Long-term unemployment scarring and the role of labour market policies : The case of Sweden in the 1990s

Sammanfattning: The experience of unemployment puts individuals at risk of long-term negative scarring and the longer the unemployment spell, the greater the risk of negative scarring. In Sweden, labour market policies aim at reducing such risks in the form of unemployment benefits, active matching and active labour market policy programmes (ALMPs). However, there is frequent discussion regarding the extent to which these kinds of policies actually reduce the risk of negative scarring. It is often argued that the programmes are of poor quality, particularly during economic downturns, and participants are often not motivated for the task. Likewise, it is claimed that unemployment insurance tends to counteract a quick return to the regular labour market. One problem related to labour market policies is that it has been difficult to examine the impact of such policies. Studies often present results that appear scattered due to differences in what is actually being measured and methodological problems. The uniqueness of this thesis is that it is based on a large-scale longitudinal register of data that has provided important empirical information regarding the long-term effects of labour market policy investments. The quality of data has also enabled the use of evaluation techniques which largely can help to reduce the uncertainty of the findings. More precisely, the research questions examine (1) in what way the level of unemployment benefit functions as protection against unemployment scarring, (2) in what way the ALMPs protect long-term unemployed people from long-term unemployment scarring, (3) at what point in a business cycle the ALMPs are efficient and finally, (4) for whom do the ALMPs function to reduce the risk of negative scarring. In this thesis, scarring effects are measured as the risk of labour market exit, the risk of labour market instability and the risk of future negative wage trajectories. The methods used in most studies are Cox regressions in combination with instrumental variable analysis (the Heckman two-step procedure). The empirical findings indicate that ALMPs worked well to reduce such negative effects both in times of booms (1999) and recessions (1993) and particularly among the youngest and oldest actors on the labour market. They also function particularly well for people with a low level of education. However, it is important not to exclude unemployed people who have a high level of education, in the belief that ALMPs have nothing to offer them, since such people are particularly helped by ALMPs as regards reducing the risk of future labour market instability. It was also found that generous unemployment benefit helped to reduce the risk of future negative wage scarring. In addition to these findings, some mechanisms were identified which proved to be important tools for transforming policies into valuable resources for the unemployed. In this thesis, the value of the findings of these mechanisms is discussed from the perspective of the capability approach. Even if the same investments were made in all unemployed persons, the participants would respond differently to the investment. Some reasons for the inequality in outcomes were found within the programmes and were due to heterogeneity in the unemployment group but some reasons can actually be explained by the converters (mechanisms) that were identified in the studies. Thus, the results emphasise the importance of investing in labour market policies, particularly during economic downturns. This is the time when cuts in unemployment benefit do not help the unemployed back to the labour market since there are very few available jobs to apply for. It is also the time when the long-term unemployed should participate in ALMP-training in order to be prepared for new challenges when the labour market improves again. As a matter of fact, the results show that skills from ALMP-training have a bridging effect which indicates that these skills will be valuable on the labour market for at least another five years after the year of investment. The findings in this thesis are controversial since they differ from most research findings from the beginning of the 1990s which point to poor micro level outcomes. However, the long-term approach of this thesis is the main explanation for these new and different results.  It is argued here that a long-term approach is needed to find out the long-term effects because ALMP participation, particularly ALMP-training, is meant to be a long-term investment in human capital. A long period of time needs to pass between ALMP-investment and evaluation before the effects can show. Reported effects from ALMP investments at the beginning of the 1990s have often been measured on a short-term basis. It is not suggested that short-term effects should be ignored but it is argued that a short-term analysis provides only a fragmental description of reality, and long-term effects should be given greater priority than is usually the case since they affect the labour market prospects of the individuals over a long period of time. This thesis dispels the “myth” about the negative effects generated from ALMPs during the 1990s.