Measurement error in schooling data: the OECD case

The effect of human capital is difficult to estimate in cross-country analysis due to measurement errors in schooling data. Using data for OECD countries over the period 1960–1990, the reliability of the widely used Barro and Lee's data set and also the new De la Fuente and Domenech's data set is analysed. Results show that both suffer from measurement errors, but the latter seems to reflect the rate of growth of schooling more accurately, especially when taking long differences.