Quantified Effect Sizes for Bleemer (2021) UC Mismatch Findings

What the evidence says

The core quantified evidence on this topic comes from Bleemer's (2021) SSRN paper on major restrictions inside the University of California system, using a dynamic difference-in-differences design with placebo bootstrap tests. When a major imposes new restrictions on declaration (typically mechanical GPA cutoffs), the share of underrepresented-minority (URM) students who go on to declare that major decreases by an average of 20 percent, and the average URM proportion among declarants falls by 3.3 percentage points. Given a roughly 10 percentage point decline in all major declarations and a 13 percent pre-restriction URM share, this means URM students exit the major at about 2.1 times the rate of non-URM students (about 25 percentage points versus 12 percentage points). This decline tracks a pre-existing gap: URM students in these soon-to-be-restricted majors had entered with introductory course grades about 0.3 standard deviations lower than non-URM students in the same fields, which sets up the mechanical restriction to remove them disproportionately.

The composition of who remains after a restriction shifts accordingly. Average SAT scores of remaining declarants increase by almost 40 points on the 2400 scale, and average normed first-term GPA of the remaining declarants increases by about 0.75 standard deviations relative to students who exit — a result that holds even when GPA is calculated using only courses outside the major's own discipline, consistent with exiting students having SAT scores as much as 300 points (about one nationwide standard deviation) below the major's average.

Displaced high-intention URM students land in different majors, and it costs them. High-intention students who intended the restricted major become 15 percentage points less likely to successfully declare it (a result the source itself flags as only marginally reliable statistically over the one-to-four-year post-implementation window). Among these high-intention students, URM students end up declaring majors with wage premiums about 1.2 percentage points lower on average than similarly-intended non-URM students, though this estimate averages over both high- and lower-premium restricted majors and has limited power for the high-premium subset alone.

Isolating the mechanism, mechanical (GPA-based) restrictions account for the entire measured gap in URM representation among graduates of restricted majors: 11 percent of graduates from lucrative unrestricted majors were URM, versus 8 percent from restricted majors (a 25 percent relative gap), and this gap is wholly attributable to majors with mechanical restrictions — there is no measurable relationship between discretionary (case-by-case) restrictions and URM graduate share. At a system level, URM graduates hold majors with about 3 percent lower average wages than non-URM graduates' majors in recent years, and within-institution stratification — not sorting between institutions — accounts for about 65 percent of the growth in ethnic stratification of college majors since the 1990s.

A separate secondary source (Wyness 2023), summarizing Bleemer's broader body of work on the Prop 209 affirmative-action ban rather than the major-restriction paper specifically, reports declining educational outcomes and lower wages for URM students after the ban (the wage decline attributed mainly to Hispanic students), no measured effect on STEM course scores or persistence for lower-attaining students reallocated to less-selective universities, and no measured change in educational or labor-market outcomes for marginal nonminority students displaced from UC Berkeley by the ban. None of these three results carries a quantified magnitude in the source — only direction. This is a real gap in what this page can report: the size of the Prop 209 educational-outcome and wage effects, and the test statistics behind the STEM null result, are not available in the reviewed material.

Where the evidence splits

Wyness (2023) states that Bleemer's finding of harm to minority students after Prop 209 stands "in contrast with" prior studies by Arcidiacono et al. (2014) and Arcidiacono et al. (2016), which found no mismatch effects. The claims on this page do not include the Arcidiacono findings directly, and the source gives no detail on what in the study design, population, or period produces the disagreement — the contrast is unexplained by the sources on this page.

A second split concerns whether enrollment at a selective college benefits marginal admits. Wyness (2023), summarizing Bleemer's regression-discontinuity result at UC Berkeley, reports that nonminority students just below the pre-Prop 209 admissions cutoff were relatively unaffected in educational and labor-market outcomes, which the source reads as suggesting limited value of selective education for these marginal students. Leukhina (2024) reports the opposite pattern in general terms: students who enroll at selective colleges are more likely to graduate and earn more, on average, than untreated students with similar SAT scores at the same colleges. Leukhina's claim does not specify whether it addresses the same population or paper as Bleemer's Berkeley discontinuity, and neither source explains the difference — this split is unexplained by the sources on this page.

Thresholds and dose-response

No threshold established in the current sources.

Open empirical questions

• What are the quantified point estimates behind Wyness's (2023) qualitative claims of "declining" educational outcomes and "lower" wages after Prop 209 — the reviewed source gives direction only.
• What is the quantified test statistic behind the null STEM course-score and persistence result for reallocated lower-attaining students?
• What in study design, population, or period explains the disagreement between Bleemer's finding of harm and the earlier Arcidiacono et al. (2014, 2016) no-mismatch findings?
• What in study design or population explains the disagreement between the Berkeley marginal-admit null result and Leukhina's (2024) finding that selective-college enrollment increases graduation and earnings?