Exchange rates as well as relative price level and output movements are decomposed into components associated with nominal shocks as well as shocks to aggregate supply and aggregate demand. In contrast to previous analyses of such decompositions based on statistical vector autoregression (VAR) analysis, this study takes as a starting point a simple textbook model of exchange rate determination, augments it by allowing for suitably defined random shocks and transforms it into a triangular format resembling the identification procedure of the VAR methodology. Applied to major bilateral exchange rate series, the decomposition suggests that exchange rate variability is mostly driven by shocks to aggregate demand, partcularly in the longer run. Overall, the evidence is roughly in line with previous decompositions obtained from statistical VARs.
exchange rates vector autoregression nominal and real shocks