something similar with economic
statistics. Arguably, nothing moves very fast demographically and, therefore,
the commentators need not worry that what he/she says will change population
trends. However, in the case of economic data, some ill-chosen turn of phrase
could affect behaviour in the financial markets, raise serious questions about
an economic policy adopted by the government, and altogether bring about what a
statistical office sets out never to do—to affect the real world with its
apparent opinions rather than with objectively estimated figures.
43. There is not much
controversy around the fact that, at a minimum, a statistical agency should
comment on salient facts and use its inside knowledge to influence the
impression created by aggregate figures if that impression is not supported by
fact. For example, hours worked in an industry fell precipitously in one of the
summer months when it was not habitual for them to fall. It turns out that the
dominant firm in the industry decided to close down its plants earlier than
usual for some technical reason and to send its workers on paid holiday. In a
case like this, it is proper for the statistical agency to remove the alarm
from the aggregates by commenting—without divulging who was responsible—on what
caused the precipitous fall.
44. It
is improper for a statistical agency to make normative judgments. For example,
in commenting on recent measures of income distribution, a responsible
statistical agency should not attach editorial comments about the latest
changes in the tax structure and their regressive effect on the shape of the
distribution curve. Nor would it be proper to couch the normative judgments in
controversial assumptions and shaky evidence of causality. But readers would
probably benefit greatly if the statistical agency, in reviewing the condition
of the housing market, for example, were to remind readers that the number of
starts had fallen and that this fall coincided with a dramatic rise in interest
rates, including mortgage rates. These examples are not prescriptive; rather they
illustrate the posture of a statistical agency that is consistent with its
vocation of objectivity and neutrality.
45. If the chief statistician
decides to routinely add analytical comment to the release of figures—social or
economic—some unit within the agency must be put in charge. If the efforts of
the agency are at all effective, the media will get into the habit of
reproducing the official comments and keep them separate from evaluations of
the figures produced by other makers of public opinion. In order to ensure that
it is in a position to explain what the agency’s comments mean, the media is
likely to seek a contact within the agency. If the chief statistician decides
not to take on the additional burden of being the media’s regular interlocutor
(although in smaller offices this might be advisable) he/she should designate
very clearly who is the official spokesperson. This will prevent any confusion
that might result from different opinions being given by various professionals
within the same agency.
46. a survey or related bodies
of data—responsibility might be assigned not to one person but to an entire
analytical unit.
47. In
spite of precautions and training administered to the official spokespersons,
errors in judgment may occur. In order to avoid embarrassment, the statistical
agency should take additional precautions. For releases that require swift
publication to be useful, a process of collective review that engages the more
senior officials of the organization and fosters cross-subject review and
criticism should be put in place. For example, a release of statistics on
employment and unemployment can be effectively reviewed by those responsible
for industry and trade statistics, the national accountants, and so on. These
are the types of reviews that, although performed on a regular basis should not
incur excessive amounts of time for any one of the reviews.
48. A
second process would be reserved for longer-lived efforts—for example, those
connected with the series of analyses generated by a new survey of family
incomes and expenditure, a new economic census, or a new census of population.
Given the time scales used for such exercises, it is appropriate to institute a
formal process of review. Persuading members of the academic profession to take
part in the process would be of added benefit. The mission of the review
process would be exclusively to judge whether the statements made are fully
supported by evidence; whether the most important inferences on the basis of
the new data available were taken into account; and whether the methods used
stand up to scrutiny in the face of current knowledge. There is, however, a
component of this review that should not be overlooked. This consists in making
certain that nothing improper is said in the analysis, given the statistical
agency’s political and legal context. This latter qualification is important.
For example, some statistical offices are responsible both for the compilation
of statistics and for economic studies (or analytical exercises). Other offices
do not have this extended responsibility. It follows that what would constitute
administrative “trespassing” of responsibilities in one country would be the
expected duty of the organization in another. The judgment of what can be said
is one that must rest squarely with the chief statistician or someone directly
delegated by him/her.
“In view of the approaching [creation of a great many new
provinces], and the prospect of their extending their commercial relations with
each other and with foreign parts, a hand-book of common information respecting
them seems to be required.”
Year-Book and
Almanac of British North America for 1867
49. The physical entity called
a Yearbook may turn out to be a twentieth-century relic, made obsolete by the
possibilities opened up by modern computer and telecommunications technology.
But the two different approaches conveyed by quotations above are as much of an
issue today as when they were first stated a century ago.
50. For many decades, the
statistical picture of a single country was available from its Yearbook and the
corresponding picture of the concert of nations was available from the set of
United Nations Yearbooks. If
one wished to learn all that is important about a country in a quantitative
sense, one could consult these yearbooks, the value of which lay in the breadth
of information and variety of points of view presented.
51. The opposing point of view
is that the activity of an office should consist not in speculation but rather
in providing relevant answers to pointed questions.
52. Leaving aside the matter
of whether the statistical agency publishes a volume of statistics called the
“Yearbook” or creates a well-designed Web site in which all relevant
information describing the country can be found, the merits of producing a
Yearbook include the following:
v It
provides an occasion for a review of the relevant information that describes a
country.
v It
clarifies the need for integration in the statistics that will be selected for
the Yearbook.
v It
reveals gaps in the available information, thereby suggesting what new
initiatives should be taken to complete the description of the country’s social
and economic fabric.
v
In the case of decentralized systems, it is yet another
means of promoting coordination among statistical units in different government
departments.
v It
is an ideal pedagogic device to introduce children to the physical, political,
and human geography of their country.
53. Against these considerations, there are
the fears that the Yearbook will become an end in itself, of little value to
the constituency for whom it is intended, and that its contents are bound to be
excessive for those in a hurry and insufficient for those who wish to deepen
their knowledge of a particular subject.
54. To a great extent, the development of the
Internet has changed the basis for many of these fears. The economics of
producing a Yearbook (the daunting aspect of the initiative for many years)
have changed radically. The greatest innovation is that in many instances it is
no longer necessary to print the Yearbook or it can be printed on demand. Other
considerations are almost as important. The pages of the Yearbook can be
updated much more quickly. And the Yearbook can easily be produced in a number
of modes depending on how much coverage and detail are required, as are some of
the better-known dictionaries.
55. Whether the statistical agency opts for a
Yearbook or for an abridged volume, made available in all of the country’s
embassies abroad and designed to inform the potential foreign investor or
tourist, or whether the matter is handled through the Internet, the initiative
to produce a Yearbook requires an organization. The structure of the
organization is not all that different from what was envisaged for the
marketing of statistical publications. It consists of a coordinating body
entrusted with the responsibility of defining contents, setting deadlines,
handling distribution (or delegating that responsibility to a dedicated
marketing unit), planning the future evolution of the Yearbook, pointing out
contradictions or ambiguities in the data, and, if required, ensuring liaison
with other departments of government, where other producers of statistics can
be found.
56. Perhaps
the most difficult challenge for the coordinating unit is to ensure that the
contributing departments do not dismiss the exercise as someone else’s
responsibility for which they are not accountable. A system of links operated
through the Yearbook’s Web site might place users in more direct contact with
contributors and stimulate their interest.
A statistical agency must strike a
balance between an increasingly general policy on the part of government that
calls for users to pay for certain classes of services and its duty to make
official statistics as accessible to the community as the budget permits. There
is a possibility of sub-contracting some or many of the functions associated
with marketing but this still leaves open the matter of sales versus free
distribution—there is no international consensus on where the boundary should
be drawn.
The publication of a
Yearbook—in conventional or electronic form—is the best method for selecting
the most important statistics required to describe a country and its economy,
society, and environment. A Yearbook is also helpful in examining existing gaps
in available information.
Chapter 12
RESPECTING PRIVACY AND PRESERVING CONFIDENTIALITY:
HONOURING THE CONTRACT
12.1. Respondent policies
1. Perhaps the most important issue in
developing a respondent policy is that a statistical agency earns the public
confidence by treating respondents with respect -- not just as the means to
reach its statistical goals. It is important to remember that, even in the
presence of laws which make response to one or more data collections mandatory,
participation by the public in a statistical agency’s surveys is a largely
voluntary process. Even when the survey process is not voluntary an agency
still has an obligation to treat respondents in an ethical manner – that is,
minimizing the burden on their time, respecting their privacy, and maintaining
the confidentiality they were promised when they provided the information.
2. Respect for privacy is an
acknowledgement that it is the individuals who “own” information about
themselves. The concept of privacy is that it is the individual who decides
what information is made available, when it is to be released, and to whom it
is released. Laws requiring the disclosure of this private information for
statistical purposes are enacted only when there is an overriding public need
for the information and with provisions to protect from disclosure the
identifiability of the data. (This concept also to applies legal entities such
as corporations. Public corporations, of course, agree to the regular
disclosure of information such as sales, profits, etc. in order to participate
in a regulated stock market. Private or closed corporations are required to
make far fewer disclosures.)
3. Ensuring confidentiality is the
response of the statistical agency when it obtains private data. It is a pledge
to honour the contract between the respondent and the agency when the
respondent provides private data. In this Chapter we consider the elements of
this contract and the means to protect the data when it is obtained from
respondents.
4. It is critical to develop policies
designed to create a cooperative frame of mind on the part of the would-be
respondents. Listed below are elements of such policies, followed by an
interpretation of each of these elements
5. With respect to the two classes of
respondents – business and individuals/households – there are some principles,
which apply to both and some, which are unique to the business community. We
shall first consider those, which are applicable to both cases:
The purpose of the data collection must be clear and meaningful to the
respondent
The application of this element is particularly
difficult in the case of small business. The overall framework within which
data is collected prevents the statistical agency from taking frequent
censuses. Indeed, in the interest of efficiency, it is best to take samples no
greater than what is strictly necessary to yield aggregate totals quickly and
accurately. But small business is typically interested in the locale in which
it operates and in its narrowly defined type of activity, neither of which can
be adequately represented by most statistical undertakings, with the exception
of a census or a comprehensive large-scale survey. Even so, any explanation of
the purpose should be as clear and convincing as possible.
The statistical agency must be
perceived to hold in the strictest confidence all individual records,
protecting them from any other party in or outside government.
This
assurance has to be delivered in a way that makes respondents feel confident.
The assurance itself consists of two elements: respondents must be made aware
that the information held by the statistical agency cannot be accessed by
someone with malicious intent, and
that the law recognizes that confidential information held by the statistical
agency cannot be shared with the political authorities of the country, with
regulatory agencies, or with the civil service engaged in policy development.
In a number of countries, specific reference is made to the fact that
information submitted to the statistical agency cannot be subpoenaed by the
judiciary.
The statistical agency must be
seen as willing to accommodate respondents, either by providing additional
explanations or by accepting legitimate substitutes for a traditional
questionnaire.
The statistical agency not only should be aware of how businesses keep
records, but it also must be willing to go out of its way to recover the
information it needs with minimal effort on the part of the business. For
example, a shareholders report that includes all or most of the variables
sought by the data collector should be an acceptable substitute for a completed
statistical questionnaire.
The professionalism and objectivity of the statistical agency, as well
as its freedom from political interference, must be established, accepted, and
continually advertised.
This is a general recommendation. In
approaching respondents, a statistical agency ought to request no more
information than is necessary, and only after it has determined the least
burdensome way of acquiring the information. The request is backed by the
agency’s reputation, specifically in the sense that the methods it employs to
achieve results protect its autonomy and freedom from political interference.
The statistical agency should be
perceived as thoughtful and concerned in matters relating to response burden;
that is to say, it should be committed to finding means that will simplify the
paperwork burden.
Generally, it should be understood that the
statistical agency will approach a respondent with a request for data only as a
last resort, all alternative avenues having been reviewed and found
insufficient. An agency that desires to be known for a reasonable policy on
matters of response burden will see to it that data collectors have understood
thoroughly the meaning of this last element in the policy. In particular, the
agency must demonstrate:
v that
at all times the statistical agency closely monitors its information gathering
activities;
v that
the statistical agency has taken into account the intrusion and the violation
of privacy necessarily implied by its efforts and that the result is a careful
balance between respect for privacy and the minimum information required to
enlighten public discussion
v that
the information collected will not be misused; and
v
that it can measure the extent of the burden it
imposes, and that, through the application of its policies, this burden will be
reduced.
6. In addition there are two principles,
which apply only to the business community:
The way in which the information is collected must reflect the ways in
which businesses keep records.
Successful
collection involves the capacity to transform the records held by businesses
into the standard records required by the data compiler. But it is the
statistical agency that must make the effort, not the businesses. The extent to
which the information can be transformed without distortion can be gauged if
the data collector understands how businesses keep records. In general, it is
advisable that a statistical agency have a sufficient number of employees with
a thorough understanding of business accounting and bookkeeping. Ultimately,
effort should be made to convince businesses that it is safe to share
confidential information with the statistical agency, not to dictate how information
such as purchases, sales, and profits should be recorded.
The information collected should employ the same terminology used in
daily business operations.
It is important to distinguish between the language
used by statisticians to communicate among themselves and the language required
to communicate with would-be respondents. Such expressions as “kind of activity
unit” and “net profits” have little meaning outside the context of a
statistical agency. The language to be used in order to elicit information must
be the language that is familiar to the respondent—household or business.
7. A statistical agency, however, is not
the only information-collecting agency. Indeed, government collects detailed
information from the public in a variety of ways and for many reasons. For
example, detailed balance sheet information is collected in order to levy taxes
and administer subsidies; detailed commodity information is collected in order
to administer laws and regulations that deal with health and safety; and so on.
The statistical agency should be among the most vigorous pursuers of paperwork
reduction, using the information collected by others whenever there is an
opportunity created by law and by overlap in content. It should be able to
demonstrate to the public that it communicates with other government agencies
regarding the means available under the law to prevent duplicate requirements,
particularly from those business that are least equipped to fulfil them.
8. The nature of the law differs from
country to country. In some, compliance with statistical collection is simply
obligatory. If respondents— businesses or heads of household— do not provide
information in the form in which it is requested and in a timely fashion, they
are in violation of the law. In other countries, requests for certain classes
of information are supported by legal requirements, whereas others are made on
a voluntary basis. And finally, there are cases where the law is ambiguous on
the subject. When this is true, the statistical agency may be fearful of
demanding too much information, because if it were to be challenged, the law
might rule that no information is to be demanded compulsorily (other than that
sought by the Census), and the resulting publicity might adversely affect
response rates.
9. The situation in most countries is
probably representative of the second case, in which the law recognizes a
restricted set of compulsory surveys. Whatever the legal basis, all agencies
find that the most important objective is to secure a cooperative attitude on
the part of the respondents — particularly small business and households.
10. Where there is an option, the statistical
agency may adopt a moderate stance, wherein economic inquiries are compulsory
and all others are voluntary. Irrespective of the legal posture, it is
important to remember that without a cooperative attitude, no amount of
compulsion will alleviate the response problem.
11. A problem, which has been increasing in
many countries, is that of low response rates in data collections. Among the
potential solutions that have been tried is that of providing incentives to
respondents. However, incentives cannot always solve the problem. In some
countries the use of financial incentives is illegal. In other cases, the
budgets are too tight to allow for the provision of significant incentives.
Moreover, if incentives are offered to householders, for example, there must be
some relationship between the time required of them and the nature of the
compensation. For some inquiries the matter is trivial. For example, in the
case of attitudinal surveys or continuous surveys, the cost to the householder
is the periodic intrusion into the household’s private affairs. In other cases,
particularly surveys of consumer income and expenditure, the effort required
from the household is considerable.
Finally the payment of incentives may set up expectations on the part of
respondents making it difficult to conduct surveys without incentives or to
require ever increasing incentives. The following are points to consider:
v The
kind of incentive chosen must be significant but not excessive, so that it is
neither treated with scorn by respondents nor looked upon as waste by those who
determine the statistical agency’s budget. The field organization must play an
important role in providing advice on this matter.
v Whatever
option is chosen, it is best to keep the householders who are surveyed informed
about the reasons why incentives were or were not provided, about the image of
the statistical agency, and of course, about the civic duty involved in the
provision of information to aid in the discussion of matters of public concern.
12. While incentives are most commonly
offered to householders, they can also be useful in surveys of small
agricultural operators and even of small, informal businesses. The same
considerations do not necessarily apply to larger businesses, for which the
provision of relevant information may be a forceful incentive in itself. For
example, if the business information compiled in typical cases is comprehensive
enough to be used in a regional breakdown, businesses operating on a regional
rather than a local level will be interested in how their regional results
compare to those of their competitors working on the same scale. Accordingly,
the incentive of being provided with special tables making those comparisons—so
long as the data have the right amount of detail and are reasonably current—may
be sufficient to elicit constructive cooperation between the business community
and the statistical agency.
13. In all cases, however, the statistical
agency must give uncompromising and explicit assurances that the information
supplied will not disclose the identity of respondents. For businesses of any
size, the combination of strong assurances and the recognition that the
information solicited is necessary for the country to manage itself in an
orderly fashion is sufficient. In other words, providing incentives in the case
of business surveys is generally unnecessary (and, for budgetary reasons, is
not in the statistical agency’s best interest).
14. There are many precedents of sharing
detailed administrative information with statistical agencies. The most
noteworthy is the case of customs administrations. But income tax collectors
are also collectors of massive amounts of information and are just as mindful
of integrity and confidentiality. If sharing is to take place, tax collectors
must be assured that statistical agencies are at least as scrupulous in the
treatment of individual information. Sharing information, particularly with tax
collectors, is a delicate matter. The following are some of the factors that
the statistical agency ought to bear in mind:
v whatever
information is shared between statisticians and tax collectors must move in
only one direction—from the tax authorities to the statisticians
v in
order to get the tax collectors to share information willingly and cooperatively,
the statisticians should render some service to the tax authorities, one that
does not sacrifice any of the vital safeguards on confidentiality
v information
collected by the tax authorities is the single most powerful resource for
reducing response burden for small businesses
v tax
authorities are interested in classifying the forms submitted by tax filers
according to their branch of economic activity but not necessarily in the same
way as statistical agencies
v there
is a substantial strategic gain if the industry classification of tax records
is aligned with that used for statistical purposes
v the
public perception of statisticians classifying tax records might offset any
assurances given by statisticians about the confidentiality of the information
submitted to them
15. The challenge for the statistical agency
is to find the best possible compromise among these factors without
jeopardizing in any way statistical confidentiality. Whatever is found to work
vis-à-vis the tax authorities is bound to work, with the occasional variant,
for the other suppliers of administratively generated information.
16. In addition to the various guarantees
that suppliers of administrative information require and that can be agreed
upon by protocol and by verbal assurances, the statistical agency should give
signs that the physical holdings of sensitive information are especially well
guarded. In many statistical agencies, where access to the premises is
controlled, the floors on which tax documents are held require special
authorization for outsiders to gain access.
17. These days statistical agencies have two
perimeters that must be protected. The first is physical—the actual location of
the documents, computer records, microfiches, photographs, and so on. In this
respect, those offices whose records are housed in more than one building not
only face a higher cost in protecting the records but also in persuading the
public that what they do is consistent with the security of individual records.
The second perimeter is virtual—the electronic perimeter traced by the agency’s
internal communications system, which we must presume is connected to its
stores of individual data.
18. Both perimeters need to be protected from
malicious or unauthorized intrusion. As we have already mentioned, in many
offices, access is strictly controlled (e.g., employees have to wear ID tags),
and additional security measures are taken where the most sensitive records are
housed.
19. The electronic perimeter must be equally
defended with password clearances and other types of security, but this is more
difficult to accomplish, particularly as technology continues to demand
upgraded defences. The popular media has given great publicity to successful
attempts by hackers to penetrate the most secure communications systems and
gain access to the most sensitive information, in some cases involving national
security files. Since the public is more apprehensive about computer security
than that of conventional documents, and this trend is likely to increase, a
few statistical agencies have chosen to completely insulate the internal system
of communications from the outside world, including from its own local offices.
20. There is no direct link between the
stores of data and communications to the outside world. Furthermore, whatever
confidential data are transmitted from local offices to the central office are
protected by the best encryption system available. Whether or not this solution
will provide sufficient protection in the face of rapidly evolving technologies
cannot be foretold. One of the agency’s systems analysts should be placed
permanently in charge of data security, with responsibility for such matters as
encryption; removal of identifying attributes; and protection against malicious
tampering with data files. In addition, someone should be assigned
responsibility for the physical security of the premises and data holdings.
21. Some official regulation must outline the
sanctions imposed for security breaches—as a deterrent and for demonstration to
the public that the matters of confidentiality and integrity of data holdings
are taken seriously. The most natural way of dealing with this is through the
law itself. An ideal situation is one in which legislation clearly lays out the
various types of infraction. These, in increasing order of severity, include:
v carelessness: for example, an interviewer leaves an envelope containing
identifiable completed questionnaires on the bus, where it could be discovered
by a third party and picked up by the press as an example of public sector
laxity
v improper behaviour: for example, a
subject matter expert brags in public about the income declared by a particular
family when interviewed in the course of a household income and expenditure
survey (the possible consequences are the same as above)
v behaviour with malicious intent: for
example, an employee wishing to
embarrass the management of the statistical agency sends the Prime Minister’s
census form to the press
v use of confidential information for
personal profit: for example, an employee offers to provide confidential information of one company to a
competitor.
22. These,
of course, are merely examples; the law can be more comprehensive about what
constitutes a security crime. But such examples illustrate the kinds of
infractions that should be listed and against which there ought to be credible
penalties.
28. While one person or household may be much
like another, large corporations like Sony, Royal Dutch Shell, IBM, De Beers,
Gasprom, or General Motors are quite unlike small businesses. For this reason,
a number of differences between social and economic statistics factor into this
discussion. In social statistics, the notion of dominance is relevant only in
some surveys of household assets, income, and expenditure. Also, while in the
domain of economic statistics, any survey sample will typically include a
certain class of units because they make up a large proportion of the variable,
families and individuals need only selection with non-zero probability. Lastly,
with the exception of special surveys (e.g., the financial balance sheet of
households), random perturbations can be introduced in the results of household
surveys to mask individual cases without affecting the estimates of any of the
moments of distribution. However, this is rarely the case with business
surveys.
29. Recently, longitudinal surveys have
raised a number of interesting issues connected with confidentiality research.
These surveys, in which a selected unit (generally an individual or family) is
tracked over time so that changes in its attributes can be observed, are
becoming popular among some of the more advanced statistical agencies. The
temporal aspect of this type of survey greatly increases the likelihood that
the sample will be identifiable. However, minimum necessary suppression may
render useless the delicate data analysis and presentation that such surveys
demand. Clearly, research into confidentiality of household statistics is
vastly different from research on the business side, and should be treated as a
separate item on the research agenda.
30. In terms of public image, it is in the
best interest of the statistical agency to position itself as the guardian of
response burden, continually demonstrating its desire to minimize paperwork
where possible. Accordingly, the statistical agency should aggressively pursue
opportunities to use the regulatory powers of other parties to gather
information, rather than duplicate queries administered by government. However,
eliminating duplication may imply the sharing of information, which can work in
two ways; either the statistical agency obtains its information from what is
collected by the government body, or vice-versa. When given a choice,
respondents usually prefer to report to the statistical agency for a variety of
reasons, the most important being its reputation for discretion and attention
to detail. If it is possible that information will be shared, respondents must
give consent (preferably in writing) to have their forms reviewed by a third
party, and be made to understand why such sharing might occur. Respondents
should also be made aware that while their refusal is absolute, the result
might be that they have to provide the same information twice.
31. There are other circumstances in which
written consent for disclosure may be sought. In highly industrialized
countries, industry analytical tables (the Input- Output tables for example)
can lose their value if disclosure rules are interpreted to the letter. Before
accepting the sacrifice in analytical value of its compilations, the
statistical agency may wish to seek consent from the businesses that run the
risk of having their operations disclosed. These businesses would be asked to
agree to a form of publication that affords imperfect disclosure protection. Of
course, taking this step on any significant scale may risk the agency’s
reputation.
32. In certain instances, there is no
feasible choice but to disclose the activities of a particular business
enterprise. Mostly, this occurs with public utilities, state monopolies, and
industries that are dominated by a single firm, as is often the situation in
smaller countries. Where the situation is well-known and in fact precedes the
creation of the statistical agency, a special clause dealing with it is written
into the law. In annual statistics this is less of an issue, because state
enterprises and large public monopolies are usually requested to report their
activities and financial circumstances in great detail, and these reports are
made public.
33. It is not always possible to prevent all
breaches of confidentiality. The most obvious case is international trade
statistics. In this case, the data collected is pertinent to recorded
transactions rather than the individuals or businesses responsible for them.
Because these statistics constitute a census of transactions over the period of
reference, are published in extremely fine detail, and include several
attributes (what is exchanged, its destination, means of transport, points of
entry and departure, and so on) it is not realistic to assume that total confidentiality
can be maintained without having serious effects on the timeliness and utility
of the information provided. The customary approach is to opt for a mixture of
active and passive disclosure protection. For all cases where it is known that
there is only one importer or exporter, or a
case of dominance, suppression measures are devised and implemented in advance.
For all other cases, suppression measures are only taken if a business or
individual takes the initiative and complains to the statistical agency. If the
agency is accused of failing to act on the matter in a timely fashion, the
respondent should be made aware that, unlike situations in which the
statistical agency designs and administers its own surveys with the means to
detect disclosure a priori,
international trade statistics employ records compiled through an external
administrative process, and are not the direct responsibility of the agency.
Therefore, it can only act when notified.
34. Under certain conditions, and by their
own decree, statistical agencies may order the publication of the names,
addresses, industrial activity (expressed in the form of an industrial code),
and size (expressed by a code denoting an employment size class or some other
agreed variable also by size class) of selected business respondents.
Generally, such an exception is made for research purposes.
Increasingly,
there are perfectly legitimate applications that not only employ statistical
aggregates, but also depend on individual data for their success. For example,
the study of complicated interactions between consumer income, expenditure,
household savings, and taxation through the examination of published statistics
tends to suffer greatly from aggregation bias. The testing of formal hypotheses
can be more readily conducted on the basis of individual longitudinal records.
Few offices are equipped to conduct this kind of research themselves. It
follows that in this situation, the common interest is best served if the
statistical agency provides researchers with the information they need and
allows them to formulate hypotheses and to conduct the necessary research
effort.
35. Nowadays, circumstances exist in which a
statistical agency cannot avoid providing microdata. But an accessible database
with individual information should be constructed in such a way that the
probabilities of identification with certainty are extremely low or
non-existent. Of course, all records must be made anonymous, and in addition,
there may be grounds for introducing random perturbations so long as the
moments of the original distribution are preserved.
Even after all these precautions, it is required that an officer oversee access
to the database and work derived from it. This means that at all times an
agency official knows who is using the database and why, what they are looking
for, and what results, if any, have been achieved so far. The agency must
coordinate with users whose work is dependent upon particularly detailed sets
of data, so that the general public is ensured that such detail is provided
only in support of meaningful advances in the social sciences.
36. It is difficult to envisage a publicly
accessible database for business information, because in each industry there is
generally an establishment or company big enough to dominate results and be
recognized by the extent to which it affects the data. However, with small
businesses the situation may be different and there are questions relating
exclusively to small business, the assembly of a publicly accessible database
is justified. Note that while numerous examples of successful public-use
databases exist on the household side, few have been constructed for business.
37. The
research community places special demands on data, which cannot be ignored,
particularly where those demands are deemed to serve the public interest. Many
of these legitimately require access to individual records or to cells in
tabulations where the number of respondents is less than the threshold. Since
such research is not carried out for personal gain or with the intent to secure
commercial intelligence that might hurt established interests, it is difficult
for the statistical agency to disallow it. Conversely, it is just as difficult
for the agency to make exceptions for certain classes of users despite all the
assurances given to respondents. Rarely does a statistical agency achieve an
effective balance between the two pressures. The following section describes a
solution that has worked in the context of a few agencies.
38. One method that has met with limited
success in several countries is the formal recruitment of academic researchers
who require access to microdata, even though their remuneration is merely
symbolic. This procedure ensures that the researcher agrees to abide by the
rules of the statistical agency, and understands that failure to do so will
result in official sanction. The following conditions would warrant this type
of formal offer:
v The
researcher has proper credentials.
v The
research plan concerns matters that the statistical agency considers to be in
the public interest or for the advancement of social science, and there is no
alternative source of data.
v The
researcher agrees to uphold the protocol and posture of the statistical agency
for the duration of his/her employment.
39. Prior to making an offer of employment
the statistical agency should ensure that the three conditions stipulated above
are met, and may wish to consult its legal adviser. It is also important not to
use this practice to circumvent the rules of confidentiality. Clearly, overuse,
or use for projects that would appear trivial could compromise the agency’s
credibility. The issue of who controls the intellectual property resulting from
the researcher’s work is beyond the scope of this Handbook.
40. Another method, tested at the US Bureau
of the Census, consists in creating a “sterile chamber” where researchers can
work. While they do not acquire the status of employees, researchers act in
full knowledge of the statistical agency’s confidentiality policies and accept
the statutory sanctions should these be broken.
Statistical
agencies are committed to guard information that obviously reveals the
operations, belongings, attitudes, or any other characteristics of individual
respondents. The principles for so doing are fairly clear, but specific
applications require ongoing research in order to certify that there is no
breach of confidentiality. Theoretical research, while invaluable in that it
can be shared and used repeatedly, is costly and is conducted in only a few
agencies, but even where it is not possible, the more practical work of
suppressing revealing information without compromising the integrity of
statistical aggregates should go on at all times. These activities should be
carried out openly as a means of reassuring the public.
There do exist exceptions to the rule of absolute confidentiality; where
these occur, the statistical agency should be candid with the public and
explicit about the conditions that warrant exceptional treatment. Household and
business records are not treated in the same way, since the former are easier
to disguise, whereas circumstances often arise in which the latter cannot be
disguised without deleterious effects on the quality of published data. In such
cases (e.g., in the case of natural monopolies or dominant enterprises), either
a specific provision for disclosure is included in the statistical legislation,
or else the statistical agency should not act without the explicit consent of
the potentially identifiable party.
Chapter 13
PUTTING IT ALL TOGETHER
13.1 Introduction
1. All functions required for a
statistical agency to perform effectively have been examined in the preceding
chapters. Clearly, one of the most important points is that the activity of a
statistical agency is constantly under conflicting pressures. Prudence and
experience suggest that there is no ideal solution, but an effective model is
one that seeks to minimize the impact of negative elements rather than one that
assumes they can be eliminated. Another major point is that most of the
features discussed so far apply to all statistical agencies, be they
centralized, in research departments of Central Banks, or embedded in
Ministries that rely heavily on quantitative evidence. This concluding chapter
brings all elements together into a schematic framework, further illuminating
some of the arguments reviewed earlier on. It is also intended to strengthen
the understanding that all parts of a statistical agency are interrelated, and
that the agency cannot work effectively without a sense of interdependence
among all members of its staff.
2. The
notion of organization conjures up ideas of hierarchy. In order to be effective
over time, organizations must clearly and unambiguously assign
responsibilities. Indeed, the very creation of an organization chart—with the
limits that it imposes on each set of functions and responsibilities—suggests
both interdependence and exclusivity. These interplay of these two principles
favours a certain amount of stability and guards against excessive autonomy.
And yet in the preceding chapters there has been a great deal of discussion of
why a simple hierarchy is insufficient, why it must be complemented by a
non-hierarchical organization (for example an internal committee), and why the
statistical agency must be willing to sacrifice strict adherence to the
hierarchical structure to adapt to external changes.
3. The
fact of the matter is that statistical agencies must adjust to a changing
environment and must do so in a timely fashion. To accomplish this, agencies
must closely monitor technical, environmental, and legal changes and should
engage in an ongoing discussion about how their structure should adapt to those
changes over which they have no control. Just as change is constant, so should
be the adaptation of priorities.
4. Descriptions of organization
necessarily look at each component individually and examine its attributes and
functions, as well as the factors that suggest that it be organized in one way
or another. However, the more detailed those descriptions, the less one
understands their part in an interrelated system.
5. In an effective statistical system,
where all activities are successfully integrated, there are bound to be groups
championing different sets of priorities. For example, one group may uphold the
view that the key activity is to ensure that users’ needs are correctly
interpreted and provided for. Another group may insist that no effective
performance is possible unless the appropriate respondents are identified and
sufficient care is taken to make sure they
understand the questions and give the best possible replies. And a third
group may argue that however successful the activities championed by the
others, they are futile unless estimates are accurate and correct rules of
aggregation are observed. In a truly successful agency the tension generated by
these three conflicting views gives
rise to a system of checks and balances that ensures the production of the best
possible data.
6. This is a static view of an effective
organization. In practice, a statistical agency should be simultaneously
adaptable and stable. Technical change seldom proceeds in a continuous and predictable
fashion. For this reason, effective statistical organizations must scan the
technical horizon and keep abreast of technological advances. For, if
forthcoming technical changes are likely to affect the way statistics are
collected, compiled, and used, there will barely be time to overhaul the system
before it is overtaken by those changes.
7. The functions of a statistical agency
can be described under four comprehensive headings:
v
those
functions required to ascertain user requirements and ensure that they are met
(this implies sustained communication with those who are, or who could be,
interested in statistical information)
v
those
required to ensure that statistical information meets quality standards (this
implies the ability to gauge the minimum standard of reliability necessary to
guarantee both the usefulness and credibility of statistical data)
v
those
required to determine, in conjunction with respondents, the least intrusive and
most convenient ways of gathering basic information (directly, as well as from
other agencies when possible)
v
those
that allow the statistical organization to review and learn from its own
practices
8. These should be supplemented with
functions that ensure adaptability and uniformity of the products of the
statistical process. Statisticians can use these criteria in evaluating the
soundness of their organizations.
9. The most efficient internal
organization is not one that strictly categorizes its staff by their various
backgrounds and approaches to statistical activities. While some grouping of
staff by function and professional background is necessary, total separation is
inappropriate. When certain activities are not respected or careers get blocked
on the basis of previous experience, the tensions that arise can severely
hinder a statistical agency’s performance.
10. Review and evaluation are crucial
activities that should be undertaken by a dedicated part of the organization.
The purpose of these two activities is to learn from past mistakes; to help
purge the organization of attributes that have become obsolete or undesirable;
and to impress upon users that the agency has not succumbed to inertia.
11. In order to increase the credibility of a
systematic evaluation of a statistical agency’s activities while introducing
new perspectives, it is best if outside resources are co-opted for both
reviewing and evaluating. This can be done through consulting contracts or,
where applicable, as part of the regular activities of the network of advisory
committees.
12. There is no prescription for how many
hierarchical levels there should be in any one organization. This is determined
by those who design the organization and depends on the scale of operations,
what is considered manageable, the specialization of the staff, and so on.
13. There is no suggestion of a hierarchical
set-up in the diagrams at the end of this chapter. Unless otherwise indicated,
arrows represent the flow of information, not levels of authority. (For an
examination of how internal committees fit into this hierarchy, see chapter 5)
14. There is no suggestion that all the
facilities (substantive and service departments) mentioned must fall within a
single agency. This is preferable from a number of points of view, but in many
situations, the cost of bringing it about is not commensurate with the
resulting benefits to the user community.
15. There is no mention of initiatives that
mobilize resources for a relatively short period of time and are dissolved once
their objectives are attained. It follows that there is no mention of how one
should organize the census of population beyond what is discussed in chapter 9.
16. The office of the chief statistician
should carry out two kinds of activity:
v
leadership
and management of the statistical agency by giving direction to its financial
administrator, director of personnel, and those responsible for its substantive
programmes
v
management
of external relations, the most important of which are those concerning the
ministry responsible for the agency, the key government counterparts, the
country’s legislative authorities, the media, and the chief statistician’s
peers abroad
17. Whether the chief statistician conducts
these activities directly or through advisers is dependent upon personal
preference and scale of operations. There are large offices where a number of
authorities are delegated to the chief statistician’s principal advisers, and he/she
is only called on when the matter at hand concerns the integrity of the office
or of its database, related policy issues, or the ministry in charge of the
office.
18. Sometimes functions normally associated
with the head of a statistical agency are put in the hands of a principal
adviser. The adviser then takes on all matters related to personnel management
(other than appointments at the highest levels) and matters associated with the
day-to-day financial management of the agency, thereby allowing the chief
statistician to tend to other concerns, such as the office’s external
relations. The chief statistician, however, remains responsible.
19. The function of a legal adviser tends to
be linked to the chief statistician, irrespective of how the office is
organized. Of course, whether the legal adviser is on staff or is consulted on
a regular basis depends on preference and availability. All cases involving
matters of integrity (e.g., exceptions to confidentiality provisions, or
granting access to data before they are published in exceptional circumstances)
where the law is ambiguous require the judgment of the chief statistician,
supported by the most reliable legal advice. Figure 3 in Appendix 3 represents
a possible structure for the chief statistician’s office.
20. Advances in computer technology and
communications have affected the way statistical agencies disseminate data. The
statistical agency should constantly gauge market preferences regarding
content, the medium chosen for delivery, and what explanatory detail should be
included with the data. To do so, it must maintain relations with principal
user constituencies. The needs of users must then be communicated to the
producer departments so that they may assemble data in the form best suited to
user needs. Marketing concerns must also be taken into account, including
determining the price of publications, maximizing their usefulness, and
ensuring the widest possible dissemination. In addition, there is the function
of maintaining special relations with the media, which is the principal
instrument for disseminating statistical information to the general public.
21. There are many possible ways to divide
the tasks of the department in charge of dissemination. Figure 4 in Appendix 3
suggests one possible model.
22. The field organization is best organized
on a functional basis, and should serve all the subject matter departments in
the statistical agency. A sensible alternative to this form of organization is
difficult to imagine. It is possible, and in some instances natural, for the
field organization to have an internal structure that manages household and
business inquiries separately.
23. In cases where the statistical agency
relies heavily on administrative data, the field organization may look after
those records as well, in the interest of efficiency. Obviously, in this case
the principles behind functional organization and subject matter organization
come into partial conflict. Figure 5 in Appendix 3 includes collection
activities based on administrative records, but shows them as a separate
department within the area of field organization.
24. The field organization performs the
following functions:
v
maintaining
contact with respondents
v explaining credibly to respondents
the purpose of a statistical inquiry, securing cooperation, and retaining
respondents’ goodwill throughout
v
providing
feedback to the rest of the statistical organization regarding the quality of
the frame after it has been tested in the field and the receptivity of
respondents to a particular statistical inquiry and the instruments used to
conduct it
v
taking
charge of the initial editing of the information collected. There is no set
boundary between the responsibility of the field organization and that of the
subject matter departments. Efficiency, training, individual perceptions of the
subject matter departments, and so on, all play into how work is delegated. But
it should be kept in mind that the closer one is to the respondent, the easier
it is to correct errors.
v
researching
effective methods of data collection, including necessary adaptations of
questionnaire and classification standards, techniques of persuasion,
incentives and deterrents, etc.
25. A possible structure is shown in Figure 5
in Appendix 3.
26. Subject matter departments perform a
number of unique functions that are difficult to delegate precisely because
they are specialized, concern a well-defined set of users, and often an equally
well-defined set of respondents. In broad terms, the functions of subject
matter departments are as follows:
v
to
assess user needs and the most cost-effective way to meet them, as well as to
forecast the direction in which those needs are evolving
v
to
have an action plan on the best method for collecting relevant information
given the additional constraints imposed by limits on paperwork generated by
government agencies
v
to
balance the special requirements of users with the constraints imposed by the
use of standard definitions, international norms of classifying and accounting,
and the specific needs expressed by those who compile the macroeconomic
accounts
v
to
document the quality attributes of the data produced and make the results
accessible to both researchers and the general public
27. In practice, relations with key users in
government should be considered a distinct function simply because of the
disproportionate amount of contact they require. What follows is a discussion
of the structures necessary for the subject matter department to carry out its
functions in a number of different statistical domains (see Figure 6 in
Appendix 3).
28. Basic statistics constitutes the raw material
a statistical agency works with. The subject matter unit can be organized in a
variety of ways. One is by source (business, household, institutions, and
public sector); this method of organization has the merit of coinciding with
the structure suggested for the field organization. Another alternative is to
organize the subject matter area by user constituency. This possibility has the
merit of coinciding with the structure of that part of the agency responsible
for marketing, anticipating users’ needs, etc. Or, the subject matter unit can
be structured by process. For example, it can be made up of a department that
specializes in surveys of business accounting records, another that deals with
brief monthly surveys administered to businesses and households alike, and so
on.
29. The most stable and robust structure
embodies all these considerations. Accordingly, it is typical to find
structures in which a distinction is made between household and business
statistics (source of information) as well as between microeconomic and
macroeconomic statistics (user constituency). The proportions of the mixtures
vary but it is unlikely that any structure can be based on a single criterion.
30. There are additional alternatives on how
to divide each of the major substructures. For instance, one way of dividing
basic economic statistics is by industry (with departments in charge of
forestry, agriculture, mining, manufacturing, construction, transportation,
distribution, communications, etc.). While such a division is appealing, it
leaves some major issues unresolved. For instance, consumer prices; exports of
goods and services; investment in machinery and equipment; industrial and
commercial employment; and expenditure on research and development are all
important to the explanation of changes in GDP; but they cut across all
industries.
31. The difficulty in using a single
criterion to create substructures exists for social statistics as well. For
example, one might conceive of a division based on the source of information
(institutions concerned with education, health, welfare). However, such a
distinction would not work well by itself. Within the category of health there
is interest simultaneously in the activity and effectiveness of the health
provider (hospitals, clinics, etc.); in the outcomes of health activities
(patients as part of households); and in the technologies used (results of
research and development expenditures). Clearly, the most effective way of
dividing labor in the area of social statistics is to take into consideration
several criteria.
32. If the major effort, when organizing an
office, is to emphasize relevance, adaptability, and demonstrable understanding
of user needs, the most effective division of labor is one in which the
departments are defined by user constituency (in this case the division can
mirror the way the government is structured). Such a classification is
particularly important for industries such as those involved with the
production and distribution of energy, transportation, or agriculture. But in
some cases there is hardly any room for choice. Where the information source is
unique and dominates all other activities relative to the subject (e.g.,
international trade in goods, where the national Customs administration is the source
of the vast majority of documentation), there is little opportunity for
combining it with other matters. In other cases the possibility does exist; for
example, price indexes in general, and the CPI in particular, are based on a
combination of method of compilation and source of information.
33. For certain statistics that cut across
all sectors and all industries there is some choice in how to best organize
them. One example is statistical information on the demand for labor and fixed
equipment. The statistic on the demand for labor comes from business and public
sector surveys, but its supply is estimated through household inquiries. If the
two are split according to the source of information, much of the significance
of the broad division of statistics into social and economic would depend on
where the household surveys of the labor force were placed.
34. It is difficult to take sides on this
matter since the arguments are so complex; in the end, personalities and a
sense of internal balance play deciding roles. It is possible to imagine an
organization in which the heads of each department depicted in Figures 5, 6,
and 7 in Appendix 3 are at the same hierarchical level and are all under the
supervision of the chief statistician, or one in which the departments belong
to different groups, and only the head of each group reports to upper
management.
35. Figure 7 in Appendix 3 gives a more
detailed view of what the department of basic economic statistics might look
like. The distinctions drawn are merely put forward as an example. The
following considerations underlie the suggestions:
v
With
respect to activities represented under industry statistics, what is sought is
that their cost structure, account for their gross value of production and for
the elements that make up value added. The adjusted sum of the values added
should be a subtotal (after a few conceptual adjustments) in the national
accounts.
v
Some
industries are direct counterparts of a ministry, which may have its own
statistical branch and/or generate information for statistical purposes as a
by-product of its administrative or regulatory activities. Those industries are
grouped as a suggestion under the heading Industry Statistics Group 1. The
others are part of Group 2.
v
The
three activities grouped under the heading Economy-Wide Basic Economic
Statistics are sui generis. They rely
on a unique source of information (e.g., international trade) or a unique
technique (e.g., price indexes), or on a wide variety of sources and special
circumstances (e.g., installation of capital equipment).
36. Figures 8 and 9 in Appendix 3 indicate
possible structures for social and household statistics and for the department
dedicated to labor. The considerations that underlie the models are similar to
those stated above. In other words, there may be strategic reasons to
consolidate all labor statistics into one organizational unit, not because of
homogeneity of methods or sources but rather to interact more effectively with
the Ministry of Labor. Similar thoughts may dominate the decision to
consolidate the production of all statistics related to health and education so
as to simplify contacts with powerful Ministries that may have their own
statistical branches.
37. There is no standard way to organize
macroeconomic statistics. In many cases, they are estimated by the research
departments of central banks or by specialized departments in national
Ministries of Finance. But they possess certain unique attributes that should
influence how they are organized:
v
Macroeconomic
statistics are not the results of direct measurement. Rather, they are jointly
derived from basic economic statistics, public accounts, and accounting
identities.
v
The
chief concerns of the national accountant and of the balance of payments
compiler include the balancing properties of the system, differences between
the various accounts, estimation procedures for incomplete elements, and so on.
v
Balancing
supply and demand for each industry, required for the estimation of total
supply (GDP), will necessarily result in unexplainable differences and gaps.
Review of these differences constitutes essential feedback for the managers of
basic economic statistics.
v The relative strengths and
weaknesses of the basic data as revealed through the balancing exercise are
critical elements in the formulation of any development programme for basic
economic statistics.
38. Given the above characteristics, two
propositions follow:
v
Compiling
the macroeconomic accounts is effectively a full-scale analysis of the quality
of the basic economic statistics that comprise them. Accordingly, it is best
managed by an independent department.
v
The
concerns of those compiling the national and international accounts are so
analytical in nature that it may not be the best of use of their talents if
they take part in conducting statistical surveys or negotiating with other
government departments over the provision of administrative records.
39. The department in charge of macroeconomic
statistics does have a number of straightforward duties. It is the architect of
the national economic accounts in the broad sense of the term (with the scope
defined in the United Nations “System of National Accounts 1993”). It should be
equally responsible for certain derived estimates, such as the index of
industrial production (usually the estimate of the value added at constant
prices of the sectors responsible for mining, manufacturing, and part of public
utilities) and the level and rate of change in productivity, where such
calculations exist.
40. Since there are many instances in which
the statistical agency is not responsible for the compilation of the national
accounts, the consequences of dividing the responsibility for macroeconomic
statistics should be mentioned. The producer of basic statistics is often less
concerned with estimates of GDP than with loss of quality in economic
statistics. Accordingly, coordinators in systems where responsibilities are
split should be especially mindful of the importance of ongoing critical
evaluation from the perspective of those who compile the national accounts.
41. Whether or not the macroeconomic accounts
department should also be responsible for such statistics as the flow of funds
table or the national balance of payments depends on the capacity of the
central statistical agency to access and interpret financial transactions.
Because this capacity varies so much, in many cases the responsibility for the
balance of payments or at least for the capital account of the balance remains
with the Central Bank.
42. In general, a statistical system has much
to gain from close contact between producers of basic statistics and compilers
of the macroeconomic accounts. Such contact should move in both directions. The
system benefits if there is prompt submission of basic results and a critical
review and assessment of the quality of those results once they have been
incorporated into the national accounting tables.
43. Figure 6 in Appendix 3 assumes that the
responsibility for all the
macroeconomic accounts lies in one department.
44. The last two decades have witnessed
increased interest in a number of non-traditional statistics, the most
significant of which are those related to the environment and natural
resources, the diffusion of scientific knowledge, and the acquisition and use
of new technologies. In addition, the dependence of many offices on tax and
pension-related administrative records has increased substantially. Tax and
social security-related statistics developed as a by-product of new statistics
on consumer behaviour, which resulted from linked survey and administrative
records. In the near future, as the use of the Internet for business and
consumer related-matters and for research, education, etc. becomes universal,
it is likely that the realm of statistics will expand even further.
45. None of these statistics fits perfectly
with the traditional split between economic and social subjects. Some fall into
both categories. Others—particularly those related to science and to the
natural environment—represent a new field of inquiry. Whether or not they
require their own departments (particularly in smaller offices) cannot yet be
determined. These statistics are still in a phase of development; there are few
internationally agreed-upon standards for collection and compilation, and in
many instances the chief responsibility is still in the hands of other
ministries, and not necessarily in the statistical branches within those
ministries.
46. One possible solution is to house new
statistical developments in the research department of the statistical agency,
where such a department exists. Another is to create a research facility within
the national accounting department. Clearly the proper solution will vary from
country to country.
47. The departments responsible for the
statistical infrastructure have three main functions, which are reflected in
the statistical agency’s organization:
v
development
of scientific methods and their application in the measurement of variables
characteristic of social and economic processes
v
drafting
of codes that ensure both internal consistency in the way variables are defined
and classified and a certain amount of inter-country comparability (compliance
throughout the national statistical system is assumed)
v
creation
and upkeep of databases required to sample businesses, households, and physical
facilities
48. In more ways than one, the department in
charge of the development and application of scientific methods also acts as
the professional conscience of the statistical system. As such, it should
adhere to the strictest quality standards.
49. The department in charge of standards and
classifications should possess sufficient authority to set legal standards
against which there is limited appeal. In general, review of the correctness
and soundness of a particular standard or method should take place prior to its
implementation.
50. The department in charge of the
statistical agency’s sampling frames should maintain the records’ physical and
functional integrity and keep them up-to-date; provide frame extracts to those
conducting sample surveys or censuses; and make sure all parts of the
statistical agency use official sampling frames rather than unauthorized
surrogates.
51. These responsibilities presuppose an
agency-wide conviction that it is best to have a single comprehensive sampling
frame for each major application (e.g., all business surveys) rather than
separate frames for each survey, and that survey taking departments should pool
their knowledge, updating a collective frame rather than maintaining individual
frames.
52. Commonly, mathematical statisticians are
concentrated in the department in charge of sampling and sample survey design.
In addition to its other responsibilities, the department should be able to
substantiate the following principles:
v all surveys conducted by the
statistical agency should be based on sound designs
v
the
information necessary to gauge the reliability of the statistics produced by
the agency must be intelligible and accessible to all users
v
all
members of the statistical agency should be thoroughly convinced of the need to
base their measurement activities on sound scientific principles as the only
means to establish the quality of the statistics produced
53. So long as these three principles are at
the forefront of the statistical agency’s agenda, less importance is placed on
whether the department in charge of statistical methods should be physically
distinct from all others, whether its staff should be dispersed among the user
departments within the agency, or whether to adopt alternative solutions. The
design of the infrastructure will depend largely on country-specific
circumstances.
54. In practice, two pressures should be
balanced. On the one hand, mathematical statisticians can contribute most to a
healthy statistical agency as part of multidisciplinary project teams, in which
they defend quality; advise on scientific methods; and work toward increased
efficiency in the realization of projects. On the other hand, they must be part
of an organization that supports the development of their technical skills and
provides avenues for their career advancement. Achieving a balance can be
challenging. Figure 11 in Appendix 3 shows a possible organization for this
area.
55. Four advances, familiar to all, have
greatly expanded the capabilities of a statistical agency:
i.
Powerful, relatively
inexpensive computer equipment has been made available to all staff members in
many statistical offices.
ii.
User-friendly applications
software have given staff members control over a number of key statistical
functions ranging from questionnaire design to collection, editing, tabulating,
mapping and publishing. Readily available off-the-shelf components have made
in-house application programming easier and component reuse within the
organization has become more common.
iii.
Computer networking has facilitated
internal access to data and metadata through established tiered client/server
environments.
iv.
Internet technology has made
timely access to the outside world feasible for staff members, thereby making
it possible to conduct research, gather general information, etc., at all
levels of an organization, rather than solely through top management.
These
and related changes have given rise to new concerns regarding how to manage the
information technology (IT) environment.
56. In light of these changes, the department
in charge of the technical infrastructure should perform the following
functions:
v
managing
the information technology model of the organization as described in 7.2.
v
administering
the technical infrastructure of the agency
v determining the extent to which
outsourcing is to be used and managing the relationship with vendors
v
planning
the replacement of capital equipment
v
advising
on how IT can be used to advance the mission of the statistical agency.
57. The analytical function must be truly
pervasive, occurring in all areas of a statistical agency. However, an agency
bears a heavy operational burden, and so its staff cannot engage in pure
analysis all the time. An analytical approach to problems is a necessity, and
should be omnipresent. However, analysis here has a different connotation.
58. There are two major analytical activities
to be considered. One, which can be categorized as research, seeks to improve
the efficiency of operations; to refine the concepts underlying measurement
activities; and to innovate with methods that minimize error. The other
analytical activity seeks to make sense of the results obtained; to place them
in meaningful social or economic contexts; to relate them to other events and
processes; and generally to make them more valuable to users. Naturally,
feedback on any inadequacy or ambiguity of numerical results is a by-product of
this analytical activity.
59. This section primarily pertains to the
second set of analytical activities. In terms of organization, the agency
should decide whether this activity takes place in all subject matter
departments or whether it should be concentrated in one department or area that
will serve the entire agency. An effective agency generally incorporates
elements of both approaches.
60. A possible boundary could be conceived
along the following lines. Explanatory material accompanying the release of new
results (particularly those of ongoing surveys or more complex statistical
activities) should be handled by the subject matter department. However,
analytical material designed to stand on its own (e.g., a study of demand for
health facilities in light of the latest demographic projections, or the
relation between survival rates of small firms and their propensity to innovate
through new production processes) should be handled by a dedicated department
capable of economic, sociological, and demographic analysis.
61. One possible location for the unit in
charge of analysis is in a subject area such as macro-economic statistics or
social statistics depending upon the analytical emphasis in the agency. Another
alternative is to place this function in the office of the chief statistician.
62. Planning is essential to a successful statistical
agency. A sound plan proposes a series objectives and offers reasonable
arguments on how they are to be met and why the resources allocated are indeed
adequate.
63. The planning function deals with
establishing a process that produces a plan, defines the conditions of its
implementation, monitors its execution, and advises on fallbacks and
alternatives when necessary.
64. There are two ways to organize the
planning function. One technique is to appoint a planning officer or dedicate a
planning office to conduct this exercise annually and to be responsible for
maintaining the plan. The other is to engage the management of the statistical
agency to cooperate in a collective effort and produce a balanced plan.
Clearly, an effort involving all parts of an agency will receive more
acceptance, support, and commitment than a specialized effort.
65. The time frame is an especially important
element in planning. Developmental statistical activities should not be limited
to a single-year plan; yet, planning should be realistic and not look too far
into the future. The longer the time frame, the shakier are the planning
assumptions necessary to ensure a proper allocation of resources. A possible
balance between the two considerations is planning according to a three-year
time frame.
66. A detailed description of horizontal
mechanisms is included in chapter 5. There are a few differences between
vertical hierarchies and committee structures. The vertical (or line) structure
of a statistical agency broadly follows the hierarchy of public administration;
is in many instances ordained in the agency’s legislation; and often
corresponds to the financial allocations made by the government to the
statistical agency. As such it cannot be easily altered, though total
inflexibility should be avoided.
67. The horizontal structure of a statistical
agency (i.e., its ad hoc committee
structure) exists to solve management, subject matter, and procedural problems
inherent to the official structure of the agency. Because of the nature of the
problems, the effectiveness with which these committees resolve conflicts, and
the respect they gain as a source of dispassionate advice, they provide a sense
of cohesion, which the vertical structure at times unwittingly undermines.
Playing these two roles requires that the committee structure of a statistical
agency be highly flexible so that it can take full advantage of the personal
attributes of its staff in responding quickly to problems as they arise.
68. No statistical agency can function
effectively without the systematic help of outside contacts. The more
prestigious they are, the more they will bolster the office’s credibility.
69. A statistical agency can be challenged by
members of its various constituencies on several fronts. Users may demand
explanations of how priorities are determined; assurances regarding the quality
of results (i.e., use of objective methods and the absence of political
tampering); and the confidentiality of individual records.
70. These questions can be difficult to
answer, particularly those regarding quality and priority ranking. The agency’s
position is immensely strengthened if it can demonstrate that a well-chosen,
knowledgeable group of outsiders reviewed the decisions and agreed, or
suggested modifications, which were taken into account in the final
determination. Likewise, the agency’s position on the objectivity and the
timeliness of chosen methods can be strengthened if it can point to the review
and approval of the best technical experts available.
71. For these reasons, while a senior council
or commission is typically charged with a periodic review of statistical
policies and priorities, a number of agencies have built a network of
supplementary advisory bodies. Their reports are made available to the public
and the government authorities responsible for the agency.
72. Supportive measures will allow the agency
to face all the challenges that arise out of the exercise of its mandate.
Whatever mechanisms are put in place, the following are crucial:
v
proper
legal basis
v
distinguished
and interested outsiders (national and international) who serve as members of a
network of commissions and committees, both formally and informally
v
public
status commensurate with the responsibilities of being chief fact finder for
the government and for the nation
73. In the end, the agency should demonstrate
that it has managed its resources wisely, kept the government and the public
informed, served the international statistical community, and communicated
important information impartially and effectively.
Appendix 1: Annotated
model of a National Statistics Act
Introduction
Section 2.6 of the Handbook
discusses some general principles of statistical legislation, but contains few
specifics. This annex presents specific texts that might be useful for
countries that want to introduce a general statistics law, or modify an
existing one. It should be stressed that this ‘model of a statistics act’ is
not in any way meant to be prescriptive or normative. It is just an
illustration of the issues that are dealt with in actual statistical
legislation in various countries; several of the texts were indeed derived from
actual statistical laws.
The ‘model’ contains
two types of elements:
·
Issues
that should be dealt with in all statistics acts (printed in bold
typeface), because they are about matters of principle. In this regard, reference
may also be made to the Fundamental Principles for Official Statistics, adopted
by the Statistical Commission of the United Nations in 1994. These issues often
directly relate to what is said in the Principles, as will be explained in
annotations to the text.
·
Alternative or optional elements. ‘Alternatives’
relates to solutions that are different, but not necessarily inferior to
the ones described in the issues mentioned before. ‘Optional’ elements are more
country specific; they may work for some countries, but may be unacceptable for others.
Statistics Act of Numberland
Note: the Act consists
of three main sections: General (Handbook section 2.6.2, about the main actors
of the statistical system, their roles, responsibilities and interactions),
Statistical Operations and Data Collection, including some provisions for
regionally decentralized systems; see Handbook section 2.6.7) and Data
Confidentiality (including access to information protected by other laws, see
Handbook section 2.6.5).
GENERAL
1. Definitions (see also section
2.6.2 of the Handbook)
In this Act,
·
"Chief
Statistician" means the Chief Statistician of Numberland;
·
"department"
means any department, board, bureau, agency or other division of the government
of Numberland or of the government of a region or any agency of either;
·
"Minister"
means a member of the Cabinet of Ministers of Numberland,
·
"respondent"
means a natural or legal person in respect of whom or in respect of whose
activities any report or information is sought or provided pursuant to this
Act;”
·
“Council”
means National Statistical Council of Numberland.
2. Chief Statistician
The President of Numberland, at the recommendation of the
Council of Ministers, shall appoint an officer called the Chief Statistician of
Numberland to hold office for a renewable, fixed term of five years.
Note: Appointment of
the Chief Statistician for a fixed term helps to guarantee his professional
independence and to prevent political interference with official statistics
(see also Handbook section 2.4.5).
Alternative 1
The Prime Minister of
Numberland shall appoint an officer called the Chief Statistician.
Alternative 2
The Parliament of
Numberland shall appoint an officer called the Chief Statistician.
Alternative 3
At the recommendation
of the National Statistics Council, the President of Numberland shall appoint
an officer called the Chief Statistician.
Option
The President of
Numberland shall appoint a Minister for Statistics, who shall act as the Chief
Statistician.
2. The Chief Statistician shall,
(a)
advise
on matters pertaining to statistical programs of the departments of the
Government of Numberland, and confer with those departments to that end.
(b)
decide
on the manner in which data for statistical purposes are collected, how they
are compiled and when and how statistics are published.
(c)
supervise
generally the administration of this Act and control the operations and staff
of Statistics Numberland.
(d)
represent
Numberland in international statistical meetings or designate one or more staff
members of Statistics Numberland to do so.
Note: the provisions
in this article support the coordinating role of the Chief Statistician (a),
his professional autonomy (b, see also the Fundamental Principles of Official
Statistics - appendix 2), as well as his international role (articles 9 and 10
of the Fundamental Principles of Official Statistics – appendix 2).
4. Statistics Numberland
There shall be a statistics bureau, to be known as
Statistics Numberland, the duties of which are
(a) to collect, compile, analyze and publish statistical
information relating to the commercial, industrial, financial, social,
economic, environmental and general activities and condition of the people;
(b) to collaborate with departments of government in the
collection, compilation and publication of statistical information, including
statistics derived from the activities of those departments;
(c) to promote the avoidance of duplication in the
information collected by departments of government; and
(d) generally, to promote and develop integrated social and
economic statistics pertaining to the whole of Numberland and to each of the
regions thereof and to coordinate plans for the integration of those
statistics.
5. Work plan, release calendar and annual report
(a) Three months before the beginning of each new fiscal
year, the Chief Statistician shall submit to the Council of Ministers
a work plan for the next fiscal year, setting out all major statistical
collections and planned publications, as well as estimates of expenditure and
revenues related to this.
(b) At the beginning of each new fiscal year, the Chief
Statistician shall publish a calendar of the most important releases of new
statistics in that fiscal year.
(c) Within three months after each fiscal year the Chief
Statistician shall present a report to the Council of Ministers
with regard to the activities of Statistics Numberland in the preceding fiscal
year.
Note: this article is
mainly about transparency and accountability, two important aspects of the
Fundamental Principles. The publication of an advance release calendar (b)
prevents political interference with statistical outcomes.
6. National Statistical Council
(a) There shall be a National Statistical Council whose role
is to:
·
Advise
the Chief Statistician on statistical work plans; its advice shall be added
when the work plan is submitted to the Council of Ministers.
·
Comment
on the annual report that the Chief Statistician submits to the Council of
Ministers; its comments shall be added to this report.
·
Advise
the Chief Statistician on any other statistical issues.
(b) The Council shall be chaired by the Chief Statistician.
(c) The Council shall meet at least four times a year.
(d) The Council may set up subcommittees and ad hoc advisory
groups.
(e) The Council shall have at least 15 members, representing
the most important user groups of official statistics, in particular government
departments, the business community and academia.
(f) Members of the Council shall be appointed by the
Minister, at the recommendation of the Chief Statistician. They will serve for
renewable terms of four years.
(g) The Council’s Rules of Procedure shall be approved by
the Minister.
Note: the principal role
of a National Statistical Council is to be an interface with the users of
statistics, and thereby enhance the relevance of official statistics (see the
Fundamental Principles of Official Statistics – appendix 2, article 1). Of
course there are many different ways to set up such a council. Therefore, some
alternatives to the above are listed below.
Alternatives:
·
Minister chairs the Council, Chief Statistician
is vice-chair.
·
Independent chairman, Chief Statistician ex
officio member.
·
Council has fewer or more members.
·
Council approves statistical work plan (as
opposed to just advise on it).
·
Recommendation for membership through
co-optation (Council itself proposes candidates).
·
Membership is specified by law (e.g. ‘a
representative of the Ministry of Finance, a representative of the Central
bank, a representative of the Council of Economic Advisers’, etc.).
·
Longer or shorter terms of service.
STATISTICAL OPERATIONS AND DATA COLLECTION
Note: this part of the
law is mainly about practical arrangements for data collection, including
access to government data sources that may be relevant for statistical
purposes. It is important that statistics are collected in the most efficient
way and that respondents are not unduly burdened (see also article 5 of the Fundamental
Principles of Official Statistics – appendix 2).
7. Statistical personnel
The Chief Statistician may employ such commissioners,
enumerators, agents or other persons as are necessary to collect for Statistics
Numberland such statistics and information as the Chief Statistician deems
useful and in the public interest relating to such commercial, industrial,
financial, social, economic and other activities as the Chief Statistician may
determine, and the duties of the commissioners, enumerators, agents or other
persons shall be those duties prescribed by the Chief Statistician.
Options
Public servants
The Minister may, for
such periods as it may determine, use the services of any employee of the
public service of Numberland in the exercise or performance of any duty, power
or function of Statistics Numberland under this Act or any other Act, and any
person whose services are so used shall, for the purposes of this Act, be
deemed to be a person employed under this Act.
Note: this implies
that the Minister who is politically responsible for official statistics has
the power to recruit civil servants outside Statistics Numberland to perform
statistical duties; among other things this may be important for censuses.
Contracted services
Any persons retained
under contract to perform special services for the Chief Statistician pursuant
to this Act and the employees and agents of those persons shall, for the
purposes of this Act, be deemed to be persons employed under this Act while
performing those services.
Options
Cooperation with
regional authorities
The Chief Statistician
may enter into arrangements with the government of a region providing for
matters necessary or convenient for the purpose of carrying out or giving
effect to this Act, and in particular for all or any of the following matters:
(a) the execution by
regional officers of any power or duty conferred or imposed on any officer
pursuant to this Act;
(b) the collection by
any regional department or regional officer of statistical or other information
required for the purpose of this Act; and
(c) the supplying of
statistical information by any regional department or regional officer to the
Chief Statistician.
All regional officers
executing any power or duty conferred or imposed on any officer pursuant to
this Act, in pursuance of any arrangement entered into under this section,
shall, for the purposes of the execution of that power or duty, be deemed to be
employed under this Act.
The
Chief Statistician may enter into an agreement with the government of a region
for the exchange with, or transmission to, a statistical agency of the region
of
(a) replies to any
specific statistical inquiries;
(b) replies to any
specific classes of information collected under this Act; and
(c) any tabulations
and analyses based on replies referred to in (a) or (b).
An agreement with a region for the purposes
of this section shall apply only in respect of a statistical agency of the
region
(a) that has statutory
authority to collect the information that is intended to be exchanged or
transmitted pursuant to the agreement from a respondent who is subject to
statutory penalties for refusing or neglecting to furnish information to the
agency or for falsifying information furnished by him to the agency;
(b) that is prohibited
by law from disclosing any information of a kind that Statistics Numberland,
its officers and employees would be prohibited from disclosing, if the
information were furnished to Statistics Numberland; and
(c) whose officer's and
employees are subject to statutory penalties for the disclosing of any
information of the kind described in paragraph (b).
Where
any such information is collected by Statistics Numberland from a respondent,
Statistics Numberland shall, when collecting information, advise the respondent
of the names of any statistical agencies in respect of which the Chief
Statistician has such an agreement.
Note:
there are many different ways of interaction and cooperation between the national
statistical agency and regional bodies. In some countries the national
statistical agency has regional offices that are fully under its control (often
called the ‘vertical system’), while in others the regional statistical offices
are part of regional government (the ‘horizontal system’). Mixtures of these
two systems also exist. In federally structured countries, the relationships
between the regional and central governments are different yet.
8. Sharing of information
The Chief Statistician may enter into an
agreement with any department or municipal or other corporation for the sharing
of information collected from a respondent.
Such an agreement shall provide that
(a)
the
respondent be informed by notice that the information is being collected on
behalf of Statistics Numberland and the department or corporation, as the case
may be; and
(b) where the respondent gives notice in
writing to the Chief Statistician that the respondent objects to the sharing of
the information by Statistics Numberland, the information not be shared with
the department or corporation unless the department or corporation is
authorized by law to require the respondent to provide that information.
Note: see also article
5 of the Fundamental Principles of Official Statistics – appendix 2.
Option:
Access
to records
A
person having the custody or charge of any documents or records that are
maintained in any department or in any municipal office, corporation, business
or organization, from which information sought in respect of the objects of
this Act can be obtained or that would aid in the completion or correction of
that information, shall grant access
thereto for those purposes to a person authorized by the Chief Statistician to
obtain that information or aid in the completion or correction of that
information.
9. False or unlawful information
Every person who, without lawful excuse,
(a) refuses or neglects to answer, or willfully answers
falsely, any question requisite for obtaining any information sought in respect
of the objects of this Act or pertinent thereto that has been asked of him by
any person employed or deemed to be employed under this Act, or
(b) refuses or neglects to furnish any information or to
fill in to the best of his knowledge and belief any schedule or form that the
person has been required to fill in, and to return the same when and as
required of him pursuant to this Act, or knowingly gives false or misleading
information or practices any other deception thereunder is, for every refusal
or neglect, or false answer or deception, guilty of an offence and liable on
summary conviction to a fine not exceeding ........or to imprisonment for a
term not exceeding ......or to both.
10. Refusal to grant access to records
Every person
(a) who, having the custody or charge of any documents or
records that are maintained in any department or in any municipal office,
corporation, business or organization, from which information sought in respect
of the objects of this Act can be obtained or that would aid in the completion
or correction of the information, refuses or neglects to grant access to the
information to any person authorized for the purpose by the Chief Statistician,
or
(b) who otherwise in any way willfully obstructs or seeks to
obstruct any person employed in the execution of any duty under this Act
is guilty of an offence and liable on summary conviction to
a fine not exceeding ........or to imprisonment for a term not exceeding
.........or to both.
Note: although
enforcement by law may not be the ideal way to obtain basic data for statistics
(see Handbook section 2.6.4), most statistics acts contain some provisions to
make data collection for statistics statutory.
Options
Coding system for goods
The Chief Statistician
shall establish a coding system for goods imported into and exported from
Numberland to enable the collection, compilation, analysis and publication
statistics in relation to those goods.
Forms for statistical data
Instead of or in
addition to using agents or employees for the collection of statistics under
this Act, the Chief Statistician may prescribe that a form be sent to a person
from whom information authorized to be obtained under this Act is sought. A
person to whom such a form is sent
shall answer the inquiries thereon and return the form and answers to
Statistics Numberland properly certified as accurate, not later than the time
prescribed therefore and indicated on the form or not later than such extended
time as may be allowed in the discretion of the Chief Statistician.
Returns under Income Tax Act
For the purposes of
this Act
(a) the
Chief Statistician or any person authorized by the Chief Statistician to do so
may inspect and have access to any returns, certificates, statements, documents
or other records obtained on behalf of the Minister of National Revenue for the
purposes of the Income Tax Act , and
(b) the
Minister of National Revenue shall cause the returns, certificates, statements,
documents, or other records to be made available to the Chief Statistician or
person authorized by the Chief Statistician to inspect the records.
Return of exports and imports from Customs
For the purposes of
this Act, the Minister of National Revenue shall cause to be sent to the Chief
Statistician returns of imports and exports into and from Numberland and
details of the means of transportation used therefore.
Additional options:
POPULATION CENSUS
A census of population
of Numberland shall be taken by Statistics Numberland in the month of .... in
the year ....., and every tenth year thereafter.
The census of
population shall be taken in such a manner as to ensure that counts of
the population are
provided for each district of Numberland, as constituted at the time of each
census of population.
AGRICULTURE CENSUS
A census of
agriculture of Numberland shall be taken by Statistics Numberland in the year
..... and in every tenth year thereafter.
Census questions
The President of
Numberland shall, by order, prescribe the questions to be asked in any census
taken by Statistics Numberland.
Note: in some
countries regulations for censuses are part of the general statistics law, in
others censuses are ruled by separate legislation.
DATA CONFIDENTIALITY
Note: Clearly,
confidentiality of individual data is one of the main concerns of the
Fundamental Principles. In addition to a general provision in the Statistics
Act (article 11 below), an oath of office for statistical personnel (article 12
below) is often part of statistical legislation as well.
11. Prohibition against divulging
information
Except for the purpose of communicating information in
accordance with any conditions of an agreement made under this Act and except
for the purposes of a prosecution under this Act but subject to this section,
(a)
no
person, other than a person employed or deemed to be employed and sworn in
under this Act, shall be permitted to examine any identifiable individual
return made for the purposes of this Act; and
(b)
no
person who has been sworn in under this Act shall disclose or knowingly cause
to be disclosed, by any means, any information obtained under this Act in such
a manner that it is possible from the disclosure to relate the particulars obtained
from any individual return to any identifiable individual person, business or
organization.
12. Oath of office
The Chief Statistician and every person employed or deemed
to be employed pursuant to this Act shall, before entering on his duties, take
and subscribe the following oath or solemn affirmation:
I, ........, do solemnly swear (or affirm) that I will
faithfully and honestly fulfill my duties as an employee of Statistics
Numberland in conformity with the requirements of the Statistics Act, and of
all rules and instructions thereunder and that I will not without due authority
in that behalf disclose or make known any matter or thing that comes to my
knowledge by reason of my employment.
Where a person retained under contract to perform special
services for the Chief Statistician pursuant to this Act is a legal person, the
chief executive officer thereof and such other officers, employees and agents
thereof as are used to perform the special services shall, before entering on
any of the duties required under the contract, take and subscribe the following
oath or solemn affirmation:
I, ........., do solemnly swear (or affirm) that I will
faithfully and honestly fulfill my duties as an employee of (name legal person)
in respect of my employment in carrying out ......
in conformity with the requirements of the Statistics Act, and of all rules and
instructions thereunder and that I will not without due authority in that
behalf disclose or make known any matter or thing that comes to my knowledge by
reason of my employment as described herein.
Options
The Chief Statistician
may, by order, authorize the following information to be disclosed:
(a) information
relating to a person or organization in respect of which disclosure is
consented to in writing by the person or organization concerned;
(b) information
relating to a business in respect of which disclosure is consented to in
writing by the owner for the time being of the business;
(c) information
available to the public under any statutory or other law;
(d) information
relating to any hospital, mental institution, library, educational institution,
welfare institution, public utility
or other similar non-commercial institution except particulars arranged in such
a manner that it is possible to relate the particulars to any individual
patient, inmate or other person in the care of any such institution;
(f) information in the
form of an index or list of individual establishments, firms or businesses,
showing any, some or all of the following in relation to them:
I. their names and
addresses,
II. the telephone
numbers at which they may be reached in relation to statistical matters,
III. the products they
produce, manufacture, process, transport, store, purchase or sell, or the
services they provide, in the course of their business, or
IV. whether they are within specific ranges of
numbers of employees or persons engaged by them or constituting their work
force.
13. Information is privileged
Except for the purposes of a prosecution under this Act, any
return made to Statistics Numberland pursuant to this Act and any copy of the
return in the possession of the respondent is privileged and shall not be used
as evidence in any proceedings whatever.
No person sworn under this Act shall by an order of any
court, tribunal or other body be required in any proceedings whatever to give
oral testimony or to produce any return, document or record with respect to any
information obtained in the course of administering this Act.
14. Disclosing secret information
Every person who, after taking the oath under this Act:
(a) willfully discloses or divulges directly or indirectly
to any person not entitled under this Act to receive the same any information
obtained by him in the course of his employment that might exert an influence
on or affect the market value of any stocks, bonds or other security or any
product or article, or
(b) uses any information described in paragraph (a) for the
purpose of speculating in any stocks, bonds or other security or any product or
article
is guilty of an offence and liable on summary conviction to
a fine not exceeding ....... or to imprisonment for a term not exceeding
........or to both.
Options
Personation of Statistics Numberland employee
Every person who
(a) personates an
employee of Statistics Numberland for the purpose of obtaining information from
any person, or
(b) represents himself
to be making an inquiry under the authority of this Act when the person is not
an officer, employee or agent of Statistics Numberland,
is guilty of an
offence and liable on summary conviction to a fine not exceeding ........or to
imprisonment for a term not exceeding six months or to both.
Application of fines
All
fines imposed pursuant to this Act belong to the Government of Numberland and
shall be paid to the Receiver General.
Appendix 2: The fundamental principles of official
statistics
Extract from the report of the
Statistical Commission on its Special Session, held in New York 11-14 April
1994. Official Records of the Economic and Social Council, 1994, Supplement No.
9 (E/1994/29, E/CN.3/1994/18).
Action taken by the Commission
59. The Commission adopted the
fundamental principles of official statistics as set out in ECE decision
C (47), but incorporating a revised preamble. The preamble and principles,
as adopted, are set out below:
FUNDAMENTAL PRINCIPLES OF OFFICIAL
STATISTICS
The Statistical Commission,
Bearing in mind
that official statistical information is an essential basis for development in the
economic, demographic, social and environmental fields and for mutual knowledge
and trade among the States and peoples of the world,
Bearing in mind
that the essential trust of the public in official statistical information
depends to a large extent on respect for the fundamental values and principles
which are the basis of any society which seeks to understand itself and to
respect the rights of its members,
Bearing in mind
that the quality of official statistics, and thus the quality of the information
available to the Government, the economy and the public depends largely on the
cooperation of citizens, enterprises, and other respondents in providing
appropriate and reliable data needed for necessary statistical compilations and
on the cooperation between users and producers of statistics in order to meet
users' needs,
Recalling
the efforts of governmental and non-governmental organizations active in
statistics to establish standards and concepts to allow comparisons among
countries,
Recalling also
the International Statistical Institute Declaration of Professional Ethics,
Having expressed the opinion
that resolution C (47), adopted by the Economic Commission for Europe on 15
April 1992, is of universal significance,
Noting that,
at its eighth session, held at Bangkok in November 1993, the Working Group
of Statistical Experts, assigned by the Committee on Statistics of the Economic
and Social Commission for Asia and the Pacific to examine the Fundamental
Principles, had agreed in principle to the ECE version and had emphasized that
those principles were applicable to all nations,
Noting also that,
at its eighth session, held at Addis Ababa in March 1994, the Joint Conference
of African Planners, Statisticians and Demographers, considered that the
Fundamental Principles of Official Statistics are of universal significance,
Adopts
the present principles of official statistics:
1. Official statistics provide an
indispensable element in the information system of a democratic society,
serving the Government, the economy and the public with data about the
economic, demographic, social and environmental situation. To this end,
official statistics that meet the test of practical utility are to be compiled
and made available on an impartial basis by official statistical agencies to
honor citizens' entitlement to public information.
2. To retain trust in official
statistics, the statistical agencies need to decide according to strictly
professional considerations, including scientific principles and professional
ethics, on the methods and procedures for the collection, processing, storage
and presentation of statistical data.
3. To facilitate a correct
interpretation of the data, the statistical agencies are to present information
according to scientific standards on the sources, methods and procedures of the
statistics.
4. The statistical agencies are
entitled to comment on erroneous interpretation and misuse of statistics.
5. Data for statistical purposes may
be drawn from all types of sources, be they statistical surveys or
administrative records. Statistical agencies are to choose the source with
regard to quality, timeliness, costs and the burden on respondents.
6. Individual data collected by
statistical agencies for statistical compilation, whether they refer to natural
or legal persons, are to be strictly confidential and used exclusively for
statistical purposes.
7. The laws, regulations and measures
under which the statistical systems operate are to be made public.
8. Coordination among statistical
agencies within countries is essential to achieve consistency and efficiency in
the statistical system.
9. The use by statistical agencies in
each country of international concepts, classifications and methods promotes
the consistency and efficiency of statistical systems at all official levels.
10. Bilateral and multilateral cooperation in statistics contributes to the improvement of
systems of official statistics in all countries.
Appendix 3:
Figures representing possible organizations
of the functions of a statistical office
This appendix shows a selection of
twelve diagrams that purport to illustrate the most important functions carried
out by the average centralized statistical office. The diagrams do not
illustrate organizational charts. The arrows shown represent flows of
information rather than hierarchical relations. The latter are thought to be too dependent on the circumstances
of individual offices to be shown in a Handbook such as this one.
There is a distinction
made among the various flows of information that connect one group of
activities to another. In some cases the flow is of directives, policies,
metadata, and advice but not of data, raw or processed. Such flows are
generally shown using narrow tipped arrows on broad shafts. Broad tipped arrows
normally illustrate flows of data. And there are feedback flows shown by very
narrow black arrows. The text appended to the bottom of each diagram should
make clear what flows and from whom to whom. There are other flows, which are
not represented in the diagrams in order not to complicate them unduly.
The diagrams bring out three key themes of the Handbook.
Firstly, all viable organizations are articulated information processing
systems. Secondly, there is no viable organization that has no feedback loops.
And thirdly there is no organization that should be constituted without an
analytical and research function.
Naturally, nothing is said about relative sizes or who is in control of
what. Individual managers will determine how to implement the spirit of the
diagrams so as to fit their circumstances
Figure 7
Possible
breakdown by subject of programme area
concerned with economic statistics
Figure 9
Possible
analytical structure for the department
concerned
with the labor market
Figure 11
Possible
structure of the department
responsible
for the statistical infra structure
Figure 12
Possible functions for a department
infrastructure of a statistical agency
[2] Very few
projects with the exception of the census have the need, time, or resources to
document their own history and negative experiences. However, without a
tradition of documenting experiences, a project cannot benefit from previous
successes and failures. Oral history, though sometimes helpful, can also be a
means of miscommunication through anecdotes and distorted recollections.
[4]
Strict methodological guidelines should be in place to ensure that such
replacements do not compromise the propriety of the original choice.
[5] Failure to
address the respondent in his/her language, or with the accent or intonation
prevalent in the region or province, could ruin the chances of success for even
the most unobtrusive of surveys. Excessive zeal in pursuing recalcitrant
respondents in a small community can just as easily ruin a survey conducted in
the area.
Arthur Harvey ed., Montreal
1867
