Financial Education and Saving Outcomes for Low-Income Individuals in IDAs: An Application of Propensity Score Analysis

Background and Purpose: Evidence suggests that Americans lack adequate knowledge and skills to make sound financial decisions in an increasingly complex financial environment. Social work interventions offering financial education have become common components of efforts to improve the financial outcomes of marginalized populations. These efforts are informed by financial-capability theory, which focuses on building financial literacy and the ability to enact desired financial behaviors. Individual Development Accounts (IDAs) are one such intervention. These matched-savings accounts  provide low-income populations with savings incentives, financial education, and a connection to mainstream financial institutions. This study focuses specifically on identifying the marginal impact of financial education on several savings outcomes. We also assess whether the relationship between financial education and saving varies by age.

Methods: As in many quasi-experimental program evaluations, this study’s primary empirical challenge is to address inherent endogeneity. Namely, self-selection into a level of financial education cannot be assumed to be random. To address this issue, we employ a propensity score weighting technique that explicitly models and controls for the choice of treatment level. In most propensity score techniques, treatment condition is a binary value. This study’s approach is novel in that treatment is a dosage: the percentage of required financial-education hours completed. Propensity scores are estimated using a multinomial logit model and are then converted into sampling weights for uses in regular ordinary least squares regression. To test the role of participant age, we stratify the sample into two groups (above and below median age) and run propensity score analysis by group. Analysis utilizes a sample of roughly 2,000 IDA participants.

Results: Overall, we find that financial-education dosage matters even if the analysis controls for selection bias. Compared with participants who fail to complete the required education, those who meet or exceed an IDA program’s financial-education requirements have higher monthly net deposits into development accounts, higher deposit-to-income ratios, and a greater percentage of months with a deposit. We also find that education’s impact on savings varies by age. For the older cohort of participants, increasing the dose of financial education is always associated with an increase in saving behavior. However, savings among younger cohort members who exceeded the required education hours by two or more times are not higher than that among members who just met the requirement or exceeded it by a lower amount.

Conclusion and Implications: This work provides additional evidence that financial education can play an important role in design of social work interventions to improve the financial outcomes of marginalized populations. The results suggest that effects of financial education vary by participants’ ages and, thus, that interventions will be more efficient and effective if they target the specific needs of subgroups. This paper also contributes to social work research by demonstrating improved analytical techniques for confronting the selection-bias problem present in much nonrandomized evaluation research. Social work researchers can employ these techniques to improve the validity of findings.