Two problems have occurred in the recent literature with regard to the empirical testing of the Balassa-Samuelson productivity based explanation of deviations from Purchasing Power Parity. First, empirical work has commonly focused only on the overall prediction of the model, instead of the underlying mechanisms. Second, the countries examined have, in almost all cases, been inappropriate given the underlying theory.
This study focuses on two papers that combat these problems. Their approaches are combined and extended to construct a more logical and thorough procedure for testing the Balassa-Samuelson model. Cointegration analysis is used in a three-stage testing procedure examining eight Asian Countries relative to the United States. First, the validity of Purchasing Power Parity is tested, then the general prediction of the Balassa-Samuelson model, and finally the underlying mechanism of the model is tested.
Support is found for the general prediction of the model, however in contrast to other recent studies no support is found for the underlying mechanism. This suggests two things: the theorised mechanism is not the true causative factor, and the general relationship tested here, and in several prior studies, is in fact misspecified. This relationship was consequently re-specified, the result being that less support was now found for the overall prediction.
