Serkan Ozbeklik

The Effect of Teacher Gender on Student Achievement in Primary School

Using data from a randomized experiment, we find that having a female teacher lowers the math test scores of female primary school students in disadvantaged neighborhoods. Moreover, we do not find any effect of having a female teacher on male students’ test scores (math or reading) or female students’ reading test scores, which seems to rule out explanations pertaining to the unobserved quality differences between male and female teachers. Finally, this negative effect seems to persist only for female students who were assigned to a female teacher with a limited math background.

Risk Taking in Competition: Evidence from Match Play Golf Tournaments

We test hypotheses regarding risk taking behavior of competitors in settings characterized by one-on-one, single elimination tournaments. We draw data from 579 professional golf matches and over 18,000 holes from 2003 to 2013 in tournaments where match-play scoring is used rather than stroke-play. Because of the uniqueness of the data, we are able to provide clean empirical tests of how risk taking is affected by horizon effects (holes remaining), peer effects arising from heterogeneity in player abilities, match status (whether behind or ahead), and the difficulty of the task/project (hole). The findings are applicable to corporate settings where only a few rivals compete for a prize, such as a winning bid, a promotion, market share dominance, and patents. Other applications include litigation contests and political elections.

Estimating Heterogeneous Takeup and Crowd-Out Responses to Existing Medicaid Income Limits and their Nonmarginal Expansions

We use a switching probit model and the income-limit-based structure of Medicaid eligibility for children to estimate treatment effects of nonmarginal Medicaid expansions on Medicaid takeup, private insurance coverage, and crowd-out, as well as crowd-out for those eligible for Medicaid under rules already in place. Many of these estimates are not found in existing work on public insurance and cannot be calculated with the linear probability model used by previous work in this literature. We provide an estimation approach that is straightforward to implement yet yields precise treatment effects.

THE EFFECT OF ABORTION LEGALIZATION ON CHILDBEARING BY UNWED TEENAGERS IN FUTURE COHORTS

This article examines the long‐term impact of legalized abortion on childbearing by unwed teenagers in the United States. I find that the 1970 legalization of abortion in the repeal states led to about a 6% reduction in unwed childbearing rates for white women aged 15–20 who were born in these states immediately after abortion became legal. I find a larger long‐term impact for African‐American women of the same ages: a 7.5%–13% reduction in unwed childbearing. My estimates are smaller and not as precise for the effect of Roe v. Wade. This outcome is not surprising given that I am able to estimate only a potential lower bound of the effect on unwed childbearing rates. On the other hand, when I estimated a Difference‐in‐Difference regression for the non‐repeal states assuming that there was no national trend that affected the childbearing behavior of the treatment age groups and their respective control age groups separately, I found that the true effect of Roe v. Wade on childbearing by unwed teenagers was about an 11% and 3% reductions for white and African‐American teenagers, respectively.

Leadership Activities and Future Earnings: Is There a Causal Relation?

This paper revisits the effect of high school leadership activities on young men’s earnings. Using several data sets and extending a recently developed econometric technique, we show that even a small amount of selection on unobservables explains the entire high school leadership effect on earnings. We also show that the use of observables to address nonrandom selection bias may yield misleading results if the fixed effects are not dealt with properly.