This paper presents an econometric analysis of curtailment costs of renewable energy systems in Germany. The study aims at explaining and quantifying the regional variability of renewable energy source curtailment, which is a measure to relieve grid overstress by temporarily disconnecting RESs from the electricity grid. We apply a Heckit sample selection model, which corrects bias from non-randomly selected samples. The selection equation estimates the probability of occurrence of RES curtailment in a region. The outcome equation corrects for cross-sectional dependence and quantifies the effect of RESs on curtailment costs. The results show that wind energy systems connected to the distribution grid increase RES curtailment costs by 0.7% per MW or, respectively, 0.2% per GWh, respectively, in subregions that have experienced RES curtailment over the period 2015-2017. The implication of this finding is that policymakers should set price signals for renewables that consider the regional grid overstress.