Despite extensive research efforts, the magnitude of the effect of employment protection legislation (EPL) on unemployment remains unclear. Existing econometric estimates exhibit substantial variation, and it is therefore difficult to draw valid conclusions. This paper applies meta-analysis and metaregression methods to a unique data set consisting of 881 observations on the effect of EPL on unemployment from 75 studies. Once we control for publication selection bias, we cannot reject the hypothesis that the average effect of EPL on unemployment is zero. The meta-regression analysis, which investigates sources of heterogeneity in the reported effect sizes, reveals the following main results. First, the choice of the EPL variable matters: estimates that build on survey-based EPL variables report a significantly stronger unemployment-increasing impact of EPL than estimates developed using EPL indices based on the OECD's methodology, where the latter relies on coding information from legal provisions. Second, we find that employment protection has a small unemployment-increasing effect on female unemployment, compared with a zero impact on total unemployment. Third, using multi-year averages of the underlying data tends to dampen the unemployment effects of EPL. Fourth, product market regulation is found to moderate the effect of EPL on unemployment.
Unemployment labour market institutions employment protection meta-analysis