Working Paper: NBER ID: w28343
Authors: Chao Fu; Junjie Guo; Adam J. Smith; Alan T. Sorensen
Abstract: We estimate a model of high school students' college choices, allowing for rich heterogeneity in students' preferences for college attributes. We use data on students' enrollment decisions and application decisions—i.e., the sets of colleges to which they applied—to identify the distribution of students' preferences. We use our estimates to quantify differences in a student's expected value upon college application that result from the uneven spatial distribution of colleges. As with other aspects of economic opportunity, we find that place matters: students with otherwise identical characteristics can have very different expected values depending on where they live. The importance of location reflects differences across states as well as differences across counties within a state. For students with low parental incomes and low SAT scores, over 70% of the variation is within-state across counties, while for students with high parental incomes and high SAT scores, 66% of the variation is across states.
Keywords: college choice; spatial distribution; student preferences
JEL Codes: I21; I23; I24
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| geographic location (R12) | expected values upon application (C29) |
| geographic location (R12) | disparities in welfare outcomes (I38) |
| geographic disparities in college access and quality (I24) | student welfare (I38) |
| tuition policy (I22) | college choice behavior (I23) |
| parental income and SAT scores (I24) | expected welfare variation (D69) |
| geographic variation in college availability (D29) | students' expected utility (D11) |