Real exchange rates and fundamentals: robustness across alternative model specifications

Konrad Adler and Christian Grisse



JEL classification
C11, C33, F31, F32, F41

Equilibrium exchange rates, model uncertainty, model combination, panel data


This paper explores the robustness of behavioural equilibrium exchange rate (BEER) models, focusing on a panel specification with Swiss franc real bilateral rates as dependent variables. We use Bayesian model averaging to illustrate model uncertainty, and employ real exchange rates computed from price level data to explore robustness to the inclusion or exclusion of fixed effects. We find that the estimated coefficients - and therefore also the implied equilibrium values - are sensitive to (1) the combination of explanatory variables included in the model, (2) the set of currencies included in the panel and (3) the inclusion of fixed effects. Increases in government consumption and net foreign assets and improvements in the terms of trade in Switzerland relative to foreign countries are associated with a Swiss franc real appreciation, as predicted by economic theory. By contrast, several macroeconomic variables commonly thought to be linked to real exchange rates are found not to exhibit a robust relationship with Swiss franc real rates. Our findings can help policymakers in understanding the uncertainty associated with estimates of equilibrium exchange rates.