The Self-Sufficiency Project (SSP) is a large, innovative social demonstration and research project in Canada that tests an employment alternative to welfare. It makes work pay by offering a generous earnings supplement to long-term, single-parent welfare recipients who find full-time jobs and leave the Income Assistance (IA) welfare system. SSP seeks to answer this question: If work paid better than welfare, would welfare-dependent single parents take jobs and leave the welfare rolls?
The project was also designed from the outset to learn about the potential unintended effects of welfare-based work incentives. This special study, described here, answers the question: If SSP is effective, will its generous earnings supplement encourage otherwise ineligible people to apply for or remain on welfare in order to qualify for its benefits?
Findings in Brief
The delayed exit effect was small. Despite widespread concern about entry effects among both policymakers and researchers, results from this first-ever experimental test of entry effects indicate that very few welfare recipients prolong their stay on IA in order to become eligible for SSP's earnings supplement. Thirteen months after random assignment, 57.2 percent of new recipients assigned to the program group had received welfare benefits in 12 of the first 13 months after entering IA, versus 54.1 percent of the control group, a just barely statistically significant difference of 3.1 percentage points. There were no comparable labour market effects.
Effects grow only slightly over time. We had expected the delayed exit effect to grow over time because the cost to a recipient of staying on welfare longer would decline as the eligibility mark was approached. For example, people who had been receiving IA for 10 months would, we thought, be more likely than shorter-term recipients to delay their exit from welfare because they would have to wait only 2 extra months to qualify for SSP — substantially less time than someone who is thinking about leaving IA after only 1 or 2 months. The size of the impact — that is, the difference in the percent of program versus control group members on IA — does grow slightly over time, but no sharp jumps are evident in the last few months before establishing eligibility.
Most eligible people understood the offer. To provide a good test of entry effects, the study designers had to make sure that program group members had information about the program so that they could decide whether or not they should stay on the rolls. But in designing the study, researchers had to walk a fine line between providing more information than would be available in the typical "real world" program and providing too little information. As a typical program benchmark, the Entry Effects Demonstration was designed to provide information comparable with what program and control group members would know about the IA system's work incentives, which disregard (do not count) some earnings when welfare benefits are calculated. By that measure, program group members were generally well informed about the supplement offer. As many as three-fourths of the program group recalled relatively precise information about the SSP program, including the fact that it would provide extra income and that the key eligibility requirements were receipt of IA for a year and then leaving IA and holding a full-time job. By contrast, about half of both program and control group members knew similarly precise information about the lA program's work incentives.
Even among those who were most knowledgeable about SSP's future earnings supplement offer, impacts remained fairly small. A separate estimate of delayed exits was made for the subset of program group members who were well informed about the earnings supplement program. Because this estimate did not involve a pure "experimental" comparison of program and control group members (and used only sample members who responded to the survey), it is not as reliable as the full sample experimental estimate. But it does suggest a possible upper-bound estimate of the expected effect if everyone in the program group was fully informed. Some 60.9 percent of "informed program group members" remained on IA for 12 of the prior 13 months, versus 56.1 percent of control group members — an impact difference of 5 percentage points, still a fairly small effect.
IA recipients explained that they were reluctant to remain on IA longer just to gain eligibility for SSP because they disliked welfare and because it was difficult to find work. Focus groups held with program group members suggested several reasons for the small effects observed here. The stigma of welfare coupled with a preference for work, plus the difficulty many recipients say they have finding work, provides strong impetus to take a job when one can be found. Moreover, single parents' daily lives involve a delicate balance between the demands of child-rearing and the demands of work. These forces preclude many recipients from planning their welfare behaviour around the timing of SSP eligibility.
SSP's one-year eligibility restriction effectively limits both delayed exits and new applicant entry effects. On balance, the results suggest that the one-year eligibility restriction for the SSP program successfully limits the size of the overall entry effects generated by the supplement offer. Moreover, the finding that delayed exit effects among new recipients were small, and that they did not emerge until recipients had been on welfare for several months, suggests that the new applicant entry effect generated by SSP may also be negligible. If new recipients — who have already borne the stigma and cost of welfare — are unwilling to prolong their time on IA in order to qualify for a future SSP earnings supplement, then working poor people who would not typically apply for welfare are even less likely to alter their behaviour. That is, they would be unlikely to enter the welfare rolls and wait the required year just to qualify for an earnings supplement.
Given the widespread concern about unintended effects of new social programs, the SSP Entry Effects Demonstration results are encouraging. They demonstrate first that policymakers can design policies that limit the likelihood that people will alter their behaviour to qualify for a new benefit program — for example, quitting a job or staying on welfare longer than they otherwise would have. And, the small size of the effect, the fact that it doesn't grow substantially over time, and the absence of any effect in the first four months following random assignment all suggest that concerns about entry effects may be somewhat overstated.