Imported Inputs and Productivity Spillovers from Foreign Direct Investment

Abstract

This paper considers how input market liberalization affects host country productivity spillovers from multinational corporation (MNC) investments. The standard “Backward Linkage” measure used to estimate technology and learning spillovers to local upstream suppliers — pioneered by Javorcik (2004) and replicated across several influential papers — implicitly assumes domestic and foreign firms share the same input structure. I show that this assumption constitutes an omitted variable bias of imported inputs in TFP spillover estimation. Using a novel Colombian firm panel that isolates imported from domestic inputs, mean backward linkage productivity spillovers reduce in half when the share of locally sourced inputs is adjusted to reflect MNCs’ observably higher propensity to import inputs. However in some industries, productivity spillovers increase in response to the adjustment. I demonstrate that the sign and magnitude of this bias are proportional to the elasticity of substitution between imported and domestic inputs. The results highlight how input market liberalization (usually associated with increased FDI inflows) can have important feedback effects on local productivity spillovers from MNCs.