Research
Working Papers
Foreign Peers in Higher Education: The Effects of Internationalization on Native Students’ Academic Outcomes with Melisa L. Diaz Lema - SSRN Paper (Under Review)
Internationalization is widely promoted in higher education, yet it may entail tradeoffs for native students. While greater diversity can enhance the educational experience, changes in peer composition may reduce learning quality when students differ in cultural and academic backgrounds. This paper examines the effects of internationalization on the academic outcomes of native students. Using administrative panel data from a large public STEM university in Italy, we follow over 33,000 students across 17 programs and six cohorts. We exploit the introduction of an English parallel track in Civil Engineering using an instrumented difference-indifferences approach. The English track more than doubled the share of foreign students in the Civil Engineering program relative to the other primarily Italian-taught programs. We then examine the overall effect of expanding the English track and find that native students, regardless of whether they enroll in the Italian or English track, are less likely to drop out and more likely to graduate on time, although they achieve a slightly lower GPA. Further analysis of mechanisms reveals that international student inflows drive graduation gains, while larger classes and other changes to the learning environment explain the GPA decline. Our findings indicate that internationalization policies can improve native students' persistence and timely degree completion in higher education.
Identifying a Cumulative Learning Technology: Evidence from Online Learning - NBER Conference Paper
An inherent feature of learning in most disciplines, especially STEM, is its cumulative structure, which makes developing advanced skills challenging. This paper credibly estimates a dynamic learning technology of student effort in a foundational university course with a cumulative structure. Doing so is incredibly difficult because effort inputs are typically unobserved and are dynamic endogenous choices. To address this, I use rich panel data on nearly 3,700 undergraduates in an online introductory programming course, which precisely tracks study time throughout the course. Then I carry out a field experiment which generates period-by-period exogenous variation in effort allocation, enabling me to identify dynamic interactions across effort inputs in the learning technology. I find evidence of dynamic learning complementarities as the marginal benefit to studying in each learning period is increasing in prior knowledge accumulated. I then develop and estimate a multi-stage behavioral model of effort supply, using the experiment to identify benefit and cost parameters at each stage. The model informs policy simulations of grading schemes, showing that decreasing assignment weights across the course—rather than equal weights—boosts effort early on when foundational skills are acquired. The findings in this study have implications for effective learning strategies and approaches to course design.
Language Proficiency and Persistence: Evidence from Rental Housing Markets with Cornel Nesseler and Jiaqi Zou - Paper
Immigrants often face ethnic discrimination and language barriers. We estimate both components in the rental housing market, and compare the benefits of language acquisition with those of lower-cost actions. In a large-scale field experiment, we send apartment‑viewing requests in four Central European countries, randomly varying name origin, proficiency in the local language, and a signal for learning the local language. Inquiries in fluent English from foreign names receive only half the viewing invitations as inquiries in fluent local language from native names. Language barriers account for almost 75% of this gap, and the foreign-name penalty for the remainder. Functional local-language proficiency yields no measurable benefit versus fluent English, nor does signaling local-language learning. By contrast, a simple follow-up inquiry yields benefits as large as the foreign-name penalty. Our results suggest that discrimination measured in audit experiments largely understates the barriers immigrants face in practice.
Provision of Online Public Goods: Evidence From a Peer Discussion Board - Paper
How can policy designers sustain active participation in online communities despite free-riding incentives? This paper addresses the question in the context of voluntary online student discussion boards -- a prominent feature of large classrooms and distance education used to support learning at scale. Using survey and administrative data from nearly 1,200 undergraduates in a foundational programming course at a large public university, I measure students' altruistic attitudes and track their reading and posting behavior. Two randomized informational interventions are implemented: one informing students about the discussion board’s availability, and another encouraging them to internalize the spillover value of peer discussion. Both interventions increase sign-ups and contributions, respectively. I find that access to and participation in the board significantly improve learning outcomes. Using a behavioural model of student contributions, I argue that having targeted bonus credit for instructor-endorsed posts can effectively reduce free-riding, enhance the value-added of the discussion board.
Don’t let your fruits rot: An experimental investigation of leadership training in a grocery store with Simone Haeckl and Mari Rege (Paper available upon request)
Despite widespread use, evidence on the effectiveness of leadership training is limited. We study the effects of a state-of-the-art online leadership program in a retail chain with 159 stores and over 1000 employees. We use an event study approach to test the effects of the training, controlling for seasonal effects. We find no impact of the training on customer satisfaction, food waste, or a measure of store profitability. In addition, we run an RCT to test whether we can improve the effectiveness by addressing two behavioral barriers to learning motivation. First, leaders may doubt their own and their employees' capacity for skill development. Second, they might underinvest in training due to present bias. All stores received the training, with random assignment to the standard or enhanced version. The enhanced version also showed no greater effect on productivity than the standard one. Leaders in the enhanced group reported being more supportive several months post-training. Still, we find no effect on employees’ work satisfaction, intention to quit, or turnover.
Supportive Leadership and Employee Satisfaction and Retention: Causal Evidence from a Large Grocery Store Chain with Jon-Sander Amland and Mari Rege (Paper available upon request)
Employee turnover imposes substantial productivity losses and recruitment costs, yet existing evidence provides limited causal guidance on investments that improve retention. We estimate the causal effects of supportive leader behaviors on employee satisfaction, quit intentions, and realized turnover using three-wave employee–leader survey data matched to administrative employment records from a large retail organization. The data cover approximately 1,850 employees across 202 stores operating under a common organizational framework yet exhibiting substantial variation in local leadership behaviors. To address endogeneity and selection concerns, we implement two complementary identification strategies: an individual fixed-effects design that exploits within-employee variation in exposure to leader behaviors over time, and a jackknife instrumental-variables approach that instruments an employee’s exposure with coworkers’ exposure to the same leader. Across both strategies, supportive leader behaviors have economically large and robust effects. Increases in supportive behaviors raise employee satisfaction, reduce quit intentions, and lead to meaningful reductions in realized turnover. We translate these estimates into policy simulations comparing the retention gains from improving or replacing low-support leaders with the losses associated with failing to retain highly supportive leaders. The results indicate that targeting low-support leaders generates retention gains comparable to—and in some cases exceeding—the losses from losing highly supportive leaders. Overall, the findings identify supportive leader behaviors as a scalable and actionable managerial lever for reducing turnover in multi-unit service organizations.
Work in Progresss
Psychological Safety and Employee Satisfaction and Retention: Causal Evidence from a Large Grocery Store Chain with Jon-Sander Amland, Simone Haeckl, and Mari Rege
This paper examines the effect of psychological safety—the belief that employees can voice opinions and explore ideas without fear of negative consequences—on well-being and retention. Using panel data from a large Norwegian retail chain with 202 stores, we address endogeneity through two complementary identification strategies: individual fixed effects and a jackknife instrumental variables approach. The fixed-effects model controls for time-invariant heterogeneity, while the jackknife-IV strategy addresses time-varying confounders in separate analyses. Our most conservative estimates indicate that a one–standard deviation increase in psychological safety raises job satisfaction by 0.23 SD, reduces quit intentions by 8 percent, and lowers actual turnover by 3 percentage points. These findings highlight the importance of psychologically safe work environments for enhancing employee well-being and retention.
Encouraging High-achieving Students to Enroll in Advanced Courses, with Robert McMillan and Linda Wang
Misinformed high-achieving students may enroll in less challenging university courses, limiting future options and hindering their human capital development. This paper examines whether such students can be guided toward more ambitious academic paths. We conduct several field experiments across four cohorts of first-year economics students at a large Canadian research university, varying how students are informed about advanced upper-year courses. The interventions differ in their scalability. We find that an intensive in-person information session substantially increases enrollment in the most rigorous second-year economics courses, particularly among first-generation students who are initially less aware of these options. In contrast, delivering the same information via email or online sessions has no significant effect. The results suggest that personalized, resource-intensive nudges can meaningfully influence course choices, with implications for university advising and policy design.
From Autopilot to Co-Pilot: Guiding Effective AI Use in Higher Education with Ajinkya Keskar, Ozlem Tonguc, and Case Tatro
An ongoing lab experiment investigating the efficacy of an AI training module in encouraging students to use AI productively.