Lightnews — Scholar-powered news

Yue Li @nkliyue.bsky.social · Jun 26

🚨Policy implications:
- Shift attention from whether to when mismatch in affirmative action occurs
- Potential for information and academic interventions for disadvantaged students.(7/7)

Yue Li @nkliyue.bsky.social · Jun 26

💡Up to top 10% within-school cutoff (46th percentile of grade 10 test scores), no tradeoffs between more selective college degrees and increased dropouts.(6/7)

1

Yue Li @nkliyue.bsky.social · Jun 26

✅RDD suggests mismatch at the margin.
- More selective college degrees for some, while increasing dropouts for others, especially for the most overconfident.
- Earning gains for women, but losses for men (5/7)

1

Yue Li @nkliyue.bsky.social · Jun 26

🎯RCT shows that large preferential admissions benefit long-term outcomes of targeted students.
- More selective college degrees without increasing dropouts
- Earning gains concentrated among women (higher take-up), with men’s earning remaining flat (null effects on higher education) (4/7)

1

Yue Li @nkliyue.bsky.social · Jun 26

Combining administrative and survey data, they employ RCT and RDD to estimate the effect of preferential admissions on targeted students and students marginally eligible for admissions, respectively. (3/7)

1

Yue Li @nkliyue.bsky.social · Jun 26

asking: What are the education and labor market impacts of AA on targeted disadvantaged students further down the achievement distribution? (2/7)

1

Yue Li @nkliyue.bsky.social · Jun 26

📢Michela Tincani (UCL), with Michela Carlana and Enrico Miglino, presents “How Far Can Inclusion Go? The Long-term Impacts of Preferential College Admissions” (1/7) @stoneeconucl.bsky.social @michelatincani.bsky.social

1 1

Yue Li @nkliyue.bsky.social · Jun 26

💡Academic program intervention + Fuzzy RD show that relatively cheap tweaks to the first-year experience can improve graduation rates and encourage early major switching.(5/5)

Yue Li @nkliyue.bsky.social · Jun 26

✅Survey data + Model of re-enrollment under uncertainty show that knowledge frictions are contributing to achievement gaps across first-generation status.
- FG students have less accurate prior beliefs, higher GPA uncertainty, and less information. They respond more strongly to grade signals. (4/5)

1

Yue Li @nkliyue.bsky.social · Jun 26

🎯Administrative data + Coarsened exact matching show that FG have worse academic outcomes in college, even conditional on the joint distribution of many observables. (3/5)

1

Yue Li @nkliyue.bsky.social · Jun 26

They combine administrative data, panel survey on student expectations and quasi-experimental evidence from academic program intervention to understand the mechanisms behind the divergence of academic outcomes by first-generation status in college.(2/5)

1

Yue Li @nkliyue.bsky.social · Jun 26

📢Basit Zafar (University of Michigan), with Esteban Aucejo, Jacob French and Paola Ugalde A., presents “Understanding Gaps in College Outcomes by First-Generation Status.” (1/5) @stoneeconucl.bsky.social

1 1

Yue Li @nkliyue.bsky.social · Jun 26

🚨Policy implications:
- Shift attention from whether to when mismatch in affirmative action occurs
- Potential for information and academic interventions for disadvantaged students.(8/8)

Yue Li @nkliyue.bsky.social · Jun 26

💡Up to top 10% within-school cutoff (46th percentile of grade 10 test scores), no tradeoffs between more selective college degrees and increased dropouts.(7/8)

1

Yue Li @nkliyue.bsky.social · Jun 26

✅RDD suggests mismatch at the margin.
- More selective college degrees for some, while increasing dropouts for others, especially for the most overconfident.
- Earning gains for women, but losses for men(6/8)

1

Yue Li @nkliyue.bsky.social · Jun 26

- More selective college degrees without increasing dropouts
- Earning gains concentrated among women (higher take-up), with men’s earning remaining flat (null effects on higher education)(5/8)

1

Yue Li @nkliyue.bsky.social · Jun 26

🎯RCT shows that large preferential admissions benefit long-term outcomes of targeted students.(4/8)

1

Yue Li @nkliyue.bsky.social · Jun 26

Combining administrative and survey data, they employ RCT and RDD to estimate the effect of preferential admissions on targeted students and students marginally eligible for admissions, respectively.(3/8)

1

Yue Li @nkliyue.bsky.social · Jun 26

asking: What are the education and labor market impacts of AA on targeted disadvantaged students further down the achievement distribution? (2/8)

1

Yue Li @nkliyue.bsky.social · Jun 26

and (3) allow for different admission and student sorting rules in the post-Prop 209 period, and then run a counterfactual simulation model.

💡The work is still in progress.(5/5)

Yue Li @nkliyue.bsky.social · Jun 26

✅They extend the model of AAH to (1) explicitly model admission and how it changes with the removal of racial preferences, (2) incorporate and make assumptions of outside options,

1

Yue Li @nkliyue.bsky.social · Jun 26

- Bleemer (2022) uses pre and post-Prop 209 administrative data with DiD and shows that after Prop 209 banned race-based affirmative action, science and overall graduation rates fell for URMs relative to non-URMs. Negative effects on earnings, driven by Hispanics, also appear.(4/5)

1

Yue Li @nkliyue.bsky.social · Jun 26

🎯Motivation: Opposing results from the following two papers.
- AAH (2016) use pre-Prop 209 FOIA data with structural model and show that reshuffling URM students according to non-URM rules within the UC system would increase graduation rates, especially in STEM fields, due to a better match.(3/5)

1

Yue Li @nkliyue.bsky.social · Jun 26

They extend the structural model of Arcidiacono, Aucejo, and Hotz (AAH) (2016), using Post-Prop 209 data, modeling admissions, and constructing graduation rates for outside options to reconcile the results of AAH (2016) and Bleemer (2022).(2/5)

1

Yue Li @nkliyue.bsky.social · Jun 26

📢Peter Arcidiacono (Duke University), with Anh-Huy Nguyen and Zachary Bleemer, presents “Reconciling Results on Proposition 209.” (1/5)
@stoneeconucl.bsky.social

1 1