We reappraise the relationship between productivity and equilibrium real exchange rates using a panel estimation framework that incorporates a large number of countries and importantly, a dataset that allows explicit consideration of the role of non-traded, as well as traded, sector productivity shocks in exchange rate determination. We find evidence of significant correlation between real exchange rates and productivity differentials in both sectors. But our finding of a significant role for the non-traded sector in exchange rate determination, and of a relatively larger correlation between exchange rates and productivity shocks of a given size emanating from this sector, represent clear contradictions of the widely cited Balassa-Samuelson hypothesis. Our findings remain valid in the face of a number of robustness tests, including the exchange rate regime and numéraire currency.
Balassa-Samuelson Emerging market economies exchange rate panel data productivity