Carolyn Tsao

This paper quantifies the extent to which teachers earn rents by estimating both the pay gap and the working conditions gap between teaching and teachers' next-best jobs. Using a regression kink design and event studies applied to administrative data from Kentucky, I find that teaching pays a premium of around $20,000 per year over teachers' outside options. However, a choice experiment reveals that teachers—especially inexperienced ones—would pay a large share of their salary for the better working conditions available elsewhere. Combining these estimates, experienced teachers earn a moderate 16% rent, while inexperienced teachers earn no rent at all.

Married women in the early twentieth century U.S. faced "marriage bars," a form of employer discrimination that barred them from paid employment. However, because the end of marriage bar use coincided with shifting social norms and labor market conditions, it is unclear how the end of marriage bars affected women's employment. We study the effects of the state-level prohibition of marriage bars in teaching during the 1930s. A difference-in-differences design shows that the prohibitions increased the share of married women teachers by 3.5 p.p. (20%), partly by leading unmarried women to switch from teaching to other occupations.

To what extent do public school principals affect student outcomes, and how do management practices differ between more and less effectiveness principals? Using administrative data on teacher-principal-student links in two US states, I estimate principal and teacher effectiveness using a two-way manager-worker fixed effect framework. Variance decompositions show that principal effectiveness explains less of the variance in student outcomes than teacher effectiveness does. That said, switching to a more effective principal improves school outcomes: event studies around principal moves and retirements show that receiving a more effective principal improves student outcomes (increased test scores, decreased absenteeism) and teacher outcomes (greater teacher retention, increased student test scores within teacher). Importantly, novel survey evidence suggests that principal effectiveness is attributable to particular management practices: linking the estimated principal effects to survey data on teachers' perceptions of their school leadership, I find that at schools that employ more effective principals, teachers are more likely to report the use of data-driven instructional practices but also a lack of trust and mutual respect between administration and staff.

This paper studies how a manager's tone when giving feedback to workers affects individual productivity and output quality. We construct a novel panel dataset that links software engineers and managers to their email communications and code contributions on the largest open source software project, the Linux kernel. To identify tones used in the emails (e.g., toxic, polite, encouraging), we use natural language processing and machine learning techniques. Additionally, to control for the informativeness of feedback, we fine-tune a large language model (GPT-2) and use the token probabilities it generates to construct measures of entropy and bits. We find a strong negative relationship between manager toxicity and engineer productivity. Using an instrumental variables design to address endogeneity in a manager's choice of tone, we find that receiving toxic feedback from a manager reduces the likelihood that an engineer completes a programming task, increases the amount of time to task completion, and decreases the likelihood that an engineer completes more tasks in the next 30 days.