In this paper we apply the Behavioral Equilibrium Exchange Rate (BEER) approach developed by Clark and MacDonald (1998) to derive equilibrium real effective exchange rates and currency misalignments for the US and its major 16 trading partners over a sample from 1986Q1 to 2006Q4. Cointegration and panel cointegration techniques are applied to derive fully country-specific measures of misalignment and country-specific measures based on panel estimates. We do not only apply the popular first generation panel unit root and panel cointegration tests, but also two recently introduced classes of tests of the second generation: the CIPS (cross-sectionally augmented IPS) panel unit root tests by Pesaran (2007) as well as the error-correction-based tests for panel cointegration byWesterlund (2007), which both account for possible cross-sectional dependencies among the units included in the panel. Using the estimates obtained over a restricted sample, forecasting tests are conducted to assess the relative forecasting performance of pooled vs. heterogeneous estimators. We find that pooling the data delivers more reliable results when calculating equilibrium exchange rates though the implicit homogeneity restriction is statistically rejected. This result is especially remarkable, since we have given the heterogeneous estimator an ’unfair’ advantage by choosing the country-specific model (of up to 21 possible ones) with the best out-of-sample performance prior to comparing it to two final panel specifications. Based on this result, we re-estimate our preferred specification and calculate real effective exchange rate misalignments over the full sample. While we find strong evidence in favor of the Balassa-Samuelson-effect, evidence in favor of other commonly hypothesized fundamentals is weak.
|No. of pages: