Academic journal article Financial Services Review

Drop-Out from Individual Development Accounts: Prediction and Prevention

Academic journal article Financial Services Review

Drop-Out from Individual Development Accounts: Prediction and Prevention

Article excerpt

Abstract

Individual Development Accounts (IDAs) provide matches for savings by the poor used to build assets. But IDAs cannot help if participants drop out. What predicts drop-out? For IDAs in the American Dream Demonstration, drop-out is less likely if participants already own some assets. In contrast, income and welfare receipt are not linked with drop-out. Drop-out is strongly associated with aspects of IDA design such as match rates, time caps, and the use of automatic transfer. Because drop-out can be predicted, IDA programs can keep costs down while targeting preventative assistance to the most at-risk enrollees. © 2005 Academy of Financial Services. All rights reserved.

JEL classifications: C53; 138; G21; O16

1. Introduction

Development-i.e., sustained improvement in well being-requires saving to build human, financial, social, and physical capital. Many U.S. policies use tax breaks to subsidize saving, but tax breaks are weak incentives for poor people (Woo, Schweke, & Buchholz, 2004; Seidman, 2001; Howard, 1997; Sherraden, 1991).

Individual Development Accounts (IDAs) are a new policy instrument designed to help the poor build assets (Sherraden, 1988). Instead of tax breaks, IDAs provide matches for savings used to build human capital (via post-secondary education), physical capital (via home purchase), or business capital (via micro-enterprise). IDAs also build human capital via financial education and social capital via support from program staff.

Saving is difficult for anyone, and it is especially difficult for the poor because they already have lower consumption, they have fewer opportunities to increase income, they have fewer existing assets available to shift into IDAs, and they suffer more frequent shocks to income and expenses. Thus, some IDA participants drop-out, saving little or nothing. For example, 48% of the 2,350 participants in the American Dream Demonstration (ADD) "dropped out" with net IDA savings of less than $100. Drop-out is costly all around; IDA programs lose their investment in participants, and participants lose potential matches. Drop-outs may also become discouraged with saving in general.

Based on the characteristics of participants and on aspects of IDA design, this paper attempts to predict drop-out and to suggest ways to prevent it. Participants are less likely to drop out if they already have assets, whether human (education or age), financial (checking accounts), physical (homes or cars), or social (marriage). In contrast, participants with debt are more likely to drop out. Unlike assets and debt, income and receipt of welfare are not associated with drop-out. Overall, asset poverty-but not income poverty-is linked with greater risk of drop-out.

Aspects of IDA design also predict drop-out. This is useful for prevention; even if policy cannot change participant characteristics, it can change IDA design. In particular, drop-out risk can be reduced by setting higher match rates, helping participants set up automatic transfers to their IDAs, and increasing the number of months eligible to make matchable deposits.

This paper first describes IDAs in ADD. It then reports on a Probit regression that predicts drop-out based on participant characteristics and aspects of IDA design. After checking the model's profiling accuracy, the final section presents a summary and discusses implications for saving and asset building in general.

2. IDAs in ADD

The American Dream Demonstration (ADD) ran from 1997 to 2003 at 14 IDA programs across the United States. ADD was open to people with household income under 200% of the federal poverty guideline. One-half of participants were below 100% of poverty, and one-fifth was below 50%. Compared to the general low-income population, IDA participants were more disadvantaged in that they were disproportionately female (80%), African American (47%), and/or not married (75%) (Sherraden et al. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.