Impact Results from the Child Support Noncustodial Parent Employment Demonstration (CSPED)

Impact Results from the Child Support Noncustodial Parent Employment Demonstration (CSPED)


LONNIE BERGER: Today it is a pleasure to introduce the long-awaited CSPED seminar Dan Meyer and Maria Cancian. I’m not surprised but really glad to see such
a good turnout and a lot of our state partners as well. I assume everyone knows Maria and Dan, and
I also promised to keep the introduction short, so I will say that they have both jointly
and independently made huge contributions to the poverty, social policy, and family
complexity literatures. They’ve both faced huge leadership roles on
campus; Dan continues to. Unfortunately for us–fortunately for Georgetown–Maria
recently started as the Dean of the Policy School there. I would say the two of them together are arguably
the top four child support policy scholars in the country if not the world. And this project has just been a huge part
of IRP’s existence for the last seven years. MARIA CANCIAN: I’m glad that Lonnie said we’re
the top four because a lot of times people think of us as one of the top–between the
two of us. So, Dan may or may not ever let me flip a
slide–the last time we did this he controlled it–but we’re going to go back and forth. We don’t have mics, so if at any point we
get too quiet or we are hard to hear please let us know. (It’s not usually our problem!) You always have a lot of thanks when you have
a project; we really have a lot of thanks here. A lot of folks in this room were our partners
in either designing this evaluation or in implementing this project. And then there are a lot of people in here
who worked very very long hours cleaning data, finding things, running analyses–so huge
thanks. We also owe a lot of thanks to partners who
aren’t with us today. This was conceived of and funded by the Office
of Child Support Enforcement (OCSE) which is a piece of HHS–of ACF–and you’ll see
some folks listed there. As is usually the case with lots of the work
that Dan and I do, we had wonderful colleagues at the Department of Children and Families,
and then also our county partners, and we thank them. Mathematica Policy Research was a collaborating
institution with us, and we also worked with the University of Wisconsin Survey Center. We need to thank all of the grantees–you’ll
hear more about them and our partner staff. Nina is with us here, so we have one person
who actually suffered and will keep us honest as we go through here. And we have to thank the over 10,000 noncustodial
parents who volunteered or agreed to participate in this–half of whom received extra services,
half of whom received regular services, all of whom completed a baseline survey and agreed
to have us collect lots of information about them, many of whom also did a follow-up survey
and other things and who kind of worked through this program. Having thanked all those people; anything
wrong, any opinions, all those things are either Dan’s fault or mine. DAN MEYER: So this is just a list of the people
we remembered who participated in this project. MARIA: And we’re old! DAN: I was going to show a slide from us in
high school, and then we had a little fight about whose high school pictures looked worse–that
is just to make the point of how long we’ve been working on this particular project. It’s been a long time, and lots of folks have
participated in this. So we’re going to start with some background–it
turns out not everyone knows a lot about child support policy. We’re going to start with just some beginning
stuff. We are open to interruption, so we’ll take
that stance until it becomes untenable if it does and then we’ll change the rules, but
right now we’re happy to be interrupted. So this project starts because–many of you
know that there’s been dramatic changes in family structure over the last 50 years in
the United States: increases in single-parent households; more attention, as a result, to
the child support system–the child support program; trying to figure out ways to ensure
that noncustodial parents contribute financially to their children. There’s lots of child support policy that
some people in this room know more about than me, but the basic story is: the child support
system doesn’t work very well for lots of families–it does work well for some families–but
for lots of families it doesn’t work very well. And here’s the national data: fewer than half
of the custodial parents say they were supposed to get some money in 2015, and of that, 43%–fewer
than half–got the full amount that was due. So, a system that doesn’t work very well–why
doesn’t it work very well? MARIA: It works well for a chunk of people,
and that’s largely because the noncustodial parent–often the father–has formal earnings,
and child support is often automatically withheld from those earnings, so just like paying your
taxes every month, you pay your child support. But a lot of the folks who aren’t regularly
paying child support aren’t employed. So, many noncustodial parents have limited
earnings and ability to pay, many noncustodial parents–you’ll see in our sample, about two-thirds–
owe support, or have had children with, more than one partner, and so they just don’t have
the capacity to support all the children that they have maybe? And in addition, there have been some challenges
with the program itself. The focus has been primarily on enforcing
collections. And this is really the way the program was
designed so this isn’t kind of saying, “oh, well, these folks were supposed to be running
a social service program and they started doing the wrong thing,” the program really
was designed in a lot of ways to be a bill collecting program. And that has kind of changed over time, but
the tools–a lot of the tools that agencies were given–were about sanctions, about taking
away people’s driver’s licenses, about kind of threatening folks, and not always a lot
of resources for services that might help folks. And there was a growing sense in the literature,
and in the policy community, that you might benefit–we might get better outcomes–not
just for noncustodial parents, but also for custodial parents and children, if we had
a system that would, in the words of Ron Mincy, “enforce and enable.” You know, to allow people to do things. One other issue here that we didn’t highlight
explicitly is because of assortative mating–because folks tend to marry people who are like them–the
folks that are struggling, the noncustodial parents who are struggling the most to pay
child support, often owe that child support to the custodial parents and children who
need that money the most. So, it’d be bad enough if we had a system
that randomly didn’t work for a bunch of people, but it particularly doesn’t work for the folks
where it’s most important that it work. DAN: So, [there is] this sense that the current
system may not be working very well, but [it is] unclear so far what might work better. In the fall of 2012 when we were younger–about
20 years younger–OCSE issued a FOI; so they competitively awarded grants to eight states–child
support agencies in eight states–who would try a new approach, and the new approach would
be targeted for folks who were struggling to meet their obligations and who had employment
difficulties. And they were going to be given enhanced services. So, grants to eight states that would provide
those services, and then a cooperative agreement to Wisconsin DCF who would be the evaluator–who
would manage an evaluation of this intervention–and DCF chose us (the IRP) along with our partner
Mathematica to do the evaluation, and then this demonstration ran itself from 2013 through
2017, and we’re now able to share the results of the impact and benefit/cost evaluations. But, before we get to the impacts, we need
to talk a little bit about how the program was designed. MARIA: So, the basic idea here is that you
have a child support-led program–and that’s really innovative and we’ll talk a little
bit more about that in a minute–and where the child support agency provided leadership
and oversight and coordination, and also provided enhanced child support services, and some
screening for domestic violence. Child support agencies were encouraged to
partner with partners who had expertise in providing parenting services and employment
services. And then there was also case management that
was provided either by the grantee or by a partner agency that included needs assessment
and kind of coordinating services across the three. Now this configuration looked a little different
across the different sites, so here’s a map that shows you: we had eight partners in eight
different states, but in some places like California it was one county, some places
like Wisconsin it was two (Brown County and Kenosha), some places like Colorado we had
five. So we had altogether 18 sites across eight
different grantees, and we won’t talk a lot about differences across grantees, and we
aren’t going to talk at all about differences across sites today. We mentioned that this was child support as
a lead agency, so here’s just some pictures from different child support agencies. They also worked with partnering agencies,
so on the next slide you’ll see some of the fathers’ partnership and other agencies. Now the reason that we focus on this is not
just because bureaucratically it’s kind of interesting to try to coordinate across three,
but because part of this project is really focused on culture and mission change CSPED
was really a project–not to put words in OCSE’s mouth–but I think that was motivated
in some ways by an interest in understanding whether this different, more holistic approach
to child support would be more productive. And part of the reason that’s so important
– there’s a couple quotes from the implementation analysis, and Jennifer Noyes who led that
effort is here and Lisa Klein Vogel who’s also played the leading role. We did a lot of interviews. We didn’t; they did a lot of interviews, and
this gives you a sense of some of the challenge of a child-support-led program. This is saying the perception is nothing good
comes from child support. That kind of gives you one sense. Child support has such a negative rep for
decades upon decades upon decades as a collection agency, so I won’t read the rest of that quote. But really, there was a sense in a lot of
these communities, and it varied across, but you know if you’re getting a letter from child
support, it must be a sting. You know? It must be. They claim they’re offering you these services
but really they’re bringing you in to arrest you. Remember, everybody pretty much who we were
recruiting, was having trouble complying with their child support orders so they were folks
who were not succeeding. So you might say; okay, given that, why would
you organize this as a child support-led? DAN: Here’s some of the advantages of doing
that. Child support agencies have several things
that they can do really well and, in some ways, are uniquely able to. They have access to the target population. If you think about what services there are
for low-income men in the community, really there’s no sort of key agency and child support
is one place where you can reach the target population. They have information in the child support
agency about the full-family context, so not just someone’s current partners and children
but the history of them and the history of their obligations. Then they can take direct action, if the child
support program itself is part of the problem, then maybe the child support agency could
undo some of those things. So maybe that the simple example is one of
the things that happens when you’re behind in your child support payment is they can
take away your driver’s license. Who in this room thinks the having your driver’s
license taken away enables you to be successful and pay your child support? Okay, so nobody. Right, it’s sort of a threat that you hope
motivates people to do something but you don’t ever want to carry it out. Incarceration: the same way. If you don’t pay, we’re going to put you in
jail. Well, if we do have to put you in jail, that’s
probably not going to lead to a steady stream of child support payments. This notion that child support itself might
be part of the problem and so, if you have a child support agency running a new, more
holistic intervention, than they cannot take those actions quite as quickly or maybe not
take them at all. MARIA: That gives you a sense for kind of
the context. DAN: And then here’s the evaluation. MARIA: Many people will be in this room will
be familiar with this kind of a design. All of the grantees and all the sites are
part of a regular kind of standard RCT. This is a place where we really did rely on
the expertise and experience of Mathematica in implementing a random assignment protocol. For those of you who know, this is kind of
one of their main business. One of the main things that they do is implement
these multi-site evaluations and we had done that in Wisconsin, but we hadn’t done a national
version, so it was very helpful to us to work with them on that piece. We have an impact analysis, a benefit cost
analysis – we’ll talk very briefly about that, and an implementation analysis that has been
really important in helping us to put the other results in context. Our goals are – there’s two main ones, but
I also highlight something else. We wanted to determine how the programs operated
and whether they improved outcomes and whether whether benefits outweigh costs. But we also wanted to increase our understanding
of noncustodial parents lives. There’s been some work on this. There is there are a number of projects – Fragile
Families, other projects – that have engaged related populations. Dan and I have done some projects in the child
support space but, typically, we’re interviewing moms or the focal sample is moms, that’s where
we’re starting, or custodial parents. This was really a project built around dads
and, if we had had all the money in the world, our initial of hope was to interview both
moms and dads. But unlike other projects where that’s what
we hoped for and when reality set in we only talked to moms, that was what we hoped for
and when reality set in we only talked to dads. We have a lot of information about 10,000
noncustodial parents, about 90 percent of whom are fathers. That’s part of what we’re trying to leverage
for this project. Our key question of interest – the key question
for the evaluation and for the program – is whether CSPED increased the reliability of
child support payments. Not the level, but the reliability. We’ll talk a little bit more about that in
a minute. DAN: This slide is linked to lots of panic
and difficulty and trauma. The grantees recruited in different ways. Sometimes they sent letters inviting people
to come in; sometimes they did recruitment within the court setting when people were
in court for child support. When they identified that someone was eligible,
they explained the study to them, went through an IRB consent process, and then told people
they had a 50/50 chance of getting extra services, and then they did a baseline survey over the
phone and at the end of the survey people were randomly assigned . The great thing is
we have this baseline survey on everybody – that’s definitional to being in the study. We used it for the implementation analysis,
we used it for the impact analysis. There is a report that’s kind of what all
we learned about the characteristics of low-income NCP’s [non-custodial parent]- that’s available
on the website, we’re not going to talk about that much today. We did a follow-up survey and this is really
where the trouble came. We have administrative records from eight
different states, sometimes multiple sites within States that did things differently,
and we’re trying to make something common out of those different ways of doing things
and that’s a whole seminar in itself. I’ll just say trouble. Then we had a bunch of other things that were
used mostly for the implementation analysis. MARIA: A key threat in this was the possibility
of cherry-picking. There are so many questions you would legitimately
want to ask. There are 18 sites; we have something that
was done over a bunch of different years; we have dads that have lots of different characteristics;
we’re measuring all kinds of stuff about parenting, about employment, about child support. Some folks involved in this process came up
with a list of a hundred and some things that we should be testing. All of you who’ve taken a stats 101 class
will say that if you know you have probability of less than 0.05, you’re going to find some
stuff if you do that. We needed to really shrink that. Our approach. We didn’t quite – now it’s sort of trendy
to pre post what you’re going to do on the web, but we did have very detailed documents
that we shared with our funders that said these are the things we’re going to measure
and this is how we’re going to measure them and these are going to be the outcomes. We had 14 primary measures across seven domains. This is what we’re going to talk about today. Child support compliance. That was our measure of the reliability of
child support. If you’re a custodial parent and you are owed
$200 a month, what percentage of that $200 a month did you get? We had lots of other measures like “how many
months did you get at least 70?,” but we had to pick one that we would do. What we did was child support compliance:
total amount paid over the total amount due. We did that in the first year and the second
year and we used administrative records – we committed to that. DAN: For those of you not knowing the lingo,
sometimes people pay on their back stuff they owe. Current just means you’re supposed to pay
this much this month – how much of that did you pay? MARIA: We used up some of our extra measures
by also having child support paid in child support orders. You might ask – why not just have compliance? Part of the reason is you could imagine people
are interested in each of those pieces differently. You cold imagine I could guarantee a hundred
percent compliance for everybody by just setting their order to zero. There are jurisdictions where in order to
get better performance, they’re making their orders very very low or they’re zeroing out
orders for people they think won’t pay. We are interested independently in what happened
to orders, what happened to payments – which is, after all, what kids the money that kids
see – and then we’re also interested in compliance because not complying with an order has negative
consequences for noncustodial parents. We care about all of those things. DAN: The next thing we wanted to highlight
was an attitude towards the child support program – more satisfaction with child support
services. There’s lots of ethnographic evidence and
almost no quantitative evidence on the levels of satisfaction or how noncustodial parents. . .
CHRISTOPHER TABER: If I miss month one and I make full payment every other month, I am
always a month behind, so current payment would be zero, is that right? MARIA: No. Putting aside tax intercepts for now; if you
make a payment, the first thing it goes to is current support. If you owe $200 a month and you pay $150,
it will all go to current support. If you pay $300, $200 will go to current support
and the $100 left over will go to back support. CHRISTOPHER: But if every other month, I pay
a full two months; or if I pay twice the full amount every other month and pay zero the
other months, that would be a payment rate of 50 percent? DAN: No. I’m going to accidentally give a more technical
answer. Child support workers have the opportunity
in some states to date the payment with what they think it should be. Probably you are correct – fifty percent is
what would show and then in our measure of payments toward current and arrears, then
we’d show the whole thing. But it’s possible. The way I think about it: people who miss
a day or two, that usually gets fixed within the discretion of the system. MARIA: And we did a lot of work to try to
make sure that we weren’t mis-measuring when you made one payment on the first of the month
and then a second payment on the 30th because things get taken out of paychecks and paychecks
come at beginnings and ends sometimes of periods. LESLIE HODGES (RESEARCH TEAM MEMBER): And
here, we’re summing what they owed over 12 months and what they paid over 12 months. If they paid twice as much in one month. . .
DAN: When States gave us information, they said whether this was for current or for arrears. What we would have done is totaled whatever
the state said was for current for those twelve months. MARIA: The bottom line is we tried it different
ways and it doesn’t matter. DAN: It’s a complicated wrinkle that turns
out not to matter very much. MARIA: But it gives you some sense of why
it was really hard to get comparable data across all these different sites because the
practice varies and the information systems vary. So for employment, we had UI records and there’s
a national system that gives you relatively comparable data. But for child support and, less so for benefits,
but especially for child support, there’s quite a bit of variation in what we actually
had to work with okay. DAN: Attitude towards the program. This is from the survey and it’s about your
level of satisfaction with child support services. Then, as Maria just suggested, we do have
some administrative records on employment and earnings. This comes from the national data systems. There’s also survey information that we use
for hours worked, the proportion of months. The survey compared to the administrative
data: with the survey, you might pick up more informal earnings, informal work, other things
too. And the administrative records: you might
miss that, but measures are not subject to memory problems or desirability biases, et
cetera. Then we got one measure of parenting and so
what we pick – then you’ll see our operationalization of this in a minute – is the sense of responsibility
for children. This is a standard method here. We’re just trying to first figure out if the
two groups are equivalent at random assignment and, if so, we’re going to do regression adjusted
differences. We’re going to do it within grantees and then
we’re going to show the average across all eight grantees. We’re doing intent-to-treat. For those of you who are familiar with that
language, I don’t have to say anything else. For everybody else, this is about the offer
of services, not whether people actually received services. Turns out that because of what I just described
about when people were randomly assigned, almost everybody got something because we
had them at day zero or day one, so almost everybody got at least some initial assessment
and some initial services. MARIA: This is false advertising because we’re
going to talk a little bit about enrollment, then we’re going to skip all the slides on
baseline characteristics because I looked at the clock, and then we’re going to get
to the results. Enrollment: we enrolled, between 2013 and
2016 our goal was to have 12,000 participants, and we got 10,161. As required by OCSE with respect to the programs,
all the participants had established paternity for at least one child, had one or more cases
that were receiving child support services, and had difficulty paying or were expected
to have difficulty paying child support due to lack of regular employment. JEFF SMITH: How many people who would have
been eligible to participate in the study declined to do so. DAN: Very few. NINA (PRACTIONER): Yes very few. DAN: I love having you here, Nina. Nina ran the program in Kenosha County, one
of our two counties in Wisconsin. STEVEN COOK: But there were people who would
have been eligible to participate but did not come in. MARIA: Yes, so we’re careful in talking about
some of these results – when we’re talking about the participants, you can’t generalize
to all low-income noncustodial parents who are having trouble with child support. In different locations you could generalize
more than others because of the way the recruitment went. Also, we’re going to be making some strong
claims about child support culture change – you cannot arguably generalize from these
eight grantees to. . . from Nina to all of her colleagues. Wisconsin is a very innovative place on the
child support space and not everybody volunteered or put together a grant proposal to go after
CSPED. There were quite a few [that] did, but eight
were selected for a variety of reasons. You can’t generalize that what was accomplished
in these particular jurisdictions could necessarily be accomplished elsewhere. This is the only other baseline side we’re
going to talk about given time, just to give you a sense of the folks in this. DAN: Ninety percent men, average age thirty-five,
lots of people of color, pretty low education levels and – this is brand new, so nobody
really had a number for this before, but that’s a pretty high level of depression. As Maria already said, lots of people who
had had kids with more than one partner. This maybe surprised us a little – about a
third had a kid living with them in addition to kids living elsewhere that they were supposed
to be supporting. Lots of people with criminal justice histories. MARIA: I hate it when people come and give
seminars and don’t get to the results, so we should get to the results. Three preliminary questions before sharing
the impact. Question 1: Did randomization work? I’m just going to say, yes it did. Trust us, it’s in the report, because we got
to move on one. DAN: 2) Was a RCT necessary? And we say yes, and here’s what we saw. This is the unemployment rate in the eight
different states over our period and mostly as a result of this, the people who didn’t
get extra services – so the control group – did better every year right. Employment rates up, earnings up, annual child
support payments up, everybody doing better. Just a little story – the people running the
programs thought that the people they were serving were doing better, and they were doing
better than they had been doing; it’s just that you need the random assignment design
to see if they’re doing better than someone just like them who didn’t get the same services. Maria: Third question – did the extra services
group actually get more or different services? Was there a treatment? The answer is yes; they reported more child
support, employment, and parenting services. These are from participants reports. One thing to recognize is you’ll see they
report an average of 14 additional hours which includes, for example, one additional hour
of child support services. They may have received – there may have been
workers working a lot more than an hour to help them – they wouldn’t necessarily observe
that the way that they would observe how many hours of parenting services they got. Employment Services is kind of a mix, but
we do see a difference. There’s the difference for E’s [extra services
group] and C’s [control group], so there was a treatment. We’ll talk a little bit more later about whether
it was a modest treatment or more. Another thing that’s really important to recognize
in this context is that part of the treatment was less treatment in some sense. Participants in the extra services group were
less likely to be subject to the kind of punitive enforcement actions – they were less likely
to have a contempt hearing, a warrant issued, a license suspended. This really was a less punitive approach to
child support. JENNIFER NOYES (RESEARCH TEAM MEMBER): This
is from administrative? DAN: This is from admin. data. These are small [differences], that’s right. MARIA: Results. DAN: The first one. All the results are going to look like this,
with extra services or E’s in this purplish color and the orange is going to be the regular
services or the C’s. You remember that we’re looking at first year
and second year separately. Orders lower for the CSPED group. These are monthly amounts, in the order of
$15 a month. This is perhaps not surprising. You look at somebody’s order – these are people
selected after all who were disadvantaged and having difficulty paying – you look at
their order you compare it to – every state has some formula as to what orders should
be – and mostly we think of this as probably right-sizing orders, making orders more consistent
with what the rules say they should be. MARIA: Here’s the impact on child support
paid. Just in case you didn’t notice; child support
owed, this is three stars – this kind of result is pretty robust to different ways of doing
thing. The decreases in child support paid are smaller
in magnitude, they are less significant in this particular measure, which is the confirmatory
measure we had. They are also less robust in the sense that
we had quirky data from one of the grantees – when we throw that grantee out and do it
over seven instead of eight, this loses significance. If you adjust for multiple comparisons, this
loses significance. It’s small. There it is – not much. Maybe not a big surprise that we see no impact
on child support compliance. If there were one measure that were the central
kind of hope for this, [it] is that you would see an increase in compliance. I think it’s worth thinking about this. You’re providing services, which you hope
will help people to pay more and comply more. You’re also being a lot less aggressive with
enforcement, so you could imagine that what you would see is a big decline in compliance
because we’re not suspending as many licenses, we’re not arresting people, we’re not harassing
people as much. You might see that you would have expected
compliance to actually go down – we don’t see that either. It’s kind of a half-empty, half full. On the other hand. . .
DAN: In case you’re leaving right now, this is the biggest and the strongest result – is
how people how satisfied they were with the services they got. This is very satisfied or satisfied with the
services of the child support system and you see 22 percentage point difference. This is large, this is consistent – every
grantee. In the grantees where there’s enough sample
size – where we can see site differences – it’s true in every site. Something happened. A transformation in how people felt about
the way they were interacting with the child support system. MARIA: This is part of why we gave the context
we did for the child support kind of leadership here. This is an important theme in the academic
literature. This is an important part of the policy dialogue. This is a real challenge for child support
agency leadership that, as we’ve come to understand fathers’ roles as being maybe more than just
providing money, and child support agencies as doing more than bill collecting. It’s great if we decide we’re doing something
different, but if the folks that we’re serving can’t be convinced that we’re going to try
to do something different, then that’s really going to be an issue. One of the real open questions behind this
was: if you had this, arguably somewhat modest, change, would you actually see people changing
their mind about what you were doing? I guess I was kind of surprised to see this
much of a difference. I was particularly surprised because we’re
comparing people’s reports and this is a culture change and I expect that some of that culture
change bled to the rest of the agency. If we really could measure what regular service
– what no intervention looked like – we would actually see lower satisfaction. This is purely speculative on my account. But this is the difference, comparing people
that were in an agency engaged in this project but were just getting regular services, to
people in an agency engaged in this project who were getting extra services. It’s a big difference. JENNIFER: Some of what we were seeing and
hearing was around some concern on the part of the custodial parent that “you’re giving
the noncustodial parent a break and yet I’m not getting any money. Not only has the order gone down, but compliance
isn’t going up. So what’s in it for me?” MARIA: I think we have problems on both sides
because I think, on the one hand, some custodial parents would feel “why are you being nice
to this guy who’s behind?” On the other hand, when you look at the CPS
[Current Population Survey] question about why don’t you have an order and why aren’t
you pursuing an order, one of the most common – I think the second most common – response
is “because the noncustodial parent is doing what he can” and, in some sense people will
tell you, “I’m not going to. . . it’s just going to make things worse if
I get you guys [child support agency/government] on him.” Some CPs [custodial parents] were probably
not happy about the NCP being dealt with in this other way. Other CPs arguably might have been happy but
we weren’t able to measure that, but you’re right. DAN: A measure of satisfaction with services
could be done on their [agency’s] entire population [including CPs and NCPs] and we don’t have
that. MARIA: If we were 20 years younger, we would
say we wanted to redo this with. . . no, we wouldn’t. DAN: 40 years. MARIA: But we’re not doing it for either reason. To summarize: reduction in child support,
smaller reduction in child support payments but they were smaller on a smaller base, no
impact on compliance, substantial increase in satisfaction. The report reports on a bunch of secondary
measures. Just a couple things to highlight: in the
first year, we see that orders were less burdensome, where burden is a measure of the percentage
of your income that orders account for. There was less owed on arrears at the end
of 24 months and on state-owed arrears at the end of the second year. Now to talk about employment. DAN: Number of months employed, number of
quarters employed – no difference. Now to talk about earnings. MARIA: I thought it was unfair for me to have
the clicker and stand in front of that. DAN: I was going to take the clicker away
if I could do it without being caught on videotape. MARIA: We fight like this all the time. On earnings, we have kind of mixed. Total earnings – no statistically significant
difference in the survey. These are self reports – pretty detailed set
of questions. We had NDNH (National Directory of New Hires)
administrative data on earnings for the first and the second year and we see: not significant
at the point one for total earnings [in survey data], higher earnings in the first year but
not in the second [in administrative data]. In summary, no effect on employment, mixed
results on earnings. We see small impacts on any employment over
the two-year period and some in quarters but, again, if you want to look at the – how long
is the impact report? it’s very long – there are lots of tables [in the report.] Dan carries it around; I do not. You can find some things there, but not in
our takeaway. DAN: I mentioned earlier we had a scale for
“sense of responsibility for children.” There’s the four questions. The news is that those in the experimental
group had a higher score, but the difference is fairly small – so increased sense of responsibility
for children. JEFF: Here [this scale] is a four or five-point
Likert scale? DAN: Yes, we’re taking the average, so these
are pretty high scores. This is a higher than a four out of five. That’s not the only thing we did on parenting
– that’s just the confirmatory measure. There’s some other parenting measures – ease
with increased contact with non-resident children by about a day a month. [With a] p less than ten [p-value significant
at the .10 or 10 percent level], evidence of decreased harsh discipline strategies. Then we looked at several other parenting
things and don’t see any significant differences. MARIA: We looked at a bunch of other impacts
and basically don’t find much. No criminal justice. Some impacts on economic well-being. A little bit of stuff on benefits. Nothing on custodial parents. No differences [by subgroup] – we had four
subgroups that we looked at. We had eight grantees. We did grantee-specific analysis; no grantee
with substantially different impacts across all the domains. The bottom line of no impact on compliance
holds across all the grantees. The bottom line, as Dan said, of big impacts
on attitudes holds across all grantees and, so now break my own rule, and both of those
also holds across all the sites where we can do it [that analysis]. DAN: I just want to say one thing here. Interesting that with increases in earnings
or zero increase in earnings, depending on which of those measures you believe, there
is increased SNAP [Supplemental Nutrition Assistance Program – food assistance] benefits
in one year. Partly that’s a result of the intervention
– the case manager talking with people about their situations. Some of this we think is the information effect,
rather than people having lower earnings. MARIA: A few bottom lines. CHRISTOPHER: Maybe you get to this in the
cost-benefit, but you’re finding not significant results, but how large an effect candidates
would [we expect]. If we compare this to a job training program
or something like that in terms of the amount of money. MARIA: We’re getting to that. Well, go ahead, I’ll let you finish. CHRISTOPHR: You can’t reject zero, but can
you reject numbers you would have expected based on that literature. DAN: Great, we’re going to do that in the
benefit-cost. MARIA: A few bottom lines. Can child-support lead an intervention that
has integrated case management, employment, and parenting services? We would say yes. We would point to the implementation analysis,
which really adds – and I can say this because I did none of the work – added tremendous
amount to our understanding of how the child support system works, the ways, the challenges,
and some of the things that partners were able to manage in this process. So that’s one bottom line. DAN: Another bottom line. Can the child support program be changed to
be less punitive? Yes, although as Tim pointed out, the differences
were not huge. Does a new approach change attitudes? Yes. Now here, I think ,the differences are large,
substantial. Does a new approach substantially increase
or decrease payments and compliance? No. MARIA: Our third bottom line. Why were most impacts modest? It’s a very disadvantaged population, it is
a relatively modest intervention. For this population, if you really wanted
to move earnings, you might be thinking about – there are folks in this room who know more
about what you would be thinking about – more than what was provided. It’s hard to evaluate programs that change
culture. We talked a little bit about that. There are reasons to be hopeful that part
of the payoffs of this will be in the longer run. If part of what we see with noncustodial parents
getting in really bad positions is there kind of unwillingness to interact with the system,
even when the system is trying to help them. You can have your child support order adjusted,
but many noncustodial parents you know won’t engage in that. We know from other work – a report that Lisa
did – that in a lot of counties if workers know that the reason you aren’t paying your
child support is you lost your job or you have a health issue, they will hold back on
punitive enforcement actions. But if you won’t talk to them, then they can’t
do that. This change in attitudes could have impacts
down the line. DAN: We get one slide for benefit costs. Modest additional costs: about $2500. Modest additional benefits: about $1670. They’re distributed across – most of them
going to CPs and children. In the short term – $2500, $1663: costs outweigh
benefits. Two things to say. One is we only have the services that were
delivered, this year’s benefits, this year and next year, and so we have a measure of
how much decline there was. We don’t have any data on whether there’s
any small or declining benefits in the next year and the next year and the next year. Reasonable assumption, or what I think of
as reasonable assumptions, the benefits will outweigh the costs in the longer term. I’m happy to say more about that, but that’s
my bottom line. From what we can see, costs higher than benefits;
but taking this out into the future – probably benefits will make up for the costs. We didn’t really answer your question, did
we Chris? Or did I do okay? If you think about the amount of service – hours
of employment services – that the people in the extra services group got, it’s not an
intensive intervention, so you might not have expected a large increase in earnings and,
surprise, that’s what we see. TIM SMEEDING: One thing this brought to mind
is – I’m thinking of the Paycheck Plus experiment, which gave the EITC to fathers and subjected
them to, at least in New York City, to the same child support rules and – what’ the word
I’m thinking of? Garnishment. They seem to like it. What I understand of the results, they had
a better sense of responsibility too. The money’s going over to the kids. Maybe there’s something about that. Then the second thing is: do you get a feeling
from this that the tighter labor market helped everybody, it helped the control and. . . I mean, it seems to me, there were some
labor market effects here that they cut across both groups no matter what the service system,
whatever work, and you gotta give. . . Again, this could be another reason why
[muffled] DAN: I think very much – in the last point,
increases in the macroeconomy made for improvements among that control group and it might have
been hard to show differential improvements among the Es. On the other end, they were getting more services. We did see – I’m going to make a point that
maybe Nina would be a better person to make than me but – in some ways the characteristics
of people interested in the program changed a little bit over time. At the beginning of the period – still in
recessionary period – people were anxious for employment help. By the end of the period, things going a little
better, people were maybe less gung-ho, committed to employment services. Does that fit or no? NINA: We still had a large caseload so we
didn’t really see that. In the beginning, we saw that they didn’t
want anything from us, it was very combative, at times it was a little scary. After about six months, word started to spread
and people started coming in as they needed services. MARIA: I just wanted to put up this slide,
which is from the baseline that we didn’t talk about, which was “what was the most important
reason why participants said they wanted to participate?” What I think is interesting is that the yellow
and blue is very or extremely important, and you’ll see: child support debt, so being behind
on your child support; my job situation; and my relationship with my kids; are just really
important things. That’s good because we were providing child
support services, employment services, and parenting services, so good match there. DAN: Good thing that this was, because we
weren’t really providing relationship services. This was not a directly targeted at your relationship
with your partner. AUDIENCE: The measure of services is from
the survey. MARIA: We have survey measures; that’s what
we had here – the hours. We have some measures in the administrative
data for some things, like enforcement actions. AUDIENCE: I guess I was surprised by the strength
of the survey measure of services, especially employment-related services, because services
are not thin on the ground. Everyone is eligible for WIOA (Workforce Innovation
Opportunity Act), yadda yadda. MARIA: I will let Lisa or Jennifer or somebody
else say more, but I will say that one of the things that I really took away from the
implementation analysis was that one of the concerns that grantees expressed the most
was the lack of colocation of services. That even if you were in the extra services
group and you were getting case management, getting the dad from the child support agency
over to workforce services even if they were just kind of around the corner or and especially
if they were across town, was really a challenge. So the fact that things are in the community,
it doesn’t surprise me that having case management and having that act of connection really made
a difference to people’s uptake. But other folks spent more time in the field
than I did. AUDIENCE: Child support agencies are not required
to be the one-stop? MARIA: No. AUDIENCE: Oh, that would seem to be a policy
recommendation. JENNIFER: I would say even in the cases where
they were close to each other, it didn’t necessarily mean the culture was in place for them to
be working together. That was another culture shift, even in terms
of – co-location doesn’t mean integration and cooperation. MARIA: Think about the Social Science building
[Laughter]. AUDIENCE: Performance management system for
the child support people that is related to the payments get made, and they could be referring
people to WIOA, it would surprise me that they are not doing that. Following up and harassing people until they
do that. AUDIENCE: I would just add that there is referral
and then there is uptake. In many cases, it can be challenging to get
someone to engage in those services or to make the decision to go and then follow through
on those services for a prolonged period of time. There are a number of barriers that we heard
about that got in the way of doing that – sometimes that’s related to the characteristics of the
program themselves. Not everybody was eligible for all things
available in the community and sometimes it’s difficult to access, and sometimes people
feel like they don’t need help. MARIA: I would also say, and we didn’t talk
through this quote, but one of the differences here is that will talk to many people – I
just came back from a Child Support Enforcement Association meeting – there are people who
feel like the child support program is mostly about collecting child support. Kind of like the tax, the Department of Revenue,
is about collecting too. They don’t embrace necessarily – everybody
doesn’t in the child support system embrace – all of this kind of broadening or mission
creep or watering down the program or whatever you have. Part of what this is is one of the project
managers saying “you know you’re having a more engaged conversation with the NCP about
his life situation while you’re preparing his order. You aren’t just checking information, filling
in a dollar amount, and slapping it over there. You’re having a conversation.” Really this was a change in the way that folks
have done things. One of the things again I’m going to point
to some work that Lisa’s done recently – part of it is about caseloads. It is easier to fill out a form and send it
on than it is to talk to somebody who’s having challenges and try to connect them with services. One of the things that we’ve seen in some
of the work that Lisa’s done is that places that have the time by virtue of kind of the
size of the agency and the size of the caseload are more often providing that kind of more
customized service. Not taking away your driver’s license if the
reason that you’re missing your payments is you just lost your job or you’re having a
health issue. But that takes an amount of time that is more
resource intensive than is available for a lot of agencies. HILARY SHAGER: Trying to ask this in a way
you as researchers can answer. If you were to test a new intervention where
compliance with child support was the key outcome, what tweak or difference would you
want to test? DAN: I think we have several ideas on this,
but one thing that I would do is run a similar kind of program. We have the observational evidence and the
implementation analysis mostly suggests that the employment services and the parenting
services could have been effective but weren’t robust enough. I would do a program where I gave grantees
more resources to deliver a deeper set of services. I would also run it for longer. If the main change here is that we see this
change in attitude, I would not expect immediate behavioral responses to changes in attitude. We could have a long conversation about this. I would want to see these people, who now
were very satisfied in and even saw the Child Support Agency as a partner with them, had
different child support outcomes down the road. Now, perhaps I’m too optimistic on this but
those are a couple things I would think about. MARIA: I think the one other thing is there
were some challenges that we identified and that grantees identified that we they really
weren’t in a position to address. Mental health issues is a really big one. Substance abuse issues is a really big one. Some legal challenges in terms of
two-thirds of the participants, roughly, had a criminal record, had been convicted. That presents some challenges. There were some things that would have been
nice to add to this. JENNIFER: Parenting time. MARIA: Parenting time. Lots more to do. That’s why I had I had flipped past; we wanted
to say thank you to everybody who’s in this room. There are many people in this room who did
a lot of work on this. We should note that Lonnie – oh I did the
wrong thing. Oh well. There’s a website. This is why Dan doesn’t let me do it, because
I get impatient and it’s wrong, but there’s a website that has already some additional
reports on it, but there’s a bunch more work going. Those of you who know that Dan and I finished
another random assignment experiment in 2001 that we just published something off of will
not be surprised to know that we aren’t finished with this one yet. There’s a lot of PhD students in this room
who have worked on this project who are doing dissertations and other things, and a lot
of other colleagues we should note. Lonnie was in charge of all the parenting
stuff, in addition to contributing to lots of other things. We mentioned Lisa and Jennifer, and I should
stop because I’m going to fail to mention everybody who’s in the room. Katherine’s [Magnuson] not here, but she was
also a Co-PI on this. Please do go and look at the things that we’re
already doing and many more things to come and thanks for your comments [Applause]

Leave a Reply

Leave a Reply

Your email address will not be published. Required fields are marked *

Copyright © 2019 Toneatronic. All rights reserved.