Unemployment Duration and Business Cycles in Finland
The recession of the early 1990s caused a serious unemployment problem in Finland. The number of unemployed increased dramatically and the share of long-term unemployed became nearly tenfold. This study analyses the background of variation in unemployment duration using individual data from 1987 to 2000. The dataset consists of a large random sample of Finnish workers entering unemployment. The main focus is on the relative contribution of macroeconomic conditions and compositional variation to unemployment duration. Generally, it is assumed that individuals who are laid off during a recession have weaker characteristics considering re-employment possibilities compared with other times. In addition, it is studied how the different type of people are influenced by the business cycle, and have the factors affecting unemployment duration changed during the period. Also the question of duration dependence is examined. Duration depedence means how the elapsed length of unemployment changes the possibility of employment. In most countries, the effect is negative. The magnitude of duration dependence is important for the design of the labor market policy. The analysis is done using a proportional hazard model with a non-parametric specification for the baseline hazard. The regional unemployment rate is used to capture the macroeconomic conditions. Separate models are estimated for shorter time periods in order to study the changes in the parameter values as well as to take into account the non-proportional effects between the time periods. The relative effects of compositional and business cycle components are identified following Rosholm’s (2001) approach. Based on the analysis it is concluded that there are asymmetric effects between the periods. The shape of the baseline hazard changes and some individual parameters have trends in the effects. This is probably explained by the structural changes in the economy that took place due to the recession. There seems to be strong negative duration dependence and only a relatively small effect of heterogeneity on the baseline hazard. Evidence against the ranking model is found because the duration dependence is not affected by the level of unemployment. The effect of compositional variation is found to be important. The results suggest, surprisingly, that the characteristics of unemployed became better during the recession. Again, the severity of the recession might explain this if it caused workers to be laid off more randomly compared with other times.
Keywords: Unemployment, duration, business cycle