While there are numerous studies on the measurement of the technical or cost efficiency of local governments, the analysis of the main drivers of this efficiency has attracted far less attention in this literature. However, information about the main sources of (in) efficiency are not unimportant, since they can provide useful information for (local) policy-makers. Therefore, the aim of this paper is to add to the existing literature by studying the determinants of local governments' efficiency. For this purpose we focus on one particular area of public goods provision, namely the construction and maintenance of (county) roads, instead of the local government as a whole. This avoids, at least to a certain degree, the problem of defining a comprehensive set of reasonable input and output indicators fully describing local governments' activities, and therefore reduces potential sources of measurement errors. Our empirical analysis is based on a broad panel of German counties rather than a crosssection used in the most previous studies which investigated the determinants of efficiency. The German counties are an interesting case to study, since one of the main tasks of these counties is the construction and maintenance of county roads. In order to determine the main drivers of efficiency in road maintenance, we first calculate efficiency indices using non-parametric (Data Envelopment Analysis) and parametric (stochastic frontier analysis) estimation techniques. In a second step, we examine how these efficiency indices can be explained by a number of exogenous (e.g. socio-economic, fiscal, political) variables. More specifically, we apply and compare four different estimation approaches to detect the potential sources of (in) efficiency. This broad variety of estimation approaches allows us to test and check the robustness of the derived results. The results of our analysis show that (controlling for numerous characteristics of the counties) the disposable income of the counties' citizens, intergovernmental grants for county roads, and the payments to the counties influence technical efficiency negatively. For the political variables we find weak evidence that efficiency decreases with an increasing share of seats of left-wing parties in the county council; on the other hand, the hypothesis that efficiency decreases with the degree of political concentration in the county council could not be confirmed. Moreover, the analysis shows that not all of the four estimation approaches are consistent concerning the sign of the coefficients as well as the (statistical) significance. Therefore, focusing on just one estimation method, as it was done in the most previous studies investigating the determinants of local governments' efficiency, may be deceptive. To test the robustness of the results, different estimation approaches should be applied and compared
