Measurement error in cross-country productivity comparisons: Is more detailed data better?

Relative productivity levels are used intensively in analyzing cross-country growth, but often based on crude measures and with little information on their reliability. In this paper, we provide a new framework to estimate purchasing power parities and productivity levels with associated standard errors using the country-product-dummy (CPD) method. For a set of OECD countries, we show that productivity levels in manufacturing industries are measured with sizeable error. We also show that cruder productivity measures are often poor proxies for the data-intensive measure presented here. This evidence can be used to deal with the problem of attenuation bias in cross-country regressions.