Should Welfare Administration be Centralized or Decentralized? Evidence from a Policy Experiment

ZEW Discussion Paper No. 10-106 // 2010
ZEW Discussion Paper No. 10-106 // 2010

Should Welfare Administration be Centralized or Decentralized? Evidence from a Policy Experiment

Germany’s 2005 welfare reform introduced two competing organizational systems for the administration and activation of welfare recipients. In the majority of the 439 German districts, a centralized organization was established where local employment offices and local authorities had to form a joint venture which is subject to the central directives and guidelines of the Federal Employment Agency (FEA). Within the joint venture, the FEA is in charge of the administration of benefits, job placement and the application of the main instruments of active labor market policy. The local authorities have the task to administer payments for housing costs and special needs. Moreover, they provide counseling in case of specific contexts such as lone parent status, home care for elderly or disabled relatives, or addiction to alcohol and drugs. In a policy experiment, 69 out of the 439 districts were allowed to opt out for a decentralized organization of welfare administration. In the decentralized agencies the local authorities autonomously organize welfare administration. They are responsible for the entire activation process, including counseling, the disbursement of benefits, job placement and the allocation of benefit recipients to active labor market policy programs. Thus, the FEA is not involved and decentralized welfare agencies can decide on their own on how to activate the welfare recipients. The decentralized welfare agencies are legally and organizationally independent from central directives and guidelines. All other components of public welfare and labor market policy – such as benefit entitlements, the taxbenefit system in general, and labor market institutions such as minimum wages and employment protection – apply equally to the centralized and the decentralized system of welfare administration. In this paper we evaluate the relative success of both systems. Our empirical analysis is based on a unique data set compiled from in-depths surveys of welfare administration, register data of the FEA, comprehensive surveys of welfare recipients and extensive regional information. The data contain information on organizational and strategic features of welfare agencies which is usually unavailable. They contain information about a sample of individuals who received welfare benefits in October 2006 and who are followed until December 2007. We find that decentralized welfare administration has a negative effect on the chances of male welfare recipients to take up employment providing a sufficient living income and leading to exit from welfare receipt. Given the low transition intensity from welfare receipt into employment in general, the magnitudes of the effects are substantial. Compared with centralized welfare agencies, the integration rate of decentralized welfare agencies is up to 25 percent lower during the observation period. For women, we also find negative effects, which are, however, statistically insignificant. No significant differences are found when considering employment with supplementary welfare receipt as the outcome. Based on these results, we conclude that centralized standards and routines work effectively for individuals without further obstacles to employment. Once additional problems, such as a lack of child care facilities, come into play, standardized placement loses its effectiveness. In particular, tailormade solutions to individual needs often require the cooperation of local actors. This is why, in our opinion, the centralized model has no significant advantage over the decentralized model in case of female welfare recipients.

Boockmann, Bernhard, Stephan Lothar Thomsen, Thomas Walter, Christian Göbel and Martin Huber (2010), Should Welfare Administration be Centralized or Decentralized? Evidence from a Policy Experiment, ZEW Discussion Paper No. 10-106, Mannheim.

Authors Bernhard Boockmann // Stephan Lothar Thomsen // Thomas Walter // Christian Göbel // Martin Huber