The Dog That Didn't Bark: Educational Content Restriction Laws and Teacher Labor Markets

Abstract

Between 2021 and 2023, twenty-three U.S.\ states enacted laws restricting classroom instruction on race, gender, and "divisive concepts," prompting claims of an impending teacher exodus. I test this prediction using Census Quarterly Workforce Indicators for K–12 schools (NAICS 6111) in a staggered difference-in-differences design with the estimator. The overall ATT for log employment is 0.023 (SE = 0.020), statistically indistinguishable from zero. Separations, hiring, earnings, and female workforce share are similarly null. The conventional TWFE estimator spuriously finds a significant positive effect (0.109, $p < 0.05$), illustrating how naive two-way fixed effects can produce misleading results under staggered adoption. One margin responds: turnover rises significantly (0.0048, $p < 0.05$), suggesting increased churn without net employment loss. The design can detect effects of 5.5% or larger with 80% power.