Total unemployment and registered unemployment
If unemployment data from household surveys or population censuses are unavailable, administrative records might be an alternative. However, a national count of either unemployed persons or work applicants that are registered at employment offices is likely to be only a limited sub-set of the total persons seeking and available for work, especially in countries where the system of employment offices is not extensive. Unemployment registration often uses eligibility requirements that exclude those who have never worked or have not worked recently, or other discriminatory impediments that preclude going to register. On the other hand, administrative records can overstate registered unemployment because of double-counting, failure to remove people from the registers when they are no longer looking for a job, or because it allows inclusion of persons who have some work. Registered unemployment data can serve as a useful proxy for the extent of persons without work in countries where data on total unemployment are not available and time-series of registered unemployment data by country can serve as a good indication of labour market performance over time, but due to mentioned issues the data should be used with care. Moreover, due the limitation in comparability to “total unemployment”, the two measures should not be used interchangeably.
Cross-country comparability issues
The issue of comparability of unemployment rates is particularly complex when looking at indicators for a large number of countries throughout the world. In an effort to resolve this issue for its member countries – and building on work carried out by the United States Bureau of Labour Statistics in the 1960s – OECD publishes “standardized unemployment rates” adjusted to ILO concepts. The ILO has further extended the OECD series in country coverage and number of labour force measures. These unemployment rates are from national labour force survey estimates that have been adjusted to make them conceptually consistent with the strictest application of the ILO statistical standards. This implies that participating countries and territories have provided detailed information on the composite elements of their labour forces. The unemployment rates obtained are based on the total labour force including the armed forces, while OECD standardized rates are now civilian-based. The rates are calculated from annual average estimates (or the period considered most representative over the year), thereby avoiding the variances that would occur if different reference periods were used. These unemployment rates, based on official national information, should provide the best basis currently available for making reasonable international comparisons and assumptions.
A significant amount of research has been carried out over the years in the important area of producing unemployment rates that are fully consistent conceptually, in order to contrast unemployment rates of different countries for different hypotheses. There are a host of reasons why measured unemployment rates may not be comparable between countries. A few are provided below, to give users some indication of the range of potential issues that are relevant when attempting to determine the degree of comparability for unemployment rates between countries. Users with knowledge of particular countries or special circumstances should be able to expand on them:
1. Different sources.
To the extent that sources of information differ, so will the results. Comparability difficulties result firstly from the already mentioned difference between sources measuring registered unemployment and total unemployment. But even when this is taken into account, the labour force surveys, official estimates and population censuses can still pose issues of comparability in cross-country analyses. Official estimates are generally based on information from different sources and can be combined in many different ways. A population census generally cannot probe deeply into labour force activity status. The resulting unemployment estimates may, therefore, differ substantially (either upwards or downwards) from those obtained from household surveys where more questions are asked to determine respondents’ labour market situation.
2. Measurement difference.
Where the information is based on household surveys or population censuses, differences in the questionnaires can lead to different statistics – even allowing for full adherence to ILO guidelines. In other words, differences in the measurement tool can affect the comparability of labour force results across countries.
3. Conceptual variation.
National statistical offices even when basing themselves on the ILO conceptual guidelines may not follow the strictest measurement of employment and unemployment. They may differ in their choices concerning the conceptual basis for estimating unemployment, as in specific instances where the guidelines allow for a relaxed definition, thereby causing the labour force estimates (the base for the unemployment rate) to differ. They may also choose to derive the unemployment rate from the civilian labour force rather than the total labour force or economically active population. To the extent that such choices vary across countries, so too will the information.
4. Number of observations per year.
Statistics for any given year can differ depending on the number of observations – monthly, quarterly, once or twice a year, and so on. Among other things, a considerable degree of seasonality can influence the results when the full year is not covered.
5. Geographic coverage.
Survey coverage that is less than national coverage – urban areas, city, regional – has obvious limitations to comparability to the extent that coverage is not representative of the country as a whole. Unemployment in urban areas may tend to be higher than total unemployment because of the exclusion of the rural areas where workers are likely to work, although they may be underemployed or unpaid family workers, rather than seek work in a nonexistent or small formal sector.
6. Collection methodology.
Sample sizes, sample selection procedures, sampling frames, and coverage, as well as many other statistical issues associated with data collection, may make a significant difference. The better the sample size and coverage, the better the results. Use of well-trained interviewers, proper collection and processing techniques, adequate estimation procedures, etc. are crucial for accurate results. Wide variations in this regard can clearly affect the comparability of the unemployment statistics. When viewing the unemployment rate as a gauge for tracking cyclical developments within a country, one would be interested in looking at changes in the measure over time. In this context, the definition of unemployment used (whether a country-specific definition or one based on the internationally recommended standards) does not matter as much – so long as it remains unchanged – as the fact that the statistics are collected and disseminated with regularity, so that measures of change are available for study. Still, for users making cross-country comparisons it will be critical to know the source of the data and the conceptual basis for the estimates. It is also important to recognize that minor differences in the resulting statistics may not represent significant real differences.
7. Differences in age-groupings
Although less important than other factors, mention should be made of differences in the age groups utilized, because the age limits applied for both youth and adults may vary across countries. In general, where a minimum school-leaving age exists, the lower age limit of youth will usually correspond to that age. This means that the lower age limit often varies between 14 and 16 years (and for some countries is even lower than 14, for example, Haiti at 10 years), according to the institutional arrangements in the country. This should not greatly affect most of the youth unemployment measures. However, the size of the age group may influence the measure of the young unemployed as a percentage of total unemployment. Other things being equal, the larger the age group the greater will be this percentage. In a few cases there is a larger discrepancy in the age limits applied. Six countries use 29 as the upper age limit: Colombia (1989-90), Costa Rica (1980-86), Honduras (1991-98), New Caledonia (1996), Panama (1983) and Suriname (1987). There are also differences in the operational definition of adults. In general, adults are defined as all individuals above the age of 25, but some countries apply an upper age limit. The upper age limit would obviously affect only the ratio of youth-to-adult unemployment rates and the effect is likely to be very small. Finally, mention should be made of the reference period of the information reported. Because there will be a substantial group of school-leavers (either permanently or for the extended holiday break) in the reported figures, the level of youth unemployment is likely to vary significantly over the year as a result of different school opening and closing dates. Most of the information reported relates to annual averages. In other cases, however, the figures relate to a specific month of the year (as with census data). The implications of the particular month chosen will vary across countries, owing to differences in institutional arrangements.
Household labour force surveys are generally the most comprehensive and comparable sources for youth unemployment statistics. Other possible sources include population censuses, official estimates and administrative records such as employment office records and social insurance statistics.
The ILO has made an intensive effort to assemble data on the indicators for as many countries, areas and territories as possible. Where there is no information for a country, it is usually because the country involved was not in a position to provide information for the indicator. Even when information for an indicator was available, it may not have been sufficiently current or may not have met other qualifications established for inclusion in the Key Indicators of the Labour Market, on which the information for youth unemployment is based.
In compiling the KILM, the ILO concentrates on bringing together information from international repositories. In other words, the KILM team rarely collects information directly from national sources, but rather takes advantage of existing compilations held by various organizations, such as the following:
International Labour Office (Bureau of Statistics)
United Nations Statistics Division
Organisation for Economic Co-operation and Development (OECD)
United Nations Industrial Development Organization (UNIDO)
Statistical Office of the European Union (EUROSTAT)
United Nations Educational, Scientific and Cultural Organization
United States Bureau of Labor Statistics (BLS)
Information maintained by these organizations has generally been obtained from national sources or is based on official national publications.
Whenever information was available from more than one repository, the information and background documentation from each repository was reviewed in order to select the information most suitable for inclusion, based on an assessment of the general reliability of the sources, the availability of methodological information and explanatory notes regarding the scope of coverage, the availability of information by sex and age, and the degree of historical coverage. Occasionally, two data repositories have been chosen and presented for a single country; any resulting breaks in the historical series are duly noted.
For countries with less-developed labour market information systems, such as those in the developing economies, information may not be easily available to policy-makers and the social partners, and even less so to international organizations seeking to compile global data sets. Many of these countries, however, do collect labour market information through household and establishment surveys, population censuses and administrative records, so that the main problem remains the communication of such information to the global community. In this situation, the ILO Labour Market Indicators Library (LMIL) programme can help. The LMIL is a system for sharing information between the ILO regional offices and headquarters. ILO regional offices are closer to the original micro-sources of data and have therefore been successful in filling in numerous gaps where data at headquarters – used in the production of the KILM – had not existed. It is an ongoing programme that continues to assist the KILM and other ILO publications and research programmes in the expansion of its country and yearly coverage of indicators.