Productivity spillovers from multinational corporations: vulnerability to deficient estimation

Abstract

Evidence on productivity spillovers from FDI to domestic firms is ambiguous. Incorrect estimation procedures may be one of the sources for the contradictory results obtained in empirical studies on this subject. We observe that inadequacy of the estimation procedures leads to a severe underestimation of the spillover effect. An appropriated econometric methodology is discussed taking into consideration the possible simultaneity of FDI and other explanatory variables and endogeneity related to firm unobserved heterogeneity. Robust inference is also addressed. Our findings for the Portuguese case seem sufficient clear to warn about spillover results obtained with a non-judicious application of the classical panel data methods.