Two essays on real exchange rate movements

Abstract

This thesis explores the PPP for tradable goods at the sectoral level for OECD countries and identifies whether the failure of PPP for traded goods is responsible for the weak correlation between the overall real exchange rate and the relative prices of nontraded-goods (Betts and Kehoe, 2008). In addition, we investigate the comovement of real exchange rate and the relative price of non-traded goods by using various cointegration techniques. The study involves 19 sectors in 12 OECD countries over the period from 1980 through 2005. PPP for tradables is first tested following the classification in a paper by De Gregorio et al (1994). Further, the tradability of a sector is determined in a more precise way - based on how much the volume of sectoral trade is accounting for the total amount of trade - and then PPP is examined according to this new classification. Finally, according to the new classification, an implicit price index is constructed for examining PPP at the aggregated level. I first test the long-run equilibrium relationship between the domestic price of traded goods and the foreign price of traded goods measured in the domestic currency by the conintegration test. More than one third of results suggest that both price series are cointegrated, but the hypothesis of the unit cointegrating vector is rejected in most of the cases. Secondly, the unit cointegrating vector is imposed so that it’s the stationarity of the real exchange rate for tradables being tested in the sectoral level. There are less than one fifth of the sectors in which the unit root hypothesis is able to be rejected. In order to increase the power of the test, I further apply various panel unit root tests which provide stronger evidence of the stationarity of the real exchange rate. Finally, in order to examine the anomaly proposed by Betts and Kehoe (2008), I divide the country sample into two groups: one belongs to EU and the other does not, and make comparison of the results between two groups. I am unable to find significant evidence of the anomaly in the more disaggregated data. The results based on the new classification are more or less the same as those on the conventional classification. Nevertheless, in the aggregated level, the evidence is less supportive. One possible reason may come from the aggregation bias, which induces a positive bias in persistence estimate. Taking account of the empirical validity detected, it's possible to investigate the long-run comovement of the real exchange rate and the relative price of non-traded goods. Instead of the variance approach, the cointegration test is employed to investigate the long-run evolution of the real exchange rate. The results suggest that there exists a long-run relationship between the real exchange rate and the relative price of the non-traded goods.