Modeling heterogeneity in country-industry-year panel data: Two illustrative econometric analyses

Abstract

The macroeconomic empirical literature based on three-dimensional country- industry-year panel data, such as the widely used OECD STAN and EUKLEMS databases, has become extremely abundant, providing more observations for studies previously investigated on two-dimensional country-year panels. However, a large part of this literature does not take advantage of the development of panel data methods to deal with heterogeneity and dynamic and non-stationarity issues. We explain in this chapter how one can put them into practice and circumvent the lack of variability left in the data once one controls for simple and two-way interacted fixed effects in these industry-country-year panels. We illustrate what to do in the context of two econometric analyses, in which we first try to estimate the productivity impacts of ICT and R&D, and second the productivity impacts of product market anticompetitive regulations.