Research
Links:
Research Fields:
Education, Public, Labor
Peer-Reviewed Publications:
Matthew Naven and Daniel Whalen. The signaling value of university rankings: Evidence from top 14 law schools. Economics of Education Review, 89:102282, 2022
Media: The New York Times
Reports and Policy Briefs:
The Nation's Fiscal Health: Road Map Needed to Address Projected Unsustainable Debt Levels. Government Accountability Office. 2024.
Michal Kurlaender, Sherrie Reed, Kramer Cohen, Matthew Naven, Paco Martorell, and Scott Carrell. Where California High School Students Attend College. Policy Analysis for California Education. 2018.
Working Papers:
Scott Carrell, Michal Kurlaender, Paco Martorell, Matthew Naven, and Christina Sun. Do Schools Matter? Measuring the Impact of California High Schools on Test Scores and Postsecondary Enrollment
This paper explores high school quality in California after the transition to Common Core State Standards (CCSS). Using a longitudinal panel of students' standardized test scores, we estimate high school test score value added in English and mathematics for the 2015-2018 cohorts of 11th grade students. We then link these student-level data to college enrollment records to estimate college enrollment value added. We decompose the college enrollment value added into two components: the persistence of test score value added and non-test score factors (e.g., college counseling services) that influence college enrollment. Results show that there is substantial variation in school quality as measured by both test scores and college enrollment. A one-standard deviation increase in school quality is associated with a 0.15 standard deviation increase in standardized test scores and an 8-percentage point increase college enrollment. Importantly, our results show that both the persistence in test score value added and other non-test score factors within a school are important determinants of college-going value added.
Media: The Education Exchange
Matthew Naven. Human Capital Formation During Childhood and Adolescence: Evidence from School Quality and Postsecondary Success in California
This paper investigates the role of school quality in human capital formation. Specifically, I investigate how the timing of school quality differentially affects long-run outcomes. Using individual-level data on the universe of public school students in California, I estimate elementary, middle, and high school quality using a value added methodology that accounts for the fact that students sort to schools on observable characteristics. I then determine the impact of school quality on future K--12 and post-secondary outcomes. I find that high school quality has the largest impact on postsecondary enrollment, while elementary and middle school quality play a larger role in college readiness. In other words, early human capital investments are important for future postsecondary success, but the unique timing of the college decision process allows for later human capital investments to also play a significant role.
Matthew Naven. Within-School Heterogeneity in Quality: Do Schools Provide Equal Value Added to All Students?
Low-socioeconomic status (SES), minority, and male students perform worse than their high-SES, non-minority, and female peers on standardized tests. This paper investigates how within-school differences in school quality contribute to these educational achievement gaps. Using individual-level data on the universe of public-school students in California, I estimate school quality using a value added methodology that accounts for the fact that students sort to schools on observable characteristics. I allow for within-school heterogeneity by estimating a distinct value added for each school's low-/high-SES, minority/non-minority, and male/female students. Standard value added models suggest that on average schools provide less value added to their low-SES, minority, and male students, particularly on postsecondary enrollment. However, value added models that control for neighborhood, older-sibling, and peer characteristics suggest that schools provide similar value added to low-/high-SES students and minority/non-minority students but more value added to female students. Within-school heterogeneity accounts for 6% of the test-score achievement gap and 22% of the difference in postsecondary enrollment between men and women.
Media: The Thomas B. Fordham Institute, The Education Gadfly Show
Matthew Naven. The Effects of Affirmative Action Bans on Postsecondary Education
Since 1965, affirmative action programs have required public universities to give extra consideration to underrepresented minority applicants. I estimate the impact of eight states banning affirmative action on postsecondary school enrollments and graduation rates using an event study methodology. I also employ a generalized difference in differences strategy that applies the event study analysis and adds nonselective universities in treated states as a control group. Estimates suggest that banning affirmative action significantly decreases minority enrollments, especially at selective universities. These estimates are large in magnitude --- enrollment at selective universities dropped by over 20\% for minority students. While less conclusive, results suggest that graduation rates were essentially unaffected. Falsification tests show that private universities and community colleges, which should not be affected by affirmative action bans, were not similarly affected.
Matthew Naven, Tim Murray, and Valentina Dimitrova-Grajzl. Design of Core Curriculum in Liberal Arts Education
Liberal Arts education has come under scrutiny in recent years both in terms of its design and its applicability to the modern world. This has provided an impetus for re-evaluation of the core curriculum requirements of liberal arts schools. We propose an alternative approach to structuring the core curriculum that is centered around the guiding question "what kind of society do we want to live in and how do we design it?" Core curriculum classes would satisfy seven sub-questions addressing specific aspects of how to design a society one wants to live in as well as a first-year seminar on civil discourse that provides students with the skills to debate various societal designs.
Works in Progress:
Michel Grosz and Matthew Naven. What's My Age Again? The Dynamic Effects of Your Age Relative to Your Peers
Throughout primary and secondary education, students perform a number of tasks where they directly compete with their peers. Older students, however, may have an advantage over their younger peers. Using individual data on California public 2nd--11th grade students from the 2002--2003 to 2012--2013 cohorts and variation in assigned kindergarten entrance age due to the existence of a universal entrance cutoff date in the state of California, we employ a reduced form regression discontinuity design to estimate the impact of being older than the peers in one's cohort. Estimates suggest that students born just after the entrance cutoff score 0.13 standard deviations higher on English language arts exams in the second grade, although this effect fades out over time to 0.03 standard deviations in the 11th grade. Relatively older students are also more likely to participate in gifted and talented programs, less likely to be classified as an English learner, and less likely to be diagnosed with a disability that is subjectively diagnosed. These findings have important implications regarding the "redshirting" of prospective kindergarteners in which parents delay their child's entrance to kindergarten in an effort to increase their likelihood of success.
Resting Papers:
Michel Grosz and Matthew Naven. Event Studies with Secular Time Trends
Many papers in applied microeconomics rely on an event study identification strategy, leveraging variation in the timing of treatment across units and allowing for dynamic effects. We show that estimation of these dynamic treatment effects is sensitive to how the researcher constructs the panel dataset, especially in the presence of unit-specific linear time trends. We consider three data construction approaches: using all the data, using a balanced sample with a subset of event years, and a specific "Loaded" combination of the two. We show that this "Loaded" approach outperforms the other two, even when no time trends are present.
Anujit Chakraborty, Lester Lusher, and Matthew Naven. Recognition and Performance: Evidence from a Regression Discontinuity in College Football
Affiliations:
California Education Lab: Affiliated Researcher
Academic Freedom Alliance: Member
Heterodox Academy: Member