Exchange rate predictability with nine alternative models for BRICS countries

Abstract

This paper seeks to add to the literature on short-run exchange rate predictability by focusing on BRICS exchange rates. We utilize both time-varying and constant parameter models, and account for a variety of macro fundamentals, including those suggested by Taylor rules. For the sample of countries, a Taylor rule model with homogeneous coefficients without interest rate smoothing as well as PPP- and UIRP-based models offers a better exchange rate predictability than the random walk model. For this sample of countries, constant parameter models appear to predict exchange rates better than time varying approaches for monthly data but not quarterly. We provide a range of sensitivity analysis, which on some occasions influence the paper's conclusions. We rationalize these results as being partly due to relatively stable relationship among variables (fixed over time-varying) and sample size (monthly over quarterly).