Can the Balassa-Samuelson theory explain long-run real exchange rate movements in OECD countries?

This study tests empirically the Balassa-Samuelson (BS) hypothesis using annual data for 12 OECD countries. New panel data cointegration techniques recently developed by Pedroni (2000) are applied and the results are compared with those obtained with conventional Johansen (1995)'s time series cointegration tests. Whereas standard time series approach turns out to be unable to put in evidence a significant long-run relationship is largely accepted for all countries using recent advances in the econometrics of non-stationary dynamic panels methods. This result doesn't mean however that the BS is uniformly supported by data for all OECD countries, since actually four of them (Australia, Belgium, Canada and the USA) are proved not to follow the BS path. Closer examinations of the three key components of the BS hypothesis enable one to identify clearly the causes of this empirical failure. It is found that the absence of a positive long-run relationship between real exchange rate and the relative prices of non-traded goods is the reason for this rejections. A possible explanation is that the PPP may not be confirmed for tradable goods in these countries.