The BAFPLAN-system is a static microsimulation model that has been used on the German Federal law on Education (BAFÖG) by the Federal Ministry for Education and Research (BMBF) for 40 years. It was developed by the Society for Mathematics and Data Processing (GMD) (nowadays called the Fraunhofer Institute for Applied Information Technology FIT) between 1975 and 1977 and has been constantly cultivated since then. The model mostly focuses on estimation of costs and financial needs, legislation amendments and the number of sponsored people.
BAFPLAN’s basis is a special sample that includes all aspects that are relevant for BAFÖG so that a precise prediction of fiscal impacts is possible which would not be feasible with any other existing data source. The sample consists of approximately 300 characteristics for every data point all of them being necessary to calculate whether a specific person is entitled to receive the educational grant. The characteristics include but are not limited to the income both of the parents and the applicant, various aspects about the educational institution and the accommodation of the applicant and his or her siblings.
Apart from information about sponsored people, the data base also contains information about applicants whose applications were unsuccessful under the current legislation. This allows us to forecast how many currently unsuccessful applicants would be entitled to the educational grant under various legislation amendments. Furthermore, the coverage of non-sponsored people makes it possible to assess to what extent the existing entitlements of the grant are exhausted. It appears that the exhaustion of the grant is positively correlated with its amount. BAFPLAN considers this correlation which is why the model goes beyond the scope of a classis static microsimulation model that only examines »first round« effects.
The sample is constructed by using anonymized data on sponsored people from federal data centers that contain all the relevant characteristics. Hence, the microsimulation model currently draws on a sample of approximately 660 thousand data points.
In the context of the BAFPLAN model, the parental tax burden is determined for every data point. Consequently, we can approximately examine the effects of tax cuts. In theory, every tax abatement increases net income which leads to a decrease in the claims of the grant. In addition, the detailed number of parameters allows us to precisely assess the effects of changes of rental costs, educational participation or the extent of exhaustion.