The Measurement of "True" Unemployment in Austria
With EU membership Austria has become obliged to adopt the international definitions and concepts for the calculation of the unemployment rate. In 1994 the questionnaire of the microcensus was adapted to the labor force concept used by international organizations. Since the unemployment rate according to the traditional Austrian sources (registered unemployment and social security employment) differs greatly from that according to the household survey (3.7 percent versus 5.9 percent in 1995), a closer analysis of the differences is warranted. The large difference in unemployment between the household survey and administrative data (1995 215,700 versus 143,700) turns out to be the result of differences in demarcation lines between unemployment, employment and inactivity. If one excludes layed off workers (51,000) who have a job to return to after some waiting period (usually seasonal workers) and casually employed registered unemployed from unemployment (13,600), as is required by the labor force definition of unemployment, the difference between the microcensus and administrative data becomes rather minor. There remains, however, an aspect which neither administrative data nor the household survey can take into account satisfactorily, i.e., the development of employment on one's own account (1995 371,000 versus 537,200). The rising number of persons doing casual work with or without social security coverage, who work on a contract basis or in some other alternative form of employment, a common feature of labor markets in developed western industrial societies, is difficult to capture in administrative data like the Austrian which is based on social security coverage. The microcensus, due to the small sample size and the relatively volatile population in question, is also not well suited to offer reliable information. Survey design, methodology and interviewer briefing have weaknesses which so far have not been assessed properly. Thus, the emergence of marked discrepancies in employment and unemployment between administrative data and the microcensus on the one hand and the census of 1991 on the other have not prompted a redesign of the household sample and weighting scheme. This suggests that the fluidity of the employment pattern at both the bottom and top end of the income and skill scale cannot be followed up efficiently by the current statistics. Therefore the unemployment rate calculated on the basis of the labor force concept – an exercise also possible with administrative data, which has the advantage of timely availability – may well underestimate the labor resources which may be activated by an improvement of economic conditions and by the same token underestimate the extent of poverty and social discontent.