The Balassa-Samuelson Hypothesis Forty Years Later

It is well known that there are big cross-country differences in aggregate TFP. Are these differences uniform across sectors or are they driven by even larger TFP differences in specific sectors? Some forty years ago, Balassa and Samuelson hypothesized that the biggest TFP differences are in the tradable sectors. Providing empirical support for this hypothesis is hard because of the lack of data on sector inputs and outputs, at least outside the OECD. We get around this problem by employing economic theory to infer the missing information from the expenditure and price data of the 1996 Benchmark Study of the Penn World Tables. We distinguish between tradable and nontradable consumption and investment. We find that Balassa and Samuelson were right: the cross-country TFP differences are much larger in the tradable sectors than in the nontradables ones.