The American Dream Demonstration: Testing Individual Development Account Policy in the U.S

Background: In the last several decades, policy makers have sought to develop policies that increase economic security and build prosperity among low-income households. These programs aim to develop stable assets and complement long-standing social welfare programs that provide short-term income supports. In 1998, influenced by research findings from non-experimental components of the American Dream Demonstration (ADD), Congress enacted the Assets for Independence Act (AFIA) which awarded grants to administer Individual Development Account (IDA) programs. IDAs are asset-building tools that provide low-income households with financial education and matched savings accounts. Withdrawals from these accounts are matched if they are used for purposes such as education, home purchase, or entrepreneurship.

ADD launched a large-scale, longitudinal randomized control trial (RCT) in 1998 to test the impact of IDA programs more rigorously. The proposed study describes the design and implementation of this RCT, the challenges posed by such an effort, findings from the experiment, and how these findings impacted policy development in the US.

Methods: The ADD experiment interviewed 1,103 clients of a community development agency in Tulsa, OK. After the baseline interview, respondents were randomly assigned to treatment and control groups. Treatment group members were given access to IDAs and financial education classes. Control group members were restricted from participation in any matched savings program during the program period. Follow-up interviews were conducted at 18 months (Wave 2) and 48 months, approximately at the end of the program period (Wave 3). Recently, ten - year follow up interviews (Wave 4) were completed to assess the long-term impact of IDAs. The survey measured a variety of outcomes with a special focus on housing and financial outcomes. Multiple techniques are applied to address item non-response and attrition from the survey panel. Data are analyzed with techniques including logistic regression, marginal effects probit regression, mediation models, and difference in differences techniques.

Results: Findings from an Intent-to-Treat (ITT) analysis indicated that among those who were renters at baseline, IDA participation significantly increased engagement in clearing debt activities in preparation for home purchase at Wave 2, and homeownership rates at Wave 3. In addition, preliminary results from Wave 4 indicate that treatment group members are still significantly more likely to own their own home than are control group members ten years after the intervention started. At Wave 4, treatment group members may be as much as 45% more likely than control group members to be homeowners. The inclusion of RCT protocols and longitudinal data collection allows the experiment to extend and strengthen non-experimental findings.

Conclusions: The findings from the RCT component of the ADD help to inform the continuing development of national and international asset building policy. Because of the numerous advantages of RCTs compared to non- and quasi- experimental research designs, evidence from this experiment has been of particular interest to policy makers, funders, and practitioners in the field. ADD also demonstrates the need for and utility of large scale experimental research in the evaluation of social policy.