Abstract

(Joint work with Christian Belzil, Jorgen Hansen, and Julie Pernaudet)

Using a unique field experiment implemented among Canadian high school students in 2008-2009 combined with data on early life-cycle outcomes collected 10 years later, we estimate the causal effect of randomized financial aid on: (i) university enrollment; (ii) field of study, conditional on university enrollment; (iii) graduation from university; and (iv) earnings at age 30. We estimate a random coefficient model where the effects of financial aid are allowed to differ across types of students. We find that a one-time $1,000 grant: (i) increases the probability of enrolling in university for a third of our student population; (ii) increases the probability of enrolling in a two-year post-secondary education for another third; and (iii) has no impact on enrolling in either type of higher education for the remaining third.

Among those enrolled in university, we find that a $1,000 grant reduces the probability of enrolling in STEM by 0.075 across all types and increases enrollment in Social Sciences and Humanities by 0.06. Similar to the results on university enrollment, some students are unaffected by financial aid in their field choice. Moreover, across all types, the same grant raises the graduation probability from two-year degrees and reduces it from four-year degrees. This result is partly due to the composition effect, where financial aid induces certain students to enroll who experience lower graduation probabilities than students who are unaffected by aid in their enrollment decision. Lastly, we find that the extensive labor supply margin appears to be inelastic to financial aid.