In many settings, market designers must contend with the presence of firms who participate in the broader game surrounding a market but do not participate in the portion under the designer’s control. In this paper, we study the empirical relevance of the configuration of on- and off-platform options in the context of a centralized college-major choice system. We quantify significant negative externalities generated by off-platform options and measure the aftermarket frictions that contribute to generating them in practice. Our empirical application uses administrative data from the centralized assignment system for higher education in Chile and leverages a recent policy change that increased the number of on-platform slots by approximately 40%. We first present a policy analysis which shows that expanding the centralized platform leads students to start college sooner and raises the share of students who graduate within six years. We develop an empirical model of college applications, aftermarket waitlists, and matriculation choices. We estimate the model using students’ ranked-ordered applications, on- and off-platform enrollment, and on-time graduation outcomes. We use the estimated model to quantify welfare impacts, decompose different mechanisms and to con-duct counterfactual exercises. We find that when more programs are available on the centralized platform, welfare increases substantially. These externalities are driven by students who receive and decline on-platform offers, and are amplified by substantial frictions in waitlists. Our results indicate that expanding the scope of a higher education platform can have real impacts on welfare and human capital. Importantly, the effects are larger for students from lower SES backgrounds, suggesting the design of platforms can have effects on both efficiency and equity.
This paper introduces a simple school choice model in which all students have the same ordinal preferences over schools but only some have access to an outside option. Our model predicts that, under a manipulable school choice mechanism, students with the outside option are more likely to apply to popular schools. We show that while students with the outside option beneﬁt from manipulable systems, students without the outside option may experience either welfare gains or welfare losses. We evaluate the positive predictions of the model using a diﬀerence-in-diﬀerences design that leverages a change from the Boston mechanism to a deferred acceptance mechanism in the New Haven, Connecticut school district. Consistent with the theoretical predictions, students with an outside option are more likely to list popular, highly-rated schools under the manipulable mechanism, but this gap disappears after the switch to the deferred acceptance mechanism.
This paper studies how welfare outcomes in centralized school choice depend on the assignment mechanism when participants are not fully informed. Using a survey of school choice
participants in a strategic setting, we show that beliefs about admissions chances differ from
rational expectations values and predict choice behavior. To quantify the welfare costs of belief
errors, we estimate a model of school choice that incorporates subjective beliefs. We evaluate
the equilibrium effects of switching to a strategy-proof deferred acceptance algorithm, and of
improving households' belief accuracy. Allowing for belief errors reverses the welfare comparison
to favor the deferred acceptance algorithm.
We study the efficiency of a centralized college admissions platform that operates jointly with a decentralized “off-platform” aftermarket. We exploit a policy change in 2012 in which a significant number of off-platform higher education options joined a centralized assignment system for higher-education programs in Chile. We show evidence that this policy change had impacts on real outcomes finding that the share of students declining their placed spot de-creased by 8% and dropout rates at the end of the first year of college dropped by 2 percentage points (a 16% drop) following this event. To quantify the welfare impacts of the aftermarket on the efficiency of the match, we develop and estimate an empirical model of college applications, aftermarket waitlists and matriculation choices, using individual-level administrative data from Chile on almost half a million applications, including test scores and enrollment decisions at all on- and off-platform higher education options. According to model estimates, welfare increases substantially, students begin their studies sooner, and fewer students drop out by the end of the first year of study when top off-platform programs join the platform. These benefits are greater for less advantaged students and for women. Counterfactual analysis suggests that more desirable options generate larger negative externalities when not on the platform. These externalities are mostly driven by students admitted on on-platform options that decline their spots to pursue off-platform programs; and are amplified by frictions in waitlists that yield socially inefficient allocations. Our results indicate that platform design can have real impact on outcomes of interest. Specifically, our findings suggest policymakers need to consider the implications of off platform options and their characteristics when designing regulation surrounding centralized assignment systems.