From Danny Yagan:
The consequences of banning affirmative action depend on schools’ ability and willingness to avoid it. This paper uses a seventeen-year sample of law school applications to estimate how completely UC law schools avoided the 1996 UC affirmative action ban. Controlling for selective attrition from applicant pools, I find that the ban reduced the black admission rate to 31%—well below the 61% pre-ban rate but still four times higher than the 8% rate that would prevail under observed white admission standards. Observed black admission advantages at intermediate credential levels were as large as 99 percentage points before the ban and 63 percentage points after the ban. The results have important implications for modeling affirmative action bans, sustaining racial diversity under a ban, affirmative action constitutionality, the effectiveness of mandating nondiscrimination, and identifying discrimination in cross-sectional data.