Working Papers:
A Nonparametric Approach to Rare Event Binary Outcomes: Evidence from Prescription Opioid Use in Canada
Abstract: Logistic regression is widely used in applied economics to model binary outcomes, yet it performs poorly when the event of interest is rare. This paper explores a nonparametric mixed-kernel estimator as an alternative classification strategy for imbalanced binary data. Using a large sample of Canadian survey data from 1996/97, we examine predictors of prescription opioid use, which was a rare event observed for less than 4% of respondents. The nonparametric estimator significantly outperforms logistic regression in terms of classification accuracy, particularly in precision-recall space. While the application focuses on opioid use, the results have broader relevance for empirical research involving highly imbalanced binary outcomes in health, labor, and social policy settings.
Link to paper.
The Impact of Mandatory Universal Pharmaceutical Insurance On Prescription Opioid Use: Evidence from Canada
Abstract: This paper utilizes a natural experiment and robust nonparametric estimation methods to examine the impact of mandatory universal pharmaceutical insurance on prescription opioid use. A policy evaluation of Quebec’s implementation of mandatory pharmaceutical insurance to complement the existing universal public health insurance plan, that provides physician and hospital services, is conducted using data from the Canadian National Population Health Survey (NPHS). The results show that, among the general population, the policy led to a significant increase in pharmaceutical insurance coverage and a small in magnitude but statistically significant decrease in prescription opioid use. Additionally, the analysis does not find statistical evidence that the increase in pharmaceutical insurance coverage led to a substitution effect away from over-the-counter pain medications and towards prescription opioids for pain treatment.
Link to paper.
The Effects of the COVID-19 Pandemic on the Earnings Distributions of Canadian Immigrants
Abstract: As the Corona Virus (SAR-CoV2) spread across the globe in 2020, many government bodies were forced to implement restrictions to slow down the spread of the virus; this included the shutdown of non-essential businesses and services, the cancellation of in-person events and entertainment, school closures, and the start of work-from-home orders. Many sectors saw a drastic drop-in economic activity, resulting in job losses and reductions in hours worked. This paper uses Canadian microdata to analyze the labour market effects of the COVID-19 pandemic on Canadian immigrants. Trends in employment status and aggregate hours worked are examined by gender and immigrant status and we find evidence that the labour supply of immigrants, especially immigrant women, was more affected than the labour supply of their non-immigrant counterparts.
Work in Progress:
The Effects of Fixed vs Growth Mindset on Student Mental Health
Karen Ugarte Bravo, Annabel Thornton
Abstract: This study investigates whether a brief online growth mindset intervention can improve mental health outcomes among undergraduate economics students. A randomized controlled trial was conducted during the 2024–2025 academic year in ECO220Y1Y: Introduction to Data Analysis and Applied Econometrics at the University of Toronto, a large, required second-year course enrolling over 1,000 students annually. Participants (N=450) were randomly assigned to either a treatment group that completed a 30 minute PERTS growth mindset module mid-year or a control group that did not. All students completed four online surveys linked to course performance data, providing repeated measures of mindset, academic expectations, performance attributions, and self-reported mental health across the academic year. Student Experience in the Research University (SERU) survey data confirms that University of Toronto undergraduates exhibit higher levels of anxiety and depression than peers at comparable research institutions. Results indicate that the mindset intervention did not improve course performance but did buffer against declines in growth mindset beliefs over time. Students in the treatment group were significantly less likely to exhibit declining growth mindset responses on two of three standard measures. Moreover, while no academic performance effects were detected, treatment participants were substantially less likely to report deteriorating mental health over the academic year (a 16-percentage-point reduction relative to control) and more likely to report improvements (an 11-percentage-point increase). Taken together, the findings suggest that brief, scalable mindset interventions may foster resilience and well-being among students in quantitatively intensive economics courses, even when measurable academic performance gains are not observed. The results highlight the potential role of psychological supports in improving student experience and retention within quantitative economics education.