Catalina Franco
Google Scholar

Welcome!

I am a researcher at the center for Applied Research (SNF) at NHH Norwegian School of Economics. I am affiliated with the Center of Excellence FAIR and the Development Learning Lab (DLL). I completed my PhD at the University of Michigan.

My research focuses on education challenges in industrialized and developing countries, often using behavioral economics as an approach. I use both administrative data and design my own randomized studies. Currently, my focus is on studying the implications of generative AI on education, as well as the role of stress on gender differences in high-stakes exam performance.

Email: catalina.franco@snf.no

You can find my CV here


Publications

Does Re-Imprisonment for Technical Violations Prevent Crime?
(with David Harding, Jeffrey Morenoff, and Shawn D. Bushway)
Journal of Human Resources, forthcoming

| Paper

We find that individuals involved in low-level crime who receive prison sentences are more likely to be re-imprisoned for technical violations during their post-release supervision, rather than for new offenses, compared to those who receive non-prison sentences. We identify the extent and cost of this incapacitation effect among individuals with similar criminal histories using exogenous variation in sentence type from discontinuities in Michigan Sentencing Guidelines. Higher re-imprisonment adds 15% more prison days to the original sentence while only low-level crimes appear to be averted. These results suggest that re-imprisonment for technical parole violations does not prevent serious crime.

Behavioral Dynamics in Transitions from College to the Workforce
(with Meera Mahadevan)
Journal of Economic Behavior and Organization, August 2021

| Paper

This paper examines how decision-making may change when individuals face a permanent change in financial resources after a major life transition. We experimentally elicit preference and cognitive measures from Colombian students on the job market, as well as from a comparison group of college peers in lower years, over 8 months. This period encompasses the job search process while students are in college and ends after they graduate and begin their full-time jobs. Using a difference-in-differences setup, we find that students entering the job market perceive greater financial liquidity and take on more responsibilities. We do not find any evidence of an increase in the take-up of credit or of students moving out of their parents' homes, features commonly associated with this transition in other countries. Regarding preferences, we find suggestive evidence that the students become less present biased and more prosocial during this transition to the workforce. We do not find significant changes in risk and ambiguity preferences or cognitive performance. These findings help us document the changes experienced during a universal transition, one that is achieved through own effort rather than cash transfers or government policies.

Latin American immigration in the United States: Is there wage assimilation across the wage distribution?
Hispanics in the US Labor Market, Richard R. Verdugo ed., Centre for Demographic Studies, Barcelona, Spain, 2013.

| Paper

This paper estimates wage differentials between Latin American immigrant males and U.S. natives along the wage distribution using quantile regression and matching methodologies. The hypothesis of wage assimilation is tested by exploiting the differences by cohort of arrival. The main findings indicate that Latin Americans’ wages do not assimilate to those of their native counterparts and that the gaps are wider for the lowest deciles of the distribution. For the cohorts of immigrants who arrived before 1979 the differential is explained almost completely by education, with a negligible effect that cannot be explained by observable characteristics.

Earnings differentials in Colombia: A study of young and rural workers, 2002-2009
(with Johanna Ramos)
Economic Analysis Review, Vol. 25, No.2, Special Issue: Inequality and income mobility in Latin America, 2010.

| Paper

This paper examines the trends and magnitude of earnings differentials among urban and rural workers, and young (18-24 years) and old (25-65 years) workers from 2002 to 2009 in Colombia. Using household surveys data and constructing cells for comparing only workers with the same characteristics, the results from time series and matching decomposition methodologies show that earnings in the groups of interest have not diverged over time. However, the earnings differentials are high at around –50 percent for rural and –40 percent for young workers, of which 14 and 19 percentage points, respectively, remain unexplained after controlling for demographic and job-related characteristics.

Working papers

Will Artificial Intelligence Get in the Way of Achieving Gender Equality?
(with Daniel Carvajal and Siri Isaksson)

| Paper

We conduct two surveys with preregistered experiments to examine gender differences in generative AI adoption and potential labor market consequences. First, we document a substantial gender gap in AI adoption among students at a top business school in Norway, with female students, particularly top students, opting out of AI use. Second, a survey of managers shows that acquiring AI skills would significantly enhance job prospects for top female students currently avoiding AI. Finally, we provide causal evidence on policy tools to close the gender gap in AI adoption. Our findings show that generative AI could widen existing gender gaps in the labor market, but appropriate policies can help prevent this outcome.

Does AI Help or Hurt Learning?
(with Natalie Irmert and Siri Isaksson)

| Paper available upon request

As AI reshapes how students learn, it raises pressing concerns about ensuring equitable learning opportunities and outcomes. A key question is who benefits from AI and who may be left behind. We address this question through a preregistered lab experiment (N=572) examining AI’s impact on learning. Students were randomly assigned to one of three conditions: (1) Control (access to Google Search only), (2) AI-assisted (AI access), or (3) AI-guided (AI access with guidance), and were tasked with learning a novel topic that they had no prior knowledge of. At the end of the experiment, participants completed an exam without AI access, allowing us to causally estimate the effects of AI on learning outcomes. While AI has no overall effect on exam performance, this null masks a significant heterogeneous effect: high-GPA women benefit, while low-GPA men perform worse, raising concerns about whether AI will widen pre-existing educational inequalities. By analyzing participants’ prompt data, we gain insight into the mechanisms behind these effects and identify how different groups interact with AI and how this in turn affects learning outcomes. Our findings reveal that AI may exacerbate learning gaps, and provides key insights for designing policies that can mitigate this risk.

Innocuous Exam Features? The Impact of Answer Placement on High-Stakes Test Performance and College Admissions
(with Erika Povea)

| Paper

We exploit randomness in college entrance exams in Colombia to study how the placement of answers impacts multiple choice test results and access to college. Using administrative data, we find that: first, applicants are 5% less likely to answer correctly when the correct answer is the last in the choice set (option D). And, second, that one SD higher share of correct answers in D in the math section reduces applicants’ overall performance and their preferred major admission rate by 3%. Considering lifelong college access implications, we show how seemingly innocuous exam features disproportionately affect unlucky test takers.

Bridging the Gender Gap in Exam Performance through Nudges and Stress Reframing
(with Marcela Gomez-Ruiz)

| Paper

Why women underperform relative to men in high-stakes exams while excelling under lower stakes remains puzzling. Previous research suggests differential responses to pressure and omitted questions as explanations. We evaluate a unique randomized trial with 4,658 applicants to a coding program in Uruguay, introducing two treatments: a \emph{Nudge} to answer all questions and a \emph{Nudge + Stress} intervention, reframing stress as performance-enhancing. Treated women omit fewer questions, boosting performance by 0.08 SD and 0.18 SD, respectively. There are no effects on men, and 9\% more women gain admission. The key insight is that reducing omitted questions narrows gender performance gaps.

Strategic Decisions Have "Major" Consequences: Gender Differences in College Major Choices
(with Amelia Hawkins)

| Paper

Does the way students navigate college admissions processes contribute to gender gaps in major choices? We study the major choices of applicants to Colombia's largest public university who marginally miss their intended major’s cutoff. We find that women scoring just below the cutoff submit additional, less-preferred majors compared to similar men intending the same major. While women enroll sooner into these less-preferred majors, men instead reapply later for admission. As a result, a gender gap in potential earnings of 4.5% emerges at the cutoff. Our findings highlight that differential responses to just failing can influence enrolled majors and earnings potential.

Gender Differences at the Top: Incentives, Performance and Choking under Pressure
(with Ingvild Skarpeid)

| Paper

Are women more likely than men to choke under pressure? We provide novel experimental evidence on this question from a large-scale study of online workers taking part in two IQ tests. We manipulate the level of pressure by varying the performance requirement or the size of the incentives. We complement the experimental analysis with survey evidence that help us confirm that the increase in incentives is substantial compared to the size of incentives typically offered to these online workers. Our results show that increasing pressure does not significantly affect overall performance, nor does it differentially impact women. However, an exploratory analysis reveals that high incentives reduce performance among female test takers who may be ``at risk'' of choking: women who performed close to the performance requirement in the first test and had correct beliefs about being a high performer. Our findings suggest that understanding how high-performing women respond to pressure may be a key step in closing the gender gap in domains with high levels of pressure such as high-stakes examinations.

Stress, Exam Performance and Gender
(with Ingvild Skarpeid and students in my course "Stress and mental health in education")

| Draft available upon request

This paper examines the hypothesized relationship between performance and stress, commonly known as the Yerkes-Dodson Law (YDL). Analyzing exam performance and post-exam survey responses from 35,000 university applicants taking a high-stakes college entrance exam in Colombia, we find empirical support for the inverted-U pattern posited by the YDL. Our results indicate that female applicants are significantly more likely to experience high, sub-optimal levels of stress compared to male applicants. We estimate that high-stress levels result in a performance penalty of 5\% of a standard deviation. Additionally, our results suggest that, in high-stakes settings, interventions aimed at reducing the importance students place on the outcome of the exam may be more effective for optimizing stress than alternatives like stress reappraisal or controlling worry.

Willingness to Pay for Formal Job Attributes: A Discrete Choice Experiment in Colombian Mom-and-Dad Stores
(with Amalia Rodríguez-Valencia)

| Paper

Informal workers are vulnerable to economic shocks since their jobs lack health and retirement benefits. Are workers informal because they do not value these benefits offered by formal jobs? Using a discrete choice experiment on a random sample of 2,900 "mom-and-dad" store owners and employees in Colombia, we estimate their willingness to pay (WTP) for formal employment benefits, namely health and retirement plans. We find that on average workers are willing to forego up to 28% and 37% of their earnings to access formal health and retirement benefits, respectively. In contrast to previous research, which suggests that workers’ low WTP induced them to become informal under expansions of free or subsidized health insurance programs, our findings suggest that workers value these benefits and point to inflexibilities in the labor market and the rules governing the health and retirement systems as more important hurdles for workers to become formal.

How does Relative Performance Feedback Affect Beliefs and Academic Decisions? Evidence from a Field Experiment

| Paper

Previous literature documents gender differences in high-stakes exam performance. It has been suggested that these differences may be partly due to women "choking" under pressure. However, no causal evidence conclusively shows that choking under pressure causes underperformance in high-stakes settings, or that women are more affected than men. To address this gap, we designed an experiment to test whether women face a gender-specific penalty when performing under pressure. Over 2,600 online participants completed two IQ tests. In the first, stakes were constant across all groups. In the second test, we varied pressure levels by introducing a performance cutoff and, in one treatment group, significantly increasing the stakes. We also conducted a survey providing novel and detailed information on typical earnings on the Prolific platform, confirming that the increase in incentives was substantial compared to standard rewards for these online workers. Our results show that increasing pressure does not significantly affect overall performance, nor does it differentially impact women. However, an exploratory analysis reveals that the large incentives reduce performance among female test takers who may be ``at risk" of choking: women who performed well in the first test and accurately believed so. Our findings suggest that understanding how high-performing women respond to pressure may be a key step in closing the gender gap in examinations.

The Value of a Signal: Information Processing among Students Outside the Lab

| Paper

This paper examines the hypothesized relationship between performance and stress, commonly known as the Yerkes-Dodson Law (YDL). Analyzing exam performance and post-exam survey responses from 35,000 university applicants taking a high-stakes college entrance exam in Colombia, we find empirical support for the inverted-U pattern posited by the YDL. Our results indicate that female applicants are significantly more likely to experience high, sub-optimal levels of stress compared to male applicants. We estimate that high-stress levels result in a performance penalty of 5\% of a standard deviation. Additionally, our results suggest that, in high-stakes settings, interventions aimed at reducing the importance students place on the outcome of the exam may be more effective for optimizing stress than alternatives like stress reappraisal or controlling worry.

Work in progress

Gender Differences in Recognition of Depression and Seeking Help
(with Sara Abrahamsson and Akshay Moorthy)

Encouraging Women Leadership: Evidence from Village Savings and Loans Associations in Uganda
(with Kjetil Bjorvatn, Shyamal Chowdhury, Danila Serra and Munshi Sulaiman)

Mentors for Minority Language Students
(with Astrid Ervik, Julian Johnsen and Kjell Salvanes)

Getting Employers to Hire Workers in Inclusion Policies through Nudges and Motivation
(with Kjetil Bjorvatn, Mathias Ekström and Ranveig Falch)


Teaching

Stress and Mental Health in Education
Summer 2024 - Bergen Summer Research School 2024, Bergen, Norway

Personnel Economics
Spring 2023 - NHH Norwegian School of Economics, Bergen, Norway

Microeconomics I (Consumer Theory)
Summer and Fall 2018 - Universidad del Rosario, Bogotá, Colombia

Introduction to Applied Microeconometrics (One-week Course)
Spring 2016 – Universidad Nacional de Colombia, Bogotá, Colombia