The evaluation of the experimentation clause in par. 6c SGB II aims at analysing the different forms of administering benefits for the long-term unemployed (“ALG II” recipients) by a consortium of the local labour office and municipalities (“ARGEn”), an approved municipal authority (“zkT”) or with separate responsibility of the local labour office and the municipality. Specifically, the evaluation will have to show which of these administrative arrangements is best suited to integrate benefit recipients into the labour market, to improve or preserve their employability and to stabilize them socially. The research agenda for the Ifo Institute was to evaluate the macroeconomic effects of different strategies employed by the different types of administration.
The evaluation project has been divided into four fields:
The Institute for Social Research and Social Policy (ISG) in Cologne was coordinating the overall project.
A meaningful comparison of the performance of the different forms of activating long-term unemployed has to take regional differences in the economic environment into account. In a first step of the evaluation, a regional comparison and benchmarking therefore was carried out. In a second step, the determinants of the performance were to identify in a given region. In a third and final step, we answer the central research question: “What would have happened at an aggregate level if either the ARGE-type organisation or the zkT organisation had been implemented nationwide?”
Since a fundamental labour market reform such as Hartz IV, which affected several million individuals, creates not only incentives for each individual to change his or her behaviour, but also leads to changes in the economic conditions, macroeconomic consequences and potential substitution and displacement effects had to be taken into account. Additionally, interactions between regions had to be addressed in the analysis.
The choice of an econometric identification strategy to evaluate macroeconomic effects was co-determined by the availability of data and specific features of the indicators for the different administrative arrangements that could be used for the evaluation. The following strategies were employed to identify the causal effects.
Due to the relatively small number of regional units (ca. 440) we abstained from using matching techniques to identify effects of the different administrative arrangements. Instead, we used an instrumental-variable approach taking care of the potential selection of regional units into one administration type or another.
Furthermore, we took regional interactions between the different regional units into account using up-to-date methods applied in current research. Most importantly, regional interactions have to be modelled explicitly, since the construction of larger regional labour markets would fail to capture the different administrative arrangements effectively used in each region.
We used all municipal-level statistics available at the Federal Statistical Office and the Federal Labour Agency. In addition, we utilized data on a variety of organisational characteristics of ARGEn and zkT surveyed in field 1 on an annual basis.
Descriptive regional comparison and regional benchmarking
Since the development of central indices for the measurement of “integration into employment” and the “improvement of employability and social stability” depends not only on the performance of the local job centres but also on several economic and geographic factors, regional differences have to be taken into account. As a starting point of our analysis we describe the regional differences for central indices differentiated according to the legal organisational form (ARGE and zkT) and according to different regional types.
Income perspective of long-term unemployed
Looking at the employment probability of a long-term unemployed would only be a partial measure of well being. A long-term unemployed measures his/her welfare by taking the income prospects of likely jobs as well as the expected duration of such jobs into account. Based on labour market matching models, the Ifo Institute developed the concept of the income perspective of a long-term unemployed that measures the weighted average income of a long-term unemployed over the incomes of different labour market states (long-term and short-term unemployment, regular employment and short-time employment) weighted by the probabilities of entering and exiting the respective states. Differences in income perspectives of long-term unemployed are mainly caused by differences in the probabilities of finding a regular or a short-time job. The average income perspective of a long-term unemployed in Germany is around 723 euros a month.
Improvement of employability and social stability
The analysis of the improvement of employability and social stability of long-term unemployed is based on a concept developed by field 3, where employability is measured by several individual characteristics. For the analysis on a regional level we used the aggregated data of 154 regions. The descriptive statistics show that on average all local job centres managed to improve the employability of the long-term unemployed.
Macro analysis using multivariate regressions
In the macro analysis we identify the casual effects that determine the development of the different labour indices under consideration. We take into account a wide variety of organisational features that are employed by the regional job centres.
The regression results for the number of long-term unemployed that find a regular job show a negative but insignificant indicator variable for the zkT organisation. A positive impact was found for the general case management, mainly used by zkT, where all clients are treated equally compared to a specific case management where only specific groups of workers are targeted. The ARGE organisations fare better by activating the long-term unemployed faster and monitoring them more intensely and if necessary sanctioning them more frequently. All these factors appear to be significant in the regressions and are able to explain the differences in the performance between ARGE and zkT organisations. Besides the analysis on the matching probability, we also look at layoff rates as a measure of sustainability. The impacts on the different transition rates help to explain the development of the total number of long-term unemployed. The estimates show that the zkT are, on average, less successful in reducing the total number of long-term unemployed. We also consider the development of the number of regular and short-time employed and find that zkT organisations are, on average, less successful in increasing the number of employed.
Analogously to the regressions on the impact on the integration rate into employment we look at the improvement of employability and social stability on the basis of the 154 regions for which aggregated individual data from field 3 is available. We find a small but positive effect for the indicator variable of zkT organisations after controlling for the different organisational features analogously to the regressions above. Thus zkT organisations seem to focus more on improving the long-term employability of the unemployed and focus less on fast integration into the labour market.
By using the above regression results we simulate the counterfactual cases of all local job centres in Germany having either been organised as zkT or ARGE. We do so for the different matching and layoff probabilities as well as for the number of long-term unemployed and the number of regular and short-time employed. The simulated difference in the number of long-term unemployed between the ARGE and the zkT counterfactual amounts to roughly 84,000 individuals within a year. The simulation results are then used to estimate the income perspective and the fiscal effects for the respective counterfactual cases.
Given the better performance of the ARGE organisations in bringing long-term unemployed into work it is not surprising that ARGE organisations are also able to provide a higher income perspective of around 30 euros per month (740 euros for the ARGE counter factual and 710 euros for the zkT counterfactual). This result is not driven by regional income differences, since we hold the incomes in all regions constant for the simulation. The only income differences that arise due to the active labour market policies of the zkT or ARGE organisation are the income losses due to stricter sanctions or the additional income due to a higher use of workfare programmes.
In the computation of the fiscal effects we first consider the state revenue losses or gains in labour taxes and social security contributions from lower or higher regular employment. The higher number of regular employed in the ARGE counterfactual lead to additional revenue of 2.3 billion euros a year compared to the zkT scenario. Secondly we compute the expenditure differences for the long-term unemployed by considering the living expenditure as well as the expenditure for housing and heating. The lower number of long-term unemployed in the ARGE scenario compared to the zkT scenario is estimated save around 800 million euros a year.
M. Werding, T. Büttner, H. Hofmann, C. Holzner and S. Munz (2009): ifo Beiträge zur Wirtschaftsforschung Band 33, Makroevaluation der SGB II - Grundsicherungsstellen, Munich (Abstract in German)
ifo Institut für Wirtschaftsforschung und Institut für Angewandte Wirtschaftsforschung (2008), Evaluierung der Experimentierklausel nach § 6c - Untersuchungsfeld 4: Makroanalyse und regionale Vergleiche, Abschlussbericht, München, 2008.
Kurzfassung: ifo Institut für Wirtschaftsforschung und Institut für Angewandte Wirtschaftsforschung (2008), Evaluierung der Experimentierklausel nach § 6c - Untersuchungsfeld 4: Makroanalyse und regionale Vergleiche, Abschlussbericht, München, 2008 (Download, 88 KB).
RSS