Job Market Paper
Child maltreatment is a serious and common issue in the United States, with nearly 4 in 10 children reporting experiencing maltreatment by adulthood. Education personnel (teachers, school staff, administrators, etc.) play a critical role in the early detection and reporting of such cases due to their repeated interactions with children, but the frequency and quality of such interactions may have been reduced during school closures and virtual instruction during the COVID-19 pandemic. In order to investigate how virtual instruction during the pandemic has affected the child maltreatment reporting process, I employ a difference-in-differences strategy leveraging variation in the level of virtual instruction across different counties in the United States during the 2020-21 school year. My estimates indicate that a fully virtual county is associated with a 36.4% decline in child maltreatment reports made by education personnel relative to a fully in-person county. This decline in reporting is approximately half of the decline in education personnel reporting during the earlier nationwide school closures near the end of the 2019-20 school year (March 2020). In contrast, there are no significant differences in maltreatment reports made by non-education personnel. I also find that substantiated reports decline more than unsubstantiated reports, suggesting that more severe cases were disproportionately affected by under-reporting. Subsequently during the 2021-22 school year, education personnel reports rebounded as the previously virtual schools returned to in-person instruction. These results underscore the importance of in-person interaction with education personnel for the detection of child maltreatment cases, emphasizing the need for appropriate policies to counteract this potential drawback of schooling arrangements where such interactions are limited.
Working Papers
with Nolan Pope
During the COVID-19 pandemic, virtual instruction reduced in-person support that may have helped high school students transition to college. Using national school-level data on FAFSA submissions, ACT participation, and first-year college enrollment, we estimate a difference-in-differences model that exploits crossschool variation in virtual instruction during the 2020/2021 school year. A fully virtual school year reduced FAFSA submissions by 4.2 percentage points, ACT participation by 4.8 percentage points, and first-year college enrollment by 2.5 percentage points. FAFSA submissions partially rebounded after reopening, but ACT participation and college enrollment did not. Effects were substantially larger in disadvantaged schools.
with David Blazar, Seth Gershenson, Ethan Hutt
(Under Review)
Presented at AEFP 2026
Researchers, educators, and policymakers have long worried about the consequences of student absences for educational achievement and attainment—concerns that have grown with the significant rise in absenteeism during and following the Covid-19 pandemic. Using administrative data from Maryland, North Carolina, and a large urban school district, we find that the impact of absences on test scores was modestly (about 5 to 20%) smaller in 2022-23 than in 2018-19 but still practically and statistically significant. Consistent with prior research, these harmful effects of absences are approximately linear and exhibit little heterogeneity across race and gender pre-Covid. In Maryland, the impact of tenth-grade absences on high-school graduation and 2-year college enrollment was much (about 40%) smaller after the pandemic than before, but the impact of absences on any (2- or 4-year) college enrollment increased slightly. Post-Covid reductions in the harmful effects were larger for white students on test scores and larger for Black students on graduation.
Works in Progress
with David Blazar
(in progress)