The Minnesota Family Investment Program (MFIP) originated as a new vision of a welfare system that would encourage work, reduce reliance on public assistance, and reduce poverty. The program began in April 1994 in seven Minnesota counties and differed from the existing Aid to Families with Dependent Children (AFDC) system in two key ways: It included financial incentives to “make work pay” by allowing families to keep more of their welfare benefit when they worked, and it required longer-term welfare recipients to work or participate in employment services. Both policies are now key elements of most state welfare programs under Temporary Assistance for Needy Families (TANF), the successor to AFDC.

MFIP was evaluated using a random assignment research design; that is, families were assigned at random either to a program group that was eligible to receive MFIP or to a control group that was eligible for the AFDC system. MFIP’s effects have been measured as the difference in outcomes for the two groups. Effects after three years, or roughly through 1997, are presented in an earlier report.¹ A modified version of MFIP — with a 60-month time limit, less generous financial incentives, and a stricter work requirement — became Minnesota’s statewide TANF program in January 1998.

This report updates the MFIP story in two ways. First, it examines whether the program’s effects held up in the longer term, through six years after study entry. Although MFIP was not designed to be temporary, Minnesota’s statewide implementation of the modified version of MFIP (its new TANF program) effectively ended the original version of MFIP in 1998. A primary question of interest in this report is whether MFIP somehow provided families with a permanent advantage, increasing their employment or self-sufficiency in the long term, or whether MFIP’s effects faded after the program ended. The report also presents new findings on MFIP’s effects on outcomes that were not available or that could not be reliably measured at the three-year point, such as school records data to measure children’s school achievement.

Findings in Brief

The report presents the effects of the original version of MFIP in three areas: economic outcomes, marital and fertility outcomes, and children’s reading and math test scores in the third and fifth grades.² Results are presented separately for single-parent families and for two-parent families. In addition, findings are broken out for particular subpopulations in two cases: (1) where MFIP’s requirements differed, as for long-term recipients, who faced participation requirements immediately, and recent applicants, who faced these requirements only after two years on welfare; and (2) where MFIP’s effects are particularly striking, as for the most disadvantaged single parents.

Single-Parent Families

• For the full sample of single-parent families, MFIP increased employment, earnings, welfare receipt, and income up through the fourth year of the follow-up period, after which MFIP’s effects on economic outcomes dissipated; MFIP’s effects varied across different subgroups of single-parent families, however. The lines in Figures ES.1 through ES.4 present MFIP’s effects — or the difference in an outcome between MFIP families and AFDC families — on employment, earnings, welfare receipt, and income, respectively, for each of three subpopulations: recent applicants, long-term recipients, and the most disadvantaged single parents. Among recent applicants (represented by the dotted lines), MFIP increased employment only modestly through Year 4 and had no effect on earnings, but it did increase income for this group of families, because MFIP redesigned the welfare system to make work pay as earnings rose. Recent applicants in MFIP were also more likely to receive welfare benefits up through the fourth year of the follow-up period. Turning to long-term recipients (represented by the heavy solid lines in the figures), MFIP substantially increased employment, earnings, and income just beyond Year 4. Compared with the control group, single-parent long-term recipients in MFIP were also more likely to combine welfare and work (not shown). The employment and earnings effects for this group faded over time, however, because many parents would have eventually gone to work on their own.³ In contrast, MFIP’s effects on welfare benefits and income for these families appear to have ended when the program ended and was replaced by the statewide TANF program.



• MFIP’s effects persisted up until Year 6 for several of the most disadvantaged groups of single parents, including those with little employment history, long-term welfare receipt, and no high school diploma or General Educational Development (GED) certificate and those with a combination of these characteristics. Effects for the small group of the most disadvantaged single parents who had a combination of these characteristics are shown by the lighter solid line in Figures ES.1 to ES.4.⁴ In contrast to the findings for recent applicant and long-term recipient single-parent families, MFIP’s effects on employment, earnings, and income persist for the most disadvantaged single parents. A primary reason for the persistence of MFIP’s effects over time is that these most disadvantaged single parents were the least likely to have eventually gone to work on their own. The continued earnings gains over the six-year follow-up period suggest that, for this group, the benefits of MFIP may eventually outweigh the costs, in large part because, by the end of Year 4, welfare was no longer being used to supplement earnings. Nonetheless, these families continue to have substantially lower levels of earnings and income than their more advantaged counterparts (not shown).



• By Year 6, marriage rates were similar for MFIP and AFDC single-parent families overall, but MFIP did increase marriage somewhat for some subgroups of single-parent families. MFIP led to a small increase in marriage, primarily among single-parent long-term recipient families through Year 4 — and among several other subpopulations of single-parent families through Year 6.



• Among the full sample of single-parent families, MFIP had no effect on the elementary school achievement of very young children; positive effects did occur for several subgroups of young children for whom data are available. Among families of long-term recipients, for children who were age 2 to 9 at study entry, MFIP had positive effects on maternal reports of children’s school performance and behavior at Year 3.⁵ Data on third- and fifth-grade math and reading achievement were matched to children of all single-parent long-term recipients who were newborn to age 3 at study entry, and MFIP improved third-grade reading achievement (assessed five to nine years after study entry). Strikingly, among the most disadvantaged families, MFIP had large positive effects on the small sample of children who were age 2 to 5 at study entry, nearly doubling the proportion who met grade-level expectation in fifth-grade reading and in fifth-grade math. Together, these findings suggest the potential beneficial effects of large and sustained increases in income (as observed among the single-parent most disadvantaged families) as well as the potential benefits to children of short-term boosts to parents’ employment, earnings, and income (as observed among the single-parent long-term recipient families).⁶ The generalizability of the results should be approached cautiously given the small sample sizes.

Two-Parent Families

• Through Year 4 of the follow-up period, MFIP reduced employment among women in two-parent families; the reduction in family earnings was offset by higher welfare benefits, resulting in no effects on family income. MFIP’s effects were concentrated among two-parent recipient families, possibly because two-parent applicant families rotate off welfare fairly quickly. For two-parent recipient families, the reduction in family earnings came about at least in part because of the number of families who had two wage-earners.



• At the six-year point, MFIP’s effects on divorce varied by the prior welfare history of the two-parent family, leading to no overall effect. MFIP substantially increased marital stability at the three-year point for two-parent recipient families, primarily by reducing reported rates of separation.⁷ While information on separations at the six-year point is not available, analyses of public records data show that MFIP did slightly decrease divorce rates at the six-year point for these families. The pattern of effects among two-parent applicant families, however, is significantly different from the pattern of effects among two-parent recipient families, with a trend toward higher divorce rates among two-parent applicant families in MFIP.



• MFIP had no effect on elementary school achievement of young children in two-parent families. Although MFIP’s effects might have provided support for the hypothesis that marital stability (among two-parent recipient families) can improve children’s outcomes, the evidence to date is not conclusive, especially since data are not available for a broader age group of children and information is not available about children’s social, emotional, and behavioral development.

MFIP Incentives Only

• MFIP’s financial incentives are a core component of MFIP’s positive effects on income. In order to assess which component of MFIP led to its effects, the evaluation includes a third research group (called “MFIP Incentives Only”) that received the program’s financial incentives but was never subject to its participation mandate. For long-term recipients, the incentives when offered by themselves increased employment modestly but also caused some people to cut back their work hours or to take part-time rather than full-time jobs. As a result, MFIP’s financial incentives alone had no effect on earnings but still increased income, since they allowed some parents to use the more generous welfare benefits to make up the difference in the loss of earnings. The full MFIP treatment did increase earnings, because it combined the incentives with a requirement that individuals work or participate in services for at least 30 hours per week.

Policy Implications

• Financial incentives combined with mandates to participate in employment-related services can move more single parents into work but are unlikely to generate long-term increases in employment and earnings for most groups. In MFIP, as in several other similar programs,⁸ the positive effects on employment and earnings tend to fade over time, not because the program ends but because many single parents would have gone to work eventually anyway. Under this scenario, the effect of the program was to speed up their movement into work. This finding and the fact that MFIP’s effects did not fade for the most disadvantaged single parents underscore the potential value in targeting more intensive or more expensive programs toward those who are least likely to go to work on their own.



• In contrast, increases in income could be sustained into the longer term by continuing the program. MFIP probably would have continued to increase families’ incomes if it had continued providing its more generous earnings disregard to eligible families — or, in other words, if the MFIP program and control group members (who were receiving assistance) had not been moved into the new statewide program in 1998. These results are suggestive only, since they are based on nonexperimental analyses, but they are similar to findings from other earnings supplement programs.⁹



• Even temporary increases in family income can benefit children over the long term. The persistence of MFIP’s effects on the young children of single-parent long-term recipients — beyond the economic effects for their parents — is quite similar to findings reported in comparable studies¹⁰ and is perhaps not surprising, given that better academic performance in early years has been found to foster better performance in later years. These findings also suggest that longer-lasting effects on employment and income for very disadvantaged subpopulations can lead to concurrent improvements in children’s academic achievement. Understanding the broad range of potential effects across cognitive and behavioral aspects of children’s development and determining whether effects exist for older children are important matters for future investigation.



• Effects on earnings and income are largest and most sustained when financial incentives are combined with work requirements rather than implemented alone. Offering financial incentives alone can encourage some families to cut back on work. Combining incentives with work requirements can minimize this effect and reduce the overall costs of the program. The value of these cost reductions should be weighed against the potential benefits that families experience when parents are able to spend more time with their children.



• Strategies of increasing income by rewarding work, along with subsequent benefits that might accrue for families and children, should be weighed against increased government costs. MFIP’s use of welfare benefits to supplement the earnings of single parents who worked was a core component of the increased income among MFIP families, but it also led to increased costs. Among single-parent families, this increased income likely contributed to MFIP’s long-term benefits to children’s achievement in school. Among two-parent recipient families, welfare income also played an important, albeit different, role: MFIP allowed one parent to cut back on work effort, and this reduction in family earnings was offset by increased welfare payments, leading to no net difference in family income. Although these effects did not last beyond the fourth year of follow-up, they did appear to influence marital stability somewhat, by slightly reducing rates of divorce in the long term for that subgroup of two-parent families.



• Changes in the welfare system that are aimed at increasing employment and income may have few and small effects on divorce rates and entry into marriage. Although there is less evidence from other studies on this point, the results from MFIP suggest that programs of its type are unlikely to have large lasting effects on marriage rates, although they may increase marriage or marital stability for some families.¹¹ Although lower rates of separation among MFIP two-parent recipient families in the short term could not be followed up in the longer term, there is little evidence here that MFIP’s short-term impacts on separations led to substantial reductions in divorce in the long run. Programs such as MFIP were not explicitly designed to affect marriage or the quality of relationships. It might be that interventions designed to address a broader range of factors influencing marital relationships could more directly influence decisions to marry or stay married.

Notes

¹See Virginia Knox, Cynthia Miller, and Lisa Gennetian, Reforming Welfare and Rewarding Work: A Summary of the Final Report on the Minnesota Family Investment Program (New York: MDRC, 2000).

²The analyses use third- and fifth-grade school records data from 2001 to 2003. Children whose data are available for third-grade reading and math achievement were roughly 0 to 3 years old at study entry. Children whose data are available for fifth-grade reading and math achievement were roughly 2 to 5 years old at study entry. Because random assignment occurred from 1994 to 1996, third- and fifth-grade reading scores for 2001 to 2003 represent a follow-up assessment period of five to nine years.

³In experimental evaluations, the behavior of families in the absence of the program is estimated using the control group.

⁴The characteristics were combined to create a new subgroup, the most disadvantaged single parents. This subgroup consists of families in which single parents received welfare payments in 11 or more of the 12 months prior to random assignment, were not employed in any of the four quarters prior to random assignment, did not have a high school diploma or GED and had completed less than twelfth grade, and represent a subpopulation of single-parent recent applicant and long-term recipient families. Of the 415 most disadvantaged single-parent families, 344, or 83 percent, were long-term recipients.

⁵Lisa Gennetian and Cynthia Miller, Reforming Welfare and Rewarding Work: Final Report on the Minnesota Family Investment Program, Volume 2, Effects on Children (New York: MDRC, 2000).

⁶A notable exception to these favorable patterns of effects is the pattern among children of recent applicant families, for whom MFIP produced neutral effects (and one negative effect) on third- and fifth-grade reading and math achievement. Some possible reasons for this are that children of recent applicants fare better, on average, than children of long-term recipients, and thus have less room for improvement, and that recent applicant families represent a heterogeneous group, some of whom might have entered the welfare system because of family upheaval. See Lisa Gennetian and Cynthia Miller, Reforming Welfare and Rewarding Work: Final Report on the Minnesota Family Investment Program, Volume 2, Effects on Children (New York: MDRC, 2000).

⁷Cynthia Miller, Virginia Knox, Lisa Gennetian, JoAnna Hunter, Marty Dodoo, and Cindy Redcross, Reforming Welfare and Rewarding Work: Final Report on the Minnesota Family Investment Program, Volume 1, Effects on Adults (New York: MDRC, 2000).

⁸Charles Michalopoulos, Does Making Work Pay Still Pay? An Update on the Effects of Four Earnings Supplement Programs on Employment, Earnings, and Income (New York: MDRC, 2005).

⁹Michalopoulos (2005) finds similar results for the New Hope program and Canada's Self-Sufficiency Project — programs that offered time-limited earnings supplements outside the welfare system, based on the condition of full-time work.

¹⁰Pamela A. Morris, Lisa A. Gennetian, and Greg J. Duncan, "Effects of Welfare and Employment Policies on Young Children: New Findings on Policy Experiments Conducted in the 1990s," Social Policy Report 19, 2 (2005).

¹¹Lisa Gennetian and Virginia Knox, "Staying Single: The Effects of Welfare Reform Policies on Marriage and Cohabitation," Next Generation Working Paper (New York: MDRC, 2003).