Since 2004, most municipalities in the Netherlands receive lump sum payments from the state for the payment of social assistance allowances. As municipalities had no authority to change the eligibility rules for social assistance, the effects of the welfare reform are solely due to the efforts of municipalities to decrease the number of welfare recipients. Using variation in the timing of policy changes, this paper uses a difference-in-difference approach to assess the effectiveness of the incentive for municipalities. Based on individual panel data from administrative records, we show that the high-powered scheme led to a decline of the welfare caseload of 14% up till 2008. The reform has been most effective for those with the highest welfare dependency: single mothers and singles from non-western origin. In line with standard economic predictions, the reform does not give an incentive for cream skimming: the welfare caseload declined as well for easy to place recipients as for difficult to place recipients.