Assessing the Effect of Schooling on Earnings Using a Social Experiment

The implementation of the 1950 Swedish comprehensive school reform was preceded by a unique social experiment. During this experiment between 1949 and 1962 the new school system was implemented in stages. This allows us to study the same cohort of individuals going through two different school systems, one of them implying at least one year of prolonged compulsory schooling, in a very similar environment. We use this unique source of exogenous variation in educational outcomes to evaluate the impact of education on earnings. We find evidence that the reform had a very strong impact on educational achievement at the lower end. However, we show that the impact of the reform went beyond shifting people from the old compulsory level to the new one, affecting attendance at higher levels. We then proceed to estimate the impact of the reform on earnings later in life and to calculate the returns to various qualifications. For this purpose we use the variation in educational achievements, induced by the social experiment. We compare the results from this quasi-experimental approach to those obtained when we use simple OLS and when we use OLS but controlling for a set of detailed ability indicators collected before the reform was effective. Our findings show that while ability bias may be present, using observed ability indicators gives very similar (but more precise) results to those obtained by using the variation induced by the implementation of the social experiment. We also find very strong evidence that the returns to education vary with ability, the effect being stronger the higher the level of education.