How much does degree choice matter?

Using a large and novel administrative dataset, this paper investigates variation in returnsto different higher education 'degrees' (subject-institution combinations) in the United King-dom. Conditioning on a rich set background characteristics, it finds substantial variation inreturns, even within subject, across universities with very similar selectivity levels, suggest-ing degree choices matter a lot for later-life earnings. Selectivity is weakly related to returnsthrough most of the distribution but strongly positively correlated at the top end. Other thanselectivity, returns are poorly correlated with observable degree characteristics, which has im-plications for student choices and the incentives of universities.