Symposium 2001/18 6 July 2001 English only |
Symposium on Global Review of 2000 Round
of
Population and Housing Censuses:
Statistics
Division
Department of
Economic and Social Affairs
United Nations
Secretariat
New York, 7-10
August 2001
Eivind Hoffmann**
C. Alternative data-collection methods
E. Follow-up and preparations during the intercensal
period
1.
Both the
revised UN Principles and Recommendations
for Population and Housing Censuses and the Recommendations for the 2000 Censuses of Population and Housing in the
ECE Region jointly prepared by UN/ECE and Eurostat include recommendations
for a range of economic characteristics. The former makes recommendations for
eight characteristics and their value sets, and the latter does so for six
“core” and 12 “non-core” topics, where three of the latter can be derived from
the others (see United Nations, 1998, and UN/ECE and Eurostat, 1998).
2.
Some
economic characteristics have traditionally been among those most commonly
included in national censuses—for example, “source of livelihood” or whether
the persons are “economically active” and in what type of work. These topics
have long been regarded as being among the more complex and expensive variables
to include in a census in a manner which will yield reliable results with the
degree of detail which the census results can support and important groups of
users will need. The International
Labour Office (ILO) therefore has advocated that further guidance should be
provided on how to effectively implement
the recommendations for these characteristics/topics, either as part of the
recommendations or as a supplement to them.
Following the acceptance of the revised Principles and Recommendations, the UN Statistics Division (UNSD)
and ILO have cooperated on the preparation of a Guide for the Collection of Economic Characteristics as
part III of the UN Handbook of Population
and Housing Censuses. Drafts for
chapters on the formulation of questions and response alternatives and on
coding strategies and procedures for “industry” and “occupation” were discussed
at a meeting of experts convened in 1998.
To expedite availability before the publication of the complete
document, they have recently been distributed by the ILO (see Gilbert, 2001,
and Hoffman 2001).
3.
This note will briefly review from the perspective of
the ILO relevant issues, experiences and possibilities concerning the economic
characteristics in relation to the topics specified for discussion at this
Symposium.
4.
Before discussing possible strategies for involving
stakeholders in census activities, it is necessary to identify the stakeholders
particularly interested in the economic characteristics and the reason for and
nature of their interest. Both their
identity and the interest that they have are related to the fact that a census
is the only source for detailed statistics on the most important economic
resource of a locality, a region and a country, namely, the persons who live
and work there. It is the economic activities of these persons and their type
of work that are the main determinants of their economic welfare and the basis
for future improvements. Thus the stakeholders for the census characteristics
that are needed to describe in detail the economic activities and capacities of
the population are all those who need to understand the structure of local,
regional and national labour markets and resources. This understanding is needed as a basis for the planning and
implementation of investments in human and physical capital; of production,
marketing and sales of goods and services; and of policies which can support
and promote such activities.[1] The unique capacity of censuses to give
comprehensive and detailed statistics for small geographic areas is
particularly important for statistics on participation in economic activities,
because in practice most labour markets in a country are local.[2] However, because it is impossible to
identify all the many potential users of such statistics and obtain from them a
financial commitment reflecting what they are willing to pay for timely census
results of this type, the census planners should seek out those few
institutions that have a mandate to represent groups of potential users, as
well as relevant ministries and research institutions, to ensure their support
for the inclusion of economic characteristics in a manner which can ensure that
the resulting statistics have satisfactory quality. For the economic characteristics such representatives of
potential users will include chambers of commerce, employers’ and workers’
organizations as well as companies large enough to operate in or supply many
localities. In principle their support
may take the form of contributions to the financing of the census as well as
lobbying those responsible for its overall financing.
5.
The above formulation that the inclusion of economic
characteristics in the census should be “in a manner which can ensure that the
resulting statistics have satisfactory quality” is important. As discussed in
Gilbert, 2001, and Hoffmann, 2001, there are many examples of census
questionnaires and processing procedures which have been designed and planned
in a manner which has resulted in statistics on economic activities with very
limited reliability, and thus also limited usefulness. Too often these
solutions have been chosen on the basis of undocumented claims that the extra
costs of alternatives could not be accommodated by the census budget, even
though some seldom-tried
alternatives are free—for example, the specification of categories for
pre-coded “industry” and “occupation” responses which reflect the structure of
the national economy instead of the top-level categories of the respective
international classifications. The implementation
of many possible improvements to question formulations is also virtually
costless.
6.
For 95 per cent of the countries in the world, there
are no realistic alternatives to either the current or the next census if one
wants to obtain the type of comprehensive, integrated and detailed statistics
on the structure of the population’s economic activities that the census can
provide. Even for the 5 per cent of
countries, mostly small or medium-sized ones in Europe, that have established
the institutional infrastructure as well as the administrative and statistical
procedures needed to make the use of administrative records a viable
alternative, perhaps in combination with a large sample survey for some
characteristics, there is a significant quality reduction in some aspects of
the resulting statistics[3]. This is the price that has to be paid for
the resulting cost reductions to the census organization and the quality gains[4]
for some other aspects of the statistics that can be produced.
7.
In the majority of countries that cannot produce
“census-like” statistics from administrative records, the statistics that can
be produced from other available sources will always fall short of those which
can be obtained from the census in some respects which are important for
certain groups of users:
·
Even where statistics from large sample surveys of
households, such as labour force surveys (LFS), are available they will tend to
exclude from their coverage certain “hard-to-observe” population groups, such
as “nomads” and those living in institutional households. An important concern is also that the
resulting statistics cannot be produced with sufficient precision to give
statistics for the small groups and the small geographic areas which may
constitute individual labour markets.[5]
·
Even when they are designed to cover all sectors of
the economy (and few are), statistics from establishment surveys will tend to
exclude from their coverage “establishments” that are small (this is often
deliberate), “invisible” because they are indistinguishable from the households
of their owner(s), and/or unstable, particularly those in the informal sector,
as well as individual own-account workers.
Thus a significant proportion of total employment, as well as all those
persons who are unemployed or inactive, will be excluded from such
statistics. An additional consideration
is that the personnel records of establishments seldom include the information
needed for detailed statistics on the distribution of employed persons
according to important characteristics, such as sex, age, educational
attainment and occupation, and that even if they do the consequent workload on
establishments to provide even simple distributions according to them will be
prohibitive.[6]
·
Even where they exist, statistics based on
administrative records such as tax records, social security records and
unemployment registrations will tend to suffer from quality problems related to
coverage and timeliness as well as to scope, validity and reliability of the
variables which are used to produce statistics for different groups, as
indicated above.
8.
However, the fact that some users will be satisfied to
some extent by the statistics on the structure of the economically active
population produced from sources other than a census, means that whether and
how to include any or certain economic characteristics in a census will be a
question of priorities. Hopefully, the
process of consultations undertaken as part of the preparations for the census,
as well as careful analysis of the costs of including, in an adequate manner,
the possible characteristics, will have provided the information needed to
understand how the trade-offs relate to the priorities for favoured groups of
users.
9.
It is difficult to see that there are special
considerations related to the use of new technologies or new ways of organizing
census operations for the economic characteristics, with one exception: the use
of computer-assisted (or automatic) coding (CAC)[7]
of “industry” and “occupation”[8]. The manual coding of these variables has
traditionally been among the most costly and least reliable elements in the
processing of a census[9]:
error rates of 20 per cent or more have frequently been reported from quality
control studies. The experiences with
the development and use of CAC systems in several countries over the last 20
years have demonstrated (1) that dramatic gains in the productivity and quality
of the coding process can be gained from the use of such systems; but also (2)
that many of the quality gains have been due to the improved understanding of
the coding task and the improved coding indexes and procedures which were
necessary for CAC systems to be developed and used effectively, and that these
improved tools can and should be used also with manual coding procedures.[10] It is also clear that to reap the possible
productivity gains and cost reductions from the use of CAC systems, they should
be completely integrated into the overall census operations and processing
procedures.
10.
Few special considerations emerge from a concern with
economic characteristics with respect to issues concerning the retention of
institutional memory, organizational structures and archiving in intercensal
years. However, it may be worth
pointing out that a programme of regular labour force surveys (LFS) will
provide both a good reason to retain in the national statistical office
relevant competence and experience on the economic characteristics and coding
procedures developed for and during the census, and an opportunity to gain
experience and carry out relevant experiments in preparation for the next
census. Studies to establish any
systematic differences between the statistics from the two sources will be very
useful. If they are carried out both for the census reference period and at
regular intervals in the intercensal period, they may provide a basis for
projecting in intercensal years statistics similar to those from the census for
smaller areas and for groups other than those for which LFS results can be
produced with satisfactory precision. They may also assist in the preparations
for the next census.
11.
Census mapping has two objectives: (1) to facilitate
census operations, and (2) to make it possible to identify the location of
households in relation to the geographic units for which statistics from the
census are to be produced. Good,
large-scale maps are obviously important for both objectives. Concern with economic characteristics and
their use does not add anything to the understanding of how to obtain and
update such maps, nor on how to identify the location of dwellings and other
structures of interest, but this concern will stress the need to ensure that
such locations are documented on the census file in a way that makes it
possible to define alternative areas as a function of the interest of the users
of the census results. Where for
practical and economic reasons it may not be possible to allocate precise
geographic coordinates to each dwelling or structure, it will be important to
document, for example, on large-scale maps, aerial photos or satellite images,
the precise delineation of small “census tracts” which can be used as building
blocks for larger small areas for which statistics will be needed. Because the area that can be considered a
“local labour market” will depend on the presence of public transport services
and the quality of roads and tracks for trains and subways, it will be
important that areas served by transportation facilities can be approximated by
aggregating such basic census tracts.
Statistics on employment and the structure of economic activities in
such “labour markets” (or “commuting areas”) will be important supplements to
statistics for the areas administered by local governments.
12.
It is important to carry out investigations into the
quality of the census results, both on how well the census has been able to
cover the total target population and on the quality of the measurement of the
various characteristics included in the census. Whether a post-enumeration survey (PES) should be one of the
instruments of such investigations will depend on the other instruments
available and how confident one can be that the measurements resulting from the
PES will provide a more accurate reflection of the situation at the time of the
census than those that can be obtained by using the same resources to (1)
improve the basic census operations; and (2) strengthen other quality control instruments. For some important groups of the population
the actual reference period used will determine the value observed for some of
the economic characteristics. This will tend to favour the use of resources to
ensure the quality of the census itself rather than to conduct a PES. In
addition, when there is a regular LFS survey in the country it may be possible
to use it as a vehicle for a PES-type investigation, e.g., by adding a module
with census-type questions. It would be particularly unfortunate to drop a
regular LFS in the census year, because of the consequent breaks in the LFS
time series. LFS results cannot be
replaced by census results, partly because of the influence that the
differences in data-collection procedures will have (see also the above remarks
on activities in the intercensal period) and partly because of the difference
in the timeliness of the results[11].
13.
Following previous census rounds both UNSD and the
UN/ECE Statistics Division have collected and presented overviews of the
content of the national censuses undertaken.
In 1990 ILO published the first edition of its Sources and Methods. Labour Statistics. Vol. 5: Total and economically
active population, employment and unemployment (population censuses), with a second, updated edition in 1996.
Annex 1 to this note presents a sample description from the latter, as
well as a synoptic table summarizing the information about key economic
characteristics included in almost all censuses: those used to make the necessary
distinction between the employed, unemployed and the not economically active,
as well as “industry”, “occupation” and “status in employment”. Other economic characteristics have been
much less “popular”.
14.
It is certainly to be hoped on the basis of similar
information from the current census round that a number of countries at
different levels of economic and social development will be selected for closer
review of their actual census procedures and experiences. To have more systematic information on the
costs and benefits of different practices will be useful for the preparations
for the next round. The experiences and
opinions of a wide range of users should also be obtained.
NOTE
15.
This paper was prepared for the United Nations
Symposium on Global Review of 2000 Round of Population and Housing Censuses:
Mid-Decade Assessment and Future Prospects, organized by the United Nations
Statistics Division and held 7-10 August 2001. Suggestions from colleagues to a
draft have improved the paper, but the views expressed are those of the author
and do not necessarily reflect those of the organizers, the ILO or its Bureau
of Statistics. The author apologizes for all errors and omissions and would
welcome comments and suggestions for improvements and correction. Address:
CH-1211 Geneve 22, Switzerland; e-mail: hoffmann@ilo.org.
Gilbert,
R. (2001). Asking questions on economic characteristics in a population census.
STAT working paper, no. 2001/1. International Labour Office, Geneva.
Hoffmann,
E. (1995). We must use administrative records for official statistics – but how
should we use them? Statistical Journal
of the United Nations ECE, vol. 12, pp. 41-48.
Hoffmann
E. (2001). Coding occupation and
industry in a population census. STAT working paper, no. 2001/2. International
Labour Office, Geneva.
ILO
(1996). Sources and Methods. Labour
Statistics. Vol. 5: Total and economically active population, employment and
unemployment (population censuses) (Second edition). International Labour
Office, Geneva.
United Nations (1998). Principles
and Recommendations for Population and Housing Censuses, Revision 1. Statistical Papers, Series
M, No. 67/Rev. 1. Sales No. E.98.XVII.8.
UN/ECE and Eurostat (1998). Recommendations
for the 2000 Censuses of Population and Housing in the ECE Region. Statistical Standards and Studies, No.
49. United Nations, New York and
Geneva.
United
Nations and ILO (forthcoming). Guide for
the Collection of Economic Characteristics. Part III of Handbook of
Population and Housing Censuses.
Studies in Methods, Series F, No. 54.
United Nations, New York.
Sample
description, from ILO (1996):
Australia
1.
Name and address of the organization responsible for the census:
Australian Bureau of Statistics, P.O. Box
10, Belconnen ACT 2616.
2.
Population censuses conducted since 1945 (years):
1947, 1954, 1961, 1966, 1971, 1976, 1981,
1986 and 1991.
The present description relates to the
1991 census (held on 6 August).
3.
Coverage of the census:
(a) Geographical scope:
Whole country.
(b) Persons covered:
All persons of all ages, except foreign
diplomats and their families.
4.
Reference period:
The week prior to the census for full-
and part-time workers, and the four weeks preceding the census day for
jobseekers.
5.
Main topics:
(a) Total population, by sex and age: yes
Economically active population by:
(b) Sex and age group: yes
(c) Industry: yes
(d) Occupation: yes
(e) Status in employment: yes
(f) Highest educational level: yes
(g) Hours of work: yes
(h) Other characteristics: yes
Re (a): The age is defined in terms of age at last birthday.
Re (g): Employed persons, at work, were asked to specify their
actual hours of work during the reference period in the main job held.
Re (h): The census also collected information on gross income and
means of transport used to travel to workplace.
6.
Concepts and definitions:
(a)
Economically
active population: It comprises all persons aged 15 years and
over who, during the reference period, were either employed or unemployed,
according to the definitions given below. Excluded are persons who did not have
a job and did not look for work in the 4-week period immediately prior to the
census day; these persons were considered as inactive. Members of the armed forces
are included in the definition.
(b)
Employment:
Considered as employed are all persons who, during the reference period,
performed any full- or part-time work for payment or profit, or any unpaid work
in a family business. Home duties are excluded unless payment was received for
work in other households. The question used to determine if a person is to be
counted as employed was: Last week, did the person have a full-time or
part-time job of any kind? It is
reported that the following categories are included:
(1). persons doing unpaid work in family
firm or business;
(2). employed persons, temporarily absent
from work;
(3). working students with a part time
job;
(4). seasonal or occasional workers;
(5). apprentices and trainees.
Only persons belonging to categories (1)
and (2) can be identified separately according to status in employment and by
cross-classification with data on hours worked.
(c)
Unemployment:
Considered as unemployed are all persons who were without work and
seeking work. To determine if a person is to be counted as unemployed, the
question used was: Did the person actively look for work at any time in the
last four weeks? Actively looking for work means being registered with the
Commonwealth Employment Service; writing, telephoning or applying in person to
an employer for work, or advertising for work. Students seeking work are
excluded from the definition.
7.
Classifications used:
Only employed persons are classified by
“industry”, by “occupation” and by “status in employment”.
(a)
Industry:
Based on the questions: For the main job held last week, what was the
employer’s trading name and workplace address? and What kind of industry,
business or service is carried out by the employer at that address? The answers
provided information for industry coding. The industry classification is based
on the Australian Standard Industrial
Classification (ASIC) and the Industry and Destination Zone Index,
which is a listing of all establishments in Australia known to carry out
economic activity. For coding industry, 615 classes were used. Links to the
ISIC-rev.2 have been established to the group (4-digit) level.
(b)
Occupation:
Based on the questions: In the main job held last week, what was the
person's occupation? (give full title; for example, accounts clerk, civil
engineering draftsman, fast-foods cook, floor tiller, extruding machine
operator; for public servants, state official designation as well as
occupation; for armed services personnel, state rank as well as occupation) and
What are the main tasks or duties that the person himself/herself usually
performs in that occupation? (describe as fully as possible; for example,
recording accounts, preparing drawings for dam construction, cooking hamburgers
and chips, fixing cork tiles, operating plastic extruding machine). Occupation
was classified in terms of the Australian
Standard Classification of Occupations (ASCO) and coded to the unit group
level of the classification. For coding occupation 337 group codes were used
which comprised 282 unit groups, 52 minor groups, eight major groups and three
additional codes to process responses which were inadequately described. Links
to the ISCO have not been established.
(c)
Status in employment: Based on the question: In the main job held
last week, was the person: a wage or salary earner; conducting own business but
not employing others; conducting own business and employing others; a helper
not receiving wages or salary? For coding status in employment, the four
following categories were used: wage or salary earner; self employed; employer;
unpaid helper.
8.
Main differences compared with the previous census:
No major difference.
9.
Publication of the census results:
Final census data on the economically
active population and its components (employment and unemployment) were
available on a state-by-state basis beginning September 1992. Preliminary results from the 1991 Census
were released in a publication series First
Counts for Statistical Local Areas (Cat. no. 2701.1-8) on a state-by-state
basis from February to April 1992. The
ABS released final census results on a state by state basis in a publication
series Census Counts for Small Areas.
Detailed data are available on demand from ABS
Information Services. The 1991
census results are also available on floppy disk, magnetic tape, microfiche,
cartridge and CD-ROM. For further information see 1991 Census: A Guide to Products and Services (Cat. no. 2910.0) or
contact Census Marketing, Australian
Bureau of Statistics, P.O. Box 10 Belconnen ACT 2616, phone 61 6 252 7879, fax
61 6 253 1809.
|
Armed
forces included |
Students
included |
Data
on time worked |
Classification
used |
||||||||||||
|
|
|
|
Industry
|
Occupation
|
Stat.
in Emp. |
||||||||||
Country,
area or territory |
Census
year |
Age
limits |
Ref.
per. |
Regular
|
Con-scripted
|
With
a part time job |
Seeking
work |
Last
week or month |
Last
year |
Number
of groups |
Links
to ISIC |
Number
of digits |
Number
of groups |
Links
to ISCO |
Number
of digits |
Number
of groups |
Albania |
1989 |
15+ |
no |
yes |
yes |
yes |
no |
no |
. |
... |
no |
. |
... |
no |
. |
3 |
American Samoa
|
1990 |
16+ |
w/y |
yes |
yes |
yes |
yes |
ah |
w |
231 |
Rev.2 |
... |
13 |
no |
. |
7 |
Anguilla |
1992 |
12+ |
m |
yes |
. |
yes |
yes |
no |
. |
... |
Rev.3 |
1 |
... |
88 |
1 |
3 |
Antigua and
Barbuda |
1991 |
15-64 |
w/y |
no |
no |
yes |
yes |
ah |
m |
157 |
Rev.3 |
3 |
136 |
88 |
... |
6 |
Argentina |
1991 |
14+ |
w/m |
yes |
yes |
yes |
yes |
no |
. |
... |
Rev.3 |
3 |
9 |
88 |
1 |
7 |
Aruba |
1991 |
14+ |
w |
yes |
yes |
yes |
no |
uh |
. |
... |
Rev.2 |
4 |
... |
88 |
4 |
6 |
Australia |
1991 |
15+ |
w/m |
yes |
. |
yes |
no |
ah |
. |
615 |
Rev.2 |
4 |
337 |
no |
. |
9 |
Austria |
1991 |
15+ |
d/w |
yes |
yes |
yes |
no |
uh |
. |
117 |
Rev.3 |
2 |
175 |
88 |
2 |
8 |
Bahamas |
1990 |
15+ |
w/y |
yes |
yes |
yes |
yes |
no |
w |
9 |
Rev.2 |
1 |
9 |
88 |
1 |
5 |
Bahrain |
1991 |
12+ |
w |
yes |
yes |
yes |
yes |
no |
. |
17 |
Rev.3 |
1 |
... |
no |
. |
4 |
Barbados |
1990 |
15+ |
w/y |
yes |
yes |
yes |
yes |
ah |
m |
17 |
Rev.3 |
1 |
9 |
88 |
1 |
7 |
Belgium |
1991 |
14+ |
d |
yes |
yes |
yes |
no |
uh/ah |
. |
809 |
Rev.2 |
4 |
1700 |
88 |
3 |
9 |
Belize |
1991 |
15+ |
w/y |
yes pr. |
yes pr. |
yes |
no |
ah |
m |
... |
Rev.3 |
4 |
... |
88 |
4 |
6 |
Benin |
1992 |
10+ |
3m |
yes |
yes |
no |
no |
no |
. |
9 |
Rev.2 |
1 |
9 |
68 |
1 |
8 |
Bermuda |
1991 |
16+ |
w/y |
yes |
yes |
yes |
yes |
uh |
m |
18 |
Rev.2 |
1 |
... |
68 |
... |
7 |
Bolivia |
1992 |
7+ |
w |
yes |
no |
yes |
yes |
no |
. |
... |
Rev.3 |
4 |
... |
88 |
3 |
7 |
Botswana |
1991 |
12+ |
m |
yes |
yes |
yes |
no |
no |
no |
9 |
Rev.2 |
1 |
10 |
88 |
1 |
2 |
Brazil |
1991 |
10+ |
y |
yes |
yes |
yes |
yes |
. |
uh |
26 |
Rev.2 |
1 |
10 |
68 |
1 |
11 |
Brunei Darussalam |
1991 |
15+ |
w |
yes |
yes |
no |
no |
ah |
. |
10 |
Rev.3 |
1 |
10 |
88 |
1 |
4 |
Bulgaria |
1992 |
10-90 |
d |
yes |
yes |
yes |
no |
no |
. |
184 |
Rev.3 |
2 |
642 |
88 |
3 |
. |
Burundi |
1990 |
10+ |
6m |
yes pr. |
yes pr. |
no |
yes |
no |
. |
10 |
Rev.2 |
1 |
10 |
88 |
1 |
6 |
Canada |
1991 |
15+ |
w |
yes |
yes |
yes |
yes |
ah |
. |
296 |
no |
. |
514 |
no |
. |
4 |
Cape Verde |
1990 |
10+ |
w |
yes |
yes |
yes |
yes |
no |
. |
10 |
Rev.2 |
1 |
9 |
68 |
1 |
7 |
Cayman Islands
|
1989 |
15+ |
w |
yes |
. |
yes |
yes |
uh |
. |
65 |
Rev.3 |
2 |
94 |
88 |
3 |
4 |
Central
African Rep. |
1988 |
6+ |
w |
yes |
no |
no |
yes |
no |
. |
9 |
Rev.2 |
1 |
400 |
68 |
3 |
5 |
Chad |
1993 |
6-98 |
w/y |
yes |
yes |
no |
no |
no |
no |
60 |
Rev.3 |
2 |
... |
88 |
2 |
6 |
Chile |
1992 |
14+ |
w |
yes |
no |
yes |
yes |
no |
. |
... |
Rev.3 |
3 |
... |
88 |
4 |
5 |
China |
1990 |
15+ |
d/m |
yes |
yes |
no |
no |
no |
. |
13 |
Rev.2 |
1 |
8 |
68 |
1 |
. |
Comoros |
1991 |
12+ |
m |
no |
no |
no |
yes |
no |
. |
... |
no |
. |
8 |
no |
. |
7 |
Cook Islands |
1991 |
15+ |
m |
. |
. |
yes |
... |
uh |
. |
162 |
Rev.2 |
4 |
390 |
88 |
4 |
6 |
Cyprus |
1992 |
15+ |
w |
yes |
yes |
yes |
no |
uh |
. |
61 |
Rev.3 |
2 |
30 |
88 |
2 |
6 |
Czech Republic
|
1991 |
15+ |
d |
yes |
yes |
no |
yes |
. |
. |
47 |
Rev.3 |
2 |
91 |
68&88 |
1 |
9 |
Dominican
Republic |
1993 |
10+ |
w |
yes |
yes |
yes |
no |
no |
. |
10 |
Rev.2 |
1 |
10 |
68 |
1 |
5 |
Ecuador |
1990 |
8+ |
w |
yes |
yes |
yes |
no |
uh/ah |
. |
72 |
Rev.2 |
3 |
284 |
68 |
3 |
7 |
El Salvador |
1992 |
10+ |
w |
yes |
no |
yes |
yes |
uh/ah |
. |
... |
Rev.3 |
4 |
... |
88 |
4 |
8 |
Equatorial
Guinea |
1994 |
6+ |
w |
no |
no |
yes |
yes |
no |
. |
... |
Rev.3 |
... |
10 |
88 |
1 |
. |
Finland |
1990 |
15-74 |
w/y |
yes |
no |
yes |
no |
no |
no |
460 |
R.2&3 |
4 |
400 |
68 |
3 |
2 |
France |
1990 |
15+ |
no |
yes |
yes |
yes |
no |
no |
. |
600 |
Rev.2 |
2 |
455 |
68 |
2 |
4 |
French Guiana |
1990 |
14+ |
w |
yes pr. |
yes pr. |
yes |
yes |
no |
. |
100 |
no |
. |
42 |
no |
. |
5 |
Gabon |
1993 |
10+ |
w/6m |
yes |
yes |
no |
no |
no |
. |
40 |
no |
. |
350 |
88 |
3 |
6 |
Gambia |
1993 |
10+ |
m |
yes |
no |
no |
no |
ah |
. |
... |
Rev.3 |
1 |
... |
88 |
1 |
5 |
Gibraltar |
1991 |
15+ |
d |
no |
no |
no |
no |
no |
no |
23 |
Rev.3 |
1 |
125 |
no |
. |
3 |
Greece |
1991 |
10+ |
w/y |
yes |
no |
yes |
yes |
ah |
no |
159 |
Rev.3 |
3 |
284 |
68 |
3 |
4 |
Grenada |
1991 |
15+ |
w |
yes |
yes |
yes |
yes |
ah |
. |
... |
Rev.3 |
... |
... |
88 |
4 |
6 |
Guadeloupe |
1990 |
14+ |
w |
yes pr. |
yes pr. |
yes |
yes |
no |
. |
100 |
no |
. |
42 |
no |
. |
5 |
Guam |
1990 |
15+ |
w/y |
yes |
. |
yes |
yes |
ah |
w |
231 |
no |
. |
503 |
no |
. |
7 |
Guatemala |
1994 |
7+ |
w |
yes |
no |
yes |
yes |
no |
. |
... |
Rev.3 |
4 |
... |
88 |
4 |
6 |
Hong Kong SAR of China |
1991 |
15+ |
w/m |
yes |
. |
yes |
yes |
no |
. |
87 |
Rev.3 |
3 |
116 |
88 |
3 |
4 |
Hungary |
1990 |
no |
w |
yes |
yes |
no |
no |
no |
. |
294 |
no |
. |
808 |
no |
. |
8 |
India |
1991 |
no |
y |
yes |
yes |
yes |
yes |
. |
no |
462 |
Rev.3 |
3 |
512 |
68 |
2 |
4 |
Indonesia |
1990 |
10+ |
w/y |
yes |
yes |
yes |
yes |
uh/ah |
. |
47 |
Rev.2 |
2 |
334 |
68 |
3 |
5 |
Iran, Islamic
Rep. of |
1991 |
10+ |
w |
yes |
yes |
yes |
yes |
no |
. |
292 |
Rev.3 |
4 |
284 |
68 |
3 |
5 |
Ireland |
1991 |
15+ |
d |
yes |
. |
no |
yes |
no |
. |
263 |
R.2&3 |
2 |
210 |
68&88 |
2 |
4 |
Isle of Man |
1991 |
16+ |
w/5m |
yes |
. |
no |
no |
ah |
w |
21 |
no |
. |
371 |
88 |
3 |
3 |
Italy |
1991 |
14+ |
w |
yes |
no |
no |
no |
ah |
. |
60 |
Rev.3 |
2 |
35 |
88 |
1 |
14 |
Jamaica |
1991 |
14+ |
w/y |
yes |
yes |
no |
no |
ah |
m |
... |
Rev.3 |
2 |
... |
88 |
2 |
7 |
Japan |
1990 |
15+ |
w |
yes |
. |
yes |
no |
no |
. |
213 |
Rev.2 |
3 |
294 |
68 |
1 |
6 |
Kenya |
1989 |
10+ |
w |
... |
... |
no |
no |
no |
. |
no |
. |
. |
8 |
88 |
1 |
4 |
Korea,
Republic of |
1990 |
15+ |
y |
yes pr. |
no |
yes |
no |
. |
no |
90 |
Rev.2 |
3 |
286 |
68 |
2 |
4 |
Luxembourg |
1991 |
no |
no |
yes |
... |
yes |
no |
uh |
. |
500 |
no |
. |
390 |
88 |
4 |
9 |
Macao SAR of
China |
1991 |
14+ |
w/m |
yes |
yes |
yes |
yes |
no |
. |
10 |
Rev.3 |
1 |
10 |
88 |
1 |
. |
The former
Yugoslav Rep. of Macedonia |
1994 |
15+ |
d |
yes |
yes |
no |
no |
no |
. |
14 |
no |
. |
10 |
no |
. |
5 |
Madagascar |
1993 |
10+ |
w |
yes |
yes |
no |
no |
no |
. |
159 |
Rev.3 |
3 |
10 |
88 |
1 |
7 |
Malaysia |
1991 |
10+ |
w |
yes |
yes |
yes |
yes |
ah |
. |
... |
Rev.2 |
... |
... |
68 |
... |
4 |
Maldives |
1990 |
12+ |
w/3m |
no |
no |
yes |
no |
uh/ah |
. |
159 |
Rev.3 |
3 |
390 |
88 |
4 |
4 |
Martinique |
1990 |
14+ |
w |
yes pr. |
yes pr. |
yes |
yes |
no |
. |
100 |
no |
. |
42 |
no |
. |
5 |
Mauritius |
1990 |
12+ |
w/y |
. |
. |
yes |
yes |
ah |
w |
263 |
Rev.2 |
4 |
390 |
88 |
4 |
8 |
Mexico |
1990 |
12+ |
w |
yes |
yes |
yes |
yes |
ah |
. |
220 |
Rev.2 |
2 |
9600 |
88 |
4 |
5 |
Mongolia |
1989 |
see (a) |
d |
no |
no |
no |
no |
no |
. |
12 |
no |
. |
982 |
no |
. |
5 |
Morocco |
1994 |
see (b) |
d |
yes |
yes |
no |
no |
no |
. |
215 |
Rev.3 |
4 |
65 |
88 |
... |
7 |
Namibia |
1991 |
10+ |
w |
no |
no |
yes |
no |
no |
. |
307 |
Rev.3 |
4 |
396 |
88 |
4 |
8 |
Nauru |
1992 |
10+ |
w |
. |
. |
yes |
no |
uh/ah |
. |
5 |
no |
. |
... |
88 |
1 |
7 |
Nepal |
1991 |
10+ |
y |
yes |
... |
yes |
. |
. |
m |
9 |
Rev.2 |
1 |
7 |
68 |
1 |
4 |
Netherlands
Antilles |
1992 |
15-99 |
w/y |
yes |
yes |
yes |
yes |
uh |
w |
17 |
Rev.3 |
1 |
10 |
88 |
1 |
9 |
New Caledonia |
1989 |
14+ |
w |
yes |
yes |
no |
no |
uh |
. |
14 |
no |
. |
33 |
no |
. |
9 |
New Zealand |
1991 |
15+ |
d/m |
yes |
. |
yes |
yes |
uh/ah |
. |
... |
Rev.2 |
4 |
... |
88 |
4 |
4 |
Northern
Mariana Islands |
1990 |
15+ |
w/y |
no |
no |
yes |
yes |
ah |
w/uh |
... |
no |
. |
... |
no |
... |
7 |
Norway |
1990 |
16+ |
w/y |
yes |
yes |
yes |
no |
uh/ah |
m |
... |
Rev.2 |
4 |
84 |
68 |
1 |
3 |
Panama |
1990 |
10+ |
w |
yes |
yes |
yes |
yes |
no |
. |
18 |
Rev.3 |
1 |
10 |
68 |
1 |
5 |
Papua New
Guinea |
1990 |
10+ |
w/y |
yes |
yes |
no |
no |
no |
. |
no |
. |
. |
9 |
88 |
1 |
. |
Paraguay |
1992 |
10+ |
w |
yes |
no |
no |
no |
no |
. |
9 |
Rev.2 |
1 |
9 |
88 |
1 |
6 |
Peru |
1993 |
6+ |
w |
yes |
no |
yes |
yes |
no |
. |
292 |
Rev.3 |
4 |
116 |
88 |
3 |
6 |
Philippines |
1990 |
10+ |
w/y |
yes pr. |
no |
yes |
yes |
no |
. |
... |
Rev.2 |
4 |
... |
88 |
4 |
. |
Portugal |
1991 |
12+ |
w |
yes |
yes |
yes |
yes |
ah |
. |
292 |
Rev.3 |
4 |
390 |
88 |
4 |
7 |
Puerto Rico |
1990 |
16+ |
w/y |
yes |
yes |
no |
no |
uh |
w |
231 |
no |
. |
503 |
no |
. |
8 |
Réunion |
1990 |
14+ |
w |
yes pr. |
yes pr. |
yes |
yes |
no |
. |
100 |
no |
. |
42 |
no |
. |
5 |
Romania |
1992 |
14+ |
y |
yes |
yes |
yes |
no |
. |
no |
99 |
Rev.3 |
3 |
437 |
88 |
3 |
6 |
St. Lucia |
1991 |
15+ |
w/y |
no |
no |
no |
no |
ah |
m |
17 |
Rev.3 |
1 |
9 |
88 |
1 |
6 |
St.Vincent and
the Grenadines |
1991 |
15+ |
w/y |
no |
. |
no |
no |
ah |
m |
17 |
Rev.3 |
1 |
10 |
88 |
1 |
6 |
Samoa |
1991 |
10+ |
w |
no |
no |
yes |
yes |
no |
. |
... |
Rev.2 |
... |
... |
88 |
3 |
4 |
Sao Tome and
Principe |
1991 |
10+ |
w/y |
yes |
yes |
yes |
no |
uh/ah |
m |
9 |
Rev.2 |
1 |
116 |
88 |
3 |
5 |
Saudi Arabia |
1992 |
12+ |
w |
yes |
yes |
no |
no |
no |
. |
60 |
Rev.3 |
2 |
284 |
68 |
3 |
4 |
Singapore |
1990 |
no |
w |
yes |
yes |
no |
no |
no |
. |
945 |
Rev.3 |
1 |
314 |
88 |
1 |
5 |
Slovakia |
1991 |
15+ |
d |
yes |
yes |
no |
yes |
. |
. |
47 |
Rev.3 |
2 |
91 |
68&88 |
1 |
9 |
Slovenia |
1991 |
15+ |
d/y |
yes |
yes |
no |
no |
no |
no |
700 |
no |
. |
... |
no |
. |
5 |
South Africa |
1991 |
no |
d |
yes |
yes |
no |
no |
no |
. |
40 |
Rev.2 |
2 |
165 |
68 |
... |
2 |
Spain |
1991 |
16+ |
w |
yes |
no |
yes |
yes |
no |
. |
... |
R.2&3 |
2 |
... |
68&88 |
2 |
7 |
Sudan |
1993 |
10+ |
w |
no |
no |
no |
no |
no |
. |
72 |
Rev.2 |
3 |
390 |
88 |
4 |
5 |
Sweden |
1990 |
16+ |
m |
yes |
yes |
yes |
. |
uh |
. |
340 |
Rev.2 |
4 |
321 |
68 |
1 |
4 |
Switzerland |
1990 |
15+ |
no |
yes |
yes |
yes |
yes |
uh |
. |
703 |
Rev.2 |
2 |
404 |
88 |
... |
7 |
Syrian Arab
Republic |
1994 |
10+ |
w/y |
yes |
yes |
no |
no |
uh/ah |
m |
... |
Rev.3 |
5 |
... |
88 |
2 |
5 |
Thailand |
1990 |
13+ |
w/y |
yes |
yes |
yes |
no |
no |
no |
13 |
Rev.2 |
1 |
83 |
68 |
2 |
6 |
Trinidad and
Tobago |
1990 |
15+ |
w/y |
yes |
yes |
yes |
no |
uh |
no |
9 |
Rev.2 |
1 |
116 |
88 |
3 |
9 |
Turkey |
1990 |
12+ |
w |
yes |
yes |
no |
no |
no |
. |
10 |
Rev.2 |
1 |
7 |
68 |
1 |
4 |
Uganda |
1991 |
10+ |
w |
yes pr. |
yes pr. |
yes |
yes |
no |
. |
no |
. |
. |
161 |
88 |
3 |
3 |
United Kingdom
|
1991 |
16+ |
w |
yes |
. |
yes |
yes |
uh |
. |
320 |
Rev.3 |
2 |
371 |
88 |
... |
5 |
United States |
1990 |
16+ |
w/y |
yes |
. |
yes |
yes |
ah |
w+uh |
236 |
no |
. |
501 |
no |
. |
8 |
Venezuela |
1990 |
12+ |
d |
yes |
yes |
yes |
yes |
no |
. |
9 |
Rev.2 |
1 |
7 |
68 |
1 |
7 |
Viet Nam |
1994 |
see (d) |
w |
no |
no |
yes |
yes |
ah |
. |
20 |
no |
. |
33 |
68 |
1 |
2 |
Virgin Islands
(British) |
1991 |
15+ |
w/y |
yes |
. |
no |
yes |
ah |
m |
292 |
Rev.3 |
4 |
390 |
88 |
4 |
5 |
Virgin Islands
(US) |
1990 |
16+ |
w/y |
yes |
yes |
no |
no |
uh/ah |
ah |
7 |
no |
. |
39 |
no |
. |
7 |
Yemen |
1994 |
10+ |
w |
yes |
yes |
yes |
no |
no |
. |
... |
Rev.3 |
... |
390 |
88 |
4 |
5 |
Zambia |
1990 |
12+ |
w/y |
yes |
yes |
yes |
yes |
no |
m (see (c)) |
9 |
Rev.2 |
1 |
91 |
68 |
1 |
4 |
Zimbabwe |
1992 |
10+ |
y |
yes |
yes |
no |
no |
. |
no |
no |
. |
. |
109 |
88 |
3 |
4 |
. |
not applicable |
… |
not available |
d |
day |
w |
week |
ww |
weeks worked |
m |
months |
mw |
months worked |
y |
year |
uh |
usual hours |
ah |
actual hours |
pr |
if residing in
private households |
ISIC |
International
Standard Industrial Classification of all economic activities |
Rev.2 |
ISIC 1968
edition |
Rev.3 |
ISIC 1989
edition |
ISCO |
International
Standard Classification of Occupations |
68 |
ISCO (1968
edition) |
88 |
ISCO (1988
edition) |
(a) |
employment: 15
years and over; unemployment: 15 to 50 years for men and 15 to 45 years for
women (China). |
(b) |
employment: 15
years and over; unemployment: 15 to 59 years for men and 15 to 56 years for
women (Czech Republic and Slovakia). |
(c) |
16 to 59 years
for men, and 16 to 54 years for women (Mongolia). |
(d) |
7 years and
over for employment, 15 years and over for unemployment (Morocco). |
(e) |
employment: 14
to 80 years; unemployment: 14 to 64 years for men and 14 to 59 years for
women (Romania). |
(f) |
15 to 60 years
for men, and 15 to 55 years for women (Viet Nam). |
(g) |
only multiple
job holders were asked this question (Zambia). |
16.
Household survey
results as alternatives to the results from a Population Census
17.
The
differences in coverage between most Labour Force Surveys (LFS) and a
Population Census (PC) mentioned in the previous note is equally valid for other
forms of Household Surveys (HS) which may be undertaken: for operational
reasons they will tend to exclude certain population groups, in particular
nomads, persons living in collective or institutional households and persons
below and/or above certain age limits, unless one or more of these groups are
seen as target groups for the survey, e.g., as with “child labour”
surveys. The importance of this
difference in coverage will need to be considered by users looking for
alternatives to the PC as a source of statistics on the population’s economic
characteristics. Such differences will
of course be particularly important for statistics on groups with many members
among those who have been excluded: e.g., if migrant workers (national or
foreigners) tend to live in collective households, be important in particular
industries and/or occupations, and numerous enough to make a difference to
estimates which can be produced with acceptable precision given the sample size
and sampling procedures.
18.
The
limitations to the precision of estimates caused by the sample size and
sampling procedures restrict the type of statistics which can be produced from
all forms of HS, i.e., by imposing lower limits to the size of the groups that
can be estimated with a satisfactory degree of precision. In practice this means that HS results will
represent an alternative to PC results only for those users who are mainly
interested in statistics for the country as a whole or for a few large regions,
and for distributions of characteristics over other large, and reasonably
evenly distributed, categories.
19.
Specifically
concerning the economic characteristics, it is important to notice that
conceptually the UN census recommendations follow closely the international
standards for measuring employment and unemployment, and do so also for
measuring characteristics of the employment situation. The same is the case for most national
Labour Force Surveys (LFS). Thus for
users who can afford to ignore the scope and precision issues mentioned above,
as well as differences introduced by operational differences between a PC and
an LFS (see footnote 2 to annex 4 below), as well as other types of HS, it may
be possible to regard LFS-based statistics about the structure of the labour
force as a viable alternative to PC results.
For countries that cannot afford to have an LFS, the same type of users
may want to consider the possibility of instead using statistics based on other
forms of nationally representative HS as alternatives to PC-based statistics on
the structure of the labour force. This
will often be possible, at least to some extent, because the relation of
individuals to the labour market will be an important type of background
information collected in many HS which are designed to focus primarily on other
issues, e.g., poverty, health, incomes and expenditure patterns, time use
etc. However, those responsible for
such HS will tend to be less respectful of the international recommendations on
the measurement the economic characteristics than PC and LFS managers,
sometimes for good reasons of cost and/or relevance for their specific context.
Thus the users will again need to evaluate to what extent such differences will
be important.
20.
The use of
pre-coded categories for “industry” and “occupation” in a population census:
What are the issues?
21.
Both
Gilbert (2001, pages 32 and 36-37) and Hoffmann (2001, p. 6) emphasize that to
use pre-coded categories for “industry” and “occupation” will prevent the
production of the type of statistics for detailed categories which most users
will want and expect to get from a population census. However, it is also clear that the coding of write-in responses
to questions concerning these variables is among the most complex and therefore
also most costly and error-prone processing tasks for the census. Thus if census management has decided that
the available resources will not permit such coding, then it has to make a
number of other choices: the first is whether this means that these variables
should be completely excluded from the census, or be included with pre-coded
response alternatives. Arguments in
favour of excluding them are both (i) that many users will not get the type of
statistics that they need from the set of pre-coded alternatives, however
cleverly they are constructed; and (ii) that there are costs associated with
including them, with respect to space on the questionnaires, time needed for
enumerator training and for obtaining responses; and for processing. Arguments in favour of including them with
only pre-coded categories are mainly that some users will benefit from having
statistics which describe, even if it is only in broad terms, the occupational
and industrial structure of the country and of localities and individual
geographic labour markets.
22.
Having
chosen to include “industry” and “occupation” variables with pre-coded response
alternatives to be chosen either by the respondent or by the enumerator, the
first task for the census planners will be to determine the most useful sets of
categories that (i) will reflect in a useful way the structure of the national
economy and labour markets; and (ii) can be presented to respondents and/or
enumerators in a way that will make it easy for them to understand how the
respondent’s situation relates to these categories. Obviously the solutions to this challenge will have to differ
between countries as a function of their economic structure and level of
development[12].
However, the following points will always be valid: (a) the most aggregate groups in the
international standard classifications do not reflect in any meaningful way the
structure of the national economy and labour markets; and (b) the terms used to
designate the categories do not correspond to the terms which respondents and enumerators
will use and understand.
23.
The
implication of (a) is that the construction of the 10-12 pre-coded categories
which are to be presented to the respondents and numerators must be constructed
from categories defined at the lower levels in the respective classifications,
and those which are significant in the national context should be separately
identified while the rest should be lumped together as a residual
category. The implications of (b) are
(i) that the pre-coded categories should be given “labels” which tests show
will give respondents and enumerators correct signals about the intended
content; and (ii) that enumerators must be given special training on the
understanding of these categories.
24.
Why a
Labour Force Survey should not be dropped in a census year.
25.
Reasons why
the results from a population census (PC) will not be comparable with those
from a labour force survey (LFS) for the same year have been mentioned already
in the note and in annex 2: The
population coverage is different, as are the data-collection instruments and
procedures[13]. Contributing to differences in results
will also be the differences in precise reference periods in the year, and the
fact that the interviewers for the LFS on average are likely to be more experienced
and better trained than census enumerators can be.
26.
These
differences are not in themselves sufficient reason to undertake an LFS in a
census year, and thus to some extent forgo for the PC operation the use of
valuable and limited financial and staff resources. A more fundamental reason is that a national, regular LFS[14] is designed to serve an objective that
the PC cannot serve: namely, to monitor current developments in the national
economy and labour markets. Such
monitoring is one of the primary functions of a regular LFS, and both the
estimates of total employment and the estimates of total unemployment, as well
as employment and unemployment for particular groups, are among the most
important macro-economic indicators used for evaluating a country’s overall
economic performance and for formulating its macro-economic policies. This is a function the PC results cannot
perform, for two reasons: (a) The
results will be available too late to be useful for monitoring “current”
developments; and (b) the differences in LFS and PC results, which may not be
particularly important when these results are used to study the structure of a
national economy and labour market, will be extremely important if the PC
results are to be compared with LFS results from previous and subsequent years
to provide estimates of the changes that have been taking place. Differences that may be a minor nuisance
when analysing structures are likely to represent a major problem for the study
of changes over a relatively short period, e.g., from one year to the
next. The need for good estimates of
changes is why the sampling strategy for an LFS programme normally includes a
strong element of overlap between subsequent reference periods (rotation). Such overlap will reduce the sampling
variance on change estimates compared with a situation where they have to be
based on differences between level estimates from two independent samples. Timeliness and the absence of unknown
systematic differences make the LFS indispensable also in a census year.
* This
document was reproduced without formal editing
** International Labour
Office (ILO), Swizerland. The views expressed in the paper are those of the
author and do not imply the expression of any opinion on the part of the United
Nations Secretariat.
[1] Organizations
that need information on small areas to design sampling frames for sample
surveys should also be mentioned.
[2]
The meaning of “local” in this context will be outlined briefly below,
in connection with census mapping.
[3]
The reduction in quality is mostly in terms of scope as well as validity
and resolution of the characteristics measured, but also to some degree in
terms of the population covered and the reliability of the recorded
information, as functions of the procedures used for updating the registers and
the incentives which persons, establishments and register operators have to
ensure that correct information is recorded with correct dating. (See,
e.g., Hoffmann, 1995, for a brief discussion.)
[4]
Quality gains are mostly in terms of the frequency with which relevant
statistics can be produced. An
important advantage also follows from the fact that to produce such
“census-like” statistics from administrative records, one normally needs to
link individual records from several registers with the help of a “unique
personal identification number (UPIN)”.
This can be used also to link records on individuals for different
reference periods to produce statistics on the changes which they experience
over time. Such statistics are
impossible to produce from traditional censuses.
[5]
Further remarks on this issue can be found in Annex 2.
[6]
However, it should be possible to produce detailed statistics according
to “industry” and “geographic location of workplace”, subject to the usual need
to avoid revealing information about individual establishments. The quality of the coding of “industry” and
“place of work” may also be better.
[7]
Because there will always be some records that have to be coded with the
use of human judgement, the term “computer-assisted coding” is preferred and
the initials CAC will be used.
[8]
This is true as well as for the coding of “field of study” with
“educational attainment”.
[9]
For this reason many countries have limited the value sets for these
variables to a limited set of pre-coded alternatives. This solution is discussed briefly in annex 2.
[10] A
further discussion is included in Hoffmann, 2001. A list of relevant software systems is not included there, mostly
because such a list would rapidly become outdated. However, most of the systems that have proven to be effective
seem to have been developed by or for national statistical offices.
[11] This point is further elaborated in annex 4.
[12] The consequent reduction in the degree of international comparability of these census results must be considered a small price to pay for national relevance.
[13] The
dramatically smaller number of total observations in the LFS makes it possible
to design a set of questions which (i) cause the contact to last longer because
(ii) the sequence of possible questions can be formulated to eliminate
ambiguities in the distinctions that are to be made: e.g., between persons who
by producing goods (only) for the benefit of their own households are to be
classified as “employed” and persons who are to be classified as “not employed”
because the services they are providing are only for their own households and
they are not producing any goods. More
precise questions can also be formulated in an LFS to establish who among the
“non-employed” will satisfy the criteria for being classified as “unemployed”.
[14] The issue of whether to “drop” an LFS will not arise unless it is a regular, and at least annual, undertaking.