Evaluating the Impact of Financial Literacy Education

We conduct a randomized controlled trial (RCT) of a large personal finance course. Focusing on college students, a population for which there is less research on the efficacy of financial literacy education, the study has two parallel goals: i) document self-selection patterns in financial literacy education, including potential selection on treatment effects, and ii) evaluate the impact of financial literacy education. We plan to offer a random subset of students a financial incentive to take the course. We then randomly admit students with varying levels of demand into the course from the waitlist. Our stratified randomization allows us to study not only the causal impact of the course, but also the extent to which willingness-to-pay is correlated with treatment effects. We then link course participation to short-term and long-term outcomes in surveys, administrative credit bureau records, and college class records.

2025-10-01

2027-06-30

We measure short-term and long-term outcomes in surveys and administrative data after the course is complete. Our outcomes focus on measuring the treatment effect of education. Outcomes include financial literacy scores (measuring using hypothetical choices in surveys), time/risk preferences, confidence, demand for education, credit score, delinquencies, and debt outcomes.

We also look at changes in education behavior (major choices, classes taken, etc.) and education outcomes in college course records.

We conduct a randomized controlled trial to investigate the causal impact of financial literacy education on short and long-run financial behavior. Our experiment consists of the following stages: (1) recruitment and baseline data collection, (2) random assignment to the course, (3) measurement of outcomes.

Not available

The treatment condition will be randomly assigned by a computer at the individual level.

Treatment conditions will be assigned at the individual level. Within treatment status, the different versions of the education material will be randomly assigned as well. The size of the treatment and control group will not be balanced in order to maximize power.

No

N/A

We will target a sample of at least 1800 individuals, however the final sample size will depend on funding and our ability to recruit additional students. If recruiting is successful, we will increase the sample size and have an unbalanced treatment and control group given the lower cost of having students in the control group.

We will target at least 1800 participants total with at least 900 students total in the treatment group. The final sample size will depend on funding and our ability to recruit additional students. If recruiting is successful, we will increase the sample size and have an unbalanced treatment and control group given the lower cost of having students in the control group.

We conducted several simulations and power calculations, considering both balanced and imbalanced designs, and the trade-off between detecting short-term survey/credit panel outcomes and long-term credit panel outcomes. Based on these calculations, we believe a sample size of 1800 individuals will be enough to detect treatment effects and treatment effect heterogeneity for both the short-term (survey) and long-term (financial outcomes) outcomes, however these calculations ultimately depend on the success of the first stage. If we find that the first stage is too weak in the pilot, we will apply for additional funding to increase the sample size.