Estimating retail gasoline price dynamics: The effects of sample characteristics and research design

Abstract

The study shows that much of the variation in the findings of the literature on retail gasoline price dynamics is systematic rather than sample variation from using different data. Estimates of pass-through rates depend systematically on research design and features of the data, such as the sampling frequency, the choice of upstream price, whether taxes are included or not, the sample length, and the postulated lag structure. In addition, there are systematic differences between time periods and countries. Using a 20 year-long dataset of 28 European Union countries we quantify the extent of estimate variation that arises from the choice of data structure from that arising from temporal and country heterogeneity and sampling variation. We also show that country heterogeneity itself has systematic components, with wealthier countries experiencing slower adjustments. Our results inform the interpretation of results on pass-through rates derived from Error Correction Models. They are also of relevance for the broader literature estimating the transmission of price shocks in the economy.