Testing Symmetry and Proportionality in PPP: A Panel Data Approach

Abstract

This paper applies a previously unexplored econometric approach--the hierarchical regression model (Lindley and Smith, 1972)--to test symmetry and proportionality in PPP. With 23 years of data for over 26 countries, our Bayesian estimation and model comparison results reject the symmetry and proportionality conditions implied by PPP at all sampling frequencies, but are consistent with the two conditions holding for individual countries at the quarterly and annual frequencies.

Previous tests of PPP have been criticized for their low power. Using a panel data hierarchical model approach, we can increase the cross-sectional variability in the data and, as a result, obtain potentially more precise parameter estimates.

Some implications of our findings in this paper are as follows. First, our results re-confirm the role of cross-sectional variation in obtaining more powerful tests. Second, we regard the symmetry and proportionality conditions as a multi-period, multi-country proposition and, hence, our hierarchical model approach allows for countrywise deviation from these two conditions. This observation complements the point made by Hakkio (1984) that short-run deviations from PPP should be allowed for. Third, more importantly, given that the symmetry and proportionality conditions are rejected at the monthly frequency in the paper, our study casts doubt on the appropriateness of unit root tests on the monthly real exchange rates.

Note: This is a description of the paper and is not the actual abstract.

JEL Classification: F31, C11, C33