Productivity-real exchange rate nexus : revisiting the Balassa-Samuelson hypothesis for emerging Asia.
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This dissertation deals with the economics of equilibrium real exchange rates in the context of Balassa-Samuelson hypothesis. For developing economies of East and South Asia (ASEAN and SAARC), trend deviations in real exchange rate from long-run PPP are tested against their biased sectoral productivity patterns, under the theoretical predictions of Balassa-Samuelson model. The thesis undertakes a study of three inter-related dimensions of productivity-real exchange rate linkage: (a) identification of productivity as a key determinant of permanent deviations in real exchange rate from long-run PPP, (b) validity of hypothesis under alternative theoretical specifications, and (c) inclusion of demand-side shocks to see if this can buy any support for the proposed model.
The novelty of this study lies with the careful theoretical as well as empirical examination, conducted to verify the long-run association between sectoral productivity real exchange rate movements. To assess the robustness of results, distinction between traded and non-traded sectors of the real economy is made in a more definitive manner. Data inconsistencies across sectors as well as across countries in the form of uncommon data sources, inconsistent scheme of sectoral division and inadequately disaggregated sectors are addressed to ensure data reliability. Furthermore, two alternative schemes of sectoral classification are employed to examine the sensitivity of model estimates. Three alternative measures of real exchange rate are used, that dominantly comprise of nontradables prices, so that the internal mechanism of the Balassa- Samuelson model could be captured appropriately. Three alternative theoretical specifications of the Balassa-Samuelson model are tested, ranging from the most restrictive domestic version of the model to the modified version, allowing for deviations in tradables prices from long-run PPP. The proposed model is also tested for its augmented version, by including a two demand-side determinants of real exchange rate, largely advocated in literature. Finally, various time-series and pooled data econometric techniques are applied for empirical verification, ranging from single equation to multivariate cointegration approaches, to test the consistency and robustness of estimates.
The results suggest that inter-country divergent sectoral productivity patterns do not exert any significant effect on the long-run real exchange rates for the ASEAN and SAARC countries in my sample, as predicted by Balassa (1964) and Samuelson (1964). The argument of the Balassa- Samuelson hypothesis that biased relative productivity of tradables at home will influence the overall price level of the country through nontraded sector prices and contribute to the long-run movements of real exchange rates does not hold valid. My findings are highly robust and successfully survived alternative sectoral classifications, different variants of real exchange rate measures, alternative theoretical specifications of the model and different econometric techniques. Empirical results for the standard (international) version of the hypothesis reveal that relative sectoral productivity differences across countries are inadequate in explaining the trend departures in the real exchange rates away from their long-run equilibrium. However, the Balassa-Samuelson hypothesis is more convincingly rejected when the domestic version of the model is tested. The results remain unchanged when the assumption of PPP, inherent to international Balassa- Samuelson model, is relaxed in favor of inter-country traded sector prices. This modified version of the model, allowing for trend deviations in international tradable prices from its PPP value, does not yield sizeable support, in favor of Balassa-Samuelson hypothesis for the sample countries. Furthermore, the empirical results reveal a clear departure of tradable prices from the long-run PPP, suggesting that this divergence is a potential reason for the non-existence of the Balassa-Samuelson effect. Finally, two demand-side factors, i.e., GDP per capita and government consumption spending, are augmented into Balassa-Samuelson model. However, their representation in the Balassa-Samuelson could gain only marginal support for the proposed productivity-real exchange rate relationship.
On the whole, I tend to reject the Balassa-Samuelson effect for emerging Asian countries due to inadequate empirical evidence in support of the hypothesis. Irrespective of alternative sectoral divisions, real exchange rate measures, model specifications or estimation procedures, sectoral productivity patterns are rarely found to cause significant real exchange rate appreciation in long-run for Asia.