Real Variables, Nonlinearity, and European Real Exchange Rates

In this paper we carry out an analysis of European real exchange rate behavior before and after the implementation of Economic and Monetary Union (EMU), that is, the single European currency, in January 1999. In particular, we model real exchange rates for a number of EMU and non-EMU countries against Germany in an explicitly nonlinear framework and allowing for variation in the equilibrium level of the long-run equilibrium real exchange rate using either relative productivities or real diffusion indices. The relative productivity specification derives from the well-known Harrod-Balassa-Samuelson effect (Harrod 1933; Balassa 1964; Samuelson 1964). According to the Harrod-Balassa-Samuelson effect, countries with rapidly expanding economies should tend to have rapidly appreciating real exchange rates. However, while the Harrod-Balassa-Samuelson effect focuses on a few series in order to explain the equilibrium level of the real exchange rate, the long-run equilibrium real exchange rate may be affected by a wider range of real variables in the macroeconomy. Including the wide range of available real variables in an econometric specification, however, raises a number of practical problems for a modeler, notably the lack of degrees of freedomaswell as potentialmulticollinearity. One way of circumventing this approach is to construct diffusion indices or factors that capture the core variability in a set of macroeconomic time series in a parsimonious fashion (Stock andWatson 1998, 2002a, 2002b; Bernanke and Boivin 2003; Bernanke, Boivin, and Eliasz 2005). In this paper we explore both approaches. The remainderof thepaper is organized as follows. In Section IIwe examine the underlying rationale for nonlinear real exchange rate adjustment, and in Sections III and IVwe briefly discuss theHarrod-Balassa-Samuelson