Introduction

This is the second report from the evaluation of New Hope, an innovative project developed and operated in Milwaukee, Wisconsin, that has sought to improve the lives and reduce the poverty of low-income workers and their families. New Hope relied on several components and services to increase the income, financial security, and access to full-time employment of low-income workers in two areas of Milwaukee. In these target areas, all low-income workers (and those not employed, but willing to work full time) were eligible to receive New Hope benefits. New Hope began operating as a demonstration program in 1994, enrolling volunteers during an intake period that lasted through December 1995.

Reflecting its broad eligibility rules, New Hope served a diverse group of low-income people. For example, 37.5 percent were employed at enrollment, and 84.9 percent had been employed full time during their adult work life (with the average longest full-time job lasting about three years). While 59.8 percent were never married and 18.3 percent were separated, divorced, or widowed, 21.8 percent were married. Men made up 28.4 percent of the full sample, and 37.1 percent of the sample were not receiving AFDC, Food Stamps, General Assistance, or Medicaid at enrollment. Participants, on average, were 32 years old.

New Hope offered access to four distinct program components: an earnings supplement to raise participants’ income to the poverty level for their household, affordable health insurance, child care subsidies, and a full-time job opportunity for those unable to find one. (Part-time jobs also were available for those who needed to supplement an existing part-time job.) In return, the program required its participants to work full time (at least 30 hours a week) and to document their work hours in order to qualify for program benefits. Program representatives ("project reps") would meet frequently with participants to collect their wage stubs, verify their full-time employment, and discuss any needs or concerns related to participants’ employment. Thus, the project combined a requirement to work full time with the necessary supports and guarantees to enable its beneficiaries to meet this requirement.

New Hope operated outside the existing public assistance system, though it was designed to be replicable as government policy. It was funded by a consortium of local, state, and national organizations interested in work-based antipoverty policy, as well as by the State of Wisconsin and the federal government. It was designed and operated by a community-based nonprofit organization, the New Hope Project, and thus provides insights into the role nongovernmental agencies can play in income support.

One goal of the project was to provide credible information to policymakers on the implementation, effectiveness, and costs of the New Hope approach. To this end, New Hope contracted with the Manpower Demonstration Research Corporation (MDRC) to conduct an independent evaluation, which began with the start of enrollment. In order to provide a reliable test of the difference the program made, 1,357 applicants were randomly assigned in a lottery-like process to either a program group (with access to New Hope services) or a control group (with no access to New Hope services, but able to seek other services). The difference in the two groups’ outcomes over time (for example, their differences in employment rates or average earnings) are the observed effects or — in the language of evaluations — "impacts" of the program. The 678 participants (that is, the program group members) and their households were entitled to New Hope’s benefits and services for a period of three years, and the last enrolled participants ended their spell of New Hope eligibility in December 1998. To determine New Hope’s effectiveness, this report compares the experiences of these participants during the first two years of their eligibility with the experiences of the 679-member control group.

The previous report presented findings on recruitment, program operations, participation patterns, and participant characteristics.¹ Shorter working papers were prepared to convey early impressions from focus groups with participants, to describe the neighborhood context of New Hope, and to report on the program’s work opportunity component: community service jobs, or CSJs.² The present report is the first to show how the program changed the experiences and lives of New Hope participants during their first two years in it. A subsequent report will cover the last year of the program and two further years of follow-up.

This report addresses important policy questions pertaining to the lives of low-income workers and their families, the choices they make in the labor market, and the effects of financial and other supports on their material and overall well-being.

Following a brief summary of the report’s key findings, the Executive Summary introduces the New Hope Project, its context, and key policy questions. It then presents the report’s findings in detail and concludes with policy implications.

Findings in Brief

Overall, New Hope increased employment and earnings, leading in turn to increased income during the first year of follow-up and enabling more low-income workers to earn their way out of poverty. New Hope’s effects on employment and income, coupled with its provision of health insurance and child care subsidies, set off a chain of beneficial effects for participants’ families and their children. On average, New Hope participants were less stressed, had fewer worries, and experienced less material hardship (particularly that associated with lack of health insurance) than control group members. Participants’ children had better educational outcomes, higher occupational and educational expectations, and more social competence; boys also showed fewer behavior problems in the classroom.

Analyses found that New Hope’s effects varied with the employment status of its participants at random assignment. On the one hand, those working part time or not at all needed to either find a full-time job or increase their hours of work to qualify for earnings supplements, health insurance, and child care subsidies. New Hope project staff assisted them in this process, sometimes by offering CSJs when they were needed. On the other hand, those working full time (30 hours or more) could take advantage of program benefits immediately, without having to increase their work effort. Indeed, New Hope allowed these participants to make ends meet without excessive overtime or simultaneously holding multiple jobs.

Among those not employed full time at random assignment (about two-thirds of the sample), New Hope increased³ both work effort and earnings. Compared to the control group, New Hope reduced by half the number who were never employed during the two years of follow-up (from 13 percent for the control group to less than 6 percent for New Hope participants). Program group members who were not employed full time at random assignment worked in 5.5 out of 8 quarters (three-month periods covered by the earnings data for this report) compared with 4.8 quarters for control group members. The program increased average two-year earnings of the program group (including those who had no earnings) by $1,389, from $10,509 for the control group to $11,898 for the program group. This increase in earnings, boosted by New Hope’s earnings supplement and the Earned Income Credits (EICs), resulted in a substantial income gain of $2,645 over the two-year follow-up period, which made it possible for many of these participants to work their way out of poverty.

CSJs were important in bringing about the employment effect for participants who were not employed full time at random assignment. However, it is unlikely that the entire employment effect was due to this program component. For that to be the case, one would have to assume that no CSJ user would have worked if there had been no CSJs. The data suggest the opposite, because most CSJ users transitioned into unsubsidized employment once their eligibility for CSJ employment ended, and many CSJ users had both CSJ earnings and earnings from unsubsidized employment in the same quarter.

For the remaining one-third of the sample (those employed full time at random assignment), there were modest reductions in hours worked and earnings. These participants were less likely to work more than 40 hours a week and did not experience net income gains, partly because New Hope reduced their receipt of AFDC and Food Stamps. In the second year of follow-up, New Hope’s effect on income for this group was a reduction of $1,148, or 7.5 percent.

The evaluation includes a "Child and Family Study" (CFS) of family dynamics and outcomes for children. Focusing on sample members with children aged 3-12 at the two-year follow-up — 89.8 percent of whom were women, and 69.4 percent of whom were receiving AFDC at enrollment — this study found evidence that New Hope increased the use of center-based child care and other structured out-of-school activities. Among those employed full time at random assignment, New Hope increased the quality of parent-child interactions. This may reflect participants’ greater ability to achieve a sustainable balance between work and parenting by cutting down on long work hours.

To capture possible effects on participants’ children, the CFS obtained permission to survey teachers of these children. From the teacher reports, it appears that New Hope had substantial positive effects on the classroom behavior, school performance, and social competence of children in the sample. These effects occurred primarily for boys, who also showed less problem behavior and higher educational and occupational expectations than boys in the control group.

This report has important implications for policymakers and program developers who are concerned with improving the lives of low-income working families. The analyses show that a package of earnings supplements, health and child care benefits, and full-time job opportunities can substantially increase the work effort, earnings, and income of those who are willing to work full time, but need assistance to do so. Such effects are not limited to nonworkers and welfare recipients, but extend to many different groups of low-income people.

On the other hand, the analyses show that earnings supplements may lead to modest reductions in work effort among those already working full time or more than full time. Interestingly, New Hope shows that such reductions can be kept to a minimum and can actually benefit the families involved to the extent that these reductions limit excessive overtime or multiple jobs.

Finally, the New Hope evaluation shows how modest changes in income, employment, and family resources can have significant effects on noneconomic outcomes, such as family well-being and child outcomes. A narrow focus on economic outcomes may understate the effects of interventions like New Hope, whose benefits extend beyond those outcomes.

The New Hope Project

New Hope offered low-income workers in two areas of Milwaukee an opportunity to use a comprehensive set of integrated program services, designed to address longstanding problems associated with the low-wage labor market and delivered in a small-scale, friendly, and respectful environment. The program had broad eligibility rules, applying to any adult in the target areas (two zip codes) whose income was below 150 percent of the federal poverty level and who was willing to work full time. It was not limited to welfare recipients or families with children. The program had four components, which could be used separately or in any combination suiting program participants. For persons who worked at least 30 hours a week, New Hope provided the following:

• Earnings supplements, which were designed to complement the state and federal Earned Income Credits (EICs) — refundable tax credits for low-income working families — in order to raise the income of full-time workers to the poverty level. In designing the structure of these supplements, program developers tried to make sure that additional work effort or higher wages would always increase participants’ overall income. This was done by reducing the proportion of each additional dollar earned that is lost to taxes or reduced benefits. In other words, program participants were able to keep more of their earnings gains, giving them an incentive to increase their hours of work and look for better-paying jobs. At the same time, the supplements raised their income to the poverty level. On average, the 78.0 percent of program group members who received any earnings supplements received $1,165 over the two-year follow-up period. (The average for all participants was $911.)

• Affordable health insurance, which was available to any participant who did not already have access to such coverage through an employer or government-provided health plan. Lack of such insurance is a continuing source of concern for low-wage workers, one they often cite as an impediment to their trying to leave welfare for work. New Hope required a copayment, increasing with income. This service was used by 47.6 percent of participants. (New Hope spent an average of $1,464 per program group member over two years.)

• Child care subsidies, which were available to parents of children under age 13. The cost of child care is a major concern to low-income workers and their families. Although there are public child care subsidies for welfare recipients who go to work, the programs that provide these subsidies sometimes have long waiting lists. Low-income workers who have not recently received welfare have an even harder time accessing such subsidized child care. New Hope allowed participants to find their own licensed child care arrangements and then paid most of the expenses involved (the copay increased with a family’s income). This service was used by 27.9 percent of New Hope participants (38.8 percent of program group members with children). (New Hope spent an average of $2,376 per participant over two years.)

For those willing to work 30 hours a week, but unable to find such full-time employment, New Hope provided:

• Community service jobs (CSJs), which were wage-paying positions with local nonprofit organizations, available to those who wanted to work full time, but could not find a full-time job on their own. CSJs were not automatic: Participants had to apply for them and could lose their CSJ if their attendance or performance on the job was poor. Each CSJ was limited to six months in duration, but participants could work in CSJs for a total of 12 months. CSJs were used by 32.0 percent of all participants. On average, participants who worked in a CSJ earned $3,000 during the two-year follow-up period. (The average for all participants was $945.)

Program Context

The New Hope evaluation unfolds in the context of rapidly changing labor markets and welfare environments, both in Milwaukee and across the United States. In many ways, the New Hope Project foreshadowed some of these changes, and in some instances it directly influenced state and local welfare policy. During the years covered by this evaluation, active social policy and a generally vibrant economy combined to make work easier to find and more rewarding for many low-income people in Wisconsin. Since New Hope was first conceived, unemployment in Milwaukee County has fallen from 6.5 percent to as low as 3.6 percent, the minimum wage has increased from $4.25 to $5.15, and the state and federal EIC programs have been expanded twice. Since the end of the two-year follow-up period covered in this report, state Medicaid programs are being expanded to include low-income working adults even if they do not receive public assistance.

At the same time, the state’s welfare system has been dismantled, replaced with a work-based system of public assistance called Wisconsin Works (W-2). It began during the last four months of the period covered in this report. More relevant to the findings presented here was a program preceding W-2, entitled Pay for Performance, which required work and work-related activities of every welfare recipient in Wisconsin. All these changes in state welfare policy took place within the larger context of federal welfare reform. The landmark 1996 federal welfare law ended the 60-year-old Aid to Families with Dependent Children (AFDC) program and its entitlement to cash welfare assistance, placed a five-year limit on most families’ receipt of federally funded cash welfare, and required states to place an increasing share of their caseload in work or work-related activities. States now have major responsibility for designing programs for the poor, and they receive block grants of federal Temporary Assistance for Needy Families (TANF) funds.

The Milwaukee economy, and the policy changes that affect the supports available to members of both the program and control groups, makes this a conservative test of New Hope. The changes have diminished the difference between what New Hope offers and what is available outside New Hope, making it more difficult for the project to create a net difference.

Policy Lessons: What Can We Learn from New Hope?

The New Hope Project offers an opportunity to learn about relevant and innovative approaches to the ongoing problems of low-income workers. Following are some of the questions that are particularly important in the current post-AFDC policy debate about helping families, supporting work, and increasing self-sufficiency:

• With supports that make work pay, will low-income people work their way out of poverty? How much will various incentives induce people to work? Is the problem that people need some support, or are they just unable or unwilling to work?

• Can such supports foster full-time work? Many low-income people work part time or intermittently. With better supports, will they work full time?

• Is it possible to make work pay without reducing work effort? The New Hope program supplemented the earnings of its participants, which in theory is a good way of providing financial support to low-income families because it rewards work instead of idleness. However, past research involving income subsidies for low-income workers (implemented without providing work incentives like those in New Hope) has left a legacy of discouraging findings, showing that such subsidies reduced work effort. Could New Hope do better?

• Should interventions like New Hope be targeted at those not already working full time? Inclusiveness was an important aspect of the New Hope program, seeking to serve not just welfare recipients or people with poor work histories. However, what is the price of inclusiveness? Does it dramatically increase program cost? Do those already employed benefit from the program? Does being inclusive have other benefits?

• Does subsidized employment work? New Hope provided CSJs to participants who could not find full-time work on their own. This is another promising approach to helping low-income workers who may have a hard time finding their way into the labor market. But does it work? Do these jobs increase employment or do they just offer an easy alternative for people who otherwise would have found a regular job on their own? Did they set up and maintain a pool of public service jobs that are more than "make work"?

• How much do health insurance and child care subsidies matter? New Hope offered health insurance and child care subsidies. The need for these services is widely documented and proclaimed. But would low-income workers use them? Would they appreciate these benefits as making a difference in their lives?

• How important is the nature of staff-participant interactions? New Hope operated on a small scale and was based in the target areas it served. Staff developed a more positive relationship with participants and interacted with them more frequently than is typical in welfare offices. Does such an approach affect the quality of program operations and the use of program services?

• If more people work and their income increases, is their family life improved? Poverty and low-wage work can be stressful for families. Is it possible to improve family life by supporting employment and increasing available income? Could increased employment have negative consequences for family well-being?

• How do make-work-pay policies affect children? The American public wants those parents who can work to do so. But the public remains concerned about the children in poor families. How might these children be affected by policies that support work?

Limitations of This Evaluation

In this demonstration, the New Hope offer was available to program participants alongside the existing welfare system. While New Hope designers thought of the program as an alternative to this system, many participants continued to use public assistance or Medicaid, either along with or instead of New Hope benefits. Therefore, the demonstration does not fully answer the question: What if we replaced the current welfare system with a work-based set of supports like those available in New Hope? Rather, it addresses the question: What if we added the supports available in New Hope on top of existing policies and programs? In addition, the demonstration provides a definitive answer to that question only for persons like the volunteers who enrolled in New Hope and who live in labor markets like Milwaukee.

Theory and Expectations

The design of the New Hope program was guided both by practical consideration of the challenges facing low-income workers and by theoretical expectations about how people respond to financial incentives. As mentioned above, New Hope was targeted primarily at specific problems inherent in the low-wage labor market, such as "poverty wages," lack of health care coverage, intermittent unemployment, and lack of good, affordable child care. However, as the program was being developed, the expected behavioral responses of those who would benefit from program services were very much part of the discussion. As noted above, prior evaluations of other interventions targeted at low-income workers had found that income subsidies could significantly reduce the work effort of some workers, even if the same programs enticed others to seek employment. This phenomenon, discussed more extensively in Chapter 4 of the report, is potentially costly to society and to participants. In the case of New Hope, these considerations led program designers to limit benefits to those working at least 30 hours a week. This ensured that any reductions in work effort would be small, and it also provided an added incentive to those not already working full time to make an effort to reach a higher level of employment.

The goals and expectations of program designers were not limited to participants’ earnings and income. In addition to those "economic" outcomes, they targeted other aspects of participants’ lives, including their access to health insurance and affordable child care and their overall financial situation. By guaranteeing a full-time job and by supplementing participants’ earnings, New Hope was expected to reduce the stress and financial worries that are common among low-income workers. By allowing some workers to reduce overtime and drop second jobs, the program might free up more of their time for personal development and family time. And by exposing sample members’ children to subsidized, good-quality child care and after-school care, the program might improve their well-being and school readiness, just as the expected increase in family income and greater financial stability might benefit these children.

Data, Samples, and Research Methods

This report relies on a number of data sources for its estimates of New Hope’s effects. All in all, 1,357 applicants to the program were included in the study and randomly assigned to program or control group status: 678 to the program group and 679 to the control group. For each of these sample members, the researchers collected two years of earnings data from unemployment insurance (UI) records and AFDC, Food Stamp, and Medicaid data from other state databases.⁴ These administrative data were augmented with information collected from a two-year follow-up survey. This survey covered details on employment histories, job characteristics, and additional income sources. It also measured material hardship, access to health care, and sample members’ feelings about their financial situations, job security, and, in the case of program group members, their experiences with the New Hope program.

Although all program and control group members were approached for this survey, some could not be found and others refused to participate, leaving a sample of 1,086 for analyses involving survey questions.

For the 678 program group members, data from the New Hope management information system (MIS) were added to the administrative and survey data. These MIS data cover participation in the program, use of program benefits, and earnings from CSJs. Administrative data from New Hope were also used to estimate program costs.

The analyses of child and family outcomes rely mostly on an expanded version of the two-year survey, conducted in respondents’ homes. Special age-appropriate modules were added to the survey to be administered to respondents’ children. The Child and Family Study (CFS) survey was targeted at 745 adult sample members and completed by 591. In many cases, more than one child per family was included in the study, resulting in a sample of 927 children for most analyses.

Finally, if children were in school, their teachers were sent a questionnaire (with permission and assistance from the children’s parents) which contained a number of scales measuring behavior and performance in school. These assessments are a primary source of data on relevant child outcomes; they are available for 420 children in the study.

Most analyses presented in this report identify program effects using straightforward comparisons of outcomes for program and control group members. Because sample members were randomly assigned to either the program group (and thus eligible to participate in New Hope) or the control group (not eligible), the only systematic difference between the two groups is the assignment of program group members to New Hope. This means that any differences in outcomes measured at follow-up are attributable to the New Hope program; as noted earlier, such differences are called the program’s "impacts."

Program Implementation and Context

• New Hope was implemented successfully and delivered benefits and services to those who qualified. Some participants did not access benefits as often as they could have, either because they did not fully understand the program procedures or chose not to report their earnings each month as required.

Implementing a program like New Hope poses important challenges to program developers, managers, and staff. The New Hope program was designed around a set of complex rules centered on the requirement that participants work 30 hours a week on average to qualify for program benefits; they were required to submit wage stubs monthly, which then were incorporated into a management information system for calculation and distribution of benefits. New Hope program staff were successful in developing such a system and implementing it in a real-world setting. Participants were paid their benefits on time. In interviews, participants expressed their satisfaction with New Hope, comparing the program favorably with other employment and welfare programs they had experienced. They consistently rated the support received from project reps as "what they liked best" about New Hope.

However, New Hope staff and management did experience some difficulty in getting participants to understand and follow program rules. Many participants did not maximize their use of program benefits because they failed to comply with these rules, falling short of the required work hours or neglecting to submit required documentation. Participants also occasionally expressed dismay at the month-to-month variation in benefit levels, which was a function of sometimes small month-to-month changes in earnings or in the number of pay periods in a month. The fact that participants had to "renew" their commitment to the program monthly (because they were required to hand in their pay stubs) may have led some to leave New Hope even when they were still eligible for benefits. Staff also had difficulty getting participants to make full use of the EIC, which is an integral part of the New Hope benefit calculations. It is likely that some of this confusion would not occur if New Hope was an ongoing, widely available program.

• People’s initial experience with New Hope differed depending on their employment status at the time they enrolled. Those employed full time could receive benefits immediately, but those not employed full time were more focused on finding a job or increasing their hours.

New Hope’s requirement that participants work 30 hours a week made their initial program experience dependent on their employment status. About one-third of program participants entered the study working full time, attracted by the financial benefits and help with health and child care. Program staff would explain the program rules to participants and help them access health insurance and child care benefits. The remaining two-thirds of participants experienced a different initial contact, which was focused on their need to find a full-time job. After a job search of eight weeks, these participants would have access to the program’s CSJ component. In the meantime, project reps would give them job leads and advice on how to get a job. Thus, these participants would be more likely to actually experience a change in their initial employment status, either finding a job if they were not working or finding a full-time job if they were working part time.

• The New Hope program operated in two inner-city target areas with high rates of poverty and limited economic opportunities. However, the regional economy was healthy and other changes in the environment also promoted work among low-income residents of Milwaukee.

A neighborhood survey conducted in New Hope’s target areas before the program began found high rates of poverty and a large contingent of low-income workers who could have been eligible for New Hope if it had operated on a larger scale. Analyses of job opportunities found most openings to be dispersed in the suburbs surrounding Milwaukee, either difficult or impossible to access without a car. Many positions also required post-secondary educational credentials, which few low-income residents in New Hope’s target areas had. Nevertheless, the Milwaukee economy was generally very good during New Hope’s implementation, making it relatively easy for many participants to find and maintain full-time employment.

As discussed earlier, the welfare environment was changing rapidly during the time of this study. General Assistance (cash welfare for low-income adults who do not have dependent children) was eliminated and welfare rolls were reduced through new welfare-to-work programs like Pay for Performance. New federal legislation eliminated the AFDC program, replacing it with TANF, whose incarnation in Wisconsin (labeled Wisconsin Works, or W-2) took effect, however, near the end of the two-year follow-up period covered in this report. Changing welfare rules and attitudes together with an improving economy caused increasing numbers of welfare recipients to leave the rolls and enter employment, offering both New Hope participants and control group members a substantial incentive to work.

Use of Program Services

• A large majority of those assigned to New Hope (79.2 percent) received program benefits, but few received such benefits every month and for many participants program benefits were limited in scope and duration.

The design of the New Hope offer directly influenced, and often limited, the use of program benefits. First, the offer was extended only to those willing and able to work at least 30 hours a week. If, for any reason, a participant could or did not want to comply with this requirement, his or her eligibility for program benefits may have been interrupted (project reps did continue to extend help and support).⁵ Second, the value of benefits was linked to participants’ income and decreased substantially as their income approached 200 percent of the poverty level for their family, or $30,000 a year, whichever was higher. At that point, earnings supplements were quite small, and copayments for health care and child care were larger. Third, three of the four primary benefits (health insurance, child care, and CSJs) were useful only to a subset of participants. For example, participants with Medicaid or free employer-provided health insurance did not need New Hope’s health coverage or its contribution to employee copayments; those with steady employment (or good job-seeking skills) did not need CSJs; and those without children had no use for New Hope’s child care subsidy. Consequently, the program allowed participants to use the components they needed when they were ready to use them.

The consequences of this approach for participation patterns are presented in Table 1, which shows that 79.2 percent of program group members used any financial program benefit, with almost all of them (78.0 percent of program group members) receiving at least one earnings supplement.⁶ In contrast, only 47.6 percent used New Hope’s health plan (or received help in paying an employee copay), and only 27.9 percent used child care assistance. About a third of all program group members (32.0 percent) worked in a CSJ. Program rules and variation in participant needs affected not only overall benefit use rates, but also the length of time that participants used New Hope’s services. The table shows that those who received any financial benefit did so for an average of only 10.8 of the 24 months of follow-up.

Comparing program benefits across the two employment subgroups defined above shows that those employed full time at random assignment received more benefits than those not so employed, reflecting the fact that the former group could begin receiving benefits immediately, while the latter group had to secure a 30-hour-a-week job first. Also, more participants among those not employed full time at random assignment experienced significant obstacles to employment, preventing them from ever meeting the 30-hour requirement (or working in a CSJ).

In analyzing these figures, it may appear that New Hope’s effect on participants’ lives was less profound than it could have been. However, that is not necessarily the case. First, it is important to consider the program’s effects on participants’ behavior even if they did not receive an earnings supplement or child care assistance in a particular month. If, for some reason, participants failed to work 30 hours a week or were not "ready" for full-time work, New Hope still offered them an incentive to continue pursuing full-time employment, an incentive that was backed up by a CSJ when they needed it. If participants already had health insurance or child care arrangements, the availability of a reliable backup might offer some peace of mind. Second, New Hope’s project reps met with most participants on a regular basis. These meetings, and what they accomplished, are not reflected in the figures in Table 1, but the findings on social support shown in Table 7 suggest that the one-on-one support from project reps meant a great deal to participants. In fact, it may have been a key program component, setting New Hope apart from other programs and benefits available to low-income workers.

Employment and Earnings

• New Hope increased the work effort and earnings of those not already working full time.

For the two-thirds of the sample not employed full time at random assignment, New Hope provided a clear positive incentive to work and to work longer hours. The lower panel of Table 2 shows that such an incentive can increase employment, especially when backed up with CSJs for those who need them. In the two years of follow-up, New Hope reduced by half the number of sample members who were never employed (from 13 percent for the control group to less than 6 percent for program group members), and it increased the number of quarters that these sample members were employed by 0.7 of a quarter and increased earnings by 13.2 percent ($1,389). Both of these effects are substantial, especially given the high level of work effort among control group members. (Again, these data are for all sample members, including those with no employment or earnings.)

Not shown in the table is the extent to which CSJs contributed to these program effects. Although it is not possible to know how program group members would have responded to New Hope in the absence of CSJs, we do know that 32 percent worked in one and that CSJs contributed $945 to participants’ average two-year earnings. This suggests that CSJs played an important role in bringing about New Hope’s impacts on employment and earnings.

• New Hope did not change the rate of employment of those employed full time at random assignment, and while it does appear to have reduced earnings somewhat, this effect was not statistically significant.

As pointed out above, New Hope offered different incentives to those who were employed full time at random assignment and those who were not. One might expect those employed full time to reduce their work effort in response to the increase in disposable income experienced while in New Hope. On the other hand, imposing a 30-hour-a-week minimum on hours worked would limit any such reductions, and other New Hope services and guarantees might help these participants to stay employed full time throughout the follow-up period.

The upper panel of Table 2, showing impacts on employment and earnings for this group, indicates that New Hope was moderately successful in preventing reductions in work effort among those employed full time at random assignment. The very high levels of employment in the control group make program-induced increases in employment very difficult to achieve. The estimated impacts on two-year earnings are negative for this group, but this reduction is not statistically significant.⁷ It appears that New Hope’s supports may have slowed the growth in earnings of participants who were employed full time when they entered the program.

• New Hope somewhat reduced hours worked by those employed full time at random assignment. It did so primarily by reducing the number of weeks in which these participants worked more than 40 hours. There were no statistically significant reductions in full-time work.

The upper panel of Table 3 shows impacts on hours worked and on other job characteristics for those employed full time at random assignment. Aside from the program effects, it is noteworthy how high the average levels of work effort were in this subgroup. Members of the control group worked an average of almost 3,600 hours in the two years of follow-up, which translates into a weekly average of 34.6 and includes any periods of unemployment or part-time work. Thus, many control group members (and program group members) must have worked substantially more than 35 hours a week when they worked.

New Hope reduced hours worked by those employed full time at random assignment, mostly in the first year of follow-up and mostly by limiting overtime (and second jobs). In the first year of follow-up, program group members in this group worked 150 fewer hours than their counterparts in the control group, a reduction of 8.1 percent.

As expected, given New Hope’s program rules, there was no reduction in the number of people who worked at least 30 hours a week. There were also no statistically significant reductions in the number of people working at least 40 hours a week. However, program participants were less likely to work more than 40 hours in an average week.

If one looks at job characteristics, it appears that the jobs held at follow-up by program group members employed full time at random assignment might not have been as good as those held by control group members. The average hourly wage at follow-up was 46¢ lower for program group members than for control group members, an effect that may be related to the reduction in overtime, but may also reflect program group members working in CSJs (which pay only minimum wage). In addition, control group members had more fringe benefits than program group members, possibly a result of the fact that New Hope provided health insurance, reducing participants’ incentive to find a job that provided it.⁸

• New Hope increased hours worked by those not employed full time at random assignment. This effect is a combination of nonworkers becoming employed and others increasing their hours to meet the 30-hour minimum to receive benefits.

Program effects on hours worked among those not employed full time at random assignment were substantial in both years of follow-up. Overall, hours of work were increased by 285, or 12.1 percent (lower panel of Table 3). This was achieved by reducing the number of months with no work from 9.2 to 7.9 and reducing the number of months with some, but fewer than 30, weekly hours worked from 3.4 to 2.4. These effects represent a shift in the work patterns of these sample members, brought on to some extent by participation in CSJs.

There were no statistically significant program effects on characteristics of the jobs held by those not employed full time at random assignment.

• Among those not employed full time at random assignment, the strongest earnings effects were found for participants with only one of a number of potential barriers to employment.

A further breakdown of the group that was not employed full time at random assignment revealed a pattern of program impacts that depended on the number of potential employment barriers that participants had, such as having limited work experience, having very young children, or lacking an educational credential. New Hope program participants best able to translate program benefits into sustained earnings increases came into the program with one potential barrier to employment. The program made less of a difference for those with none of the potential barriers or those who had two or more. This pattern of findings (not shown in tables) suggests limits to the New Hope model, which may be less necessary for some participants and not strong enough for others.

• New Hope’s effects on employment and earnings showed similar patterns across a wide range of subgroups and did not vary between the two target areas served by the program.

In addition to the subgroups defined by employment status at random assignment, program effects were examined for people varying in family status, gender, ethnicity, welfare receipt at random assignment, and target area. None of these analyses showed significant variation in impacts. This implies that New Hope’s effects were widespread and not limited to a single group or target area. (These analyses are not shown in tables.)

Welfare Receipt, Income, and Material Well-Being

• Overall, New Hope participants did not receive fewer AFDC and Food Stamp benefits than their counterparts in the control group. However, in the second year of follow-up those employed full time at random assignment experienced larger reductions in their receipt of public assistance than control group members.

New Hope was not designed or operated as a "welfare-to-work" program, although it was billed as an alternative to welfare for working poor families; that is, the program did not emphasize typical welfare-to-work services, such as job club and job training. Although program designers expected to find indirect effects on welfare receipt by increasing sample members’ earnings or income, pursuit of such effects was not part of the original program design. New Hope program group members who were receiving welfare continued to be subject to any mandates imposed by the welfare department, such as those in the Pay for Performance program.

One might expect to see reductions in the receipt of AFDC and Food Stamps as a consequence of the increases in earnings discussed above. However, this pattern of impacts was not found. Both program and control group members received substantially reduced public assistance during the follow-up period. But rather than further reducing welfare receipt among those not employed full time at random assignment (the group experiencing impacts on earnings), the program accelerated transitions from welfare for those who were employed full time, and only during the second year of follow-up. Table 4 shows that in the second year those employed full time at random assignment received $445 less in AFDC benefits (a reduction of 37.7 percent) and $274 less in Food Stamps (a reduction of 23.5 percent).

Thus, rather than reducing welfare receipt through increased employment, it seems that New Hope effected such reductions by offering those who were close to leaving welfare anyway alternative sources of support. In other words, to some extent New Hope’s supplements and in-kind benefits replaced welfare and Food Stamps for these families.

One might have expected to see reductions in welfare receipt tied to increased work effort for those not employed full time at random assignment, but no such reductions materialized. (In fact, New Hope increased the amount of Food Stamps received by this subgroup in the second year of follow-up, a program effect that is difficult to explain.) The lack of reduction in welfare receipt in this group may be due to changes in welfare rules that would have delayed or prevented such reductions — for example, increased earnings disregards, which allow people to earn more without having their welfare grant reduced. On the other hand, all participants and control group members volunteered to enroll in New Hope, expressing their ability and willingness to work full time. This means that many would have left welfare anyway, limiting New Hope’s effects on this outcome.

• New Hope caused a modest increase in sample members’ income, an effect that was concentrated among those not employed full time at random assignment.

One of New Hope’s primary goals was to increase the income of low-wage workers and to reduce poverty among them. Table 5 documents the extent to which the program met this goal, focusing on two-year cash income and Food Stamps for the full sample and the two employment subgroups. The table shows that by increasing and supplementing earnings, New Hope increased both "earnings-related income" (income directly tied to one’s earnings) and total income. However, these effects were modest for the full sample, representing increases of $1,718 and $1,611 for earnings-related income and total income, respectively. This represents 10.8 and 7.1 percent of the income available to these participants in the absence of New Hope (as captured by the control group).

The subgroup breakdown shows that all of this effect is concentrated among those not employed full time at random assignment, for whom there was a more substantial increase in total income of $2,645 (11.8 percent), mostly resulting from an increase of $2,450 in earnings-related income (20.3 percent). No such effects were found for those working full time at random assignment, who actually lost some income in the second year owing to the aforementioned reductions in receipt of AFDC and Food Stamps.

• By supplementing earnings, New Hope increased the number of sample members whose employment yielded enough income to lift their family out of poverty.

Another way to look at New Hope’s effects on income is to focus on sample members’ ability to rise above the poverty line using only their own earnings and benefits directly connected to their work (EIC and New Hope earnings supplements). Ultimately, this outcome best captures New Hope’s underlying philosophy: making work pay so that full-time workers would not be poor. Table 6 summarizes the program’s effects on this poverty measure for the two employment subgroups. For the program group as a whole (not shown in the table), New Hope increased the number of participants whose earnings-related income was above the federal poverty line for their family by 5.6 percentage points in year 1 and by 7.8 percentage points in year 2.

Nevertheless, most program participants were unable to "work their way out of poverty" using only the regular earnings and CSJ wages of a single worker, even after New Hope supplements and EIC were included. The impacts on poverty status did not vary significantly across the two subgroups defined by employment status at random assignment.

• New Hope reduced material hardship, partly by increasing participants’ incomes, but more importantly by providing participants’ households with health insurance and subsidized child care.

The effects on income and poverty presented thus far fail to consider the contributions made by the program in providing health insurance, child care subsidies, and support by project staff. As discussed in an earlier section, New Hope spent more money on health insurance and child care than on earnings supplements and CSJ wages. Having access to these benefits and being able to afford them can greatly add to the material well-being of low-income households. The New Hope survey measured impacts on material well-being by asking respondents about a number of different material hardships that commonly affect low-income households, including unmet medical and dental needs, periods without health insurance, housing problems, and utility shutoffs. Program effects on these outcomes are shown in Table 6.

The table shows that New Hope did not produce improvements in all of these areas, but significantly reduced material hardships associated with lack of health insurance. Those effects were stronger for those not employed full time at random assignment, although differences across subgroups were not statistically significant.

Also, analyses of survey data found that program group members spent significantly less of their own funds on child care than control group members, despite the fact that they were more likely to use center-based care, which tends to be more expensive. (Impacts on child care use are discussed in more detail in a later section of this Executive Summary.)

Stress, Worries, and Emotional Well-Being

• New Hope reduced stress and worries reported by participants, but it increased time pressure in the lives of those who worked more in response to the program’s incentive. The program also increased the social support available to participants. However, New Hope did not improve participants’ feelings of depression, mastery, or self-esteem.

To gauge the less tangible benefits of New Hope, the survey asked sample members about issues like stress, financial worries, satisfaction with their standard of living, and social support. Sample members who were part of the Child and Family Study (CFS) were also asked about depression, mastery, self-esteem, feelings of agency, and time pressure. An analysis of the program’s effects on these outcomes showed an interesting pattern, summarized in Table 7.

First, sample members in the CFS reported large and significant increases in social support as a result of their participation in New Hope, probably because the program provided valuable advice, assistance, and emotional support. This effect was strong for both employment subgroups and identifies an aspect of the program that has not yet been discussed extensively, namely, the role of project reps. The frequent interaction of these program staff members with New Hope participants can almost be considered a fifth program component (in addition to the earnings supplements, CSJs, health insurance, and child care subsidies), which may have had its own effects on sample members’ well-being. These services were especially valuable to participants who were employed full time at random assignment and were less likely to find case management in other venues).

Overall, New Hope program group members reported being less stressed than control group members (effects for the full sample, not shown in Table 7, were statistically significant). Reasons for the reduction in stress may include, for example, greater financial security, less overtime work, and fewer child care hassles. Among those not employed full time at random assignment, New Hope also reduced a number of specific worries. These program group members were less worried about their medical care, about being able to afford housing, and about their financial situation in general.

Among CFS parents, effects were found on their feelings of agency (measured with the Hope Scale), capturing their sense that they could achieve their goals, but these positive socioemotional effects were accompanied by an increase in time pressure, especially for those not employed full time at random assignment.⁹

The evaluation did not find significant program effects for CFS parents on more stable, personal dispositions such as depression, mastery, and self-esteem.

Impacts on Parent-Child Relations and Child Care Use

• For parents employed full time at random assignment, New Hope moderately increased parental warmth and monitoring of children’s activities.

The Child and Family Study (CFS) component of the New Hope evaluation measured family dynamics and the interaction between parents and children using the participant survey and surveys administered to children. (See Table 8.) It was expected that changes in parental employment, material resources, and emotional well-being would play themselves out in the relationships between parents and children and in the home environments in which children grow up. Boys in New Hope families perceived relationships with their parents to be more positive (not shown in a table).¹⁰ Parents in New Hope expressed more feelings of warmth to their children and monitored their activities more. These positive effects on parents’ behavior were concentrated among those employed full time at random assignment, suggesting that New Hope modestly improved the lives of these families, perhaps by allowing parents to cut back their work hours without significantly reducing their earnings-related income.

Those not employed full time at random assignment did not experience similar effects on parenting, which may reflect increased demands and time pressure for these parents that could offset positive effects from increased resources and employment.

• Through its provision of child care subsidies and its effects on parental employment, New Hope substantially increased children’s exposure to formal child care, after-school care, and other organized activities.

The provision of child care subsidies coupled with increases in parents’ employment were expected to increase the use of child care and to allow parents to select the care they preferred. As parents consolidated their employment, many used New Hope to provide formal center-based and school-based child care for their preschool and school-age children. Although the program effect was significant for the full sample (not shown in a table), there was a somewhat stronger effect on use of center-based care during the preschool and early school years for girls and on use of school-based extended day care for school-age boys (see Table 9).

The New Hope subsidy could be used for licensed home-based child care, but there was no effect of the program on using this type of care. There was some tendency for program group members to use less home-based child care (licensed or unlicensed) than control group members; program group boys were less likely to be cared for by someone outside the household, and program group girls received less care by household members.

In addition, 9-to-12-year-old children whose parents were New Hope participants were more likely to engage in structured out-of-school activities (such as lessons, organized sports, religious classes, clubs and youth groups, and recreation centers). They also watched more TV on weekends (not shown in a table).

Impacts on Child Outcomes

• Teachers reported that boys whose parents were in New Hope had better academic performance, stronger study skills, higher levels of social competence, and fewer behavior problems than control group boys.

The Child and Family Study (CFS) component of the New Hope evaluation included a survey of teachers of children who were in school. New Hope had large effects on the behavior and school performance of boys. Using standardized scales, teachers rated their students’ academic performance, classroom skills (for example, ability to work independently and to make transitions), positive behavior (for example, social competence), and behavior problems (for example, agression). The teachers, who were unaware of the program or control group status of their students’ families, rated boys whose parents were in New Hope significantly higher than control group boys on school performance, classroom skills, and positive behavior and significantly lower on behavior problems (see Table 10).¹¹ No effects occurred for girls, but girls in both research groups scored better than boys on the above measures, possibly indicating less need for improvement.

For families in which parents were not employed full time at random assignment, these effects on school outcomes were reflected in two measures of school progress: Children in New Hope families were less likely to be receiving educational services or to have been retained in a grade than control group children.

• Boys whose parents were in New Hope reported higher educational expectations and higher occupational aspirations and expectations, implying that the program affected their ambitions for future study and careers.

The New Hope survey asked children about their educational and occupational aspirations. It was hypothesized that New Hope might change children’s feelings in this regard, following its effects on their parents’ employment and the children’s own increased participation in child care and after-school programs. Again, substantial impacts were found, but they were limited to boys. Boys whose parents were in New Hope expected to attend and finish college in greater numbers and were more likely to aspire to professional and managerial occupations with high social prestige than boys in the control group.

Two-Year Costs of New Hope

• Over two years, New Hope cost about $7,200 per participant.

It is too soon to write any final assessment of New Hope’s costs and benefits. Two years of costs had been incurred, but the ultimate benefits for families (in terms of employment, income, and poverty) and children (in terms of general well-being and school performance) are not known. Beyond this, the New Hope vision is not easily summarized in any traditional benefit-cost framework, since many of its key goals and achievements cannot be captured in dollar terms. New Hope sought to reduce poverty, improve family functioning, and improve the well-being of children.

With this caveat, the results to date do provide some information on cost effectiveness. Through two years, it cost, on average, approximately $9,000 per participant to provide the New Hope package of services and benefits. Offsetting reductions in public assistance and the value of the work produced in CSJs reduce the costs to about $7,200 per participant. In return, New Hope produced clear impacts on children, moved families out of poverty, and provided participants with about $4,600 in cash or in-kind benefits. Future reports will show the extent to which these total benefits cumulate over time.

Policy Implications

• The New Hope program represents a useful tool for improving the ability of people to earn their way out of poverty. As with any single approach based on employment, however, it is not a panacea.

It is unlikely that any effort to reduce poverty through employment could succeed for all participants, even with bolder incentives. To the degree that policymakers hold antipoverty goals, they will need to consider both employment-based solutions and other means to transfer income.

• It is insufficient to focus solely on work effort and earnings