My current research focuses on:
- Dynamic microeconometric methodology, dynamic treatment evaluation
- Crime economics, forensic economics and tax evasion
- You will also catch my attention with an interesting paper in development, labor or education.
This paper evaluates a job search assistance program for unemployment insurance recipients. The assignment to the program is dynamic. We provide a discussion on dynamic treatment effects and identification conditions. In the empirical analyses we use administrative data from a unique institutional environment. This allows us to compare different microeconometric evaluation estimators. All estimators find that the job search assistance program reduces the exit to work, in particular when provided early during the spell of unemployment. Furthermore, continuous-time (timing-of-events and regression discontinuity) methods are more robust than discrete-time (propensity score and regression discontinuity) methods.
Work in Progress:
Randomization in Dynamic Settings: Estimating Ex-Ante and Ex-Post Treatment Effects (joint with Bas van der Klaauw)
This paper considers the evaluation of a policy effect in dynamic settings when there is an initial randomization to treatment and a treatment at a later time. We assess the underlying assumptions and relative performance of several discrete and continuous-time microeconometric treatment evaluation techniques when actual treatment depends on dynamic selection. We provide a discussion on defining parameters of interest in different policy settings with ongoing entry in treatment and anticipation behavior. We compare the commonly applied 2SLS to dynamic IV methods and propose a new method to estimate ex-ante and ex-post dynamic treatment effects when agents adapt their behavior prior to entering treatment. To illustrate the performance of the competing methods we present a simulation study using a simple structural dynamic model. The results underline the sensitivity of IV methods, and in particular of 2SLS, to the choice of population entering the control group. Our dynamic non-parametric and dynamic least squares methods produce reliable estimates of ex-ante, ex-post and total effects in our simulation when agents anticipate treatment.
The Effects of Auditing Announcements on Tax Fraud (working title)
In the Netherlands, the tax authorities announce at the beginning of the tax-filing period a specific component in tax reports which will receive special attention when auditing. Theoretical models predict that rational taxpayers will reduce misreporting on these components in response to stricter auditing. We use detailed administrative data from the Dutch tax authorities (Belastingdienst) to detect unusual patterns in tax reports due to two announcements. We find increased declarations of secondary property, residual assets and freelance income not subject to third-party reporting. These increases are not always directly visible on the tax item targeted by the announcement. When there is a possibility to declare returns in different sections of the tax returns, the increases reveal themselves in other overlapping topics. The substitution patterns suggest that taxpayers try to reduce their declarations in the announcement topic whenever possible and declare previously underreported income and wealth across other topics in an effort to minimize their tax burden. When shifting declarations to other sections is not possible, taxpayers increase their declarations in the topic targeted by the auditing announcement.
Street Prostitution Zones and Crime (joint with Paul Bisschop and Bas van der Klaauw)
This paper considers the effect of legal street prostitution zones (tippelzones) on crime in the 30 largest cities in the Netherlands. Using a difference in difference model we evaluate the opening and closing effects of tippelzones on both reported and perceived crime. We find that open street prostitution zones are associated with a decrease in registered crime but an increase in perceived crime. These results are mainly driven by strong closing effects. For registered crime, the increases are noticeable after a year while the reduction in perceived crime is strongest immediately after the closing of a tippelzone.