By Yamin Ahmad, Ming Chien Lo and Olena Mykhaylova
This paper investigates the time series properties of real exchange rates series produced by DSGE models. We simulate a variety of new open economy DSGE models that incorporate features such as local currency pricing, home bias, non-traded goods and incomplete markets. We attempt to ascertain whether the dynamics of the real exchange rate in this class of models are consistent with those found in the time series literature using data from the current floating period. Although none of the basic specifications we consider match the volatility in the raw data, our findings suggest that home bias in consumption and non-traded goods are the key components of DSGE models that are able to generate persistent real exchange rates, comparable to that in the data. Moreover, we find that some of the structural micro-level nonlinearity embedded within DSGE models may be represented as macro-level nonlinearity in the form of a smooth transition autoregressive process, which has previously been found to parsimoniously characterize the dynamics of real exchange rates in the time series literature.
Explaining exchange rates has always been a challenge, even for purely statistical models. It is to be expected that standard fare open economy DSGE model will have difficulties as well, as they assume a world with far less frictions than there actually are. Previous attempts have only tried to introduced one imperfection at a time. This paper tries to look at all in one go and finds that we have still substantial ground to cover, but the paper can suggest some more promising avenues.