Year-long internships and skill formation in post-secondary education

Year-long placements, semester-long internships, and cooperative education programs are common in post-secondary education in North America and the UK. Rules vary by country, institution, and discipline, but in all these programs students leave their studies for periods from one to three semesters to work in settings related to their academic study. Program participants generally have better academic performance both before and after their internships. This paper investigates whether and how year-long internships affect academic performance. The analysis uses a sample of students from an Economics Department in South West England, and explicitly models selection into year-long internships. We find that the overall effect (or average treatment effect) of placement on performance is negligible but that the direct effect is positive. The difference arises because of the negative correlation between the error terms in the academic performance and selection equations. The negative correlation can be due to unobserved time invariant factors (unobserved heterogeneity) or due to unobserved effects of placement (time variant factors). We exploit an institutional feature of the program to explore this issue. Our results are consistent with time-variant factors underlying the negative correlation. We speculate that these factors are related to study effort upon return from the placement.