It didn't work. We found no significant differences in deposit rates across the conditions. Enrollment into automatic deposits did not vary between conditions.
It didn't work. We found no significant differences in deposit rates across the conditions.
These findings suggest suggest that timing of the ask may not affect college savings, at least in relation to tax time and payday. It may be more effective to consider more structural changes to increase deposit activity – for instance, looking for ways we might re-design the incentive structures for CSA accounts to be more appealing or motivating.
Many low- to moderate-income families file their taxes as soon as they can to receive their refunds. Our interventions, however, were sent a few weeks into tax season, Interventions may thus be more successful if they are timed when families receive these refunds.
Beyond overcoming the psychological barriers that make saving for future educational costs difficult, many CSA programs also struggle to capture the attention of parents and get to top of mind. Often finding the right moment to connect with parents about saving for college is difficult. This is especially true if CSA programs are reaching out to parents at the beginning of the school year, when their attentions are often focused on more immediate challenges.
We continued our partnership with the St. Louis Office of Financial Empowerment (OFE). In these two experiments, we attempted to couple college savings nudges with two earnings-related events: tax time – when many families receive sizeable tax refunds – and paydays.
There are several reasons why coupling college savings with tax time and with someone’s paycheck is more likely to be effective than presenting information about the CSA separately.
People are already likely to pay attention to their paycheck and their tax refund. Using these channels is a way of connecting with families when the program is more likely to capture their attention.
Parents are likely to find it easier to set aside some money for their child’s future education right when they are paid or when they receive a tax refund. A previous study we carried out with Digit–a fintech company–showed that asking an individual to pre-commit to saving a portion of their tax refund before they receive it can increase tax-time savings.
Experiment 1: Saving at Tax Time
We hypothesized that making it easier for families to contribute to a CSA at tax-time would increase CSA engagement. We utilized the IRS Form 8888, which allows US taxpayers to split part their refund for savings. Families were randomly sent one of four packets, and some families served as a control - which did not receive anything. We used an additive research design, which means each subsequent condition builds on the last condition.
Experiment 2: Quick Enroll on Payday
We hypothesized that by reducing friction in the savings process and making it easier for someone to save at payday would increase the number of direct-deposit enrollments to the College Kids program. To test this hypothesis, we piloted a program with the City of St. Louis where employees would receive a quick enroll form for automatic deposits into any College Kids CSA.
We designed an enrollment form that would be easier for city employees to complete and return. Employees are still paid by paper checks on a biweekly basis, so we distributed the forms by attaching them directly to their paychecks. We randomly assigned departments to either receive a quick-enroll form or not, making sure that both conditions included departments with different amounts of employees.
We ran two rounds of the experiment, one with each of two paydays. In the first round, we tested the presence and absence of a quick-enroll form; in the second round we added handwritten and printed sticky notes to the quick-enroll forms.
Experiment 1
In total, our sample consisted of 8,324 families, roughly evenly split between each of the five conditions. Across everyone, we observed only 11 families make at least one deposit during the study period. Of those who saved anything, total deposits ranged from $25 to a little more than $250.
We found no significant differences in deposit rates across the conditions. We believe that many low- to moderate-income families file their taxes as soon as they can to receive their refunds. Our interventions, however, were sent a few weeks into tax season. Parents may have already completed the tax filing process before receiving our intervention, which may have contributed to the low deposit rates.
Experiment 2
Enrollment into automatic deposits did not vary between conditions. The findings in this experiment are limited due to the lack of quick-enroll form returns – as in the other two experiments, we observe very low baseline savings activity.
We suspect that there was a limited response in part because few parents had or knew of children who were enrolled in the program. Nevertheless, we still expected more engagement than we saw. Together, both the findings from this experiment and from the tax experiment suggest that the timing of the ask may not affect college savings, at least in relation to tax time and payday.
One interesting avenue that did emerge from this was that several people anecdotally expressed an interest in sponsoring a child’s college savings account, similar to how one would with charitable giving. Overall though, our thinking is moving toward more structural changes to increase deposit activity – for instance, looking for ways we might re-design the incentive structures for CSA accounts to be more appealing or motivating.