Using Incentives in Welfare Reform: The New York State Child ...

Using Incentives in
Welfare Reform:
The New York State
Child Assistance

Program












Cambridge, MA

Lexington, MA
Hadley, MA

Bethesda, MD

Washington, DC

Chicago, IL

Cairo, Egypt

Johannesburg, South Africa






February 1998








Prepared by
Abt Associates Inc.
William L. Hamilton
Nancy R. Burstein
55 Wheele r Street
David Long
Cambridge, MA 02138

---

Using Incentives in Welfare Reform:
The New York State Child Assistance
Program
William L. Hamilton, Nancy R. Burstein, and David Long
Welfare reform has been one of the dominant themes of domestic policy making in
recent years and, as a result, a myriad of reform schemes has emerged. The Child Assistance
Program (CAP), which was developed by the New York State Department of Social Services
and first implemented on a test basis in seven counties in 1988, is one of these many initiatives.
CAP is distinctive, however, both in its approach to reducing welfare dependence and
in its indisputable success in actually achieving its goals.
It is very difficult to reduce both poverty and welfare dependence. If welfare benefits
are increased in order to reduce poverty, work becomes comparatively less attractive and
welfare dependence and government costs may grow. If welfare benefits are cut to reduce
dependence and costs, families on welfare may experience more severe poverty. Alternatively,
if benefits are held constant, and recipients are allowed to keep more of their earnings
in order to encourage work and self-sufficiency, welfare is provided to families with higher
incomes, which may seem unfair to low-income workers and also draw more people onto
welfare.
Many welfare reform initiatives have tried to solve this policy puzzle by altering the
existing welfare benefits structure. They typically mandate specific actions on the part
of welfare recipients (such as efforts to look for work), and focus on the single parents
(usually mothers) who head most welfare cases in the United States. These initiatives have
imposed restrictions and penalties that reduce the welfare benefits of recipients who do
not take the desired steps toward self-sufficiency, but leave unchanged the incomes of
those who do what is required. Since the enactment of the federal Personal Responsibility
and Work Opportunity Reconciliation Act in 1996, states have begun to limit the time
families can receive welfare unless they are exempted for health or other reasons.
New York’s CAP initiative stands out as different from most other welfare reform
schemes in several respects:
•
CAP is a voluntary alternative to regular welfare. Families who enroll in CAP
first leave the regular welfare system. A
1
ll single-parent families with at least one
child covered by a valid child support order are eligible to make this shift. If they

---

2
Using Incentives in Welfare Reform
do so, financial assistance is provided under completely different rules by CAP
workers located in separate CAP offices.
•
CAP strongly encourages work. Public opinion polls indicate that the most
common objection to welfare is that it discourages work. CAP, however, provide
2
s
a strong financial incentive not only to get a job, but to work steadily on at least
a half-time basis. CAP pays less than welfare when an individual is not working,
but is much more generous when a person has earnings.
•
CAP promotes parental responsibility. Rather than prescribe how individuals
should behave, CAP encourages parents—absent parents as well as the single parents
who receive welfare—to take the steps necessary to achieve family self-sufficiency.
Families must obtain a child support order in order to qualify for CAP, but there
are no further mandates or restrictions.
CAP was designed to increase family income—through the earnings of the custodial
parent, and the support of the noncustodial parent, supplemented by CAP assistance—while
minimizing both dependence on and interference from the welfare system. Table 1 contrasts
these central features of CAP with the policies of the Aid to Families with Dependent Children
(AFDC) program in effect at the time of the evaluation.
The CAP model was subjected to a rigorous, five-year evaluation based on an
experimental research design (described below). The evaluation found that CAP significantly
increased families’ employment and earnings, produced more child support orders, lowered
expenditures on public assistance, and ultimately produced net savings to government.
At the same time, families who had the chance to switch to CAP had more income and
a much smaller share of their income came from public assistance. These results are among
the most positive to emerge from any welfare reform initiative that has been rigorously
evaluated.3

---

Using Incentives in Welfare Reform
3
Table 1
Key Differences Between AFDC and CAP
AFDC
CAP
Grant Calculation
AFDC grant based on budgeting needs for
CAP grant based on number of AFDC-eligible
family size
children covered by support orders
After work expense deduction, for every $1
For earnings under the 1990 poverty level, for every
earned, AFDC grant reduced $1 (after initial
$1 earned, CAP grant reduced 10 cents; over that
4 months of earnings)
level, CAP grant reduced 67 cents for every $1
earned
For every $1 of non-earned income, AFDC
For every $1 of non-earned income, CAP grant
grant reduced by $1
reduced by $1 (as in AFDC)
Child Support
Requires cooperation with child support
Requires child support order for each child in CAP
enforcement efforts
grant
$50 pass-through of child support
No pass-through of child support
Other Financial Features
Resource limit of $1,000 plus the equity
No resource limit
value of an automobile up to $1,500
Food stamp benefits in coupons
Food stamp benefits as part of CAP check
Earned income disregard for child care
expenses
Child care stipend in advance
Medicaid eligible
Medicaid eligible (as in AFDC)
Transitional Medicaid for up to 12 months
after case closed due to increased income
Transitional Medicaid for up to 12 months after
case closed due to increased income (as in AFDC)
Administration
Income maintenance examiner focuses on
Generic case manager recruits cases to CAP, helps
eligibility and budgeting
CAP families develop economic self-sufficiency
Quarterly reporting of income (monthly
Quarterly reporting of income
when demonstration began)
Restricted payments not allowed
Restricted payments allowed (e.g., rent
payment to landlord)

---

4
Using Incentives in Welfare Reform
This paper reviews the results of the CAP evaluation, drawing on various published
evaluation reports. describes key features of the
4
CAP program design and of the evaluation.
Subsequent sections briefly summarize evaluation findings concerning operational issues,
CAP participation, CAP’s impacts on clients, its administrative cost, and its net fiscal impact
on government. The paper concludes with some considerations of the possible utility
of the CAP strategy in other welfare reform settings.
How Does CAP Work?
A single parent who decides to enroll in CAP must first have a valid court order for
child support from the noncustodial parent, and then transfer out of the regular welfare
program. Once in CAP, the parent is subject to new program rules. Two aspects of the
CAP rules are critically important:
Work incentive. Most welfare recipients who work have their benefits reduced by
nearly one dollar for each dollar that they earn after an initial period. This constitutes an
extremely high “tax rate” on earnings, offering little or no incentive to work. In contrast,
CAP recipients’ benefits are reduced by just 10 cents per dollar for the first several hundred
dollars of monthly earnings, and 67 cents per dollar thereafter.
CAP’s favorable treatment of earnings is coupled with a base grant that is about one-third
lower than the basic welfare grant. Thus, a single parent with no earnings would receive
a smaller payment under CAP than under welfare. Welfare recipients are expected to transfer
to CAP only when their earnings reach the threshold at which their CAP grant exceeds
their welfare grant. At the time CAP was evaluated, the threshold was typically about $350
in earnings per month, roughly equivalent to half-time work at the minimum wage.
Child support incentive. Unlike welfare, the CAP base grant amount is determined
by the number of children with child support orders. The family receives an amount for
the first child with a valid court order for child support, plus an additional amount for
each additional child with an order. Because children without orders are not covered by
CAP, and also may not receive welfare benefits while the family is enrolled in CAP, it is
in the interest of the single parent to establish orders for every child in the family. Participants
receive program assistance in obtaining child support orders and upgrading existing orders.
Other features. CAP has other features designed to strengthen these incentives, further
encouraging self-sufficiency. These include providing food stamp benefits in cash rather
than coupons, allowing recipients to build their savings without becoming ineligible for
assistance, and eliminating the practice of paying part of some families’ grants directly to
landlords or other vendors. CAP also deliberately projects a “non-welfare” image—for
example, by having separate, professional-looking office space—to encourage participants
to think of CAP as a step toward financial independence.
CAP case managers work with individual parents, facilitating their CAP participation
and helping them overcome obstacles to self-sufficiency. Part of the case managers’ time

---

Using Incentives in Welfare Reform
5
is devoted to encouraging welfare recipients to take up the CAP offer, and assisting them
in obtaining the jobs and support orders they need to benefit from the program. The rest
of their time is spent working with active CAP participants, helping with problems that
threaten the participants’ ability to keep working and also aiding their efforts to find better-
paying or more secure employment. This intensive case management is possible because
CAP staff have caseloads of about 50, much smaller than those of welfare workers, who
typically had caseloads of about 150 in the experimental counties.
How Was CAP Evaluated?
The main objective of the evaluation was to measure CAP’s net impacts—that is, to
determine the effects of the program on outcomes like employment, beyond the changes
that would have occurred anyway. Random assignment of eligible welfare recipients, the
most reliable method for measuring such effects, is the approach that was used. A sample
of single parents receiving AFDC was divided, entirely by chance, into a “treatment” and
a “control” group. Treatment group families had the opportunity to participate in CAP
if they qualified and were interested (as discussed below, about one-sixth of the treatment
group families actually participated in CAP). Those in the control group remained subject
to the normal AFDC rules and procedures. A total of 4,287 families were randomly assigned
to these two groups in three counties where CAP was initially implemented: Monroe
(Rochester), Niagara (Niagara Falls), and Suffolk (Long Island). The evaluation then followe
5
d
these families for five years, collecting extensive data on employment and earnings, child
support, and public assistance benefits.
As shown in Figure 1, the evaluation judged CAP’s effectiveness by comparing the
entire treatment group—including those who never participated in CAP—with the entire
control group. This comparison has two important features. First, it captures CAP’s effects
before and after families enrolled in the program—that is, it covers the period when welfare
recipients took jobs and sought child support orders in hope of qualifying for CAP, as
well as the period after families enrolled in CAP. Second, it measures CAP’s average effect
for the entire single-parent welfare caseload, which includes many people who were not
affected by CAP as well as those who were. Thus, the evaluation sought to measure CAP’s
full impact (before and after program enrollment) on the entire caseload (including cases

---

6
Using Incentives in Welfare Reform
Treatment Group
CAP participants
Do not participate
Figure 1
Full Treatment Group Is Compared to
Full Control Group
that were unaffected by the program as well as those who were).6
The evaluation also examined how the program operated in all seven counties where
CAP was implemented, including the three where the impact analysis was conducted.
The findings of this implementation analysis are important in interpreting the results of
the impact analysis. They are also useful to state and local administrators seeking to replicate
an effective CAP-like program, as well as to avoid the pitfalls encountered by the counties
included in New York’s CAP demonstration.
What Are the Key Operational Lessons From Implementing
CAP?
Involving top administrators. Instituting a full-blown alternative to welfare, which
must operate side-by-side with the existing welfare system, can present some serious
implementation challenges. One of the keys to CAP’s success in addressing these challenges
proved to be the full commitment and active involvement of high-level administrators.
This happened in most of the counties where CAP operated, including Monroe and Niagara,
two of the counties where the impact analysis was done. As might be expected, the CAP
programs encountered obstacles in developing effective interagency linkages, securing
agency resources, and handling staff communication problems and turf battles. The

---

Using Incentives in Welfare Reform
7
commitment of administrators, however, made successful implementation of the CAP
program an agency priority and kept the implementation process on track.
Two counties, however, including Suffolk (the third of the experimental sites), had
to be pressed by state administrators to become involved in the CAP demonstration, mainly
because county administrators were concerned about the program’s long-term costs. These
counties had the weakest performance records in the CAP demonstration, apparently
reflecting their less than total enthusiasm for the program.
Reaching out to the entire caseload. Another important factor in implementing
CAP was targeting and recruitment. Several counties focused on those welfare recipients
who were already working and/or had child support orders in place. In addition, some
counties, including Niagara, viewed CAP as the final stage in a logical progression from
employment and training services to self-sufficiency—that is, the program provided the
financial incentive and case management that enabled recipients to find employment and
become financially independent once the services had been provided. As a result, these
counties targeted CAP to welfare recipients who completed employment and training
activities. Other counties, including Monroe, thought CAP could potentially help a majority
of welfare recipients, and consequently recruited CAP participants from their entire caseloads.
The problem with the first two approaches, which are intuitively appealing, is that
they limit the program to “job ready” welfare recipients whose motivation has already been
demonstrated and whose employability is indicated by work experience or participation
in training activities. These are individuals who may be well on their way to finding and
holding employment without CAP. Thus, consistent with evaluation research on other
programs for welfare recipients, CAP may make less of a difference for these recipients than
it does for less motivated and prepared individuals.7
Distinguishing CAP from welfare. A third important element was physically and
stylistically separating CAP from welfare. Welfare recipients were more likely to enroll
in CAP when the program was located in non-welfare office space. State planners had
recommended that CAP be located in separate facilities in order to create a fresh new image
for the program. The counties that perceived CAP as a distinct alternative to welfare tended
to house the program in private office space that projected this non-welfare image. Two
counties, including Suffolk, placed the program in regular welfare office space, and
experienced the lowest participation rates in the demonstration.
Finally, the most successful CAP programs used close monitoring and personal contact
to assist prospective and active participants. The counties with the highest participation
rates also tended to have the most extensive and systematic procedures for monitoring
and assisting the progress of potential participants. Several counties, including Niagara,
had computerized tracking systems and used various tickler systems to remind workers
to make appropriate follow-ups. In addition, the most pro-active counties, including Monroe,
provided participants with varied services intended to speed their success. For example,
they helped single parents by providing extensive help in securing child support orders
and referring them to further child support enforcement services. In contrast, the more

---

8
Using Incentives in Welfare Reform
“hands off” sites, including Suffolk County, expected prospective participants who had
received a program orientation to return to CAP only when ready to enroll, and active
participants to maintain their employment and child support without assistance.
Who Participated in CAP?
Over the course of the five years covered by the evaluation, 16 percent of the families
in the treatment group participated in CAP at some time. It is unusual for a voluntary
welfare-to-work program to reach such a high proportion of the AFDC single-parent
population. Indeed, the 16 percent rate is only slightly lower than the rates of actual
participation achieved in “highly mandatory” programs.8
Beyond this overall enrollment rate, several aspects of the CAP participation findings
are noteworthy. First, most people who participated in CAP took sizable steps to do so.
Four-fifths of those who enrolled in CAP had no reported earnings when they entered
the program. On average, they raised their earnings by the month of CAP enrollment
to $679. Similarly, over 40 percent of CAP enrollees did not have a child support order,
but obtained one or more prior to enrolling in CAP.
Second, the requirement for child support orders clearly limited program participation.
Families who had support orders for all of their children when they entered the sample
had the highest CAP enrollment rate of any subgroup examined (26 percent). The lowest
enrollment rate (8 percent) occurred among families lacking any support order and having
a noncustodial parent who was unknown or living out of state.
Third, apart from the population groups directly affected by the support order
requirement, CAP reached most subgroups fairly evenly. In particular, clients with children
under three years old—a group that in the past was exempted from welfare-to-work
requirements—had virtually the same enrollment rate as families with older children. In
addition, program enrollment rates were not closely related to previous work experience,
education, age, or length of time on welfare. In short, there are no identifiable groups
for which CAP was either totally inapplicable or uniquely appealing.
What Difference Did CAP Make?
CAP had substantial overall impacts on earnings, public assistance payments, and child
support orders. The magnitude of these effects varied across counties, and in some instances
county-specific impacts were not statistically significant.
Increased earnings. Treatment group members earned an average of $15,882 over
the course of the five years. They surpassed the control group’s earnings by $2,613, on
average, or about 20 percent. This difference is statistically significant.9 It is important
to recall that this impact is averaged across all households in the sample, not just the ones

---

Using Incentives in Welfare Reform
9
Figure 2
Impact on Annual Earnings
that participated in CAP. Presumably, the average impact on CAP participants was much
larger, whereas most nonparticipants’ earnings were probably not affected by the availability
of CAP.
As indicated in Figure 2, CAP’s impact on earnings varied across counties. It ranged
from over $3,600 in Monroe to about $1,000 in Suffolk, where the impact was not statistically
significant. The impact began almost immediately and continued throughout the period
covered by the study, suggesting that the earnings gains will continue beyond the five years.
Earnings were higher in the treatment group, both because more people worked and
because those who worked earned more money. On average over the five years, 29.4 percent
of treatment group members were employed in any calendar quarter, compared to 26.1
percent of the control group. This statistically significant difference of 3.3 percentage points
means that the number of treatment group members working was 13 percent greater than
the number of control group workers.
By itself, this additional quarterly employment accounts for just over half of the observed
impact on earnings. The rest of the earnings gain occurred because treatment group workers
earned more money than control group workers during quarters when they worked. It

---