Exchange rate shocks and trade: A multivariate GARCH-M approach

We build on the recent literature studying the effects of uncertainty on trade by introducing a model that combines a reduced form vector autoregression for the growth rates of exports, foreign income, and the real exchange rate (RER), with a multivariate GARCH model. Up to 12 lags of several conditional standard deviations are added to relevant mean equations, and all parameters are estimated simultaneously using maximum likelihood, thus allowing us to avoid two step procedures that are common in the literature. Using a large data set of both developed and emerging countries, we find evidence that RER uncertainty negatively impacts trade for several less developed countries. We also find that RER uncertainty tends to be associated with a real currency appreciation. When we compute generalized impulse response functions to study the impacts of unexpected shocks to RER growth on export growth, the results are typically asymmetric. Positive shocks generate substantial negative responses while unexpected depreciations produce relatively smaller positive responses, especially in our developing country sample.