Modeling retirement behavior of single individuals in the Netherlands

In this paper we estimate a dynamic programming model for retirementdecision in the Netherlands. Our model provides an empirical analysis of how the Dutch pension system rules a¤ect labor supply of older single individuals in the presence of uncertainty on income, health status and lifespan. We use a panel based on administrative and survey data which has the main advantage of having new accurate information on pension entitlements. The model is able to capture the main patterns of the data. We observe that individuals declinetheir labor supply considerably after their 60s and this varies by health status and income. Health status is a key determinant for the labor force participation.Healthy individuals tend to stay longer in the labor force than unhealthy ones and this is only slightly a¤ected by gender or origin. Individuals in the two upper percentiles tend to work longer than those in the other percentiles. In this paper we also simulate a potential reform of gradually increasing the retirement agefrom 65 to 67 years old. The simulation results show that individuals optimal behavior is to postpone retirement by 0.51 years, on average.