Robustness of econometrically estimated local multipliers across different methods and data

Abstract

AbstractLocal traded‐goods employment multiplier estimates play an important role in the decision making of local policy makers. Therefore, it is important to understand the robustness of the approaches and quality of the data used in empirical studies. The local employment multiplier estimates by Moretti (AER; 2010) are a good benchmark for departure. In this paper, I find the traded‐goods multiplier falls within the range of 1.17 to 1.93, which is substantially lower than the estimate of 2.6 obtained by Moretti. I use multiple estimators in combination with two data sets, a range of controls and different ways of classifying traded industries to show the robustness and variability of local employment multiplier estimates. Finally, I demonstrate the sensitivity of Moretti's results. The most notable result being that Moretti's multiplier for separate skilled or unskilled workers in the nontraded sector is overestimated by a factor of 2.