A coding error, detailed below, affected some of the descriptive trends.All regression results, main findings and conclusions are NOT affected.The error occurred in the do-file 'over year figures', when computing the real median wage by year, age-group and education, the real log wage was used rather than simply the real wage.This error has affected Figure 2 and Figure 3 in the published paper, and 5 additional descriptive figures in the online appendix.In this document, we place the incorrect figures and the correct ones side by side below.The error has made the trend of graduate wage premium appear smoother and closer to zero.After correction, the trend is still flat over most of the sample period , with a small decline near the end of the period, and especially in the private sector.No other do-files are affected.Specifically, it did not affect any regression tables, including the analysis that rejects the two alternative models of technological changes.Hence the rest of the analysis in the paper are not affected by the error.We are extremely grateful to Professor Gianna Boero for bringing this error to our attention.
No takes yet. Share an insight, caveat, or question.
Blundell et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: