The Impact of Length of the School Year on Student Performance and Earnings: Evidence from the German Short School Years

256 state-grade-year cells for grade repetition, 70 for Gymnasium entry; survey and social-security earnings samples

West German school cohorts affected by the 1966-67 shortened school years, followed into adult earnings

Economic Journal 117(523); full text read from NBER working paper 9964 (September 2003). West Germany moved the start of the school year in 1966-67, and the affected cohorts covered the same curriculum in about two thirds of a year less schooling. Difference-in-differences across cohorts, states and secondary-school tracks. THE OUTCOMES ARE ADMINISTRATIVE AND LONG-RUN - grade repetition, entry to the academic track, and adult earnings - not test scores, which makes this the only source in the topic that prices instructional time in the labour market. The author's own scope warnings matter: the reform compressed a fixed curriculum rather than cutting content, so compensatory effort cannot be ruled out; German wages are union-regulated, which compresses any earnings effect; and 'Germany has one of the longest school years in the developed world', so this is a cut from a high base.

Losing two thirds of a school year raised grade repetition by 0.78 to 0.96 percentage points against a base rate of 3.89% - a relative increase of roughly 20 to 25%, statistically significant. It did NOT change entry to the academic Gymnasium track (0.004, 0.017 and 0.010 across specifications, all insignificant). And it did not change adult earnings: the survey estimate is -0.018, 'almost a 2 percent loss in earnings for each year less in school', but imprecise; the social-security median regression gives +0.019 (SE 0.020); and the combined estimate is 0.003 with a standard error of 0.011 and a '95 percent confidence interval [that] ranges from -0.019 to 0.026'. Pischke's own conclusion is the one to record: 'the findings are not encouraging for policy makers who wish to use lengthening the school year as a measure to boost the performance of their students... The enthusiasm of the authors of a Nation at Risk for longer school years may therefore have been misplaced.'

Low. The reform was imposed on entire state cohorts by administrative fiat and nobody selected into or out of it; the design compares adjacent birth cohorts within the same states and tracks. What remains is the usual DiD threat that affected and unaffected cohorts faced other differences, which the state-by-track structure addresses.

Unreplicated - no other country ran this experiment. The precisely estimated earnings null is difficult to reconcile with Carlsson and colleagues' finding that a school year is worth 0.14-0.21 SD of crystallised cognitive skill, and with Zhong's finding that a single retained year costs 19% of annual earnings at 26. The most likely reconciliation is the one Pischke offers: the German reform compressed the same curriculum into less time rather than removing content, so students may have absorbed the loss through effort, and the union wage structure would compress whatever residual skill difference survived.