Structural Models in Education
Friday, Jan. 6, 2017 3:15 PM – 5:15 PM
- Chair: Chao Fu, University of Wisconsin-Madison
Optimal Admission Criteria in School Choice
AbstractIn school choice programs, schools follow admission criteria to rank and accept students. In the absence of monetary transfers, these criteria determine the priority indices and thus dictate which students are allowed to choose before others. The choice of criteria can therefore affect student sorting across schools as well as student welfare.
With school choice data from Paris, our paper study the admission criteria that differ in their weighting of three factors: academic grades, random priorities from lotteries, and affirmative action favoring disadvantaged groups. For example, the current Paris criteria emphasize grades but give priorities to low-income students. The results show that random priorities would substantially lower sorting by ability, compared with the current policy, but would slightly raise sorting by socioeconomic status (SES). By contrast, a grades-only policy would substantially increase sorting by SES. This is largely because of the low-income priority in the current admission criteria. We also consider mixed priorities where the ``top two'' schools select students based on grades, while admission to other schools is based on random priorities. This policy would lower sorting by ability but increase sorting by SES. The welfare analysis highlights the trade-off between the welfare of low- and high-ability students, under the grades-only or random-priorities policies. Overall, mixed priorities appear as a promising candidate to enhance overall welfare with modest effects on sorting, while balancing the welfare of low- and high-ability students.
Heterogeneous Beliefs and School Choice
AbstractMany school districts in the U.S. offer students a choice of schools, with seats at highly demanded schools apportioned via a centralized mechanism with random rationing. This paper presents an empirical analysis of a school choice mechanism that rewards accurate beliefs and careful strategizing on the part of parents. We conduct a household survey of parents in New Haven, Connecticut, to collect data on households' information acquisition, preferences, strategic sophistication, and subjective beliefs about admissions chances. We link this survey to administrative data on applications, outcomes, test scores and enrollment records and use this data to inform an empirical model of school choice that explicitly models families' preferences, beliefs, and sophistication. We use the estimated model to evaluate the individual and equilibrium effects on the distribution of test scores and graduation rates of improving households' information about the lottery mechanism, and of switching to the strategy-proof student-proposing deferred acceptance algorithm. Given households' observed strategic sophistication and beliefs, switching to a deferred acceptance algorithm would raise total welfare and reduce inequality of welfare outcomes. However, welfare would improve further under the existing mechanism following a best-case information intervention that allowed all households to strategize optimally given their preferences.
Student Loans, Student Choices and the Average Returns to Higher Education
AbstractThis paper studies the effect of student loans on individual choices across higher education options on both the extensive and intensive margin. The empirical approach estimates a discrete choice model with preferences defined across institutions, majors, net present value as well as out of pocket expenses incurred during college. We estimate the model using administrative data before and after the introduction of government backed student loans in Chile to study how choices change over time and across an arbitrary eligibility cutoff for test scores and GPA. A regression discontinuity design is used to estimated the model of college and major choice and provides credible identification of the impact of student loans on individual choices near the threshold for eligibility. We find student loans alleviate credit constraints and enable choices that raise expected earnings on average across the threshold of loan eligibility. We also find evidence that a significant portion of students who are induced into choosing college degrees with the policy also choose low to negative return options exhibiting behavior that is consistent with a low preference for future labor market earnings relative to other higher SES students. The growth in the supply of low return fields also leads to further increase the expected default rate and difference between expected earnings across high and low SES students. Counterfactual exercises modifying the degrees that are eligible for the loan policy show college attendance can be increased substantially while lowering negative return outcomes.
- I0 - General