Wednesday 16 October 2019, 11.00AM to 12.00 pm
Speaker(s): Nikolas Mittag (CERGE Charles University)
Abstract: How potential benefit duration (PBD) of unemployment insurance (UI) affects wages and matching is crucial to evaluate PBD extensions as a policy tool and to understand the causal relation between unemployment, job search and wages. Recent studies provide evidence from quasi-experiments, but disagree even on the sign of (local average) wage effects. We re-visit the regression discontinuity design of Schmieder et al. (2016), but use more detailed data and a wage decomposition to re-analyze the effects of PBD in a framework that allows for unrestricted heterogeneity of both duration and wage effects. Our (preliminary) results indicate treatment effect heterogeneity that casts doubt on simple mechanisms and complicates learning from (local) average effects. Specifically, we first show that duration effects are heterogeneous, which confirms that treatment changes dynamic selection. Our results so far suggest that PBD almost exclusively prolongs unemployment spells ending close to exhaustion points. We examine pre-determined wage components to (partly) separate dynamic selection from dynamic treatment effects. Dynamic selection is potentially non-monotonic and may create spurious treatment effects in our sample. We find that PBD affects wages only through the firm fixed effect, which adds to evidence on the importance of firms. It suggests that wage losses are likely due to firm attributes (such as bargaining power) rather than individual attributes (such as productivity). Purging (parts of) dynamic selection from re-employment wages shows that the steep wage decline is mainly driven by the firm fixed effect, but also by time-varying unobservables. The marked exhaustion effects have at most a small impact on the effect of PBD on wages. Wage effects appear to accumulate through employment at lower-paying firms at short unemployment durations. Our current results are work in progress, but demonstrate that heterogeneity in both duration and wage effects alters the interpretation of common analyses and thereby potentially reconciles diverging findings. We develop tools to make progress on important questions in the presence of essential heterogeneity.
Admission: All welcome