 I would say to welcome back Greg Duncan, who has spent a good part of his life here at Michigan. Greg is the Edwinna S. Terry professor at the School of Education and Social Policy at Northwestern University and is also a faculty fellow at the Institute for Policy Research. I had the pleasure of being Greg's colleague for at least a few years at Northwestern and I can tell you there's not a better colleague around. Greg is an incredibly productive person. He has, in addition to the book he's talking about today, produced together with Lindsay Chase Lansdale, the book for better and for worse, Welfare Reform and the Well-Being of Children and Families, a book that those of us who are interested in the overlap between sort of child development and public policy consult often. He's produced the book The Consequences of Growing Up Poor, another book that is frequently produced by people in my area of study, frequently consulted, as well as the two volume, Neighborhood Poverty. Greg is an economist by training from here at the University of Michigan, but has spent a great deal of time co-publishing and working with people who are much more in sort of the social psych area and the child development area. And he's really unique in terms of bringing both the tools and perspectives of economics together with the understanding of some of the more developmental literature. I guess that's a very unusual trait and it's one reason I think why he's had the impact that he really has had. He's a member of the American Academy of Arts and Sciences and is the incoming president of the Population Association of America, or recently elected. I don't know quite when you get inaugurated into that position. It's just absolutely great to welcome Greg back because I know that he's been here for many years, not only because he has his degree here, but as some of you know, he was the PI on the Panel Study of Income Dynamics for many, many years. In fact, if you go to that six-floor conference room at ISR, there's this great picture of Greg and Mary Carpenter working together, I assume, on some arcane problem in the Panel Study of Income Dynamics. You can see the younger, great duck in there at work. But it is just a delight to welcome him today. He is going to talk about a book which he has worked on in one form or another for, I know, more than a decade. And this book is sort of the summary of all that work and what came out of it higher ground, which is a part of this project called New Hope, which really was, I think, one of the most exciting demonstration projects running in the 1990s and which continues to have quite a wide variety of applications from New Hope to a whole set of interesting policy issues. And that's what the book is about and that's what Greg will talk about. Great. Thanks so much. It's terrific to be here. It's great to see old friends, including Jim Morgan up in the audience. Jim Morgan is Mr. PSID. He was director of the PSID before I got involved as a graduate student. And thanks to Becky. Becky has special connections to the New Hope project. She was on the Research Advisory Board as it was being developed. I think she read two drafts of the book itself as an informal reviewer and then as an official reviewer. So thanks so much to Becky for all her contributions. I want to talk today about the book, Higher Ground, New Hope for the Working Poor and their Children, not to be confused with Mike Huckabee's book, From Hope to Higher Ground. It's an author's worst fear. It was literally two weeks before our book was coming out when I sat down for my daily dose of The Daily Show with John Stewart. Mike Huckabee was a guest. And here he comes on with the title From Hope to Higher Ground. So we'll see which higher ground ends up selling more. Let me introduce New Hope by giving you stories of three women who we feature in the book and the different circumstances that they found themselves in on the verge of their entry into the New Hope program. The first woman is Lakisha, that not her real name. But in 1994, just as New Hope was starting, Lakisha is an African American. She was 24 years of age, a single mother of three young children. Lakisha had married at 20 but was now separated from her unfaithful husband who was paying no child support. And Lakisha had received welfare continuously since the birth of her first child. Lakisha turns out to be a very stern parent, not a harsh parent, but is very concerned with her children's lives as they grow up. Her personal appearance is immaculate. Her clothes are always dressed, clothes are always well-pressed. But in terms of her prospects for gaining employment, Lakisha has relatively little going for her. When she was a junior in high school, she lost interest, said she felt lazy and just stopped going to school. She's had no real work experience during the five years since finishing high school. She lacks confidence in her skills. She's rather easily overwhelmed when things aren't going well. She wants to get a GED. In 1994, the Wisconsin Welfare Reforms were pushing everyone into work. She wanted to get a full-time job but didn't really know how to do it. So she was, in many respects, a classic case of a welfare-dependent mother who was facing an uncertain prospect as she wanted to transition into the world of work and new hope turned out to offer her some tools that benefited her substantially. The second woman, Inez, in 1994, she's Puerto Rican. She was raised by a single mother in New York. She's aged 20 in 1994 in the verge of new hope. Mother of a baby boy, father is in jail. She's now living with another man who soon becomes the father of her second child. She too is receiving welfare but is really very ambitious. She has a high school diploma. She has substantial work experience. She's working part-time and definitely wants to work full-time but she can't pull together the childcare to do that. She has great self-confidence. She tells her ethnographer in one of the early visits, I can get any kind of job I want to if I just try hard enough. If I don't have a skill, I'll learn it. I could do anything if I set my mind to it. The third woman that we track, Elena, is an immigrant from Central America. She's already working full-time as a receptionist in a social services agency. She has two children in 1994, a third is on the way. Her husband also is in jail and her mother has taken a job, making it very difficult for her childcare arrangements to work out. She comes into the program already working full-time but needs a set of supports to make that full-time work work for her. New Hope was a program developed in the Milwaukee community by community activists. It's a voluntary program. It was initiated by the community, although soon won the support of the business community as well. It was designed to help low-income adults, both men and women. It's not a welfare kind of program directed only at women with children. Use full-time work to support themselves and their families. It's a package of benefits available to people who can demonstrate that they've worked 30 hours or more per week. The offices were set up and arranged a set of benefits that families could choose from, provided that they worked 30 hours a week. The benefits included a wage supplement that brought family income above the poverty line. It involved childcare subsidies which were available so families could get decent quality childcare with what they paid geared to a sliding scale based on their income. If they didn't have health insurance from their jobs, health insurance was available through a HMO similar to what Medicaid offered. That too was subsidized on a sliding scale. If someone came into New Hope unable to get a job in the private market, there was a network of community service jobs that were available that the program arranged for that provided six months of a temporary job and additional six months if that was necessary. These were jobs that the participants had to apply for. They weren't make work jobs. They were real jobs. They paid the minimum wage. And the idea was for people who needed work experience to be successful in the private market, these community service jobs would provide that to them. And the whole package was administered by a set of knowledgeable, respectful and supportive case workers. That was the package. One way of thinking about New Hope is expressed by David Riemel, who is one of the designers of New Hope. He says, New Hope is literally you and me across the table. You're a low income adult and have some needs. I'm here to offer you some tools that will connect you with the labor market. If you'd like to take up this offer, we'll help you. If not, that's fine. You can always come back. And if there are things you're interested in that I don't offer, well then maybe I can refer you to those. So again, a voluntary program offering a cafeteria of benefits that people could take up or not, depending on their situation, intended to help make work, work for people who wanted full-time work. New Hope was tried out in two Milwaukee neighborhoods. I don't know how many of you know Milwaukee, but the Menominee River runs right down the middle of the Milwaukee City area above to the north are mostly African American low income neighborhoods. To the south, mostly Hispanic neighborhoods. New Hope opened its doors to families in one north side and one south side neighborhood based on zip code. They raised about $18 million worth of money to run the program and evaluate it, which provided a scale of operation that enabled them to enroll about 700 families altogether. And they drew from both these north side and south side neighborhoods. The evaluation of New Hope was a random assignment. MDRC was the contracting evaluator. This is one of the things that the expert panel that Becky served on insisted on. It was one of the things that the business community that ended up supporting New Hope wanted as well. It involved then recruiting about 1,400 families altogether, telling them right up front that they had only a 50-50 chance of actually getting into the program, rolling the dice, as it were, and assigning about 700 families into the New Hope program, 700 families to the control comparison. And then it tracked both groups now eight years beyond the point of random assignment. It's important to always understand what the counterfactual is with program evaluation. I'll talk more about the conditions in Wisconsin, but you always are getting a comparison between what the New Hope group is experiencing and what the control group is experiencing and what the control group was experiencing. A lot of welfare reforms at the state of Wisconsin was implementing, so it's that kind of contrast. So a lot of the results in my involvement with New Hope was in conjunction with the MacArthur Child Network on Middle Childhood. Bob Granger is now president of W.T. Grant, was at MDRC at the time and got our network group, including Alita Houston, Tom Weisner and myself, the co-authors of the book, connected to an MDRC project, New Hope. And the New Hope itself was set up to concentrate mostly on employment kind of outcomes, welfare kind of outcomes, and we extended the evaluation to include family outcomes and child outcomes. We were especially interested in the extent that an economic program like New Hope might work through the family and have effects on kids. So the model that guided our thinking about what New Hope might do begins with the actual provisions of the program itself, the 30-hour work threshold for eligibility, the earning supplement, childcare subsidy, health insurance, the service job, the CSJ, if needed, and supportive case workers, these key elements of the New Hope package. Those provisions were expected to induce certain changes in the adults participating in New Hope, increase employment, reduce welfare receipt, increase total family income. Our interest was in extending this from the adult behavior impacts to the impacts on the family and the children. So if you think about how children might conceivably be affected by a New Hope kind of program, you have to think about changes in the child's resources and the child's context that are set in motion by this program that might in turn affect how well the children are doing. So the higher income could increase the family's material well-being to the advantage of kids. The employment and higher income might affect parenting, might make parenting less harsh, which would benefit kids. The childcare subsidy of New Hope plus the higher income could well affect a family's use of childcare and community programs. Various aspects of the provision of New Hope could improve maternal mental health support for the mother, which in turn could affect the kids. And finally, the health insurance subsidy could affect the nature of the health care for the kids. So this is the kind of model that we had in mind as we went about the evaluation in assessing what the kind of bottom-line impacts are for the kids as well as the process by which those impacts might have played out. So let's think about the context in which New Hope developed. New Hope opened its doors in the fall of 1994. Think back for those old enough to think back to 1994. Welfare reform was in the works. Ron Haskins was here not long ago talking about what was going on in Washington at this time. The federal reforms passed in August of 1996, which is right in the middle of the point that New Hope was in operation. It was the dividing line between the old world of AFTC and the new world of TANF. In Wisconsin, Welfare reform started long before it started in the federal stage. Tommy Thompson, the long-serving governor in Wisconsin, started in 1986. And I think the year after he took office, he cut AFTC benefits by a token amount and put the state on notice that he was concerned with reforming welfare and that turned out to be a very popular set of actions that he took. So various reforms were introduced over the late 1980s into the early 1990s. By the time New Hope was in operation, Wisconsin had put in place a set of diversion programs. So people coming in to apply for cash assistance were diverted to a series of job search kind of programs that they had to go through satisfactorily before they could receive any sort of cash benefits. So it was a very serious pre-welfare reform attempt to get welfare recipients into the world of work. With the passage of the federal legislation in 1996, all the states had to come up with a new welfare program that was consistent with the legislation and Wisconsin's Wisconsin Works, and that officially started in 1997. So if you think about the evaluation here, we're going to be comparing families that were offered this package of work condition benefits from New Hope, but the control families were going through the turmoil of welfare reform in Wisconsin during exactly the same time. And it's the difference between those two groups that we're going to be focusing on. The evaluation of New Hope was unusually comprehensive. When people signed up for the chance to get into New Hope, they signed away their life in terms of administrative data that could be gathered about them. So we knew both before the baseline signing up, as well as up to eight years after what their earnings were according to administrative data, what their cash welfare receipt was, what food stamps looked like. So we had a great deal of administrative data that we could track on a continuous basis, people's employment and income. We conducted, or MDRC conducted a baseline survey right at the point just prior to random assignment to gather information about demographic conditions. We did a two-year survey, two years after the point of random assignment. So this is still in the middle of New Hope. New Hope was offered on a 36-month, three-year calendar basis to everybody. So this two-year survey was timed to come 24 months after they signed up for the program and started being eligible. We followed up with a five-year after baseline survey, which would be two years after the end of the New Hope program. We still had interesting results, so we could still get money to do an eight-year evaluation. So that's going to be five years after the end of the program. Not all the data are analyzed from the eight-year follow-up, but I'll provide some information about that. What's unusual in our evaluation is that we built in an ethnographic component. Tom Wisner, who's a cultural anthropologist, led this effort. We went round and round on how many families, how to pick the families, and ended up deciding my arguments based on years in the Survey Research Center thinking that sampling was a good way to do it was to literally take a random sample of 22 families from the control group, 22 families from the experimental group, and follow these families periodically over the course of three years, and then we did a follow-up in 1994. So we had field workers visit these families every six weeks or so. They collected a great deal of... it was a semi-structured kind of interview, plus a lot of observation, trying to get an idea of what was really happening to the families, the family routines, what was happening to the kids, how New Hope was playing out in the lives of these families, how the parents participating in New Hope perceived New Hope. We started this, we wish we had started earlier, but we started in the last year of New Hope operation. We were able to observe families over the transition, as they were transitioning out of New Hope to try to see what sort of changes were caused by the loss of benefits. Our ethnographers spent time in family gatherings and restaurants just trying to get an idea of what was going on with the families. And because we had a random sample from both the experimental control groups, 44 isn't enough to do a lot of quantitative analysis, but you did get a pretty broad sense of what was going on. In the book, Laquisha, Inez, and Elena are three of the families from the ethnography that all were in the program group, and we chose them in a systematic way as we possibly could. We took the basic findings from the evaluation, from the quantitative evaluation, and wrote out a list of about eight or nine lessons that were really important from the evaluation. And then we tried to select cases that collectively represented as many of those lessons as we could. So that was the selection plan, and those were the three women whose stories we featured most. In the book, we also drew from the other ethnographic cases. One thing you find is there's tremendous heterogeneity. Some families were much more successful than others. Some families couldn't be bothered by New Hope, even though they signed up for it and they discovered that it really wasn't for them. So we bring in evidence from other families as well, but we really try to spend the most time acquainting readers with these three families. If you just look at the characteristics of the New Hope families at baseline, they're about 1,400 or so altogether, half in the control group, half in the experimental group. Remember, New Hope was not just a welfare program. It was available to all low-income adults living in these two neighborhood areas. So almost a third of the people who came into New Hope were already working full-time. They saw New Hope as a chance to take that full-time work and make it more profitable for them, the earning supplement, the childcare supports, and so forth. So a third of the people were already working full-time. It wasn't restricted to female heads of households, so almost 30% of participants were males. They didn't have to have kids in the household. Again, the idea was that this should be a set of work supports universally available to full-time workers who just weren't earning enough to have a family income above the poverty line. Most but not all had kids in the household drawing from these two neighborhoods. About half of the participants were African-American, 27% Hispanic. Most but not all had received cash assistance in the past year. So I know Jason DeParel came and spoke about his wonderful book. He was really focusing on African-American long-term welfare recipients in Milwaukee. And the New Hope population is broader than that. It's somewhat more selective than that because these are the people who volunteered for this program that required them to work full-time before they could get any benefits. So it's more representative of the working poor than it is of the kind of long-term welfare recipients that Jason DeParel talks about. There was a pretty serious implementation study that MDRC did. If you think about the task of implementing a program like this, it's pretty formidable. You've got to work out an earning supplement system so that marginal tax rates don't go through the roof and they work very carefully on that. You've got to line up this childcare subsidy with childcare providers in a way that someone walks in the office and needs childcare. You want to be able to deliver that to them within a few days. You had to line up this health insurance subsidy, the HMO, which they did. And then they had to line up a set of about 300 community service jobs. So that was the number of people who ended up taking community service jobs. And that was done through connections with various nonprofits in the Milwaukee area. You really wanted work situations that were real work situations but that were supported work situations for people who really needed the work experience. You didn't want make work kind of situations. You wanted a real job that people could have on a temporary basis that would be used as a stepping stone into private employment. So all of those things were arranged and pretty much in place when New Hope opened its doors in the fall of 1994. And the big surprise was with this wonderful set of benefits being offered and though almost everyone who was eligible for those benefits at some point during their 36 months, only half of them took up any benefit in any particular month. So it was a shock to the New Hope people that this wonderful set of benefits weren't always taken up. The task was really to understand why it was that the take up was so hit and miss. Heterogeneity is the clue to understanding everything and it's no exception here. You can really think of very different kind of participants coming in and being attracted to something like New Hope. Some took up most of the benefits all the time. Some came in already working full time. It turned out their income was close enough to the cutoffs that they weren't getting that much out of the program. They had to come in once a month. That was too much of a hassle. So they just decided they couldn't be bothered with the hassle. Some, as you might expect, just had other problems that made it very difficult for them to come up to the 30 hours a week. Sometimes there were drug problems, sometimes depression problems. New Hope wasn't set up to address those kind of problems. There were referrals to community agencies that provided help with those kinds of things. But the New Hope program itself was conditioned on the fact that people were able to work 30 hours, perhaps with the help of the community service job at first and then provided the list of benefits conditioned on that 30 hours. And here's where the ethnography came in very handy. Some came in but just used one benefit. And it turned out that it was a very rational kind of process that most families engaged in, where they might have already had an acceptable childcare arrangement with a relative and what they really needed was just the health insurance because they weren't getting that in their job. Other people desperately needed the community service job in order to get work experience. Elena, who I mentioned before, was the Central American immigrant. She was already working full-time. Her mother had just taken a job. Her mother had been providing the childcare for her kids. So she was desperate for decent quality childcare for her kids. And she only took up the New Hope benefit for childcare. It was the only one that she was interested in. And some came in for the case worker support. Remember, this package of benefits was administered in the central office by case workers that were very carefully trained to support these work efforts. And in some cases, they just had to process the paperwork. In other cases, they had to spend quite a bit of time with the participants to make them aware of what sort of supports were available and to help them in various ways with the confidence they needed or advice that they needed to be able to actually get these full-time jobs and qualify for the benefits. So if you think about the nature of the program, it's really different from the typical way we think about programs where there's an EITC or there's a childcare subsidy and they're all being administered by different agencies. New Hope's model was more of a cafeteria of benefits that condition on 30 hours a week. This was available. See what you like. Let's work it out so you can take up those benefits. If there are benefits that just don't make sense for you and your family now, that's fine. We'll just concentrate on the benefits that do. So I think that was one of the great strengths of New Hope, was this flexibility and this one-stop shopping that states have tried to put in place but haven't been very successful in most cases. So let me talk about the impacts of New Hope. It was a random assignment so we can do these comparisons between the experimental group and the control group. Let me take a moment to set up this graph here. This is taking administrative data on what fraction of New Hope and control families we're showing up in payroll records in any particular quarter. We've got quarterly data here. So we're gearing the quarter here to the point of random assignment. That's quarter zero. These are the quarters prior to random assignment. These are the quarters afterwards. The shaded area starting at zero, running up to 12 quarters, is the 36 months that they were eligible for New Hope. We're showing in the solid line the fraction of New Hope participants who were showing up in the administrative data in a particular quarter. The dashed line is the corresponding between the experimental and controls. First thing to notice, always look at what's going on with the control group first. Look at the dashed line around the point of random assignment. These are families that came in interested in New Hope supports but lost the lottery and they ended up in the control group. That group had labor supply increases on the order of 10 to 15 points. That's a control group. They didn't get anything from New Hope. They were just living in the middle of the drumbeat of welfare reform in Wisconsin. They were self-selected, being attracted to New Hope with this opportunity to make the work work. They were in the process of trying to do that on their own. They just weren't lucky enough to be in the New Hope lottery. The test for New Hope isn't whether employment increased. It did by almost 20 points. But how much better did New Hope do relative to the control group? And that's the difference, which is over the course of the 36 months, almost always significant. The diamonds down here represent a statistically significant difference with employment rates 7 to 10 points more for the experimental group relative to the control group. By the end of the program period, these differences have faded to insignificance and they're essentially zero after that. If you have questions about the data that I'm going to present, please ask them. So overall, for New Hope, there were modest employment effects during the program that faded quickly after the program ended. I should say the designers of New Hope and the business community that supported the designers conceived of New Hope as a perpetual program. It's the set of supports that ought to be available to America's working poor. But for the purpose of actually running the evaluation, they had to time limit New Hope. So that's the overall result. It's always good to ask, is there a substantial subgroup within the overall group that matched the program model the closest? Where the benefits might be larger and more persistent? And this again was one of the advantages of the field work. Catherine Magnussen, who's now at Wisconsin, was one of the field workers and took on this task of trying to look through the ethnographic notes and understand what kind of families seem to be benefiting the most from the program. And she drew from the literature that Sandy Danzinger and others have contributed to that frames employment barriers is being very important. And the idea was that families that didn't seem to have any employment barriers, families that are already working full-time, weren't likely to get big benefits from New Hope. Their employment wasn't likely to be affected, they were already working. And the control group was probably going to do just as well as the New Hope group. Similarly, if you think about families that have multiple barriers where there's a drug problem, a depression problem, a serious set of barriers, they're not likely to be helped very much by a program like New Hope. New Hope didn't offer services to address those kind of serious needs. But maybe there's a group in the middle, the people who are somewhat constrained, but not overly constrained, where the family is just one barrier away from really being able to make it in the labor market. And that was the idea. And she developed a barrier index based on a diverse set of criteria where, for example, having two or more young children were counted as a barrier, not having high school diploma was a barrier, not having work experience was a barrier, having been in prison was a barrier. So you can develop an index based on various kinds of barriers that might be addressed by program components. So someone has a jail record, the community service job provides the kind of experience that might lead them into the private labor market. Obviously people with a couple of young kids are going to benefit from the childcare subsidy. So the idea was that if you could form this barrier index and look at people who had just a small number of barriers, that they might indeed be the subset of participants who benefited the most. And sure enough, that was the case. It turned out that nearly half the sample, 42%, had just one employment barrier. So here we've got the same kind of administrative data, but just for the 42% with the one barrier to employment. And for them, the impacts were larger, more consistently around the 10 percentage point level. But what's amazing is that they were enduring. They held not only during the program period, but as long as we've been able to track them five years after the program. So this group indeed seems to be the group that fit best to the program model, and their labor market success seems to reflect permanent advantages imparted by the program as a result. Let me turn to family impacts. Remember, New Hope was really about the idea that if you work full time, you shouldn't be poor. So tracking family poverty was an important outcome. During New Hope family poverty rates, this is constructed from administrative data and not a full accounting of family income. But if you just look at the sources of administrative data, you get family poverty rates that are 14 percentage points lower during New Hope for the New Hope families compared to control families. And a couple of years after New Hope, the difference was still significant, eight points less. We expected because family income was higher and poverty rates were lower, that there would be impacts on material hardships, but there weren't that many. It was as though the margin that the income advantages were working on weren't so much at the very basic levels of meeting needs as they were above the next level up. Being able to afford childcare, for example, being able to afford other kinds of things. There were no impacts on welfare receipt. But again, you have to think about what's happening in the control group. The rates of welfare receipt fell 50% among the New Hope group, but they also fell 50% among the control group. This was the peak period in Wisconsin when the welfare rolls were declining by 80% altogether. And the control group was caught up as much in that decline as the New Hope group was. If you look to other kind of family system effects, there were quite substantial impacts on family's use of formal childcare. You might expect that because of the childcare subsidy, but it extended to other kinds of formal programs like after-school programs. There were also positive impacts on questions related to social support that kind of helped that the case workers might be able to provide. What's interesting is that we also asked questions about their awareness of community resources. Not the kind of things that New Hope offered, but the kind of things that were available in the community that New Hope case workers might have referred them to. And there were significant differences between New Hope families and control families, not only during New Hope, but also a couple of years after. And then there were scattered effects less than we had expected on various measures of maternal health. So let me turn to child impacts for working our way across that flowchart. We would have expected that this combination of higher income, less poverty, more formal childcare use, more cohesive parenting, perhaps, might play through to the benefit of the kids. The kids we concentrated on, ours was a MacArthur network on middle childhood. So we were concentrating on kids who, two years after baseline, were age three to 12 years old. And we gather, I'm going to talk about some of the teacher-reported data. And these are the kids who are obviously in school, so they're six to 12 years old in the point two years after baseline. They're six to 15 years old, five years after baseline, and then have a little bit of data from eight years after baseline into their adolescence for the most part. All right, so here are the two and five year impacts. As I say, the eight year impacts are still somewhat preliminary. We, in addition to asking parents about the kids, we asked teachers about the kids. We surveyed as many teachers as we could. Had a response rate of about 80% for the teacher surveys. We asked the teachers about the individual children and some of them that they were in a control group or experimental group used a standard kind of achievement rating that's asked of teachers in the developmental literature as well as classroom behavior for the kids themselves. We conducted interviews with them and asked them about their college expectations. These are impacts. These are the differences between the new hope kids and the control kids expressed as a fraction of standard deviation for the given measure. So at two years, the teachers for the new hope kids were reporting achievement rates that were about a fifth of a standard deviation higher than the rates for the control group. By five years after an assignment, two years after the program, that difference had faded to insignificance. Positive classroom behavior was significant at two years but not five. At both two years and five though, there were significant differences in college expectations favoring the new hope group relative to the control group. The surprise came when we looked at gender differences in these impacts. Here we present the same kind of data only for the boys and new hope families. And for them, the impacts were considerably larger and considerably more enduring. So for the boys, teachers were reporting achievement at two years and five years that was about close to a third of a standard deviation higher for the experimental kids, relative control kids. Behavior differences were equally positive in favor of the new hope kids. And then these college expectations were about four tenths of a standard deviation higher for the experimental group and they also didn't fade. So the real triumph of new hope was in these child impacts for the boys. It's a very interesting kind of gender story. Why should it be that new hope is affecting the boys so positively but not the girls? One thing if you just look at the raw data, the girls are doing much better than the boys are. So you can think of the girls as being up here, both the new hope girls and the control girls. The control boys are down here and what new hope does is essentially bring the experimental boys up to the level of the girls. So there might be something of a ceiling effect going on but again we pushed our ethnographers to try to help us understand why it might be that the boys do better. We could look at some of the survey data and see were there any family system kind of impacts for boys that were more positive than they were for girls. One of the impacts shows up for formal childcare, extended day programs. New hope families tended to enroll their boys more than their girls, relative to controls in these kind of after-school programs. But we also found a lot of stories like this in the ethnography. This is Jackie, a 35-year-old mother of four who explicitly talks about the fact she has both boys and girls but that she's more concerned about her boys and spends more of her new hope money on boys. My neighborhood is infested with gangs and drugs. It's different for girls and boys. Gangs are full of older men who want these young boys to do their dirty work and they'll buy them things and give them money. So the story, again, these kids were elementary school when this ethnography was running. The parents were considerably more concerned about their boys and their girls. They reacted to that differential concern by being more likely to channel their boys into the Head Start programs, into formal childcare programs, into after-school programs. They talked about using some of the earning supplements to buy tennis shoes and the kind of things that the gangs would have gotten for the kids had they not. So it's a very... We didn't expect this, but it's a very sensible kind of story given the differential concern that the parents had in these quite dangerous neighborhoods, differential concern for their boys relative to their girls. Your impacts are still preliminary at this point. We continue to see positive impacts for experimental boys relative to control boys in terms of behavior. No impacts on test scores. This is five years after the end of New Hope. And again, what was very surprising, not so much earlier, but in eight years, there are these consistently positive effects on parent-child relations. Again, these kids started out in elementary school. They're now in adolescence. And across a host of parent-child relation kind of measures, the Y indicates a youth report, a P indicates a parenting report. You observe these adolescents now in New Hope families and their parents reporting better parent-child relationships. We swore eight years was going to be the last time. We wrote the proposal, got the money, went out and did the survey. We're kind of hoping there wouldn't be any effects at eight years to have to follow up on. But you've got to wonder, one of the results from welfare reform experiments more generally is that for kids who are adolescents at the time the treatment started, you tended to find negative impacts rather than positive impacts, even among programs that boosted income. And what New Hope might have done for the younger kids is provide them with a set of benefits, higher achievement, better behavior, better family kind of conditions that enabled them to make this transition into adolescence and perhaps beyond in a more positive kind of way. That's the hypothesis that we should probably test. So if we were to get more money to gather data, say, on crime, if you really want to try to tote up the benefits at the cost of New Hope, you might expect this better behavior, better parent-child relations to play out in ways that can generate some real financial benefits for society, if you think about it in a cost-benefit kind of context. It's easy to get excited by New Hope. I started out as an evaluator and have ended up being something of an advocate, which makes me uncomfortable, but I do think, as our nation thinks about the next step beyond welfare reform, to the kinds of work supports they should provide. They've expanded work supports, a higher EITC, more childcare subsidy. But still, there are 8 million kids living in families in the United States where our parents are working full-time and they're still living in poverty. There are about 7 million adults in that category. So despite welfare reform, plus 10 years, we still have a situation where a lot of families are working full-time and are still poor. So what sort of programs might we think about that would attack this problem? And New Hope is a potential model for those kinds of situations. If you add up the benefits, there were these employment gains, especially among the families that were just one barrier away from being able to sustain full-time work. Poverty reductions were quite substantial and these achievement gains in particular and behavior gains for the boys. The costs, if you towed up the taxpayer costs, New Hope was not a cheap program. In today's dollars, about $6,600 more than what Wisconsin was offering at the time New Hope was in operation. Again, this is always relative to what is already in place. If you look at what Wisconsin has in place now, partly because of New Hope, they've instituted a quite generous childcare subsidy system, a quite generous health insurance subsidy system. So the incremental cost of New Hope in Wisconsin today would probably be about half the $6,600, $3,300 per year, which is still a substantial amount. If you look at the nature of New Hope costs, the biggest chunk went for these childcare subsidies, which proved to be the most popular kind of component. It was a generous subsidy for high-quality care and high-quality care cost a lot of money. So 38% of the total cost of incremental costs of New Hope went for childcare subsidies, about a third for program costs, and the rest were scattered across the other components, the health insurance earnings supplements and community service job wages. So that's the range of costs, $3,300 today per year, $6,600 back when New Hope is in operation. If you add up the employment benefits, it's a fairly modest set of differences. Even for the one barrier people, you don't get earnings benefits that come close to $6,600 per year. The kicker is really to what extent these child benefits, impacts on achievement, impacts on behavior, will pay off in the way that the Perry Preschool Program is paid off in lower crime, the kind of things that really generate substantial benefits to taxpayers. So that's an unknown that we may or may not follow up with future rounds of research. Another problem we have to worry about, New Hope was a community-initiated program operating in Milwaukee for three years in the mid-1990s. Can you possibly replicate New Hope on a state level, a national level? Here I think there are grounds for optimism. New Hope wasn't the only kind of welfare reform that was tried out experimentally in the 1990s. A couple of them, Minnesota in particular, was rather like the New Hope program. Minnesota, with its MFIP program, reformed its welfare offices, retrained their case workers to support work rather than just write welfare checks. It too was evaluated with random assignment. It too boosted income about the same extent that New Hope did. It too generated positive impacts on achievement and behavior of kids. So the Minnesota program shows that you can indeed replicate some of the basic findings from New Hope in the context of a government-administrated program. There were Canadian and Connecticut experiments that also boosted work, boosted income, and boosted child achievement. In contrast, there were additional set of programs that were focused on work that did boost work but didn't have any change in family income and there were no child benefits associated with those. So from this kind of scattered experimental evidence, it appears that key elements of New Hope can indeed be replicated, including some of the key results. So who cares about whether the impacts can be replicated? Can the story be published? So let me say a word about the book itself. Russell Sage runs a scholar in residence program which I recommend to everybody. It's a wonderful year in New York. The price is that you have to write a book and we were eager to take on the challenge because we had lived with New Hope so long and we wanted to tell the story of New Hope. So I talked with Suzanne Nichols, the head of publications, and said, I want to write a book. We've got these reports from MDRC that detail all the findings. We don't want to do another technical report. We'd like to write a book that might reach a broader audience. She said, well, you've got to write that book but you can't put any tables and you can't put any figures in. Never in my life have I had an occasion to write anything without tables and figures. It was kind of fun. We used these ethnographic cases, as I described, drew them very carefully from the results of the overall evaluation and tried in a mini Jason DeParle kind of way to use the qualitative stories to drive the narrative to explain the nature of the labor supply impacts, to explain the nature of the family impacts, to give people an idea of what the participants were actually like, how they experienced the program, who were the people for whom the program seemed to work, who were the people for whom the program didn't seem to work. And that's all done in the book in 124 pages. That was another requirement that we didn't want to do a long book. We wanted to be able to tell the story not only of the program itself but of the development of the program, the people in Milwaukee that dreamed this up. They worked on it for 15 years before they actually opened their doors. The very interesting story of the Milwaukee business elite that became interested in New Hope did heavy-duty political support on behalf of New Hope. New Hope had to raise money not only from foundations in the federal government but also from the state of Wisconsin. And you can imagine Tommy Thompson is sitting there with a much more punitive kind of approach to welfare reform. New Hope is the experiment. W-2 is the control group. Any sort of positive result for New Hope is going to be showing that New Hope's better than what Tommy Thompson had put in place. So members of the Milwaukee business community used up a lot of political chips to lobby the Wisconsin state government to put money into and otherwise support New Hope, and they succeeded in that. So it's this really coalition that came about because of the work focus of New Hope, the 30-hour-per-week requirement that led both the advocates and the business community to be very enthusiastic about it and for me to be very enthusiastic about telling a story to you. Thank you very much. Community service jobs, as I said, there are about 300 of them lined up altogether. They tended to be in community nonprofits. They were receptionist kind of jobs, oftentimes. For men, they were sometimes receptionist jobs, but sometimes fairly basic kind of work. They were intended to be real jobs. People had to apply for them. They paid the minimum wage and only the minimum wage. In Milwaukee at this time, business conditions were really hot. Unemployment was very low. So the going wage was really considerably higher than minimum wage. So it wasn't attractive to take a community service job if you could actually get a private sector job. The 30 hours of community service work qualified people for the set of New Hope benefits, including the earning subsidy. So it was trying to simulate a real-world job with a little bit of sheltering to get people through the six-month period. But if they weren't showing up, they'd be fired from the job. It wasn't a coddling kind of situation. It was more of a supportive kind of situation. Yeah? So these are minimum wage jobs? Do they have another job at the side? Is it great for them to love that poverty line? Or is it just the New Hope job and no other job? These were people who could not get other jobs on the side. These are people like LaKeisha. I started explaining she had no work experience, lack confidence. She had no clue as to how to go about getting a private sector job. So the community service job that was providing this 30-hour-a-week work was really her stepping stone into the private labor market. And she was just not in a position to have any other sort of job at the same time. They were... I'm sorry. You mentioned it. Right. So did they keep that other job and do the new job? No, no, no. The community service job, like all the other benefits, were purely voluntary. Right? And most New Hope participants didn't take a community service job at all. 70% of them didn't take a community service job. But 30% did. And those are the ones who weren't already working, who needed the community service job as a stepping stone into the labor market. Yeah. George. Can you say a little bit about the business community support? What percent of the total funding came from the business community? Was it a small number of key actors? Was it a broad base? And how was it framed to get support? There was some corporation support, but not very much. Most of the support came from the lobbying efforts that the Greater Milwaukee Community, which is the community elite group that took this on, engaged in. So New Hope was developed in the community in 1988. And then they came to the Greater Milwaukee Committee around 1990 to try to get it to sign on. The Greater Milwaukee Committee went to Bob Haveman at Wisconsin and asked how they should think about this. He worked with them to think about what the evaluation, what sort of conditions that they might want to impose to be enthusiastic about this. They, over a period of a couple of years, debated back and forth. There's an archive actually in the University of Wisconsin at the Milwaukee Library that's got a lot of notes from the Greater Milwaukee Committee meeting. It's a fascinating reading. And, you know, there was some opposition, but a number of people became very excited about this. And another person that we featured in the book, Tom Schrader, was the CEO of Wisconsin Gas. And he was fairly young when New Hope started, and he was asked by one of the higher-ups in the Greater Milwaukee Committee to kind of take this on as a community service activity. And he did, and he became chair of the board for New Hope. A very enthusiastic supporter, ended up spending a lot of time. But the group itself mostly helped out by lending their seal of approval, as well selectively as using their political connections, mostly in the Wisconsin government, because it included the group of enthusiastic business people included some of the big-time Republican supporters in the state of Wisconsin. So once they took it on as a group and came to think that it was a good thing, then a number of people became very enthusiastic supporters of it. Can you say something about who volunteered for the program? Yeah, that's a good question. I mean, we can certainly describe who volunteered for the program. I put up a few statistics. You know, some fit... There was a community survey done at the same time in some of the early reports that contrasted to some extent with the New Hope volunteers who were versus others. By and large, you know, there was a subgroup that, like Lekisha, were pretty traditional welfare recipients. You know, you have to remember a third of the people were adults without children. So it's a very different group than we usually think about. But then, among the rest, it was families of kids where they were considerably more work-ready or already working than the typical of then Jason DeParle's families, for example. So it's a heterogeneous mixture like that. But on average, they were considerably more advantaged than in the other kind of MDRC welfare reform programs which were targeted exclusively to either long-time welfare recipients or new applicants. New Hope had some of those, but not mostly those. You know, if there's sort of systematic differences, I would say people suffer from depression. We haven't done those kind of... You know, we could compare to the West data, I suppose. But we haven't done that. Now, I suspect, again, it's somewhat more advantaged, but there were certainly a substantial number who had big-time problems in domestic abuse. That was one of Alena's problems with depression. The ones who didn't take up New Hope very much at all, the multiple barrier families, which were like 30% of all the families. Yeah, Arlen. Right, the results that you talked about seem really exciting. You talked about people in Milwaukee working for 15 years to put this program together. But then it seemed like it stopped. And I'd be interested in what are the... What's going on in Milwaukee? What kind of an impact are they going to take up a new program? Sort of that most kind of issues. Right. To some extent, New Hope has had an impact, I think, on Wisconsin state policy. It's always difficult to nail down exactly what influenced the state legislature to introduce a particularly generous health insurance program for low-income families, a particularly generous childcare subsidy program. Wisconsin, for all the harshness of its welfare reform, has put a ton of money into work supports. And it turns out that its childcare subsidy ended up looking very much like New Hope's when it finally went into effect in 1999. The Badger Care Health Insurance Program was provided very similar kind of coverage to what New Hope provided. So to a certain extent, some of New Hope's features have been absorbed by the state of Wisconsin. You know, to a much greater extent than is true in a lot of other states. Alitha's from Texas, and when you hear about what Texas has in place, it's a very small set of supports relative to a place like Wisconsin. So that was part of the legacy of New Hope. The people in New Hope, David Reamer, who's one of the developers, he's been a tireless advocate inside of government outside for these kind of work-based progressive reforms. So he's continued his work. Julie Kirksick, who is really the heart and soul of New Hope, has continued on consulting various New Hope type programs, work support programs around the country. So we're still hoping that we can get some sort of replication. NDRC is hoping this too. To actually see to what extent a New Hope kind of program would be as successful elsewhere. So these efforts are ongoing, but they're slow and steady. Yeah, Ryan. I was curious if you could talk just a bit more about the health insurance take-up and the health impacts. I think that seems particularly relevant in today's policy climate. And could I imagine have really large benefits along with the crime weighing into that cost-benefit analysis? Right. We didn't ask extensive questions about health care and health conditions. And kids are pretty healthy by and large. So it's often hard to find health differences in kids. I mean, there might be underlying differences that are going to emerge. Later in life that you really can't track with these kind of surveys. There were significant differences in the extent to which in the number of dental visits and the number of doctor visits, higher in the case of the New Hope family as well as the control families. In terms of the pretty short inventories of health conditions, there were as often as not no significant impacts. So there weren't big-ticket differences that were showing up that would add to the cost-benefit analysis. But I'm not sure I would expect them to, given the kids were as young as they were and given the relatively few number of questions that we had to try to detect these health differences. And yet this generated a lot of political support for the health insurance component? Yeah, mostly because, well, partly because David Riemer was particularly strong in the health insurance side of things and worked very much on that. The child care subsidy was probably more closely modeled on the New Hope health insurance, I'm sorry, the child care subsidy. The fact that New Hope was there, I think was more important than what New Hope's impacts were. So people knew about New Hope and the two-year results helped kind of put New Hope on the map. So everyone knew it was a set of work-related supports that made sense. So as the legislative process proceeded, they turned to New Hope for the possible model. One last question. Yeah, Rob Hollister spent a long time trying to work this out and they wanted to make sure that the marginal tax rates were never higher than 80%. That was the goal. And with the phase out of the subsidies, that proved very difficult. But it was the cash income that they wanted to bring about the poverty line and the value of things like child care benefits would be on top of that. So conceivably, some of the families, the couple of kids taking up the child care benefit could be substantially better than the poverty line. I think we should take break. People who want to talk more about break will be here and I think there's another session outside. So thank you, Ray.