Teacher Sorting and Preferences over School Disadvantage: Evidence from Performance Pay in Texas (Job Market Paper)
In this paper I study how performance-based compensation affects teacher mobility and sorting using the Texas Teacher Incentive Allotment (TIA), a statewide program introduced in 2019. TIA gives teachers a tiered quality designation, Recognized, Exemplary, or Master, based on performance evaluations and value-added measures. These designations are portable across schools and come with a salary supplement that rises with school disadvantage and rural status. Using administrative data on Texas public school teachers, I combine a staggered difference-in-differences event-study design with a discrete choice model of school choice that allows for heterogeneous teacher preferences over school SES. I find that designation reduces mobility on net: teachers must remain in their district to collect the first allotment, and the recovery in mobility afterward does not offset the moves foregone during that year. Designation shifts teachers toward more advantaged campuses: designated teachers become substantially less likely to leave already-advantaged schools, while exits from the highest-need schools fall only in the first designation year. This response is ordered by designation level, with Recognized teachers becoming more mobile and sorting toward advantage while Master teachers become less mobile and less likely to leave disadvantaged campuses. Preference estimates imply that the average designated teacher would require roughly $6,940 more per year to accept a school with conditions associated with a 20 percentage point higher economically disadvantaged share. While the salary gradient created by TIA falls short of that benchmark at every tier, but the gap is smallest at the highest tier. These results suggest that TIA's portable credential expands effective teachers' outside options at the levels where the pay gradient is shallowest and that the gradient becomes steep enough to hold teachers at disadvantaged campuses only at the highest tier.
The Contribution of College Majors to Gender and Racial Earnings Differences with Scott A. Imberman, Michael F. Lovenheim, Kevin Stange, and Rodney J. Andrews
Gender and racial/ethnic gaps in labor market earnings remain large, even among college-goers. Cross-gender and race/ethnic differences in choice of and returns to college major are potentially important contributors. Following Texas public high school graduates for up to 20 years through college and the labor market, we assess gender and racial differences in college major choices and the consequences of these choices. Women and underrepresented minorities are less likely than men, Whites, and Asians to major in high earning fields like business, economics, engineering, and computer science, however we also show that they experience lower returns to these majors. Differences in major-specific returns relative to liberal arts explain about one quarter of the gender, White-Black, and White-Hispanic (but not White-Asian) earnings gaps among four-year college students and become larger contributors to earnings gaps than differential major distributions as workers age. We present suggestive evidence that differences in occupation choices within field are a key driver of the differences in returns across groups. The work shines light on the roles that college major choice and returns by gender and race contribute to inequality.
Who Benefits from Teacher Performance Pay? Evidence from Texas
In 2019, Texas implemented the Teacher Incentive Allotment (TIA), a state-funded teacher performance pay program. This policy identifies effective teachers and provides a bonus ranging from $3,000 to $32,000 per year. I estimate the intent-to-treat effect of district TIA adoption on student achievement, specifically 3rd through 8th grade reading and math state test scores. Using statewide administrative data, I standardize student scores within grade, subject, and year and aggregate them to the campus-grade level. I then estimate a difference-in-differences event study that exploits staggered adoption across districts, with campus-by-grade and grade-by-year fixed effects and controls for baseline (2018) campus-grade achievement, teacher characteristics, and student demographics. Achievement effects are small in the first years after adoption and increase over time. Four years after adoption, campus-grade mean scores are about 0.08 standard deviations higher in reading and 0.07 in math. The gains differ by campus socioeconomic status. In reading, these gains are concentrated at low- and middle-SES campuses, with no lasting gain at high-SES campuses. In math, gains appear at campuses across the SES distribution by the fourth year. I find no change in student absence rates or chronic absenteeism, suggesting the gains may operate through teachers rather than through student attendance. In ongoing analysis I am examining to what extent these gains are being driven by compositional change in a schools teaching population and to what extent they are being driven by increase in teacher effort.
The Short- and Long-term Effects of School Shootings on the Labor Supply of Texas Teachers (with Kuma Okoro)
The Contribution of College Majors to Intergenerational Mobility (with Scott A. Imberman and Michael F. Lovenheim)
The Effect of Lansing SAVE on Student Outcomes