The “Make Work Pay” Experiments


Beginning in the 1990s, the “Make Work Pay” experiments tested whether offering earnings supplements would increase employment and income and improve family well-being among welfare recipients. The experiments responded to a fundamental challenge: Low-wage jobs often leave families only barely better off financially than even subsistence-level welfare benefits. As a result, requiring welfare recipients to work, even if it succeeds in increasing employment and reducing welfare receipt, still may not increase income. Earnings supplements sought a way out of this conundrum by augmenting the earnings of individuals working in low-wage jobs.

MDRC has conducted a series of random assignment studies of different earnings supplementation strategies in the U.S. and Canada. The studies have examined the cost of various approaches, as well as their effects on work, welfare receipt, and the well-being of families and children. The results have been promising.

Agenda, Scope, and Goals

The “Make Work Pay” set of experiments encompasses five evaluations conducted by MDRC, each of which included an earnings supplement component: Connecticut Jobs First, Milwaukee’s New Hope Project, Minnesota Family Investment Program, Canada’s Self-Sufficiency Project, and Canada’s Earnings Supplement Project (the latter two were conducted in cooperation with SRDC, MDRC’s sister organization in Canada). The programs provided work incentives in the form of monthly cash payments to supplement the earnings of low-wage workers. The payments were made only when people worked, and the amount of each month’s cash payment depended on the amount of each month’s earnings.

Some of the programs tested were part of broader reforms that included other components as well, from work requirements or time limits on benefit receipt to job search assistance or subsidized childcare. Most targeted current welfare recipients, although Canada’s Earnings Supplement Project targeted recently displaced workers who had filed for unemployment insurance benefits.

The results were positive. The programs increased employment, earnings, and income. Adults who were offered earnings supplements were more likely to work, earned more, and had more income than control group members. When income did not rise, this sometimes appeared to be because earnings supplements allowed parents to cut back their hours of work while maintaining their family’s income. In addition, effects of the policies on employment and earnings were larger and more persistent for a group of very disadvantaged families. The effects of the policies were generally larger and longer-lived for long-term welfare recipients with limited education and work experience.

While effects on employment, income, and well-being were positive, effects on welfare receipt varied with the structure of the earnings supplement offered, as some programs required recipients to leave welfare, while others required them to remain enrolled. In addition, follow-up studies after the programs ended found that the effects of the programs diminished over time — with effects on employment and earnings starting to lessen before the programs ended and with effects on welfare receipt disappearing when the programs ended.

The Make Work Pay experiments also offer important lessons about improving child well-being. Parents’ employment and income gains produced, in turn, modest but important benefits for young children on a range of school measures, as well as, in some cases, on behavioral and social skills. For preschool-age children of participating parents, this was in part because their parents had higher incomes; these children were also more likely to attend center-based care programs, which may also have played a role in producing the effects, although the evidence here is weaker. Remarkably, eight-year follow-up conducted as part of the New Hope study found that these positive results lasted into adolescence.

The results for children who were adolescents at the time of the program, however, are less encouraging. Programs that increased the employment of mothers of teenagers were associated with modestly unfavorable school-related outcomes for those teens, although these effects did not appear to lead to negative effects in the long-term.

Design, Sites, and Data Sources

Each of the five programs MDRC evaluated included an earnings supplement component, intended to encourage work and to boost the income of adults who worked. Each was studied using a reliable research design that randomly assigned people to a program group that was eligible for earnings supplements (generally along with other program elements) or to a control group that was not.  What additional services were offered or requirements applied to the program and to the control group varied from program to program. The size and length of the study also varied by program.

Connecticut’s Jobs First Program: The welfare reforms of the 1990s were characterized by one or more of the following three core components: broad and tough work requirements, financial incentives to make work pay, and time limits on cash benefit receipt. Launched in 1996, Connecticut’s Jobs First program was one of the first statewide reform initiatives to include all three, including, at 21 months, one of the shortest state-imposed welfare time limits in the country. MDRC completed a four-year evaluation of Jobs First in 2001. The study used a rigorous random assignment research design to track 5,000 welfare recipients over a four-year period. One-half of the targeted population was assigned to Jobs First, and the other half remained subject to the prior Aid to Families with Dependent Children welfare rules.

New Hope Project: New Hope, which was operated outside the public assistance system but was designed to be replicable as government policy, offered its enrollees three things: job search assistance, and, if no full-time job could be found, a time-limited community service job; a monthly earnings supplement that brought most low-wage workers’ incomes above the poverty level; and subsidized child care and health insurance, both of which phased out as earnings rose. All adults over 18 in two low-income areas of Milwaukee who had income below a certain cutoff and a willingness to work at least 30 hours per week could apply for the program. More than 1,300 New Hope applicants were included in the study. Starting in 1994, each was randomly assigned either to New Hope or to a control group. For those in the program group, benefits were offered for three years; for both groups, follow-up continued for eight years after random assignment, and included an in-depth study of the effects on children and families.

Minnesota Family Investment Program: The Minnesota Family Investment Program (MFIP), piloted from 1994 through 1998, combined financial work incentives and employment mandates. It included a requirement that long-term recipients work or participate in employment-focused services; financial work incentives for recipients who worked; payment of working recipients’ child care costs directly to providers, rather than reimbursement of recipients later; and simpler public assistance rules and procedures that combined different programs into one and provided food stamps as part of the cash welfare grant. MDRC’s evaluation of the initiative, conducted under contract to the State of Minnesota, included more than 14,000 welfare recipients and applicants, most of them single parents. Starting in 1994, each one was randomly assigned either to MFIP or to Aid to Families with Dependent Children, the traditional welfare program. To test the effects of the financial incentives without the employment mandate, individuals in the urban counties were also assigned to a third program that only provided the financial incentives. Follow-up continued for six years. The study included extensive analysis of the program’s effects on families' and children's well-being as well as its economic impacts.

Canada’s Self-Sufficiency Project: The Self-Sufficiency Project (SSP), a large-scale demonstration launched in Canada in 1992, was a voluntary program, targeted to single-parent welfare recipients, that offered a generous earnings supplement to those who left welfare for full-time work. The supplement roughly doubled enrollees’ monthly pretax earnings and could last up to three years. In order to qualify for the SSP supplement, enrollees had to have received welfare payments for at least a year, and then start working full time (at least 30 hours per week) within two years of becoming eligible for supplement payments. They could not receive cash welfare and the earnings supplement at the same time. The SSP evaluation encompassed about 9,000 single-parent welfare recipients and applicants in British Columbia and New Brunswick who volunteered for the study. Starting in 1992, each one was randomly assigned to either the SSP group or the control group, which was enrolled in the traditional welfare program. The main study included individuals who had been on welfare for a year and who could qualify for supplements immediately. To find out whether job search assistance would enhance the supplement’s effects, the demonstration also tested SSP Plus, a variant that offered enrollees job search services as well as the supplement. To find out whether SSP would encourage some parents to remain on welfare to qualify for supplement payments, the demonstration also included a study of new welfare recipients. Follow-up continued for six years.

Canada’s Earnings Supplement Program: Canada’s Earnings Supplement Project (ESP) was designed to encourage rapid reemployment for displaced workers who had recently filed for unemployment insurance benefits. Unemployment benefit claimants who found a full-time job within 26 weeks were offered two things: first, if the new job paid less than the old one, ESP enrollees received an earnings supplement equal to 75 percent of the difference between the two wages, up to a monthly maximum; second, earnings insurance protected ESP enrollees against a future reemployment earnings loss. About 8,000 displaced workers entered the ESP study over a one-year period from 1995 to 1996. Half were randomly assigned to the supplement group, making them eligible for the earnings supplement or the earnings insurance if they found full-time work within 26 weeks. The other half were randomly assigned to the control group, making them ineligible for the supplement and earnings insurance but eligible for standard unemployment benefits. ESP’s effects were evaluated over a 15-month period.