The Wisconsin Experience
THE WELFARE SYSTEM AS AN INSTITUTION is abhorred by society because it separates the receipt of income from the need to work.
But why do we think of work as so necessary to legitimize income?
In its unique capacity to enable individuals the opportunity to serve others by producing valuable goods and services, work fulﬁlls a basic human need. Work connects individuals to larger society—and in order for welfare’s destructive inﬂuence to ﬁnally end, work and income must be permanently rejoined.
These words, written in 1995, laid out a fundamental premise of what was to become Wisconsin Works, a welfare reform notable not only for its departure from the then existing welfare system of income support, but also in that it was devised from scratch without reference to any existing law. This account is written in the first person so as to permit the author, a participant in the matters discussed here, to describe the considerations he others made in decisions relating to the implementation of these reform ideas, in particularly in the realm of the privatization of delivery services, and to comment on what was learned from these events.
During the period Wisconsin Works was being developed, 1994 and 1995, academics and policy makers’ concern about the rise of welfare dependency and its intractability was paramount. Since the onset of President Lyndon Johnson’s Great Society welfare dependency had grown and grown, sometimes pausing, never permanently retreating. Various reform eﬀorts had been tried; these included experiments with versions of a negative income tax; required registration for job search; additional funds for education and training—all these had failed to stem the tide of increased dependency. The challenge to the designers of Wisconsin Works was to devise a plan so powerful in its eﬀects that it would be capable of stopping and reversing the persistent upward trend in dependency.
The Political Context
Wisconsin’s Governor Thompson was elected on a platform of reforming welfare in 1986 and won an upset victory. Over the subsequent ﬁve years from 1987 through 1993 Thompson proposed changes to the welfare system almost once a year. These changes came in the form of experiments, called “waivers” of federal program rules. These program change requests to the federal government would mostly be seen as modest by today’s standards, but they all struck a similar and consistent theme, namely that welfare recipients should work and accept personal responsibility.
The state legislature, controlled by the opposition Democrat party, grew increasingly frustrated by the support for reform that the Governor had created through his series of proposals. Finally, the legislators struck back by passing a law which would completely abolish the existing AFDC program. The law provided that in its place, the Governor must propose a work-based alternative. Would the law calling for abolition of the entire cash safety net for the poor be vetoed by the Governor? If so, reasoned some of the legislators, the opposition could take away his claim as the agent of reform, and they would be free to make proposals of their own.
Instead, Governor Thompson accepted the opportunity to design a work-based program from the ground up, unconstrained by existing law and he signed the bill. He then set about his task in earnest, assembling a group of his staﬀ to develop a proposal.
The philosophical underpinnings of the Wisconsin approach to welfare reform
The 1995 proposal resulting from the Governor’s planning group was unprecedented in two ways. First, it was a complete substitution of the existing approach rather than working around the edges of existing law. Then, arising from this, the program could be conceived from a set of consistent philosophical principles which were mutually reinforcing. Some of these principles and their rationale are summarized below from the plan:
- For those who can work, only work should
There are both economic and practical reasons for tying income to work. First, experience shows that entitlements to income without work have unwanted eﬀects on dependency. In addition, it is essential that parents understand they will always be responsible for supporting themselves and their families through work: this inﬂuences behavior and motivation in ongoing constructive ways. Finally, experience shows that individuals without extensive work history are usually in a stronger employment position after one or two years of actual work (at any wage) than after a comparable period of work preparation through education and training.
- Begin with the assumption that everyone is capable of some kind of
The best way to help an individual out of work get back into the labor force is to provide an actual work opportunity which matches their capabilities. This is contrary to many government subsidized “helping” programs which seek to identify barriers and limitations to work, and in so doing categorize and place individuals out of the reach of the workplace where they might very well have succeeded if given the opportunity. Only by testing the suitability of work through actual attempts to work can any true limitations which prevent full participation in the labor force be dentiﬁed and resolved
- Strengthening the ability of parents to provide for their children is a better approach than having the government intervene directly on their
In well-meaning attempts to look after the interests of children, government has, over time, participated in many of the roles that were previously the exclusive responsibility of parents. There are many calls for government to take on still further responsibility for assuring the well-being of children. However, government cannot raise children, only parents can. Government can do the most by helping to put parents in a position to meet their responsibilities, not by taking away these responsibilities for itself.
- Measure the fairness of the new system by comparison with working families
It is sometimes argued that a work-based welfare system will be unfair unless it can be shown that those formerly dependent on various beneﬁt programs will continue to receive a comparable package while working. Others argue that it is unrealistic to expect work for wages unless such wages will guarantee a high enough standard of living to make work seem worthwhile
But self-suﬃciency through work should be seen as an end in itself, quite apart from the package of beneﬁts gained or lost as a result. More important is the relationship that those who are receiving welfare beneﬁts have with comparable individuals who are working to support themselves and have not asked for assistance.
- Look to non-government organizations to deliver the
It is axiomatic that government programs, authorized by legislation, must be overseen by government as an agent of the public interest. However, for too long government has been the assumed operator of the programs it devises. A more eﬀective model is almost always for the government to set the ground rules and then let non-government entities actually operate programs under public oversight.
The Wisconsin-Works program (or W2)
What does a work-based safety net program such as W2 mean?
Under W2 there is no entitlement to cash assistance based on income alone. When an individual without work applies for help at a W2 agency, their access to cash beneﬁts is dependent upon their actions to help themselves become economically self-suﬃcient through private employment. All individuals participating in the program “earn” their income (cash assistance) through participation, measured by the hour, in one of four tiers of employment in private or public settings. These four are from highest to lowest, which make up an employment “ladder”.
The Four Tiers of the Ladder:
Number One and first option—Unsubsidized Employment
Individuals entering W2 should always be guided to the best available immediate job opportunity in the private sector. The agency is geared and focused on matching every participant’s capabilities with the best work option available when they arrive, rather than diverting them to extended education and training program.
Unsubsidized employment pays the market wage (currently about $7.00 – $8.00 per hour) along with the earned income tax credit (EITC) which contributes up to $4000 per year depending upon family size and income.
Second option—Subsidized Employment (Trial Jobs)
Where an individual could succeed in private employment but lacks work experience or in some instances skills, the W2 agency can provide wage subsidies to private employers for a temporary period in order to oﬀset some of the additional costs of integrating a new employee to the worksite. In practice, Trial Jobs are rarely used because employers say that the economics favor hiring even unskilled workers directly without the modest Trial Job wage subsidy, and they prefer to stay away from the administrative tasks of collecting the subsidy.
Third option—Community Service Jobs (CSJs)
If an individual tries and cannot ﬁnd private employment, the next option is to work in exchange for beneﬁts, in either a government or non-proﬁt organization which serves the public. Examples include simple oﬃce work or outdoor maintenance for government. These jobs are intended to be 30 hours per week, to be useful even with minimal training or background, and to oﬀer the opportunity for individuals without a work background to learn-by-doing, or in other words to experience work supervision and task orientation in an actual simulated employment setting. An additional ten hours can be used for education and training, plus job search.
In order to replicate the circumstances of the workplace, the W2 cash beneﬁt of $673 per month is reduced for each hour of scheduled work missed by the minimum wage (currently $5.85). CSJs are intended to be temporary, each assignment generally not to exceed six or nine months. The by-product of labor in such a setting is intended to contribute to the community and fulﬁll the essential civic role of work—giving to others in exchange for wages.
Fourth option—Transitional Jobs
For those unable to perform independent, self-sustaining work even in a CSJ because of limitations such as mild disability or substance abuse, Transitional Jobs are available for participation in work or vocational training and other activity such as treatment, consistent with an individual’s capabilities. The grant amount is $628 (less than that of a CSJ) for up to 28 hours of eligible activity, and up to twelve hours of additional education and training.
Summary of Incentives for Participants to go to Work
In order to receive cash assistance, an individual enrolled in W2 faces approximately the same number of hours dedicated to a community service job or a transitions job, as would that in full-time private employment. Leisure without obligations was no longer an option; cash beneﬁts are reduced at the rate of minimum wage for each hour of missed work in a community service or transitions job, to replicate the conditions of private employment.
At each step up the work ladder, from Transitions to unsubsidized employment, the net after-tax income (which includes the earned income tax credit for those in private employment) increases. Therefore participants have every reason to accept the highest employment option they qualify for, starting with unsubsidized employment if available, or if not to move up the ladder when the option presents itself.
Supportive services such as subsidized child care or free medical care are no longer reserved for those enrolled in the welfare system, but rather to anyone on a sliding scale based income alone. Therefore there is no longer an advantage to enrolling in welfare to just to receive these extra beneﬁts.
The Operational Considerations Driving Program Design and the Privatization of Services
When implementing a work-based system, the W2 designers were faced with a dilemma. The organizational requirements of W2 are higher than that of running an income maintenance agency where the primary task is determining eligibility based on income (along with making referrals to separate job training or employment agencies).
Previously welfare operations were divided into two parts. Eligibility and administration were administered only by county government, while employment and training could be operated by county government, or through county contract to any non-proﬁt or for-proﬁt organization.
The county-run welfare system could function satisfactorily where its main task was to determine eligibility. However as an operator, government has well-known limitations:
- Usually managed by process rather than outcome;
- Lack of accountability due to the diﬀusion of responsibility;
- Performance compromised by decisions subject in part to political considerations;
- Lost mission focus as energy was sunk into attempting to coordinate the multiple organizational units operating the program (e.g. the state policy making function, the county administrative function, and the non-proﬁt service delivery function);
- Inability to shed non-functioning operational units.
In particular, Wisconsin’s largest county of Milwaukee (population one million) showed resistance to mission change or operational improvement. Like many large cities, Milwaukee’s welfare agency had weak management and a labor force rarely challenged to perform. In the period leading up to W2, the state had attempted to get the welfare agency to properly refer welfare recipients to private agencies for work or training, and when not participating in such training, to reduce cash beneﬁts until recipients complied. The county agency exhibited limited bureaucratic capacity to carry out such multi-step processes, nor among agency top management an interest in the objective itself.1
In any event the requirements for W2 went beyond the task of eligibility determination and referral for employment. An agency would have to act in such a way as to limit its own size and inﬂuence in the lives of participants and instead to empower them to achieve economic self-suﬃciency. Through contract incentives, we believed we could achieve this.
We merged the two parts, eligibility and job placement, into one so that there would be a single organization wholly accountable for results. This organization, the Wisconsin Works agency owns the Job Center, provides eligibility determina- tions for cash beneﬁts and provides work supports such as subsidized child care or “job access loans” (these are temporary loans for the purpose of enabling employment, e.g. for repair of an auto so the participant can drive to work). Most impor- tantly, the role of the W2 agency is to help applicants ﬁnd the best available job opportunity at the time of application, and provide such a job itself as a last resort.
Helping individuals with marginal work histories would require closer connections among case-workers, employers and participants. We anticipated that the W2 case workers would be paid higher salaries and carry far lower caseload than traditional caseworkers (about 75 instead of about 300). They could be recruited from among the best of the existing case workers, but also from the outside, based more on personality characteristics than longevity or specialized education.
Finally, we wanted the W2 agencies to experience the maximum latitude to ﬁnd their own solutions to achieving the outcomes we wanted—namely employment and caseload reduction. In this regard state policy makers had struggled with some of the counties (not others) over the years before the implementation of Wisconsin Works.
Competition for vendor selection
In order to maximize the likelihood of Wisconsin Works mission success, our plan envisioned a competition for program operators, open to any and all organizations, county government included. As potential W2 agencies, three of the four types of potential W2 operators had strengths:
- County government, particularly in smaller or rural counties, was in certain places the only organization with the infrastructure ready to go. Looking forward, in many instances counties had dedicated staﬀ and supportive elected oﬃcials.
- Non-proﬁt organizations were sometimes well established and were already providing the kinds of services required under Wisconsin Works, particularly those expected to be expanded such as community service jobs, in which rehabilitation agencies such as Goodwill already provided on a relatively large
- For-proﬁt businesses were already providing employment and training services and could be expected to oﬀer a dynamic energy and be more capable of assuming However, in the period leading up to W2, for-proﬁt organizations were only a modest factor in the overall inventory of operating providers so this sector would need to be developed.
- Quasi-government providers, in particular the “private industry councils” with governing boards appointed by the counties along with volunteers from the local business community were the least appealing as agents of They were usually highly politicized grant making organizations to employment and training vendors, hobbled by internal conﬂicts on their boards and in no position to lead a dynamic in-house operating agency. Consistent with our open competition, they were not excluded from the RFP process.
As we looked to organize the incumbent service providers into a force for change, we attempted to maximize their freedom of action and provide the clearest possible incentives to succeed.
In the past, multiple organizations combined into consortia diﬀused responsibility, and required consensus to act. Poor performing partners could not be expelled because they were part of the management team. Further, states would ask for multiplicity of happy results—but the achievement of one would often compromise another.2 Under W2 the state’s objectives were clear—to maximize family economic self-suﬃciency through employment and to reduce the corresponding cash assistance caseload. We permitted partnerships, but only ones with prime and sub relationships so that leadership would remain clear.
Another consideration was size.3 We noted that the smaller the county agency, the higher the employment rate. Of course, the welfare population in small counties is generally easier to serve, and the culture of self-reliance is higher. But there were other diﬀerences too, having to do with span of control. In a big city the welfare department a recipient might not see the same caseworker twice, leading to anonymity. Follow-through is less likely to occur face-to-face and more likely through a computer tracking program. Responsible recipients looking to connect with the department for help often will not get through to caseworkers by telephone, with voicemails piling up. Managers and supervisors manage by numbers, paper ﬁles and adherence to regulations, not through the exertion of personal leadership.
For this reason we divided Milwaukee into six regions of about 6000 welfare cases each, so that the state would contract with multiple vendors. Each region was drawn so that it would have, to the degree possible, recipients of similar general characteristics. In addition to the expectation that smaller vendors would perform better, multiple vendors in Milwuakee reduced the state’s risk in otherwise relying on one vendor for a caseload representing about half the state, and low performing vendors could be replaced with less disruption. Finally, we hoped that smaller vendors would become identiﬁed with their surrounding neighborhoods. Prospective vendor agencies were permitted to bid on more than one, and in the event we awarded one of the applicant vendors (Goodwill) two regions, each with its own agency oﬃce.
Rather than replacing the old AFDC program with another one operated on the basis of the same regulatory “command and control” mechanism, we were looking to create powerful contract incentives which would be in alignment with the incentives that the participants themselves would experience to ﬁnd and accept private employment.
In the period the few years prior to W2 we had experimented with a simple form of performance incentives for job placement and training agencies which had functioned better than expected.4 The vendors earned part of their allocations based on their job placement success and they went earnestly to work, increasing placements by 29% the ﬁrst year. But what was equally remarkable to us at the time was the degree to which vendors had been misallocating eﬀort and resources prior to the introduction of this outcome-based payment system. One vendor director said “I had thought we were supposed to be getting recipients general equivalency degrees” (remedial education degrees). He said this despite the fact that the state had been exhorting the vendors to change their emphasis to employment for some time. It occurred to us that our means of communication (words) were deﬁcient and we had at last stumbled upon the right tools.
The initial contracts from September 1997 through December 1999 were structured so that reductions in the caseload resulting from movement off W2 cash assistance into employment would reduce costs which would then be shared with vendors as follows:
- The initial contract amount was set by county region by calculating the current total welfare expenditures in the most recent period (including cash benefits, employment services and administration), and reducing this amount by twenty percent. This total amount could be allocated by the agency according to its needs, after cash beneﬁts were first paid to eligible
- Agencies generating caseload reductions which generated savings beyond the total budgeted amount for the period, could keep the ﬁrst seven percent of the savings in the form of unrestricted income (or profit). Thereafter additional savings would be divided three ways:– Forty-five percent would revert to the state in the form of budget savingsForty-five percent could be allocated by the W2 agency to their local community for expenditures in general support of low income families, with approval of these expenditures by the state;Ten percent of additional savings would revert to the W2 agency in the form of unrestricted income (or profit).
- Agencies whose expenditures during the contract period exceeded the total allocation from the state would be obligated to provide the difference.
The Contracting Process
When the initial Wisconsin Works plan was made public in April 1995, it met with great interest. However, early discussion revolved around the provision which would open Wisconsin Works contract opportunities to any potential service provider—county, non-proﬁt or for-proﬁt. Since all state legislators have constituents employed in county welfare departments, almost all legislators heard from county employees concerned about the future of their jobs. From the perspective of the state welfare department, many or most of Wisconsin’s 72 county governments, especially in rural areas, were likely to do a good job administering W2, and we did not want to jeopardize the overall future of the plan due to objections over which we were unlikely to prevail.
Looking to make the best of the situation, we created a right of ﬁrst refusal for county governments which wished to operate W2 and met certain performance objectives in the twelve month period prior to contract selections. The performance objectives we selected were directly related to ones which would make the W2 agencies successful in helping their participants gain employment once the program went live in September 1997.5 The response upon the publication of these measurement elements by county management and employees was electric. County elected oﬃcials and managers began in earnest to work on achieving the right of ﬁrst refusal, setting up their measurement systems and organizing their agencies along the lines required. At the end of the measurement period all but about half a dozen counties had succeeded in meeting the requirements, with the most notable and expected exception of Milwaukee (where county elected oﬃcials had in any event by then decided not to compete for W2, reasoning that the ﬁscal risks outweighed the rewards).6
The state next contacted prospective vendors located outside Wisconsin in order to have a robust and competitive bidding market, especially within Milwaukee. At that time several big for-proﬁt corporations specializing in information technology were actively considering entering the welfare-to-work marketplace. They reasoned that their proﬁtable IT businesses, already serving state human service agencies throughout the country, would be natural partners gaining entry wherever they received contracts as welfare-to-work vendors. Companies we were aware of considering such line extensions of their business at that time included Electronic Data Systems (EDS); Lockheed Martin IMS; Aﬃliated Computer Services (ACS); Deloitte & Touche; and IBM. However, although some of these companies already had welfare-to-work contracts elsewhere, for various reasons, none of these giants elected to bid on Wisconsin Works in Milwaukee or in the state (EDS partnered with Goodwill, an eventual Milwaukee vendor, but provided only IT services). It is possible they were unnerved by our unusual state contract, in which vendors would be liable for making up shortfalls over the state budgeted amount (including shortfalls in cash assistance amounts) in the event they were not successful (smaller vendors, especially non- proﬁts agreeing to sign our state contract, perhaps reasoned that were they poised to go under, the state would have to bail them out or take over the program, which in fact was true although it never happened).
One large for-proﬁt that did submit a bid for several regions in Milwaukee was Maximus, a corporation which had pioneered the human service operations business in Los Angeles in the 1980’s, and at that time was the largest of its kind. Some smaller for-proﬁts also submitted bids in Milwaukee and elsewhere in the state. After the RFP process contracts were awarded in Milwaukee to the following:
- Employment Solutions, a for-proﬁt subsidiary of Goodwill Industries, a major provider of welfare-to-work services in Milwaukee and a long-time vocational rehabilitation agency with national presence; this agency was awarded two Milwaukee regions;
- Opportunities Industrial Corporation OIC), an African-American community vendor which had provided welfare-to- work services in Milwaukee and was part of a national organization;
- United Migrant Opportunities Society (UMOS), a Hispanic community vendor which had provided welfare-to-work services in Milwaukee;
- YW-Works, a for-proﬁt subsidiary of the Milwaukee YWCA and two other organizations which had provided services in Milwaukee;
- Maximus, a for-proﬁt corporation without previous contracts in
Eight counties other than Milwaukee were served by ﬁve private agencies in the initial implementation contracts.
The Program Results from the Implementation of Wisconsin Works
As we awaited the September 1997 W2 start date, we believed there would never be a more propitious set of circumstances boding for the experiment’s success:
The plan was about to open its doors “on the street” almost exactly as planned, with virtually no signiﬁcant design compromises along the way; the state economy was strong and general unemployment was going down; certain elements of the program had already been put in place in advance of implementation and the caseload was already moving in the right direction; the Right of First Refusal had organized the counties which would run the program around the program design; the Wisconsin public was well informed by the Governor and the media about the plan and was with us all the way; our natural opponents among foundation-funded advocacy organizations and academics at the University of Wisconsin were temporarily subdued (even if not supportive they acknowledged the plan’s comprehensiveness and internal consistency); welfare recipients themselves had been exposed to and absorbed several years of media and public discussion about the importance of work; soon work in some fashion was to be the only option for receiving cash assistance; and for county and vendor operators alike the stakes were exceedingly high on both the upside and the downside.
Events had created a “perfect storm”.
— THE RESULTS —
The results were more powerful than anyone, including its most optimistic planners anticipated. Immediately after Wisconsin Works opened, the welfare caseload began to collapse. In the year prior to W2, Wisconsin had about 55,000 welfare cases statewide—this dropped to 11,000 cases in the ﬁrst thirty-six months of the program (thereafter rising slowly and then falling back to 11,000 in 2005).7
At the end of the second year the state looked back at all of the Wisconsin Works cases which had been enrolled in the program at some point and were now closed (32,000 cases or 81% of all who had enrolled). Of these, fully 76% had closed for earnings and income (9% had chosen not to roll over into the new W2 program from the old one, 8% had enrolled but then closed for failing to comply with W2 program requirements such as attending work assignments, 4% no longer met basic eligibility requirements and 2% closed for other reasons). Among those working, about half were earning between $12,000 and $18,000 per year (before additional refundable tax credits), about a third were earning less and about a ﬁfth were earning more8 (the 1998 poverty threshold was $13,001 for a family of three).
As would be expected with large numbers entering the labor force, poverty dropped substantially in both Milwaukee and the rest of the state. The child poverty rate (a more sensitive measure than adult poverty) peaked at 34% in Milwaukee in 1993, declining in the period before W2 implementation, and then, in the ﬁrst single year afterward, dropped a full six percentage points to 20%. Thereafter it rose slowly along with the weakening of the national economy; by 2006 after the beginning of the recovery it had given back about half of the drop from the top (the state and national trends were much gentler, but also followed a similar curve).9
The Critics of Wisconsin Works
Wisconsin Works is about ten years old, and has almost certainly been the most studied and analyzed program of welfare reform in the U.S. (in just the ﬁrst three years the Urban Institute counted 53 studies completed or underway).10 There are numerous critics of W2 and its work-ﬁrst approach.11 Criticisms fall into two categories—shortcomings in program implementation, particularly in Milwaukee, and second, the continuation of poverty as a social problem. Issues related to the ﬁrst of these criticisms will be discussed at greater length below.
Wisconsin’s most vocal critics at ﬁrst warned of serious consequences which could be anticipated from large numbers who would be unable to cope with the work requirements of W2, dropping out or never applying. Many would end up homeless, resorting to crime, or worse, neglecting or maltreating their children. None of these alarming predictions have come to pass. Instead, the rate of homelessness has remained ﬂat12 , crime is down 12% in the state since program introduction13, and incidents of child abuse and neglect have fallen in Wisconsin as they have nationally.14
Of course, poverty has not been stricken from the face of the Wisconsin prairie. Even if every individual who enrolled in W2 had found full-time work right away and kept his or her job forever, many of those starting in entry level jobs would go through a period of income below poverty on the presumed way to wage increases over time. Yet wage growth has been stronger than academics predicted. An early study from the University of Wisconsin reported:
In part because of the substantial caseload reductions that preceded the implementation of W2, many of the ﬁrst participants in W2 had low levels of education, substantial family responsibilities, and a history of reliance on welfare. Notwithstanding these barriers, we ﬁnd higher levels of employment than have been found in other states, and substantial growth in employment and earnings over the short period considered. Average family income…increased, rising from about $12,000 in 1998 to nearly $15,000 in 1999, and the poverty rate [among participants] fell from 77 percent to 67 percent. [emphasis in the original].15
Looking over a longer period, for those who left W2 at the end of 1999 and had full or part time employment in each of the following four years, average annual income increased by 22%. Of this group 37% had exceeded poverty based on income alone, while 57% were above poverty after tax credits.16 17
Privatization After Implementation
While traditionally about half of Wisconsin’s welfare population lived outside of Milwaukee, after the sharp caseload reductions more than four ﬁfths of the remaining small number of cases reside in the county, which is fully privatized. Therefore we will devote most of our attention there. However, it should be noted that in the balance of the state, the numbers of counties which have elected to privatize has steadily increased, from eight in the ﬁrst contract, to 21 in 2005. Of the privatized counties, one is a medium sized city, three are suburban Milwaukee, and the other 17 are rural.
Within the rural areas of Wisconsin, the growth in the privatization of W2 makes sense in part because there are so few remaining cases, and some counties no longer think of welfare as one of their main lines of business (superseded by foster care and adoption, medical assistance, distribution of food stamps, subsidized child care etc.). A by-product of the early years of Wisconsin Works outside of Milwaukee is that many counties merged their related programs into “one-stop” job centers with modern oﬃces and equipment because funds from savings were plentiful.
Turning to Milwaukee, among the ﬁve contract vendors the state selected, one might consider three of these high- performing (one of these got oﬀ to a bad start and recovered), one average, and one poor (the last is no longer in business).18 The transition between the old program and the onset of Wisconsin Works under the new privatized agencies, from those who experienced it, can best be described as an intensive, compressed and creative fervor.19
The new private agencies were not only operating under a completely new set of contract provisions, but also under a new mindset. Competition for good operations directors was pronounced and in addition agencies responded by looking for line staﬀ outside the traditional social work background. Some agencies advertised in sales and marketing magazines and sites because they were looking for employees comfortable interacting with businesses in a more professional environment and willing to accept contingent compensation based on performance.
In one high-performing agency the ratio of new hires to transitioned staﬀ was high—about three new hires to one transitioned from the old program. Of the older staﬀ who had worked in county government, some transferred to other county human service jobs, some retired voluntarily, and some were hired by the new agencies as at-will employees. With competition for good staﬀ, salaries increased. Former workers earned in a salary range around $28,000 per year (1997 dollars) with sizeable beneﬁt packages; new workers in one agency earned $35,000 to $38,000 with annual bonus opportunities of three to ﬁve thousand dollars, albeit with lower beneﬁts. All agencies had their employees sign non-compete contracts, as they were concerned that the expense of training on this new program would be lost to competitors (in the event, no enforcement of these contracts ever occurred).
In addition to hiring from the outside, agencies also directed their relationships more to employers. All agencies paid special attention to assuring their job developers had solid connections with employers and access to jobs. One agency created a new title called “business service representatives” which did no casework but whose job it was to take orders from employers and act as its representative within the agency in order to ﬁnd the best employee ﬁt. Business service reps would interact with caseworkers, each with their own bonus structure and incentives for job placement, but with diﬀering internal customers.
As pressure mounted prior to the start date of W2, managers and new staﬀ worked late and the state helped by keeping its caseload computer system open nights and weekends. Each case transitioned on the computer from welfare to W2 would take two hours, and there were thousands of them to be changed. One agency set up a separate room with 20 computers and a knowledgeable programmer to allow case workers to have quiet time for the transition. All this was occurring as new oﬃces were being built-out and ﬁtted (one agency built a new professional oﬃce from the ground up in the heart of the district which had once seen race riots).
Finding good middle managers is always key in government and human service organizations (and experience shows almost always more diﬃcult than ﬁnding good agency directors or good line staﬀ ), and this task was made more diﬃcult because new middle management hires were not familiar with the technical aspects of the job, nor the welfare rules, which had to be understood in order to explain to staﬀ how to operate in the new environment. The absence of quality middle managers has been a major shortcoming of agencies which performed below expectations.20
In addition to their internal pressures, agencies had to go outside and speak the public so as to assure that existing welfare recipients came into the oﬃces and changed their enrollment. All agencies made special eﬀorts, using door hangers, church announcements, interviews on television, and in some cases doing home visits if individuals did not respond to multiple letters. In the end, about 30% of the former welfare recipients declined to enroll in the new program (some were already working oﬀ the books, others might not be eligible, still others did not want to participate under the new work requirements).
Work sites had to be prepared, and some agencies showed creativity here too. One purchased a plastics fabrication plant in order to provide easy work which would be ﬁscally self-sustaining, another used its own packaging, used clothing store, and commercial laundry service, still others used the county’s existing program with many outdoor assignments such as keeping parks maintained.21
Competition among vendors during this period was beneﬁcial. Each had its own geographic region, so in a sense they were not competing for “customers”. Each had its own performance goals and ﬁnancial targets, but at they same time none wanted to be perceived as lagging. Once a month the ﬁve agency executive directors would meet, to hammer out common concerns (and, to a degree, to act as a counterweight to the powerful state agency). Inter-agency work groups were formed around subject areas such as transitioning cases from one program to the other.
The Successive Contracts
During the ﬁrst contract period savings exceeded the most optimistic predictions. In Milwaukee, $319 million had been allocated for all expenditures, including cash beneﬁts. Of this amount 84% was spent, leaving $26 million for agency proﬁts, $13 million for reinvestment in the community and $13 million to the state in the form of savings, just as provided for in the contract.
Certain critics of W2 have said that the ﬁrst round of contracts did not contain performance criteria, but we certainly did not view it this way, nor did the vendors. These Milwaukee vendors were taking substantial risks in a completely unknown environment. Their performance would be measured by their employment success leading to caseload reductions, while leaving them the operating room to get there in the best and most innovative way. Critics also claimed that agencies could simply reduce costs by denying service to applicants, or arbitrarily categorizing them as job ready thereby excluding access to community service jobs. This objection too, is unfounded. Agencies created procedures to move applicants through their system, and while individual cases might not be processed properly, no agency would risk denying service with the stakes so high (in addition, the state could penalize an agency $5000 for each instance of denied service, although it never found it necessary to invoke such penalty).
Nevertheless there were signiﬁcant problems with vendor implementation. Two of the high-performing agencies billed hundreds of thousands of dollars improperly to the W2 program as they chased new business outside the state.22 Both these agencies reimbursed these disallowed costs and made additional payments in compensation. However, one of these two agencies declined to bid on a third contract period as a result. Its loss was to be a great detriment to operations n Milwaukee.23 24
In the face of bad publicity over the misallocation of funds, the state agency overreacted. In the third contract period the state added a surfeit of performance goals, too many, both process and outcome. It sharply reduced operating funding levels to that below what was necessary, and permitted any performance bonuses to be earned only as a reimbursement for funds the agency had already expended, putting these agencies in a very serious ﬁnancial risk were to fail to meet the standards after the fact. In short, the state had shifted to the agencies almost all risks but took away the opportunity for rewards.
In addition to pressing its vendors beyond what was necessary for good management, the state made a second mistake. After the exit of the high-performing agency which had misallocated funds, there remained two of six regions without an operator. The state had failed to cultivate other potential replacement vendors from outside Milwaukee and so doubled the contracts of two of the existing Milwaukee vendors (which was acceptable as far as this went). But in the subsequent fourth contract period the state contracted to the lowest performing vendor three of the six regions, reshuﬄing the deck once again and in a way which proved most unfortunate. The winning vendor’s executive director and others with whom he had ties were indicted for kickbacks in 2004 and sent to prison.25 The agency folded soon thereafter.
The Milwaukee operations are recovering from these setbacks. In the most recent contract period the state has at last brought in fresh vendors from outside the county, increased the contract responsibility of incumbent strong vendors, and management has signiﬁcantly improved. In an indication that the state intends to devote more of its attention to Wisconsin Works, it has organized a dedicated new agency to the function. The W2 caseload is stable and low, and participant exit for employment continues at a high rate (though slower than it was right after implementation).
Lessons from the Wisconsin Works Privatization and Contract Management Experience
Looking back from the vantage point of ten years, one sees that Wisconsin Works unleashed a wave of change on a scale unanticipated and unprecedented. Dependency was reduced far more than expected to the point that welfare is now con- sidered only a secondary public policy issue. Former welfare recipients proved far more resilient than believed, took available jobs in large numbers, increased their wages over time and reduced the corresponding poverty rate. The state saved enormous sums as beneﬁt payments to the formerly dependent were replaced by wages. The state legislature has re-allocated these funds in a variety of activities including programs for youth, literacy, nutrition, and domestic violence prevention.26
While the newly available funds was a welcome indication of program success it created a challenging political dynamic, with critics denouncing the proﬁts earned by the private agencies. The state should have been better prepared to forthrightly make its case that program savings were only available because families were newly self-suﬃcient, and that a portion of the savings were being plowed back into the community through the reinvestment provisions or through state reallocations.
The ﬁrst lesson here is that program changes on the scale of W2, combined with privatization and new vendor contract incentives, will require swift iterative actions on the part of each actor, state and vendor community, in order to maintain an equilibrium and manage through the learning process.
Further, it is evident that the state administration which took over after the election of 2002 (Governor Jim Doyle, Democrat) lost focus on the program. In a familiar turn of events in every democracy, newly elected oﬃcials must stake out their mark distinct from their predecessor’s, and Wisconsin Works was indelibly linked to the former Governor Thompson.
The new administration brought its own W2 philosophy to governing, but in a way that was inconsistent and reactive. In 2003 the state declared that the program should be “participant-friendly” and explicitly oﬀer a greater array of services. It instructed the vendors to assure that all barriers to participation in the program be identiﬁed and resolved before reductions in beneﬁts due to non-participation were taken, partly in response to program critics (because of the new rule’s ambiguity, vendors were reluctant to act, often permitting non-participating recipients to receive beneﬁts without working in community service jobs). Primarily resulting from these policy changes the caseload started to climb and dependency went up. The state reacted and reversed itself, again emphasizing early employment and work, leading to confusion on the part of agencies and vendors.27
The second lesson, then, is that the governing agency must maintain an interest in the state of the vendor marketplace and in the incentive signals it is sending. Vendors do not operate on autopilot, and will feel comfortable taking risks where they feel their footing is secure. Frequent signiﬁcant changes in stated program objectives reduces performance—smaller changes at ﬁrst are more readily absorbed.
The state initially selected good vendors in Milwaukee (along with at least one poor-performing one) but it did not tend to the creation of a vibrant marketplace for the growth of competitive services there. We take for granted that there will always be competitors and options in the private sector marketplace for goods and services. However the competitive human service marketplace remains insuﬃciently developed. The third lesson then, is that the state itself must actively create and nurture a competitive marketplace environment. It should recruit potential vendors from outside its borders even when no bids are active, it should conduct annual meetings with prospective vendors to discuss the program state of aﬀairs, and it should consider setting aside a certain portion of each contract for new entrants into the marketplace.28
What about the nature of contract performance criteria themselves? Our initial contract incentives in which agencies shared in reduced expenditures was powerful and outcome based but did not give suﬃcient information about ongoing program activity levels such that thorough reports could be made to the public and policy makers.29 On the other hand, in the third contract period the state created so many measurement standards that taken together they did not drive agency performance toward an understandable or suﬃciently narrow mission objective.30 The fourth lesson, then, is that the state should determine its most important policy objectives, organize its performance payments for these using outcome based measures (and a limited number of process measures too), and refrain from trying to achieve all conceivable desirable objectives through one contract.
The ﬁfth and ﬁnal lesson from our Wisconsin experience is that the energy and creativity generated by private vendors in conjunction with powerful contract incentives were together essential to achieving the program success. During the most intensive and creative period of the Wisconsin Works transition, the new Milwaukee vendors, the other county agencies and the state worked cooperatively at a breakneck pace to get ready. Once operations began, each Milwaukee used its operational ﬂexibility to approach operations in various ways, with diﬀering emphases.31 And when it is all is said and done, when we include all of the program’s avoidable and unavoidable mis-steps, the lives of Wisconsin’s poor have been permanently improved.
1The situation among counties outside of Milwaukee was much better, with most county agencies embracing the mission of economic self-sufficiency for its welfare population. But some objected to the state’s preferred “work-first” approach to achieving this objective rather than the traditional education and training model. Our explanations and exhortations favoring the work-first approach and explaining its rationale, including the research evidence of its efficacy, were most influential when falling on willing county ears. However, even there the state had made progress in this regard by the introduction of “Work-First” pilots. These pilots, in which applicants for assistance looked for work even while their application was pending, proved successful in achieving early employment outcomes. Even better, the success of “work-first” was touted by the pilot counties themselves, leaving them to sell the idea rather than the suspect state.
2For example, many states would ask their welfare training vendors to simultaneously increase exits due to employment and to maximize average wages at exit. These are both worthy objectives, but maximizing participant employment exits, including those with limited work experience, will depress average wages at exit (although not necessarily in the long run, as those in the labor market increase their wages over time). In this example it would be better for the state to give one assignment to welfare-to-work vendors (to get as many as possible into the labor market) and a second assignment to education and training vendors (to improve wages though skills upgrades among both the working and non-working population). The success measurements for each will diﬀer, as will the corresponding strategies.
3For example, for every 100 welfare cases in Milwaukee, there were 15 entered employments over a year, while the median mid-sized county obtained 26 and small counties 33
4In Milwaukee, we grouped the six job placement and training vendors into two teams, A and B. Each team had a lead vendor, and the proportion of referrals from the state to the respective teams was exactly proportionate to their respective total team budgets for services. Since vendors from Team A had two-thirds of the total job placement and training budget from the state, we referred to them all individuals with social security numbers whose last three digits were between 001 and 666, with the balance to Team B. This system of referrals was easy for desk clerks to learn, and at the end of the competitive period the state could easily determine employment outcomes by using the social security numbers to match against quarterly wage payments already present in the state data ﬁle.
5These performance objectives included:
Reducing the welfare caseload by ten to twenty-ﬁve percent depending on prior results;
Placing greater numbers of recipients into unsubsidized employment of at least 20 hours per week, increasing over three measurement periods; Placing those not working in private employment into the equivalent of community service or other jobs, increasing over three measurement periods; Reducing expenditures for cash assistance;
Credit given to counties which had cooperated with the state and performed well in the past.
6The right of ﬁrst refusal process showed that in certain circumstances government entities are motivated by ﬁnancial incentive structures as well as private vendors are, and that the resulting improvements in operations, if any, should not be automatically assumed to derive exclusively from privatization.
7Wisconsin Department of Workforce Development ﬁgures. Note that 9000 welfare cases without an adult (child-only) are excluded from both ﬁgures.
8Wisconsin Department of Workforce Development; Case Closure Study; 2000.
9U.S. Census, Current Population Survey.
10Urban Institute, Overview of Research Related to Wisconsin Works; March 2000.
11As an example see University of Chicago, Chapin Hall Center for Children; Issue Brief #107; May 2006.
12Reliable statistics are diffi cult to obtain; anecdotal reports from author’s interviews with the director of Hope House, Milwaukee’s largest shelter, indicates that the total number of overnight stays has remained relatively constant since Wisconsin Works was introduced; fewer individuals are entering the shelter now as compared with before, but this is off set by an increase in the average length of stay. Also see Milwaukee Continuum of Care Steering Committee, Homelessness in Milwaukee, July 2007 for a point in time survey.
13Wisconsin Offi ce of Justice Assistance; Crime and Arrests in Wisconsin 2006; p.105.
14See Urban Institute, Welfare Reform’s Effect on Child Welfare Caseloads; Feb. 2001. Also University of Chicago, Chapin Hall Center for Children; Issue Brief #107; May 2006; table 5.
15University of Wisconsin, Institute for Research on Poverty; Child Support Demonstration Evaluation, Phase I Final Report; 2001; p.2
16State of Wisconsin, Legislative Audit Bureau; Evaluation of the Wisconsin Works Program; April 2005; table 27.
17Another criticism is that Wisconsin’s work-fi rst, or labor attachment model, would benefi t from the addition of more hours of education and training, or even that education and training should substitute for community service work hours (note that non-work activities including training have been increasing as the program matures, but is intended to be an adjunct not a substitute for work). However, education and training as an intervention for the welfare population has been exhaustively studied over thirty years, including using extensive random assignment experiments, and has consistently been demonstrated to have barely detectible, or no detectible impact on wages or employment (one experiment in Portland Oregon with mixed work and training emphasis showed some impacts).
18This categorization is based on the judgment of the author of this paper, taking into consideration general performance over the entire contract periods, reputations within Milwaukee and the state, and the author’s own knowledge. The three high performers have been Goodwill (Employment Solutions), YWorks (YWCA lead agency); and Maximus. The average performer has been United Migrant Opportunity Society (UMOS) and the poor performer was
Opportunities Industrialization Corporation (OIC) which is now out of business. The high performer which got off to a bad start and recovered is Maximus. The
organization hired a poor director who was fired after a scandal, after which operations improved markedly. As of the date of this paper, the state has added
new vendors and organized the contracts differently, which is not taken into consideration here.
19The descriptions which follow are based on interviews in January and February 2008 with Milwaukee agency directors and managers who were present at that time, and the author’s own recollections.
20Good middle management is more important in government and quasi-government than in the private sector. There are more rules which must be observed, more direct supervision needed below because compensation is not directly tied to achievement, and more complex processes occur between top management directives and line staff implementation.
21However, the existing inventory of community service jobs maintained by Milwaukee County, which had been set up to accommodate single men and women under the recently terminated “general assistance” program, was insuffifficiently utilized by W2 agencies.
22The agencies were Maximus, a for-profit vendor, and Employment Solutions, a for-profit subsidiary of Goodwill Industries of Southeastern Wisconsin.
23The agency was Employment Solutions, which had contracts in two regions and performed exceptionally well.
24It should be noted that improper expenditures are not the exclusive province of the private sector. The state routinely finds significant amounts of improper expenditures resulting from audits of county operations and requiring reimbursement (for example a state audit of the Milwaukee County Hospital discovered such a high level of improper expenditures that the county resorted to selling the hospital, i.e. privatizing it). Nevertheless, unlike government misallocations, private sector transgressions are considered more serious violations of the public trust by all concerned, and probably should be.
25The agency was Opportunities Industrialization Corporation of Greater Milwaukee, now defunct.
26Department of Workforce Development, letter from DWD Secretary to State Auditor, March 31, 2001.
27State of Wisconsin, Legislative Audit Bureau; Evaluation of the Wisconsin Works Program; April 2005; p.98.
28The Netherlands has done this, creating a new vendor community for job placement of several hundred.
29As an example of process measures that the initial contract did not measure but should have, is the weekly average number of work hours assigned. Lack of attention to this measure is in part responsible for its decline from 26.5 average weekly work hours in 1998 to 17.7 in 2004. See State of Wisconsin, Legislative Audit Bureau; Evaluation of the Wisconsin Works Program; April 2005; table 37.
30Performance standards and measurements included required caseload ratios, caseworker training requirements, customer satisfaction, new required basic skills training for participants, wage increase measurements, job retention, full or part-time employment measures, among others.
31For instance, agencies allocated funds among program components differently, had variations in sanction rates, job placement and retention rates, performed differing participant assessments, used job access loans at different rates.