In order to identify Usual Environment (UE) of tourists in MPD, country/National Statistical Office could refer to International Recommendation on Tourism Statistics 2008 (Compliler’s Guide) and Feasibility Study on the Use of MPD for Tourism Statistics Report 3a. (Methodological Issues) published by Eurostat.
Identifying the usual environment. For the time span of usual environment, National Statistical Office (NSO) or other institutions also should refer to UNWTO/UNSD Manual on Tourism “International Recommendations for Tourism Statistics 2008” (IRTS2008), both “Concept and Definition and Compilation Guide and Eurostat” (2013) “Methodological Manual for Tourism Statistics: Version 3.1”.
For domestic tourism, the usual environment and place of residence are contained within the country of reference (unlike the case with inbound data). Then, in order to define usual environment within the country of reference, one approach is to use the anchor point identification model (also called meaningful places identification), which is based upon the long-term time-space patterns of the subscribers. Such a model identifies the most probable home, work-time and other important locations over time. Alternative methods for identifying residence and usual environment can be used. However, all of the methods require a longer data period before meaningful locations can be detected. In Indonesia, use one year period to determine the usual environment, previously Indonesia use 3 (three months) data, but it did not work.
In addition, a periodic recalculation should be conducted in order to be able to detect changes and carry out the reclassification of meaningful locations (e.g. in the case of historical migration detection, it is possible to detect a change of home only after a longer period of data collection has become available for subscribers).
Based upon anchors or meaningful locations that are calculated using the anchor point model, the place of residence and usual environment can be identified for a specific period of time, which is subject to being recalculated and reclassified with new data updates as mentioned previously. The specific algorithms that are used can vary depending upon the methodology and definitions that are employed. Anchor points can be identified based upon the duration and/or frequency of the stay in the specific location. For example, the home anchor could be the location at which the subscriber spends most of their workday evenings and mornings; and weekends over a period of time (one month). The combination of such home anchors over a longer period of time (one year) can be used to define differences between permanent and temporary homes (the latter perhaps being holiday or secondary homes). Anchor points can change during time, so good practices include updating anchors monthly and also applying changes to historical data. This means that the usual environment is a dynamic description of space over history (as people might change their residence and usual environment over time) and this can ‘change’ the historical data. For example, based upon the data, a new home location (migration) can only be detected from the data after a significant amount of time has elapsed (at the start the new home seems only to be a new travel destination and might not even be significant). After the change has been detected, some of the historical data has to be changed.
The results of the anchor point calculation provide the geographical area (based upon the administrative units, though not necessarily a contiguous one, that describes a usual environment within which an individual conducts one’s regular life routines. Based upon the anchor point model or other home detection model, home and work-time anchor points or model with secondary frequently visited anchor points should form one’s usual environment. For example, in Estonia, where holiday homes and secondary homes (summer houses) can be identified from the anchor point model, the treatment (within or outside of the usual environment) is a subject of interpretation (e.g. how to treat the migration to a summer house during all three summer months), whereas in Indonesia, where holiday home or secondary home are not common, the usual environment is modified. However, in Indonesia there is term ‘’circular’’ where people have two homes, with one being for work and other being a family home. Therefore, any country could either implement the usual environment in Box 4 or modify it.
Box 4. Usual Environment Model
The analysis of outbound data that can be used in order to identify any frequent trips to foreign countries is also necessary and, based upon that information, the usual environment can be extended. However, as it is not possible to identify the limited geographical area to which subscribers are travelling in foreign countries, if a subscriber travels frequently, a whole foreign country is extended as a usual environment even if the destination locations for such outbound trips are not the same (e.g. shopping in a border town inside a foreign country).
The resulting dataset for subscribers’ usual environments should be calculated based upon specific periods and should also take into account the change of the place of residence (migration). The following calculations which involve domestic tourism are based upon those trips which were taken outside the usual environment.
The data representation for the usual environment can be limited to an administrative unit or a specific geometry (a buffer zone around the usual travel routes). The current description utilises the administrative unit approach as this is usually the common practice, although it must be noted that tourism-specific trips within this unit will be unaccounted for (e.g. in large administrative units). When compared to accommodation statistics, in which domestic visitors who spend nights in hotels within their usual environment are still considered as being domestic visitors, mobile data will provide a different methodological outcome as it is not possible at a small scale to define whether a person spent the night at home or in a hotel; the night was spent within the subscriber’s usual environment and therefore this is not considered as a tourism activity.
The meaningful places identification should also identify regular trips to places that might be at a substantial distance away or in a different administrative area but are regularly and frequently visited. It is recommended, in line with the IRTS 2008 Compilation Guide, that each country should define the precise definition of what is deemed to be regular and frequent in the context of its tourism statistics and should specify the parameters and criteria that are to be applied in the usual environment identification algorithm.
Because of the peculiarities of the data, it may be impossible to identify the place of residence or usual environment for some subscribers. For this reason, many temporary, short-term or otherwise defunct data has to be eliminated, as there is no information about such subscribers. From the data contained in the sample pilot, home and regular anchors can be identified for 74% of the subscribers and the rest of the data is useless in terms of tourism.
There are other methodological possibilities that are available when it comes to identifying the usual environment for subscribers as the anchor point model is mainly data-driven and differs from the methodology used for identifying the place of residence and usual environment mentioned here.
Refer to IRTS 2008,
In Europe, the Regulation 692/2011 (Regulation 692/2011) and the methodological description by Eurostat (Eurostat 2013a) identifies the usual environment as ‘the geographical area, though not necessarily a contiguous one, within which an individual conducts his regular life routines and shall be determined on the basis of the following criteria: the crossing of administrative borders or the distance from the place of residence, the duration of the visit, the frequency of the visit, the purpose of the visit’. The main interpretation of the usual environment should be defined based upon the subjective feeling of the respondent which is not available in the case of MPD. Therefore, the proposed determination of the usual environment should be based upon the criteria like frequency of the trips, duration of the trip, crossing of the administrative border and distance from the place of residence.
Usual environment could be defined by administrative unit such as, for example, at the second or third level NUTS classification. Also it could be defined by geographical area as a distance or a radius from a residence such as, for example, fifty kilometres from a residence.
The retrospective time window used for determining usual environment could be set for a number of days, for example 45, 60, or 90 days, for at least four (4) weeks within the time window, with at least one week having more than one day present during the week within the time window.
If the parameters of the criteria are universally agreed upon, it will lead to results that have better comparability between regions and will improve the understanding of the concept. With MPD, all of the criteria can be implemented. However, this will not be without some issues.
Concerning the identification of the usual environment within the country of reference, either by using the anchor point model or other methods, the success of the model depends heavily upon the specific conditions used in the calculations. As the specific criteria for defining the usual environment is left for the authorities of individual countries, the outcome in different countries can vary largely. Mobile data is very quantitative, and the criteria applied to the identification of the usual environment can be very extensive, despite following the official guidelines. The criteria suggested firstly for the identification of the usual environment and subsequently for the identification of trips outside this area can be successfully based upon the suggested ‘cascade system’. The following criteria for the calculation of the usual environment can be applied:
- Administrative unit/level or other geographical representation (polygons, radius from residence, buffer zone). If the administrative unit or level or geographical area is very large, many trips that are subjectively considered as being tourism trips can be identified as being trips within the usual environment and therefore, not a part of the tourism category. With smaller areas, there is a threat that too many trips are being considered as tourist trips.
- The frequency and duration of visits to a specific place. The underlying criteria can include any of the following:
- Time window for measuring the presence of the subscribers in specific locations. If longer periods are used as time window, the results include the ‘future to be’ usual environments while the subscribers were not so connected to the place. Shorter periods can result in trips home being identified as tourist trips during the period of time in which the subscriber was on a longer holiday.
- Retrospective or prospective time window. With continuous data updates, the logical solution is to analyse the data up to the point of its creation (looking only back). However, when re-processing historical data, the results can improve as the retrospective frame can ‘ignore’ the new usual environments for subscribers who re-locate to new places.
- Measure of frequency. The number of days and/or weeks the subscriber visited the administrative unit. The definition provided by Eurostat is difficult to implement in terms of quantitative data - ‘less than once a week’ (i.e. not every week) is rather limited when considering the fact that, during a longer holiday, people tend to visit their homes on less than a weekly basis.
- Number of days a week. This criterion can be ignored, but it can show ‘stronger connection’ to the place if the person spends more than a day in such a place.
- Length of stay in the places - usually is overridden by the frequency (spends small amount of time, but every day - e.g. visiting kindergarten to pick up the children).
- The duration over time that the usual environment holds to be in effect during the period of time in which the criteria for the usual environment are not fulfilled. It is very difficult to exactly estimate the moment at which the usual environment ceases to be the usual environment. If the ‘effect time’ is fairly short, then longer holidays can provide an influence so that the usual environment is not in effect when the person in question returns for a short home visit during their holiday. With a longer impact, a person’s usual environment might actually cease to exist because they have migrated elsewhere but the old usual environment remains in effect.