Creating a Marketplace for Social Welfare Services

Volume 16 Issue 2Symposium on Poverty and the Law

Article 12

Creating a Marketplace for Social Welfare Services

Douglas J. Besharov

CREATING A MARKETPLACE FOR SOCIAL WELFARE SERVICES

Over the last forty years, government social welfare programs have taken on increasingly important roles in advanced industrial nations. In countries of the Organization for Economic Co-operation and Development (OECD), rising social expenditures are largely responsible for the almost 60% increase in the percentage of GDP to government spending between 1960 and 1998, from 27.3% of GDP to 44.3%. For the United States alone, the percentages rose from 28.4% in 1960 to 32.8% in 1998.1

Social welfare expenditures are sure to climb in response to various demographic changes and technological advances. Aging populations and increasing family breakdown, for example, will require higher levels of public support. And new medical procedures as well as longer life spans will raise medical costs, while the demand for a more highly skilled workforce will raise post-secondary education costs.

Around the world, however, there is growing unease about the quality of government-supported social welfare programs. Too often, government programs seem ineffectual and unresponsive to the needs of their clients, even as program costs and total spending spiral. Various administrative procedures and regulations have been imposed in an attempt to correct such problems, but results have been uneven, at best.

This paper argues that many (but certainly not all) government social welfare programs would improve if market-based systems were used to put "purchasing power" directly in the hands of consumers, that is, clients. Such systems are "bottom-up" as opposed to "top-down" in operation. This paper also outlines the situations in which particular, "bottom-up" market mecha-

nisms-lower taxes, cash assistance, reimbursed fee-for-service regimes, or vouchers-are programmatically more appropriate. However, this paper purposely avoids specific programmatic prescriptions. A wise policy in one program area might be foolhardy in another. An approach that might work in one country, state, or community might be catastrophic in another. Instead, this paper presents the general factors that should be considered in deciding whether to create or encourage a market for a particular social welfare service.

I.

Governments provide social welfare services either (1) directly, through government owned and operated agencies, or (2) indirectly, through grants and contracts with nonprofit agencies, for-profit firms, or other institutions of society. In almost all countries, old-age pensions, cash welfare, and educational services (at least at the primary and secondary level) are provided directly. The picture is more mixed for other program areas, such as medical services, higher education, child care, and social counseling. Most European countries provide the bulk of these services directly, while indirect approaches predominate in the United States.

In many countries, growing unhappiness with governmentprovided social welfare services (as opposed to income-support and old-age pension programs) has led to increased interest in using non-governmental agencies to provide various services. (See Box 1) Often called "privatization," this movement has been aided by a general belief that non-governmental entities provide higher quality services at lower costs than do government

agencies.2 Moreover, many non-governmental providers are housed in key institutions of civil society, or "mediating structures,"3 such as churches and voluntary associations, which are seen as more effective because of their moral authority and their ability to leverage other forces in the family and community.4

The United States probably uses non-governmental service providers more than any other nation. The planners of Lyndon Johnson's "Great Society" in the 1960s deliberately chose to engage the private sector in their programs in order to build political support.5 Billions of dollars became available for grants or contracts to non-governmental agencies. As a result, the last thirty years have witnessed an explosion of government-supported, private agency-provided social welfare services.6 This approach has so permeated social welfare services in the United States that it even has a name: "third-party government."'

For example, in 1989, in the United States, nonprofit agencies that provided health services, education, social, and legal services, or that promoted art and culture, had combined revenues of approximately $476 billion, according to Lester Salamon, of the Johns Hopkins Institute for Policy Studies, in one of the few comprehensive studies on the financing of private social welfare services.' (Note- All dollar figures in this paper have been adjusted for inflation, and are in 2000 dollars.) Fully 32% of this total, over $153 billion, was provided by the government,

through grants, contracts, and other forms of reimbursement. (Fees and service charges accounted for 51% of total revenues, and private giving contributed 18%.)' Different types of agencies received different proportions of their funding from these sources: Health agencies, for example, received 55% from fees, 36% from government, and 9% from private giving; education agencies received 63% from fees, 17% from government, and 19% from private giving; and social service agencies received 23% from fees, 42% from government, and 35% from private giving.

In their aptly titled book, Nonprofits for Hire: The Welfare State in the Age of Contracting, Steven Smith, of the University of Washington's Evans School of Public Affairs, and Michael Lipsky, of the Ford Foundation, describe how government in the United States "relies on nonprofits to provide social services":

In fiscal year 1989, 14 Massachusetts state agencies spent over $800 million, about 8.5% of the state budget, to purchase from over 1,150 contractors such services as alcoholism rehabilitation, family crisis intervention, instruction in English-as-a-second language, and daycare. Overall, the state recognizes over 200 distinct types of social services in its purchase-of-service system. In New York City, several municipal agencies do not deliver direct services at all but wholly depend on purchase-of-service agreements with nonprofit agencies. These include the Youth Bureau, the Department of Employment, the Community Development Agency, the Department of Mental Health, Mental Retardation, and Alcoholism Services, and the Agency for Child Development (daycare and Head Start). The total amount of money disbursed in contracts in fiscal year 1985 was over one billion dollars.'

Alas, government utilization of private, nonprofit agencies, and for-profit firms has not proven to be the remedy for inadequate and unresponsive social welfare services. Many of these agencies end up looking-and behaving-very much like the government agencies for which they substitute.

In response have come calls for using a more competitive process in the funding of contracts and grants' '-propelled in

9. 10. SMITH & LIPSKY, supra note 6, at 4. 11.

See generally OSBORNE & GAEBLER, supra note 3; see also P. Nelson Reid, Reforming the Social Services Monopoly, Soc. WORK, Nov. 17, 1972; Robert Pruger & Leonard Miller, Competition and the Public Social Services, PUB. WELFARE, Fall 1973; John Goodman, Privatizing the Welfare State, in PROSPECTS FOR PRIVATIZATION

part by the fall of the Soviet Union and the disenchantment with command economies that it spurred. Whether it is a car or a college education, most people now believe that competition, more than anything else, promotes quality and lowers costs.

As a result, many countries are experimenting with competition in various program areas. Non-governmental entities are invited to bid on contracts or apply for grants to provide various services in an effort to instill greater efficiency and innovation in the provision of services. (Sometimes, government agencies are also allowed to bid, on the theory that this will improve their costeffectiveness as well.) 12

In many situations, competitive contracting encourages greater programmatic innovation and responsiveness, as well as putting a damper on prices. But much of this competition has been for government grants and contracts, which is really just another form of "top-down" decision making. In this paper, I argue that such "top-down" decision making is inherently less reliable than "bottom-up" approaches, and it threatens the independence of private agencies, including those sponsored by the mediating structures of civil society. Let me explain.

A.

"Top-Down" Decision Making

In "top-down" decision making, money is transferred from the government (or other source, such as a private foundation) to a public or private agency that, in turn, provides services to clients. (In federal systems, the money may move from the federal government to a state or local government and thence to service providers.) "Bottom-up" funding, on the other hand, is recipient-directed. Using government provided cash, vouchers, or similar instruments, individual recipients (or sometimes communities) determine which agencies or firms get funded through their cumulative decisions of which to use (or, in the case of communities, which they endorse).

In top-down funding, then, the government chooses the service provider that will aid recipients; in bottom-up funding, the recipients themselves choose their service provider. These two concepts, and their advantages and disadvantages, are widely dis-

cussed in the academic literature concerning the "implementation" of social programs. 13

Cash assistance programs-such as income support, unem-

ployment insurance, disability payments, and old-age pensionsare by definition bottom-up since the funds are given directly to individuals. But most developed countries usually fund other social welfare programs (excluding cash assistance) top-down. For example, government housing and education programs (at least at the elementary and secondary levels) are almost always top-down programs. Health care, child care, and higher education programs tend to be more mixed, partly because they are often built on earlier, private-sector models."

Top-down funding, and especially large, government-operated bureaucracies, have a long, and often successful history. In the United States, for example, many government-operated systems-the post office and public schools-were widely viewed as successful. What has changed?

There are many explanations for what appears to be an international breakdown in at least some forms of bureaucratic government. The growing complexity of social and management problems, the growth of unions and the inflexible work rules they often secure, the increase in non-governmental employment opportunities that draw away the most skilled workers, and so forth have all contributed to the apparent decline in the effectiveness of government-provided services. But above all these factors, one stands out: the rapidity of change and the consequent ongoing need for prompt operational responsiveness.

Bureaucracies work best, writes Warren Bennis, of the University of Southern California's Marshall School of Business, "in a highly competitive, undifferentiated, and stable environment, such as the climate of its youth, the Industrial Revolution. A pyramidal structure of authority, with power concentrated in the hands of few with the knowledge and resources to control an enterprise was, and is, an eminently suitable arrangement for routinized tasks." However, Bennis continues:

The environment has changed in just those ways which make the mechanism most problematical. Stability has vanished .... One factor accelerating change is the growth of science, research and development activities, and intel-

suppliers,

is, as well, more

lectual technology. Another is the increase of transactions with social institutions and the importance of the latter in conducting the enterprise-including government, distributors and consumers, shareholders, competitors, raw material and power sources of employees (particularly managers), trade unions, and groups within the firms. There interdependence between the economic and other facets of society, resulting in complications of legislation and public regulation. Thirdly, and significantly, competition between firms diminishes as their fates intertwine and become positively correlated.15 Top-down programs are at a disadvantage in this environment because they do not change quickly-or easily. "The organizations that thrive today," writes Bennis, "are those that embrace change instead of trying to resist it. The old Weberian bureaucracies are simply too slow, too weighed down with interorganizational agendas and priorities, to compete in a world where success goes to those who can identify and solve problems almost before they have names."6

Turning to competitive bidding for grants and contracts was an attempt to create this kind of flexible responsiveness-and to lower costs. Unfortunately, even when funding decisions are based on competitive bids and proposals, government is likely to fund the wrong agencies and services. A major reason is that the operative decision makers are not the actual recipients of the services, creating a disharmony of interests between the two. Even when top-down decision makers try to reflect the interests of recipients, they have limited tools for identifying those interests and weighing them against other considerations.

Competitive bidding can help, but only when decisions can be based on easily measured and relatively reliable outcome measures. But such outcome measures are rare for social welfare services. A government official might be able to gauge whether trash was being collected properly, but measuring the quality of social work counseling is much more subjective. Even medical care is not subject to simple cost-effective measurement. Hence, top-down funding of social welfare services tends to fall victim to the vagaries of misplaced priorities, bureaucratic convenience, personal favoritism, and political calculation. Thus, for example, according to Louis Winnick of the Fund for the City of New York:

15. WARREN BENNIS, BEYOND BuREAucRAcy: ESSAYS ON THE DEVELOPMENT

AND EVOLUTION OF HUMAN ORGANIZATION 10-11 (1993).

16.

Id. at xii.

A 1993 city performance review of New York city job-training programs gave an unsatisfactory rating to 25 contractors, 22 of which had their contracts renewed. In a report highly critical of New York's social service contracting policies, Constance Cushman, the Executive Director of New York's Procurement Policy Board, comments on this practice: "Vendor screening is politicized as never before.... We pay dearly for this system in administrative costs, higher prices, and decreased competition, yet still do not succeed in screening out corruption, much less foster quality. A free market is the best safeguard from corruption, and we preclude its existence."'

To be blunt, top-down decision making in the provision of social welfare services is analogous to establishing an industrial policy: Instead of trying to pick the best manufacturer, for example, government tries to pick the best service provider.

Worse, once government funding begins, political pressures make it almost impossible to end support (and rechannel funds toward more promising approaches). Allan Meltzer of Carnegie Mellon University documented this effect in regard to government provided seed money for start-up companies. He found that non-government decision makers were more successful at picking "winners" because they were more likely to abandon an obviously unsuccessful project than was the government. In other words, government has difficulty performing the key task of good decision making: creating "losers." As Meltzer explains:

Why, in general, is government less efficient? Someone had to decide to make additional investments in companies that appear to have good prospects, thereby putting more money at risk, or to shut down companies that no longer appear promising. Government is more likely to delay closing the failures and more likely to pump additional money to try to cover mistakes or misjudgments."

In the United States, many Head Start providers, for example, were originally chosen almost three decades ago for political reasons having little to do with their ability to care for children-

17. Louis Winnick, Privatizing Social Services: Are Vouchers a Pathway? 10 (Mar. 1995) (unpublished paper, on file with The Fund for the City of New York), quoted in John Hall & William D. Eggers, Health and Social Services in the Post-Welfare State: Are Vouchers the Answer?, 192 PRWATIZATION CENTER POL'Y STUD. 3 (1995).

18. Allan H. Meltzer, Why Governments Make Bad Venture Capitalists, WALL

ST. J., May 5, 1993, at A22.

and many continued to be funded even though they were pale reflections of their former selves.'9

B.

"Bottom-Up," Recipient-Directed Funding

These problems with bureaucratically managed, top-down regimes make recipient-directed, bottom-up approaches attractive to many observers. As Stuart Butler of the Heritage Foundation writes in To Empower People, "Don't decide-let the people decide for you."20

If there were a true marketplace of social welfare programs, particular agencies and services would be funded only because recipients chose to use them. As David Osborne and Ted Gaebler point out in their influential book, Reinventing Government, "The single best way to make public service providers respond to the needs of their customers is to put resources in the customers' hands and let them choose. Putting resources directly in customers' hands may at first sound like a radical idea, but it is not."21

"Customer-driven systems force service providers

to be accountable to their customers," argue Osborne and Gaebler.22 Under such systems, service providers must win the patronage (in both senses of the word) of the clients they are meant to serve. And, the defunding of apparently unsuccessful services or programs is easier because, in theory at least, recipients will simply stop selecting them.23 In customer- or recipient-directed systems:

If service quality declines or price increases, the recipient will look to other providers for a better deal. Like any other consumer, the voucher recipient will perform this monitoring function automatically. The competitive incentive for providers to discover low-cost ways of delivering a service is a form of market-induced monitoring. Each pro-

ducer, in seeking to increase profits, will act as a monitor of the other producers' performance. Producers that cannot or will not keep costs down and quality high will be driven from the market.24 All

things being equal,

the cumulative decisions of thousands, or millions, of recipients are simply more likely to be correct than are top-down decisions by a small number of government or foundation officials, and they will drive up the average quality of services provided. Although some recipients might select inferior services,25 on average they should do better than government because they are in a superior position to determine what they need, and because their individual decisions are less likely to be determined by such extraneous factors as political favoritism.

In theory, at least, market-oriented approaches should also raise the average quality of services-because they "require organizations to 'market' themselves directly to those who will be served rather than to those who ultimately provide the financial support (typically government). "26 Thus, for example, proponents of educational vouchers argue that they will encourage schools to develop innovative programs, recruit better teachers, or do both.27 Here is how John Chubb and Terry Moe, when they were of the Brookings Institution, described the market dynamic:

While schools controlled only by the market are free to organize any way they want, then, an environment of competition and choice gives them strong incentives to move

24. Hall & Eggers, supra note 17, at 4 n.14. 25. David Stoesz, Social Service Vouchers: Bringing Choice and Competition to

Social Services, PROGRESSIVE POL'v INST. POL'Y REP. 1, 7 n.16 (1992).

The federal student loan guarantee program, for example, had suffered from trade schools and other institutions that provide beneficiaries of loan guarantees with little or no real education, thus increasing the likelihood of eventual default. It should be noted, however, that poor quality services are a danger both in purely private transaction (e.g., car mechanics) and in publicly provided services (e.g., state motor vehicle departments). The relevant question is not whether some clients may receive inferior service; undoubtedly, some will. The question is, rather, whether there is reason to believe voucher clients will receive lower service quality than if they were given cash to purchase the services, on one hand, or provided the services by public employees (or contractors), on the other.

26. Butler, supra note 20, at 118. 27. The most prominent proponent of school vouchers is, of course, the

American Nobel Prize winning economist Milton Friedman. See generally MILTON FRIEDMAN, CAPITALISM AND FREEDOM ch. 6 (1962).

MARKETPLACE FOR SOCIAL VWELFARE SERVICES

toward the kinds of "effective-school" organizations that academics and reformers would like to impose on the public schools. Of course, not all schools in the market will respond equally well to these incentives. But those that falter will find it more difficult to attract support, and they will tend to be weeded out in favor of schools that are better organized. This process of natural selection, based on ease of entry and performance-based attrition, complements the incentives of the marketplace in propelling and supporting a population of autonomous, effectively organized schools.

Markets for social welfare services, like all other markets, do not require that all buyers be wise: A well-functioning market needs just enough savvy consumers competing to buy better (or to which suppliers cheaper) products to create a demand respond. This benefits everyone in the market, not just those who are initially the most astute (unless the suppliers can somehow segment the market).29 Consider how all American cars improved thanks to competition from abroad, even though only some consumers sought the "better" foreign cars.

Economists argue about how many savvy consumers are needed and how well-informed they need to be. The consensus answer is: Not many. In fact, some economists argue that it is only the marginal consumer who matters.

Generally, in a well-functioning market in which price is determined by the intersection of demand and supply, providers are forced to sell their product at the prevailing market price in order to sell all that they have available. This is called a "pricetakers' market"-providers have to produce according to the "market-revealed value of goods."3 Especially for highly diversified goods, where consumers have a variety of options that meet their needs, markets are self-regulating-the individual provider will gain no benefit from straying from the market standard, because there are so many other providers consumers may turn to if prices rise or quality deteriorates. In the words of Nobel Prize winning economist George Stigler:

28. JOHN E. CHUBB & TERRY M. MOE, POLITICS, MARKETS, AND AMERICA'S

SCHOOLS 190 (1990).

29. The best example is the way U.S. airlines can set so many different prices for the same seats on the plane-based on when the ticket is purchased, etc., which in effect segments the markets between business and vacation travelers.

30. ARMEN A. ALCHIAN & WILLIAM R. ALLEN, UNDrERsrrY ECONOMIcs 287

(2d ed. 1964).

[F]rom the manufacturer's viewpoint, uncertainty concerning his price is clearly disadvantageous. The cost of search is a cost of purchase, and consumption will therefore be smaller, the greater the dispersion of prices and the greater the optimum amount of search. This is presumably one reason. .. why uniform prices are set by sellers of nationally advertised brands: if they have eliminated price variation, they have reduced the cost of the commodity (including search) to the buyer, even if the dealers' they otherwise margins average somewhat more than would.1 Note that this analysis does not require consumers to be especially wise, or even literate. Rather, they need to have clear tastes or preferences, and the ability to exercise them.

Recipient decision making can also lower program costs if recipients are given sufficient reason to be cost conscious.32 As described below, the most direct way to do so is through cash assistance, but cost consciousness also can be induced by requiring a meaningful copayment in either a reimbursed fee-for-service or voucher system (or by refunding the unused portion of the voucher). The point is: Recipients must directly benefit from obtaining a better price for a good or service (or suffer for not doing so). In other words, cash payments let them choose a less expensive service provider (or not to consume as much of the good or service)-and use the difference (the amount saved) for other purposes. This creates downward pressure on their individual expenditures, on total expenditures, and also on the price of goods and services purchased.

31. George Stigler, The Economics of Information, 69 J. POL. ECON. 223

(1961).

32. According to one study: Competitive markets force producers to seek low-cost solutions. Producers in a competitive market cannot increase prices much above a competitor's price. The only way to increase profits is to reduce costs. With lower costs, a producer either has a higher rate of profit or reduces price to capture market share. In either case, other producers are alerted that a better way has been found to deliver a product or service, spurring them to seek new, cheaper ways of delivering the service. As long as the recipient has a choice, these cost reductions shouldn't be achieved at the cost of reduced quality. Why? Because if service quality falls, recipients will change providers. The competitive market lets producers and consumers engage in a dialogue about cost and quality issues. The simple act of exchange allows both sides to exchange information and discover what the best price/quality combination will be.

Hall & Eggers, supra note 17, at 4 n.13.

Bottom-up systems also produce more options or flexibility in the services available to recipients. It is easier for individual agencies to modify their programs to meet the needs of particular clients or groups of clients than it is for a central, top-down authority to anticipate all the possibilities and provide for them in the grant or contract.33 Writing for the Reason Foundation, John Hall, at the time at George Mason University, and William D. Eggers, at the time at the Reason Public Policy Institute, explain:

By putting the choice of providers in the hands of consumers, vouchers also have the potential to break up producers' monopoly on the supply of social services. The result should be an overall increase in the supply, quality, and diversity of providers. Vouchers should prompt suppliers to tailor their programs and services to better meet the particular needs and circumstances of voucher holders. Providers will begin to specialize in meeting different market niches. Child care is a good example: Parents know best whether their children need small, quiet settings where providers can devote time to interacting with the children, or settings with larger groups of peers to provide social stimulation. Parents can also balance quality, location, and cost in a way that is best for their children and themselves. Current policies generally support parental choice in child care because it is believed to yield the best arrangement for both the children and the families. Over time, these choices should encourage providers to offer the types and qualities of child care that parents prefer.35

Some parents want, or need, only half-day care; some need evening or after-hours care; and others need full-day care, perhaps with extended hours. Some parents want their children

cared for by other family members; some want to use neighbors; others want a nursery school; and still others a care center, perhaps in a church. Some parents may want all their children of different ages in one place; others may not care. Some parents will want their children close to home; others will want them close to work. The variations are almost infinite. Accommodating such variation is all but impossible in a top-down regime. Only bottom-up, market-driven systems give low-income recipients a range of choices remotely comparable to those available to middle- and upper-income families. To explain further:

Using vouchers, parents can shape the child care market to provide more of the types and features of child care that they want. Vouchers expand parents' choice of providers to include relatives and informal providers-persons who are generally not included in contract systems.36

Also, because the distribution of service providers is not dependent on centralized contracting, market-driven systems tend to reduce the income segregation of clients inherent in programs based on top-down grants and contracts (unless the number of providers is restricted by administrative or statutory rule). It is impossible for public authorities to enter into contracts or make grants to all potential service providers, so there is a tendency to fund those providers who specialize in serving the economically disadvantaged or the neighborhoods where they predominate. On the other hand, low-income children using child care vouchers are more likely to be integrated with children from other socio-economic backgrounds." (Of course, the agencies or firms serving the middle class must be willing to accept the voucher.)

These are all theoretical arguments, of course, so it is comforting to see that theory translates into practice, at least sometimes. In the United States in the early 1980s, interest grew in using certificates or vouchers to provide government-funded child care. The results of various experiments and pilot

projects38 (see Box 2) led policy makers to conclude that bottomup funding was indeed feasible, and perhaps preferable in many circumstances. As a result, U.S. federal law was changed to require states to offer parents "a certificate, check, or other disbursement that is issued ... directly to a parent who may use it only to pay for child care services from a variety of providers (including center-based, group home, family, and in-home child care), or, if required, as a deposit for services."9

There is another, almost unexpected benefit of bottom-up systems. Osborne and Gaebler explain that separate institutions for the poor tend to become stigmatized, but "if we do not limit the program or institution to the poor, we have no way of ensuring equity.. .because the affluent become the most intense users of the service." In the case of public universities, tax dollars intended to make higher education broadly affordable are being used "to subsidize higher education primarily for the middle and upper middle class." Osborne and Gaebler continue, "When governments fund individuals rather than institutions, it is much easier to promote equity .... This also removes the stigma of subsidies for the poor by allowing them to participate in the mainstream-to attend any school, live in any apartment building within their means, use any health clinic."4°

The use of housing vouchers and certificates demonstrate this effect. Public housing projects, especially in the United States, are notorious for creating ghettos of immobilized (trapped?) and marginalized low-income families.

In the United States in 1996, the last year for which data were found, of $21.7 billion in total federal housing expenditures, $8.1 billion were in the form of vouchers or certificates.41 Because the housing subsidy is provided in the form of a voucher, households can choose where to live. The subsidy is tied to the household, not to the housing unit. In some programs, there used to be an incentive for families to be cost-conscious: If they find a unit that rented for less than the amount of the voucher, they got to keep the difference.

The one major evaluation of this program concluded that vouchers and certificates successfully implemented recipient choice with few negative side effects.42 Most importantly, 46% of new or recently moved voucher recipients paid 30% or less of their income for rent.43 Recipient tastes were important. Given a voucher, some households moved to better but more expensive housing; some moved to better but less expensive housing; and some moved to worse but less expensive housing. (See Box 3)

All these benefits of bottom-up approaches, of course, require a functioning, decentralized market. When one already exists, the argument for bottom-up approaches is easier to make.

(The preexisting private market for health care is what made reimbursed fee-for-service systems so easy to establish.) When a market does not already exist, however, the proponents of bottom-up approaches need to show that one will develop. (Fears that such a market will not develop for elementary and secondary education has been a major obstacle to generating more support for school voucher programs.)

Of course, one of the arguments for vouchers and other market-driven approaches is that they tend to encourage the growth of such markets, which benefit all recipients. But the bar-

riers to market entry-perhaps caused by high start-up costs, stringent licensing or quality control requirements, or a low profit potential-may simply be too great in particular program areas. There may also be differences based on population concentrations, with, for example, more potential providers in urban and suburban communities than in rural ones.44

Moreover, there will always be some recipients who are not good consumers. In particular program areas, there will be many who lack the requisite information, judgment, or interest to make wise decisions. And, in some program areas, there will be agencies that try to "cream" clients, that is, that try to attract the easiest, and least costly, to serve. The result of both factors could be educational ghettos of poor students, social service agencies serving the hardest to reach, and so forth. Thus, despite a likely overall improvement of services, the possibility that some (or many) recipients might be left behind has led many experts to reject market-oriented approaches, and is certainly reason for careful consideration before adopting them in particular instances.

Clearly, recipient decision making is not appropriate for all areas of social welfare. Recipients must have a sufficient personal interest in the effectiveness of the particular service-and function at an adequate level to make informed and responsible choices. For example, it is easier to trust, or at least accept, recipient choice for purchases of food and housing, where the recipient's own interests are clearly at stake, than in elementary and secondary education, where other considerations may predominate. Likewise, it may be easier to provide recipient choice in markets that already have a large number of private suppliers, than in others, like education, which are dominated by

government agencies or where the barriers to entry may be substantial.45

Furthermore, recipient decision making may not be appropriate for (1) involuntary interventions or authoritative services, such as those for child abusers, and (2) incompetent or dysfunctional recipients. As Stuart Butler asks: "Should drug addicts undergoing treatment have complete freedom to choose which organizations shall serve them? That may be better than the alternatives in many instances, but there would be plenty of opportunities for exploitation."46 Thus, warn Hall and Eggers: "For some services, such as court-ordered drug and alcohol or mental health treatment, the state may find it counterproductive to allow for choice because recipients may opt for the least rigorous program, thus making their choice at odds with the outcome most socially desirable."47

Top-down approaches have another advantage that bears note: They make rationing the good or service politically easier. (Some people, of course, would consider this a disadvantage.) If there are not funds to provide goods or services for all those formally eligible for them, a bottom-up regime requires explicit rationing. The central political authority, in a relatively visible manner, must decide, for example, who gets the voucher, what service is provided, and so forth. In top-down systems, the central authority's rationing is less visible-because it is more likely to appear to happen at the agency level.4" In addition, some argue that "by increasing the quality and supply of governmentfinanced services, vouchers will likely increase the demand for

the service, thereby making controlling overall program expenditures more difficult than previously."49

For all these reasons, the consideration of recipient-directed systems must be on a program-area-by-program-area basis, as described in the second part of this paper.

One final point: The foregoing argument in favor of recipient decision making is instrumental in nature; that is, it seeks to make social welfare systems more efficient. However, in many program areas, a broader, normative argument can be made: If the middle class can choose its own service providers, why not the poor? As David Stoesz, of the Virginia Commonwealth University School of Social Work, writes, "Vouchers afford social services clients the same consumer sovereignty enjoyed by those who purchase such services in the open market-the power to end services that are not useful, and the opportunity to shop for care that is more helpful."5°

Moreover, vouchers and other recipient-directed systems

can do the following:

[They] could eliminate much of the paternalism inherent in the public welfare quasi-monopoly. One of the most egregious assumptions of many direct service delivery programs is that social welfare clients are incapable of defining what is in their best interests. Instead of relying on the client's judgment, public welfare professionals have often usurped consumer authority in personal matters. While such paternalism is warranted for certain types of services, such as substance abuse treatment, or for particularly vulnerable clients, such as children in need of protection, it is patronizing to assume that an inability to choose wisely holds for all clients of all public social services.51

C. Government's 'Fatal Embrace"?

As mentioned above, contracting out for services is often called "privatization." But top-down funding also tends to increase government's influence over private agencies, including those operated by society's mediating institutions.52 For, when government chooses the service (or agency) for clients, it tends to prescribe, or at least highly regulate, the nature of the services provided by private agencies. Smith and Lipsky comment: "Instead of shrinking the role of government and making the

provision of public services subject to market discipline, contracting has actually diminished and constrained the community sector by government intervention in nonprofit organizations."53 Government, of course, has broad power to regulate the

activities of mediating structures and other private agencieseven when it is not funding them. Whether through legislative fiat or licensing rules, government can regulate almost all aspects of private agency operations. (In all but a few states, for example, church-based day care is regulated even when no public funds are involved.)

Government is much less likely to regulate private agencies, however, if it is not actually arranging for the service. Passing legislation or adopting administrative regulations is often a long, drawn-out process requiring support at the highest levels of the government agency-whose visibility often arouses important political interests. On the other hand, attaching programmatic conditions to a contract or grant is a relatively invisible and lowlevel process. It also has the special justification that, since public money is involved, it should be spent in accordance with the "public trust."

This heightened level of government control is understandable-after all, the contract is meant to serve a public purpose, not to sustain a private enterprise-but the effect is the same: Distant, government decision makers dictate the shape of local services. "As 'agents of the state,"' writes Joe Dolan, now executive director of the Achelis and Bodman Foundations in New York, "non-profits become disconnected from local mediating institutions such as schools, churches, businesses and neighborhood, family or parent-oriented groups."54

This is not some abstract or hypothetical concern. In area after area, the conditions attached to government funding have altered the nature of private agency services and raised their costs. Even successful providers are often required to alter their programs if they are to receive government funding. As Smith and Lipsky explain, " [P] rivatization has led to government penetration of the private sector through the regulations, obligations and restrictions that accompany contracts. It has created rules and regulations for private agencies that otherwise would not be subject to government control."55

The programmatic requirements imposed by government on the services it funds or purchases tend to increase the cost of the agency's services, often without improving quality. Government often dictates costly "quality" standards, for example, with less concern about whether such standards actually improve the service than would individuals (who would have to pay for the "improved" service out of their own pockets). No better example of this two-sided dynamic exists than the quality/cost differentials between publicly and privately funded child care.

Although estimates vary, and undoubtedly depend on the types of care provided, regulated child care costs from 10 to 30% more than unregulated care.56 Gormley, in his study of family day care home regulation in Milwaukee County, Wisconsin, for example, "found cost increases directly attributable to regulation. The cost of home improvements alone was $936 per provider."57 Gormley also conducted a survey of local regulation in cities with populations greater than 50,000, and "found extensive local regulation of relatively small family day care homes. For example, family day care providers who care for six children are required to have a business license in 39 percent of these cities; an occupancy permit in 43 percent; and a zoning permit in 28 percent.'58

Beside raising costs for parents, these regulatory burdens reduce the supply of providers, especially family day care homes. Gormley found that regulation was responsible both for declines in the growth of new licensed providers and declines in the actual numbers of licensed providers of family day care.59 (As a result, many providers operate as unlicensed family day care homes, further complicating government's ability to maintain some standard of care.)

But is regulated child care better? Ron Haskins, then on the professional staff of the U.S. House Ways and Means Committee, wrote that one "problem with studies of child care quality is that the observed correlations are modest and nearly all the studies

are short-term."6° Moreover, sometimes the regulation seems irrelevant to research findings about "quality." Staff/child ratios are among the most common subjects of regulation, and as Gormley notes, an "Abt Associates study of group day care centers found, not surprisingly, that lower child-staff ratios result in higher costs."61 However, "the Abt Associates study found that low child-staff ratios, though beneficial, made less of a difference than, for instance, group size."62

Often, besides raising costs, the imposition of such "quality" or "professional" standards also acts to exclude community-based or informal service providers by raising to prohibitive levels the costs for them to enter the market. On the other hand, it is not always easier to rely on unregulated or community-based programs. Stuart Butler describes the predicament in which public (and private) funders find themselves when they wish to support community-based less professionally credentialed) agencies:

(and therefore

How do funding officials discharge their fiduciary responsibility, ensuring that money goes to qualified, bona fide organizations and yet also direct funds to the most deserving new institutions? In the real world, even an insightful funding officer has tojustify his or her decisions to a supervisor, auditor, or oversight committee chairman who will see only the paperwork associated with a funding decision. The pressure on that officer is to show traditional credentials and long experience in a group he has funded. The fiduciary responsibility thus invariably runs counter to the urge to assist the more creative but least "professional" mediating structures. And this dilemma is made more difficult by the tactics of the professional organizations who see "amateur" groups as competitors for funding. Not surprisingly, these professional groups lead the demand for credential requirements to "protect the public" from "unqualified" organizations or individuals.63

Finally, there is a political dimension to large-scale, topdown funding: Private agencies tend to become dependent on government-not just on the government's money but also its good will. Because government officials are awarding the con-

60. Ron Haskins, Reform of Federal Child Care Programs, Address at the Day Care and Early Childhood Programs Under Welfare Reform Conference at the American Enterprise Institute (Mar. 24-25, 1997).

61. Gormley, Mister Rogers' Neighborhood, supra note 57, at 24. 62. 63. Butler, supra note 20, at 120.

Id. at 23.

tracts or grants, criticizing government policies becomes even more difficult.

in

Mutual dependence blurs the lines between public and private. The agencies now often depend on government financing. A 1988 survey of the Child Welfare League of America concluded that government support accounted for, on average, 59 percent of its member agency revenues. A 1991 survey of over 350 nonprofit social service agencies in Massachusetts found that 52 percent of agency income was from state contracts, and an additional 17 percent from nonstate-contract government funding such as Medicaid. A 1991 survey of 276 nonprofit social service, educational, and cultural agencies the Raleigh-DurhamChapel Hill, North Carolina, region found that 45 percent of total funding was from government, if hospitals and universities were excluded. Fully one-quarter of all nonprofits in the region received over one-half of their revenue from government. A 1989 survey of nonprofits in the New York metropolitan region found that government funding contributed almost half of total revenue.64 "Allegiance is to government-government periodicals, government conferences, government policy, more government funding, and officials who favor the expansion of government programs," decries Dolan.65 Although bottom-up systems also can create dependency on government funds, there is a big difference: the financial dependency of private agencies or groups is not nearly as subject to the government's good will, since recipients-not government officials-decide whether they get funded.

Mediating institutions are at special risk to all these forces because top-down government funding often requires them to abandon the very features, like religious activities, that make them effective.66 These entangling strings can have "a profound and damaging impact on mediating structures," according to Stuart Butler:

Invariably, it encourages, or even requires, organizations to surrender their special character and position in society as they seek to comply with government requirements. By making these organizations less concerned with satisfying

64. SMITH & LIPSKY, supra note 6, at 4. 65. Dolan, supra note 54, at 44. 66. For a dramatic example of how government aid can change the whole

character of a private agency, see Marvin Olasky, The Corruption of Religious Charities, in BERGER & NEUHAUS, supra note 2, at 94.

the local community, and more sensitive to the demands of a distant official, it weakens their relationship with that community.

That is why government aid has been called a "fatal

embrace."68

Indeed

"Unless that problem is solved," warn Peter L. Berger of Boston University and Richard John Neuhaus of the Institute on Religion and Public Life, "when such institutions are first 'discovered' and then funded by government, the very vitality that originally distinguished the institutions from government agencies is destroyed. they become government agencies under another name."69 As Berger and Neuhaus elaborate, "In the 1970s, we underestimated the degree of corruption that comes with government funding-not, of course, corruption in the sense of criminal misuse of funds (that is a relatively manageable matter), but the much more insidious corruption in which these institutions reshape themselves to continue as beneficiaries of government largesse. "70

The problem is so great that some thoughtful observers denounce all government assistance to mediating structures. However, many elements of the modern social welfare state are simply too expensive to be funded by the private sector alone. (Housing for the poor, residential care for the elderly, health care services generally, and foster care for abused and neglected children, come immediately to mind.) According to John DiIulio, writing when he was at Princeton University, "[E]ven if every charitable organization in America gave their entire endowment to social welfare causes, that would cover only one-seventh of what the public sector now spends on social welfare."7'

If mediating structures are to be a major part of modern social welfare states without losing their essential character, we

67. Butler, supra note 20, at 22. 68. Of course, the government can regulate the activities of mediating structures even when not giving them money. But the temptation to attach requirements to the behavior of mediating structures-and the political support for doing so-is greatly increased when there is direct funding, as described above. (There is also a tendency to regulate the content of services in bottom-up systems, but it is not nearly as great as in top-down regimes because the government has less palpable responsibility for the nature and quality of the service.)

69. BERGER & NEUHIAUS, supra note 2, at 150. 70. 71. John Dilulio, Am. ENTER. INST., Annual Policy Conference (Dec. 2,

Id. at 151.

1997).

must, as Berger and Neuhaus urge, protect them from the "fatal embrace of government regulation."72

Hence, the significance of bottom-up approaches: Funding based on cumulative recipient demand is less threatening to the independence of private agencies because, when individuals rather than government bureaucrats are the decision makers, institution-distorting strings are less likely to be attached.73

The operative words, however, are "less likely." With the exception of cash assistance, even bottom-up approaches increase the risk of greater government influence over mediating structures. Stuart Butler makes the point that "accepting a voucher may turn out to be a poisoned chalice for an organization, encumbering it with heavy regulation. This is of particular concern to private schools in the debate over education choice. Nonetheless, vouchers appear to carry much less of this danger than direct forms of government support.'74 Even when the bottom-up aid comes with no strings, some thoughtful observers fear that the mere acceptance of public aid in a heretofore area of private activity builds additional and deleterious dependence on the government.

There is no one answer to whether top-down or bottom-up approaches are better. The answer depends on the specific situation, and what values one holds to be most important. Box 4 sets forth the general factors to be considered in deciding between the two systems.

Bottom-up, recipient-directed systems have been used for medical services, housing, transportation, food, child care, education, job training, and certain counseling and rehabilitation services, such as those for the mentally ill or developmentally disabled. Vouchers are the most familiar market mechanism, but lower taxes, cash assistance, and fee-for-service systems can also be used to implement a bottom-up social welfare regime. Choosing the most appropriate market mechanism for a specific program area requires considering the needs and capacities of recipients, as well as local service structures or markets (or potential markets). It also requires answering two linked questions: Should recipients be encouraged to consume the particular good to be cost or service? And, should conscious?

they be encouraged

This section examines the considerations underlying each of

the major market mechanisms:

" lower taxes (including deductions and credits), " cash assistance (including refundable tax credits), " reimbursed fee-for-services, and " vouchers.

A. Lower taxes

Government aid is unnecessary, of course, if people can afford to pay for the goods or services themselves. Although this is basically a question of their having sufficient income or savings, lower taxes could free a significant number of individuals and families from the need for some types of government assistance. Consider the following: Between 1960 and 1998, the tax burden in OECD countries rose from 27.7% to 37.2% of GDP.75 In European OECD countries, the tax increases were even greater, rising from 30.9% to 39.8% of GDP.76

In the United States, for example, over the past fifty years, a greater portion of the federal tax burden has been shifted from taxpayers without dependent children to lower- and middleincome families with dependent children.77 One of the main

reasons for this shift has been the decline in the relative value of the personal exemption.78

From 1948 to 2000, the personal exemption was increased from $600 to $2,800.79 Had the exemption been adjusted to keep pace with inflation, its value would have been $4,287. For lower-income taxpayers (in the 15% marginal tax bracket), this adjustment would have reduced a family's tax liability by about $225 per child. For higher-income taxpayers (in the 28% marginal tax bracket), the reduction would have been about $415 per child. (The Earned Income Tax Credit, as described below, offsets some of this increase-but only for the lower-income taxpayers, generally with earnings below $28,000 a year.)

But this is merely an adjustment for inflation. Had the exemption been adjusted for the growth in per capita income, to negate the effect of "bracket creep" and other factors in the American tax regime, the exemption would have been a whopping $10,689. For lower-income taxpayers, the result would have been a reduction in the family's tax liability of about $1,185 per child; higher-income taxpayers would have seen a reduction of about $2,200 per child.

How to put this in context? The size of this increase in tax burdens varies by family income and size, but here is one example: According to a 1998 U.S. Department of Treasury report, from 1955 to 1996, the income of a four-person American family to at one-half the median nearly doubled, from $15,804 $28,271.80 During that same period, the tax burden on such a family rose nearly ten-fold, from $316 (2000 dollars) to $2,989. (As a percent of income, it increased from 2% to 10.57%.) Most of this growth had occurred before 1981, and, indeed, the tax burden peaked in 1986 at 13.79% of income.

C. Eugene Steuerle, "This decline in the dependency exemption, along with increases in Social Security taxes, has increased the tax burden of families with children relative to almost all other taxpayers.").

78.

See C. Eugene Steurle & Jason Juffras, URa. INST., A $1,000 TAx CREDIT FOR EVERY CHILD: A BASE OF REFORM FOR THE NATION'S TAX, WELFARE, AND HEALTH SYSTEMS (1991); C. Eugene Steuerle, The Tax Treatment of Households of Different Size, in TAxING THE FAMILY 73-97 (Rudolph Penner ed., 1983). 79. Beginning in 1997, the personal exemption is phased out for high-

income taxpayers.

80. See ALLEN H. LERMAN, DEP'T OF THE TREASURY, AVERAGE AND MARGINAL FEDERAL INCOME, SOCIAL SECURITY AND MEDICARE, AND COMBINED TAx RATES FOR FOUR-PERSON FAMILIES AT THE SAME RELATIVE POSITIONS IN THE INCOME DisTRIBUTION, 1995-1999 tbl. 3 (1998). The tax burden calculations are based on the assumption that all income is earned by one spouse and that the family uses the "married filingjointly" filing status. The effects of the Earned Income Tax Credit (EITC) are also included, assuming two eligible dependents.

This increased tax burden places added financial stress on lower- and middle-income families and creates more pressure in two-parent households for both parents to work."1 (The $500 per child tax credit, created by the Taxpayer Relief Act of 1997, the personal helps offset 82 exemption.)

the value of

the decline

in

Some experts justify this higher tax burden on the basis that low- and middle-income families now receive additional benefits from the federal government. But why take money from families in order to give it back to them? Taking money from families (or all taxpayers, for that matter) and giving it back to them in the form of categorical assistance is a way of controlling their spending decisions. As Stuart Butler explains, tax relief could:

leave more money in the pockets of ordinary Americans, letting them allocate that money to the organizations of their choice rather than taking that money and then forcto use organizations of the government's ing people choice. This philosophy is at the heart of proposals to reduce taxes on families in tandem with reductions in government funding for social services. The central issue in this approach is not about whether it is a good idea to spend money on school lunches or other services or whether nonprofit organizations should be funded. The central issue is who should control that spending and decide which institutions should receive support.83 For example, tax funds go to tax-paying families to help pay for college and other post-secondary education. In 2000, Pell Grants averaged about $2,000 and subsidized Stafford student loans averaged about $4,300.84 This is like taking money from one of the taxpayer's pockets and putting it in another-because we think that parents cannot (or will not) save the money themselves.

Forced saving, or intertemporal redistribution of wealth, sometimes makes good policy sense-particularly when the cost for the good or service is very high, cannot be adequately insured against, or involves socially useful behaviors subject to undercon-

sumption. It is, for example, difficult to envision industrial and post-industrial societies without some form of government-sponsored (and mandated) old-age pension system.

There are practical and political limits to using generalized tax cuts to fund social welfare services, however. First, the costs of some goods and services may be beyond the tax liability of all but the most affluent. (Refundable tax credits are a form of cash assistance, and are described in the next section.) The cost of education is a prime example, but even child care and medical insurance are unaffordable to many tax-payers-even with a full refund on taxes paid.

Second, a generalized tax cut (or cash assistance, as we will see) may not provide a sufficient incentive to consume the particular good or service involved. Although taxpayers may be able to afford the good or service, they may lack sufficient interest to purchase it. (Conversely, it may be difficult to target the tax relief on to purchase the good or service.) There are many examples of socially harmful underconsumption-for example, the failure to obtain immunizations.85

those who need financial assistance in order

Finally, generalized tax cuts rarely satisfy the political pressure to have government address a particular social need. Voters tend to forget that their taxes were lowered so that they could afford a particular good or service. They still want targeted aid (whether in the form of a specific tax deduction or credit or a grant program) to pay for the service. So, of course, do the advocates, because they want to encourage consumption of the good or service.

The 1990 expansion of the American Earned Income Tax Credit (EITC) aptly demonstrates the phenomena. Conceived, at least partly, as a response to calls for increased child care funding, the EITC was expanded to provide over $24 billion over the in additional aid to the working poor-so succeeding five years" that they could pay for child care or so that the mother could stay home. (In 1988, the average EITC credit per family was $770; in 1991, $1,027; in 1993, $1,225; in 1996, $1,625; in 1999, it was an estimated $1,593; and, in 2000, and estimated $1,625.)87 But, because the EITC expansion was meant to support both child care and stay-at-home mothers, it was not called a "child

care" program. Hence, at this writing, a scant twelve years later, political pressure for a "child care" bill has made the passage of another multi-billion dollar bill extremely likely. This explains the political attraction of specialized tax deductions and credits, such as the Child and Dependent Care Credit, which reduces a family's tax liability in proportion to specified amounts of spending on child care.

There is, however, substantial question about the actual impact of such deductions and credits. Unless the financial gain is very great, they end up changing behavior very little, and are more accurately considered a cash subsidy to a group of taxpayers who happen to behave in a certain way (or have certain expenditures). Thus, most economists saw the recently enacted tax deduction for college or other postsecondary education in the United States as a subsidy to the upper-middle class (whose children would likely go to college regardless of the deductibility of tuition).

In sum, it is not possible to grant sufficient tax relief to help the least affluent afford high-cost social welfare services (such as medical care) or to do more than marginally encourage the purchase of a specific good or service (such as higher education). With these two limits in mind, it is nevertheless important to appreciate that lower taxes could keep important elements of social welfare private.

B. Cash assistance

If individuals or families do not have the money for a particular good or service, why not simply give them the money to purchase it? Cash assistance maximizes consumer choice and requires no special regulation of providers. More importantly, it enables individuals and families to choose how to use the government aid to make themselves better off. Osborne and Gaebler write:

Poor women may need training, housing, child care, and Head Start, but our laws make it far easier for them to get food stamps and welfare .... Dislocated workers may want health insurance, but the programs we fund are more apt to pay for job-search clubs. Systems that put resources in customers' hands allow them to buy what they want, rather than what the legislature or city council thinks they need."8 But this, as we will see in a moment, is also the biggest draw-

back of cash assistance.

Besides maximizing recipient choice, cash assistance also tends to make recipients extremely cost conscious. This is important if one objective is to control the program costs. The best way to give recipients an incentive to shop on the basis of cost as well as quality is to let them benefit from obtaining a better price for a good or service. In other words, allow recipients to choose a less expensive service provider (or spend less than might have been anticipated on the service)-and let them use the difference in cost (the amount saved) for other purposes. Again, this tends to lower individual expenditures, total expenditures, and also the price of goods and services purchased.

The case for non-cash aid-to encourage consumption or assure "quality"-is much weaker when there are wildly different prices for essentially the same good or service, and individual recipients are much better positioned to obtain the lower-priced one than is the government. This is apparently the basis for the decision of a number of American states to allow parents to "cash out" their child care assistance, as described below.

The most familiar form of cash assistance is aid to the poor. In the United States, for example, as described above, it takes two major forms: in 1999, amounting to a projected amount of almost $31 billion and averaging about $1,600 per household,89 and (2) "welfare" grants, in the same year amounting to about $11.7 billion and averaging about $4,428 a year per family.9°

tax credit (EITC),

(1) a refundable

However, both universal old-age and disability payments are also forms of cash assistance. After all, their purpose could also be accomplished through in-kind services provided by public and private agencies. The United States makes a sharp distinction between cash assistance for the poor and for the elderly and disabled: It gives the poor a higher proportion of their aid in the form of vouchers (food stamps) and in-kind services (housing) than it does to the elderly and disabled. Thus, while in 1999 the average cash grant to a family on welfare with no income was $4,428 a year,9 the average Social Security payment for a retired worker and aged spouse was $16,906 a year.92

First, giving large amounts of aid to the poor in the form of cash can discourage work and saving-as well as being politically unacceptable. Providing a cash grant equal to the costs of food, housing and medical care, not to mention education, would add about $8,500 to each family's welfare payment.93 And its total grant (almost $13,000) would be higher than the wages of over 20% of all workers with earnings and over 8% of full-time fullyear workers with earnings.94

Even when cash assistance is phased out based on income, the resultant marginal tax rates create a palpable disincentive to work. Using non-cash these problems because it reduces the apparent size of the grant. Also, because non-cash aid can only be used for certain, delimited purposes, recipients still have an incentive to work (although it is reduced) in order to obtain money with which to purchase other things."5 In addition, the nature of the in-kind aid, itself, may be

forms of assistance mitigates

its use, and therefore sufficiently stigmatizing (Recently, some economists have estimated that dependency. cash aid to the elderly discourages savings,96 but, so far, their work has had little impact on policy.)

to discourage

The second reason for not giving more cash assistance to the poor (or non-poor for that matter) is that they are not trusted to spend the cash aid "wisely," that is, on the things government wants them to purchase. This concern is not just about recipients choosing gin (or a stereo) over food, for example; it is also about their choosing clothes over food, or education over health care, or transportation over housing, and so forth-and vice versa.

Thus, for example, experts in the nutrition field (and voters) tend to support using food stamps (which are vouchers) instead of cash because, even though food purchases can be made in the open market with minimal supervision of the consumer, the object of policy is to encourage food consumption. Other programs seek to encourage the "consumption" of health care or housing. The well-known American economist Lester Thurow refers to this as government supplantation of "consumer sovereignty," although one of the "mildest and least coercive" forms:

Governments have a whole range of public policies that can be used to supplement or supplant consumer sovereignty, but one of the mildest and least coercive of these is the public provision of goods and services in kind. Such in-kind aid can be used to influence individuals to make

those decisions that society thinks they would be making if they were "competent."98 Sometimes, the purpose of the assistance

is to force a change in the behavior of recipients. Then recipient choice is irrelevant, if not at counterpurpose.9' And, indeed, such arguments are made by some child care experts:

Offering parents their choice of any legal provider, however, may conflict with the goal of improving the quality of child care for low-income families. Specialists in child development define quality in terms of provider-child interactions, provider-child ratios, group sizes, and training. Parents, however, may value features other than those valued by specialists, such as the provider's location, and are often willing to make trade-offs among these features. Low-income parents also face budgetary constraints and other pressures that may make them less likely than relatively higher-income parents to emphasize "quality" in their search for child care. Even if quality is a priority, information on the quality of particular child care options is difficult to obtain, and parents may not have the time to research, compare, and finally select providers.100 This, then, is the second limitation to using cash assistance. While it might be possible to "trust" the poor to make decisions about how much to spend on their own food and housing, giving them the same power over health care and their children's education (by giving them cash equal to the full value of their medical insurance, about $4,000 per family, and of public education, as much as $9,000 per child) would create too large a danger of underconsumption, to put it tactfully. Many people would see this as a big problem because they want the poor to "benefit" from the program as intended by the government. Others might

98. Lester C. Thurow, Cash Versus In-Kind Transfers?, 64 AM. ECON. REV.

190, 193 (1974).

99. Here is an example given by Thurow: If public child care is designed to allow each mother an equal opportunity to go to work (regardless of her husband's income), then vouchered private child care centers may be the appropriate answer. If child care is also desired as a method of altering values and characteristics (enrichment is our current euphemism), then public in-kind provision of child care is appropriate. Since society wants to change the values that the mothers would inculcate if they were at home, it obviously cannot allow the mothers to pick the private child care centers that the public funds are to be used to support. In either case, however, restricted transfers dominate cash.).

See id. at 195. 100.

See Ross & Kerachsky, supra note 35, at 40.

be more willing to see people spend less on health care, especially if the unused money was used for other productive purposes.

Actually, it is difficult to control the spending of recipients, even with vouchers or other in-kind programs. Unless the benefit exceeds what the recipient would have spent in the absence of aid, the benefit is simply a form of income assistance since the recipient is free to substitute the aid for prior expenditures. This means that in-kind benefits are more accurately viewed as increasing the marginal propensity to consume a good or service. (This is discussed in greater detail in the section on vouchers.)

In order to regulate recipient spending decisions, the American welfare system allows a hybrid between cash and vouchers: "grant diversion." Under this procedure, an individual's cash grant is paid directly to a service provider (such as a residential drug treatment program), an employer (to supplement earnings), or a landlord (when rent repeatedly goes unpaid). Since the diversion is for designated purposes, the payment, although formally denominated because the recipient must consume the particular good or service but gets to choose the provider. As efforts to reshape the behaviors of public aid recipients grow, greater use of this type of mechanism is probable.

in dollars, is more like a voucher-

C. Reimbursed fee-for-services

Most widely used in health care, reimbursed fee-for-service systems allow recipients to choose their providers (whose fee for the particular good or service is then paid by the government or other third party, either directly to the provider or indirectly by reimbursing systems encourage consumption of the particular good or service because the recipient usually does not pay for it, and cannot use the aid for another purpose.

recipient's expenditures).

Such

the

In the United States, the major reimbursed fee-for-service programs are Medicare and Medicaid. In 1998, Medicare provided $222 billion in benefits to 38.4 million individuals, with an annual per capita cost of $5,773. Medicaid expenditures were $187 billion in benefits to 40.6 million individuals, with an annual per capita cost of $4,609.1°1

An unappreciated benefit of fee-for-service systems is that they are almost pure, recipient-directed systems. They maximize

recipient choice, and minimize top-down, government control (within the context of directed consumption, of course).

This is changing, however. In the United States and many other OECD countries, the costs of fee-for-service health care programs are escalating rapidly. As a result, various top-down efforts are being made to control costs. In the United States, there is now massive price regulation through the "Diagnostic Related Groups" (DRG) system that pays providers a predetermined amount per procedure or medical condition. The DRG system encourages providers to cut costs, since they get to keep any "profit." In addition, almost all low-income families in the American Medicaid program are now required to be in health maintenance organizations, which are often subject to a cap or other control on total expenditures. Many national programs have adopted other ways to restrain costs, such as limiting the total income of physicians or the number of approved service providers (if the number of approved providers is small enough to restrict access).02

Taken together, these cost controls appear to have stemmed the inflation of health care costs. But many observers believe that they are only temporary fixes. For example, the incentives under the American DRG system are not as great as they could be because the price schedule reflects average costs, rather than the lowest costs. In any event, per recipient health care costs, and therefore total costs, continue to rise as patients receive more medical services and technological advances make health care more expensive.

Not only do these efforts to control costs tend to erode recipient choice, they also fail to use what might be the most effective constraint on costs: cost-conscious recipients/consumers. In almost all developed countries, the current approach is for government to pay the full amount of the recipient's medical costs. But this gives recipients no incentive to economize-by either pressing for lower costs or seeking fewer medical procedures.

If recipients were simply given cash, of course, they would become cost conscious because they would personally benefit from obtaining a better price for a good or service. But, as we have seen, there are many program areas where it may not be advisable to provide cash aid. The large amount of cash that

would have to be provided to cover the purchase of some goods or services might create an unwanted work disincentive or other behavioral effect, and recipients might not even spend the money as intended. Or, we may wish to encourage recipient consumption of the good or service beyond what recipients might choose on their own.

Nevertheless, it is possible to introduce cost consciousness into a fee-for-service system-through recipient copayments. Copayments can constrain costs or ration services by making individuals feel at least some of the financial costs of their deci(As we will see, copayments can also be an important sions. aspect of voucher systems.)

In 1974, the RAND Corporation began a long-term study of the impact of copayments, what the study called "cost-sharing," on the utilization of health services. In the RAND Health Insurance Experiment, individuals were randomly assigned to four experimental health insurance plans with varying levels of costsharing: 0% copayment (free care), 25% copayment, 50% copayment, and 95% copayment. (The latter three had a $1,000 a year cap on recipient copayments.) The RAND researchers concluded that "use of medical services responds unequivocally to changes in the amount paid out of pocket," with per capita health care expenditures on the free plan "45% higher than those on the plan with a 95% coinsurance rate." (Spending rates on the 25 and 50% coinsurance plans were also lower, but not as much.)'03

Thus, copayment strategies are most attractive when there is a need to encourage recipients to set priorities among what would otherwise be discretionary purchases. However, the unambiguous purpose of the copayment is to reduce consumption of the particular good or service. But what if the good or service is deemed important to the individual's or society's well-being? In the Health Insurance Experiment, for example, the researchers found that "the reduced service use under cost-sharing plans had little or no net adverse effect on health for the average person. Indeed, restricted activity days fell with more cost sharing."104 Some groups, though, mainly the chronically sick poor, did suffer. "The poor who began the Experiment with elevated blood pressure had their blood pressure lowered more on the free care plan than on the cost-sharing plans. The effect on predicted

mortality rates-a fall of about 10 percent-was substantial for this group."15

It is important to understand the nature of the choice reflected in this behavior of the sick poor. It is not that they chose to die, nor that they could not meet the co-payment without starving. No, they preferred to have the immediate cash as opposed to having a 10% lower risk of a future heart attack. One need not disapprove of the choice to appreciate society's stake in the outcome.

In any event, if copayments are sizable enough, they can make recipients price conscious. That is, recipients who are paying a meaningful amount toward something's cost will be more sensitive to its price-and more likely to bargain for a lower price, shop for a less expensive version, reduce consumption of it or forego it entirely. Many recipients, of course, cannot afford a large copayment. That is why copayments are often set on a sliding scale, and some very poor recipients may be entirely excused from making a copayment. But, as mentioned above, not all consumers need be savvy or cost conscious for market forces to work. Copayments are sometimes large enough to help pay for a service. Thus, in child care, sliding-scale payments often require parents to pay as much as half of the cost of their children's care, and some pay more (up to the full cost). In such cases, the copayment goes beyond being a device to sensitize recipients to costs and becomes an integral element of program funding.

As the foregoing suggests,

there are many similarities between reimbursed fee-for-service systems and vouchers. The major difference is that, because fee-for-service systems reimburse providers for the particular goods or services rendered, they are more appropriate in program areas in which the types and amounts of services (and hence, costs) vary over time or between accounting periods. When there is no such variation, then vouchers are preferable because they involve much less paperwork and hence, lower overhead expenses.

This explains the growing interest in vouchers for the American Medicaid and Medicare programs. As more recipients are

enrolled in health maintenance organizations (HMOs) that have a set fee for coverage, vouchers become a more appropriate market mechanism. 106

D. Vouchers

Vouchers are a form of scrip given to designated recipients, usually by government, that recipients use to pay providers for a specified good or service.107 Vouchers normally contain an express limitation on the costs or services that are reimbursable. (The scrip, which can be in the form of a coupon, certificate, a credit card-like device, or even a simple bookkeeping entry, is then submitted by the provider to the government or other funder for reimbursement.)

Vouchers are meant to provide assistance in obtaining a particular good or service: housing vouchers cannot be used to obtain clothing; educational vouchers cannot be used for food; and so on. In the United States, the largest program to use a voucher is the Food Stamp Program. In 2000, food stamps were provided to 17.2 million people, or one in sixteen Americans, at a cost of $19.1 billion.108 The average monthly value of food stamps per household was $158, with a maximum of about $426 per month for an average household of four.°9 Actually, food stamps have many attributes of a cash grant: Clients are able to use them to purchase any number of different food items from a large array of merchants, so that there is relatively wide consumer choice and corresponding competition among providers. As described above, the major difference between reimbursed fee-for-service systems and vouchers is that the former can vary in value depending on the services provided in the particular accounting periods while vouchers are usually for a specified

value or service. Thus, because vouchers involve less paperwork, they are a preferred market mechanism when the amount of the subsidy does not vary from accounting period to accounting period (either because the cost of the service, such as housing, is constant or because the benefit is capped, as in most educational vouchers). This very inflexibility, it should be noted, means that vouchers are an easy way to control or cap costs, by limiting their value or number. "On the other hand, if vouchers can be used as partial payment, or if voucher prices are set too high, they may exert an inflationary impact on service prices, particularly where there are constraints on supply, or where voucher holders will constitute a large proportion of the market.""°

As mentioned above, vouchers are being suggested as a means of slowing spiraling health care costs, for example. Much of government-provided health care is under reimbursed or third-party, fee-for-service regimes-whose costs, as we have seen, are difficult to control without appreciable copayments. The idea is to give a voucher of fixed value that would enable recipients to choose their own health maintenance organizations (HMO), which in turn would apply cost controls because their reimbursement would be capped.

The ability of vouchers to influence what recipients actually consume, however, is limited by the possibility of substitution. For vouchers (and fee-for-service systems, as well) free up other household resources that would have been spent on the goods and services provided by the voucher. As Rebecca Blank of the U.S. President's Council of Economic Advisers observes in relation to food stamps, "A family that replaces food dollars with Food Stamp coupons will have more dollars to spend elsewhere, which may improve their ability to find adequate housing or to maintain a car."''

This raises the question of whether a particular voucher encourages consumption of the particular good or service-or is just a form of income support. The American Food Stamp Program demonstrates the blurred lines between vouchers and cash support. According to most research studies, food stamps result in only modest increases in food consumption (about 30 cents on the dollar)1"2 and there is now a thriving black market for them. If recipients are willing to accept a discount off their face

value (typically from 20 to 50%), most are able to trade their stamps for cash, thereby escaping the need to spend the entire value of their coupons on food stuffs.

Thus, vouchers may not increase consumption if their value is less than what the recipient would otherwise spend on the same good or service (with a corresponding increase in income) or if the recipient can sell all or part of the voucher in a secondary or "black" market. Although efforts have been made to police the transfer or resale of vouchers, it is hard to prevent the creation of a black market-unless the nature of the voucher makes it difficult to sell. The best examples of vouchers that are difficult to transfer are probably vouchers for housing and for center-based child care, since they require the long-term identification of the recipient to providers who have little interest in cheating.

Deciding to adopt a voucher system for a particular program area requires a prediction of the degree to which the vouchers will be sold or traded followed by a judgment about whether the (Fee-for-service syspredicted level of diversion is acceptable. tems are somewhat less likely to encounter the same level of fraud because they tend to require greater proof of the identity of the recipient.)

The nature of the market for the particular good or service, once again, should have an impact on the decision. A number of American states allow parents to "cash out" their child care vouchers for the very reason that the cost of child care can vary widely depending on the type of care the parent selects-and on whether the parent chooses a friend or relative who will charge less than the "market" rate.

Vouchers can take many forms. The following programmatic options are discussed in terms of vouchers but much of the discussion is equally applicable to reimbursed fee-for-service systems. (See Box 5).

Form of scrip? Vouchers can take the form of coupons, certificates or even simple bookkeeping entries. Coupons are the most cash-like of vouchers. They are often used when there will be multiple purchases from multiple providers. The United States has adopted a credit card-like device, called "Electronic Benefit Transfer" (EBT), to disburse its food stamp benefits. Bookkeeping entries are possible when there are a small number of providers and they have an ongoing relationship with the funder.

Limitations on the goods or services obtainable? Most vouchers place some limitation on the particular goods or services that can be purchased with them. In some programs, the limitation is relatively minor, as in the American Food Stamp program, which places virtually no limits on the specific forms of foods that can be purchased. (Excluded, for example, are alcohol, tobacco, and hot foods intended for immediate consumption.) In other programs, the limits are more substantial. Thus, the American Special Supplemental Nutritional Program for Women, Infants and Children (WIC) provides vouchers that can be used only for specific food items, such as milk, cheese, infant formula, cereals, and fruit or vegetable juices. A day care voucher, for example, might encompass all types of day care to be provided for all parents needing the service or, at the opposite extreme, might be for a single type of day care to be made available to a specified low-income population.

Designated providers? Some vouchers may be used with any provider of the specified goods or services, but most are limited to providers that are licensed, registered, or meet some other criteria. In the United States, for example, food stamps can be used only at approved grocery stores; other vouchers, such as'those for housing or education, usually can be used only with designated providers. On this issue, child care vouchers have gone through something of a transformation, from being useable only with licensed providers to now being technically useable, in most American jurisdictions, with anyone the parent chooses."3 This ability to designate or limit service providers under a voucher system increases the tendency to impose government standards on the service or program. After all, with the power to approve service providers goes the increased responsibility to assure that the voucher is not misused.

Amount? A voucher can cover the full cost of the goods or services involved,"' a set percentage of the cost (or a specific dollar amount), or a sliding-scale copayment based on income or some other determinant of need. Besides rationing aid, a partial voucher can instill cost consciousness. A voucher that covers the full cost of the service, on the other hand, gives recipients no incentive to limit expenditures or shop for a lower price. To the extent that we desire them to do so in a particular program area, a copayment is appropriate (as discussed in the section on feefor-service programs). Such copayments usually vary by the recipient's income, and they can be large enough to help pay for the service (although a sliding scale can create a hefty marginal tax rate for low-wage workers).

Refundable? For a required copayment to encourage cost consciousness, it must be large enough to "bite." What about those recipients who cannot afford to make a copayment of any meaningful size? Much talked about, but apparently little used, would be a refundable voucher. Recipients who can pocket the difference between a lower-cost provider and the dollar value of the voucher have a strong incentive to be cost conscious. This is how the "Section 8" housing voucher worked in the United States until 1998, when the "shopping incentive" credit was eliminated in Title V of the FY 1999 HUD Appropriations Act."' Under that credit, "If a household can find an acceptable unit renting for less than [the amount of the voucher], it can keep the difference."116 In essence, refundability is a reality when there is (or

can be) a black market for the vouchers, such as with the American Food Stamp Program. And, it can also develop in those systems, such as child care and housing, in which the recipient and the provider can agree to what is essentially a kick back, legal or otherwise.

Actually, the idea of refundability is not that radical. For many years, the United States government has offered a refundable education voucher to service members and veterans."17 Authorized by what is commonly referred to as the GI Bill and administered by the Department of Veterans Affairs, the federal government currently provides $672 per month to service members and veterans pursuing education or vocational training.18 (Entitlement is based on months of service-to receive thirty-six months of benefits (four academic years), participants must complete three years of continuous active duty (or two years if they signed up for less than three years), or if discharged for the Convenience of the Government, they must have completed twenty months (if they signed up for less than three years) or thirty months (if they signed up for three years or more .

The voucher is for tuition and fees, and participants receive $672 per month regardless of whether they choose to attend a private or public college or university, or a vocational training program. The amount is adjusted depending on hours in school-full-time students receive $672, half-time students receive half that amount. If the student attends an institution for which tuition is less than $672 per month, students can keep the difference and attribute it to living expenses.

Voter resistance to refundable vouchers would, nevertheless, probably be substantial. In an innovative response, John Hood

of the Reason Public Policy Institute has proposed that recipients "have the option of depositing any part of the [medical] voucher not spent on medical insurance or care not only into medical savings accounts but also in educational savings accounts, from which they could make withdrawals for theirs or their children's education, or individual development accounts, from which they could make withdrawals for housing, transportation, or other approved expenditures to help get themselves off public assistance."119 The idea is particularly apt for medical vouchers, since they would be so large, and the possible price reductions through competition so great. But this concept of modified refundability could be applied to other areas as well, such as food stamps.

Number? Limited funding often requires a trade-off between universal but small vouchers and large but targeted (or rationed) vouchers. Thus, the other way that vouchers can be used to limit costs is by limiting the number of vouchers issued.

There are no simple prescriptions for deciding when topdown or bottom-up systems are better and, if a bottom-up system is selected, which market mechanism should be used. The answers will necessarily vary from program area to program area, based on existing service structures, potential markets, and policy preferences. [[National] ]Local attitudes are also important. What might be accepted in one [ [country] ] jurisdiction might be disdained in another. Hence, there is no single, correct answer as to whether there should be a market in social welfare services or how it should be structured. Rather, policy makers should appreciate the general advantages (but also disadvantages) of bottom-up, recipient-directed systems, and the need to make choices based on local conditions.

Nevertheless, it is possible to outline the general policy considerations involved. (See Box 6) Ordinarily, recipient-directed systems are preferable to top-down ones unless it appears that a true market will not develop, the recipients will not be wise consumers, or that the uncertainties of the market may compromise other important considerations in relation to the particular good or service.120 Tax relief or cash assistance is preferable to govern-

ment fee-for-service or voucher systems because cash maximizes individual choice about how to improve their standard of living. But cash may not be appropriate (1) when large amounts are involved (because of possible work disincentives and other behavioral effects) or (2) when added consumption of the particular good or service is desired and under-consumption is otherwise likely. Fee-for-service systems are preferable to vouchers when the amount of the subsidy is likely to vary from period to period. And vouchers may be inappropriate when there are widely different prices for essentially the same good or service, and individual recipients are much better positioned to obtain the lower-priced one than is the government. Finally, in both fee-for-service and voucher systems, copayment arrangements may be appropriate, to the extent that there is a concern about over-consumption.

One final point: There is no mystery about creating a system of bottom-up funding. In fact, there are examples in most developed countries throughout social welfare programming. In the developed world, the obstacles to developing recipient-directed systems are less conceptual than political-because they often involve the defunding of deeply entrenched interests. That, as much as developing a wise and workable system, is the challenge. Most middle-income countries do not yet have the extensive network of top-down programs of more affluent countries. So,

there is one final lesson from the experience of more developed countries: If a recipient-directed system of bottom-up social welfare funding is desired, establish it before service providers become programmatically and politically established within a tap-down regime.

Although market-based systems are most easily installed within pre-existing private service systems, middle-income countries should not reject them simply because there are few private providers of a particular service. Through sequential and orderly implementation, such providers will appear.

--- FOOTNOTES ---

DOUGLAS J. BESHAROV*

INTRODUCTION

* JosephJ. and VioletJacobs Scholar in Social Welfare Studies and Director of the Social and Individual Responsibility Project at the American Enterprise Institute.

1. J. CONG. ECON. COMM., 106TH CONG., REP. OF THE J CONG. ECON. COMM. OF THE UNITED STATES ON THE 1999 ECON. REP. OF THE PRESIDENT TOGETHER WITH MINORITY VIEWS 30, fig. 3.4 (Comm. Print 1999).

"Top-DowN" vs. "BoTroM-UP" DECISION MAKING

Box 1

SOCIAL WELFARE PROGRAMS

* Cash assistance (income support, unemployment insurance, disability

payments, and old-age pensions

* Medical services " Housing * Transportation " Food " Child care " Education (elementary, secondary, and post-secondary) * Job training * Social, counseling, and rehabilitation services

2.

"By using nonprofit agencies government policy makers hoped to administer their programs with more skill, creativity, flexibility," writes Leslie Lenkowsky, former head of the Hudson Institute in America. Leslie Lenkowsky, Philanthropy and the Welfare State, in PETER L. BERGER & RICHARD JOHN NEUHAUS, To EMPOWER PEOPLE: FROM STATE TO CIVIL SOCIETY 88 (2d ed. 1996). See generally DAVID OSBORNE & TED GAEBLER, REINVENTING GOVERNMENT: HOW THE ENTREPRENEURIAL SPIRIT is TRANSFORMING THE PUBLIC SECTOR (1992).

3. Mediating structures, as defined by Berger and Neuhaus, are "those institutions that stand between the private world of individuals and the large, impersonal structures of modern society. They 'mediat[e]' by constituting a vehicle by which personal beliefs and values [can] be transmitted into the mega-institutions." BERGER & NEUHAUS, supra note 2, at 148.

4. See id. However, this is an easily exaggerated argument, since many of

these agencies have little real connection to their host mediating institution.

5. For a general discussion of this element of the Great Society, see NEIL GILBERT, CAPITALISM AND THE WELFARE STATE: DILEMMAS OF SOCIAL BENE'VOLENCE 7 (1983); see generally Lenkowsky, supra note 2.

6. "For example, government purchase of service contracting with nonprofit service agencies in Massachusetts rose from $25 million in 1971 to $850 million in 1988." STEVEN RATHGEB SMITH & MICHAEL LIPSKY, NONPROFITS FOR HIRE: THE WELFARE STATE IN THE AGE OF CONTRACTING 6 (1993).

7.

See Lester M. Salamon, Rethinking Public Management: Third-Party Govern-

ment and the Changing Forms of Government Action, 29 PUB. POL'V 255-75 (1981). 8. Lester M. Salamon, AMERICA'S NONPROFIT SECTOR: A PRIMER 26 (1996).

(Steve Hanke ed., 1987); E.S. SAVAS, PRIVATIZATION: THE KEY TO BETTER GovERNMENT (1987); DAVID LINOWES, PRIVATIZATION: TowARD MORE EirEcrrvE GovERNMENT (1988); but see Paul Starr, The Limits of Privatization, in PROSPECTS FOR PRIVATIZATION (Steve Hanke ed., 1987).

12. For examples of government agencies bidding on contracts, see

OSBORNE & GAEBLER, supra note 2, at 88-89.

13.

See, e.g., Paul Sabatier, Top-Down and Bottom-Up Approach to Implementation Research: A Critical Analysis and Suggested Synthesis, 6 J. PUB. POL'Y 21-48 (1986); Dennis Palumbo, Symposium: Implementation, What We Have Learned and Still Need to Know, 7 POL'Y STUD. REV. ANN. 91-102 (1987).

14.

See generally GILBERT, supra note 5.

19.

See Letter from Edward Zigler, former head of the Head Start Program, to Douglas J. Besharov (Apr. 27, 1998) (on file with author). "30% of Head Start centers were of such poor quality that they had to be upgraded to be effective in improving the growth trajectory of those children involved." Id.

20. Stuart M. Butler, Practical Principles, in BERGER & NEUHAUS, supra note

3, at 118.

Id. at 181.

21. OSBORNE & GAEBLER, supra note 2, at 180-81. 22. 23. Even if government initially picked allowable providers, it could drop them based on the actual selections of clients: A rule could be established that a minimum number of clients must select that particular service provider over a designated period of time; otherwise, the provider is automatically dropped from the list.

33. Stoesz, supra note 25, at 5, citing Harry Hatry. Providers, given this incentive to be more responsive, may eventually develop proficiency in addressing specialized needs within their broader service area. Similarly, vouchers offer professionals dedicated to working with the poor the opportunity to develop innovative approaches to working with this population with potentially less bureaucratic interference from the public welfare apparatus. Innovative, indigenous programming can be supported by social service vouchers.

34. Hall & Eggers, supra note 17, at 4. 35. Christine Ross & Stuart Kerachsky, Strategies for Program Integration, in ENHANCING EARLY CHILDHOOD PROGRAMS: BURDENS AND OPPORTUNITIES 39-40 (Douglas J. Besharov ed., 1996).

[Vol. "16

Id. at 56.

36. 37. Recently, a controversy has erupted over the use of housing vouchers to achieve residential integration. The opponents of doing so point out that the cost of providing housing in more affluent neighborhoods raises program costs substantially, and argue that lower-income neighbors tend to reduce housing values and the neighborhood "quality." This debate too easily raises questions of racial discrimination beyond the scope of this paper. For present purposes, it is simply worth noting that recipient-directed systems can be a powerful tool for various purposes. See Howard Husock, Voucher Plan for Housing: A Trojan Horse, WALL ST. J., Dec. 21, 1994, at A14.

Box 2

CHILD CARE VOUCHERS IN HUDSON COUNTY, NEW JERSEY

In the early 1980s, the New Jersey Department of Health and Human Services determined that there was a shortage of child care "slots" available for its low-income clients at state-owned and state-subsidized centers in Hudson County. At the same time, private centers were reporting 17% vacancy rate. Thus, instead of building more state facilities or contracting for expanded state-subsidized centers, the department conducted a twoyear child care voucher demonstration in the county.

Previously, Hudson County has provided child care for low-income families through sixz state owned and operated centers serving 448 children and an additional 454 child care slots purchased through contracts with private centers. This system was no longer adequate for the growing need for child care in the community, was facing sharply escalating costs, and provided poor access to child care for parents in nine of Hudson County's twelve communities because of "historical funding patterns which tied day care subsidies to particular localities."

Parents were given vouchers equal to 75% of the State's prevailing contract rates for child care, thus, $37.50 per child per week for center-based care and $16.90 per child per week for family day care. Parents could use the vouchers for in-home and relative-provided care, or at any licenses family day care home or center.

The evaluators concluded that the project demonstrated the successful use of vouchers by low-income parents. The parents were able to use the vouchers in the broader market for child care services. Moreover, the vouchers provided a means of providing care for those children previously excluded from subsidized care because of their geographic location. And, since the vouchers were pegged at a lower price than State contract rates, costs were 25% lower.

Source: Barbara Catteral & Carol Williams, New Jersey Bureau of Research, Evaluation, and Quality Assurance, Division of Youth and Family Services, Voucher Subsidized Child Care: The Hudson County Project 2-8 (1985).

38.

See, e.g., BARBARA CATTERALL & CAROL WILLIAMS, NEW JERSEY BUREAU OF RESEARCH, EVALUATION, AND QUALITY ASSURANCE, DIVISION OF YOUTH AND FAMILY SERVICES, VOUCHER SUBSIDIZED CHILD CARE: THE HUDSON COUNTY PROJECT (1985).

39. OFF. OF INSPECTOR GEN., DEP'T OF HEALTH AND HUMAN SERVICES, STATES' CHILD CARE CERTIFICATE SYSTEMS: AN EARLY ASSESSMENT OF VULNERABILITIES AND BARRIERS 53 (1998).

40. OSBORNE & GAEBLER, supra note 2, at 180-86. 41. U.S. GEN. AccT. OFF., SECTION 8 TENANT-BASED HOUSING ASSISTANCE:

OPPORTUNITIES TO IMPROVE HUD's FINANCIAL MANAGEMENT 36 (1998).

42.

See STEPHEN D. KENNEDY & MIREILLE D. LEGER, FINAL COMPREHENSIVE

REPORT OF THE FREESTANDING HOUSING VOUCHER DEMONSTRATION 23 (1990).

43.

See Memorandum from Paul Leonard to Christopher D. Lord, Legis-

lative Dir., Senate Banking Comm. Staff (May 1, 1997) (on file with author).

MARKLFPI4CE FOR SOCIAL WELFARE SERVICES

Box 3

TENANT-BASED VS. PROJECT-BASED HOUSING ASSISTANCE

two tenant-based assistance programs-a

The U.S. Department of Housing and Urban Development (HUD) provides two types of housing assistance, tenant-based assistance and project-based assistance. Tenant-based assistance differs from projectbased assistance in that tenant-based assistance ties the subsidy directly to the family instead of to the housing unit. Thus, participants can choose where they want to live and can move while still maintaining their benefit. Until 1999, HUD operated certificate program and a voucher program. Private owners could agree to accept the certificate or voucher as a rental payment, and the rented units were required to meet health and safety standards set by the Department of Housing and Urban Development and the state or local Public Housing Authority (PHA). In the certificate program, households could choose where to live, but the rental amount had to be within the HUD-determined fair market rent for a given area and type of housing (set at the 40th percentile-the dollar amount below which 40% of standard quality housing rents in a given area). The certificate amount was set at the difference between the rent charged and 30% of the household's adjusted income, and the household was required to pay 30% of their adjusted income as rent.

In the voucher program, households received a voucher subsidy amount determined by the difference between a payment standard (generally set between 80-100% of the fair market rent) and 30% of their adjusted income. There was no limit on the rent the household selected, but if it found a unit that rented for less than its 30% contribution plus the subsidy, it could keep the difference. Households were required to pay at least 10% of their adjusted income or $25, whichever was greater, for rent.

A 1990 Abt Associates study found that most households participating in the certificate or voucher program were able to find and secure housing. Most importantly, 46% of new or recently moved voucher recipients paid 30% or less of their income for rent. Recipient tastes were important. Given a voucher, some households moved to better but more expensive housing; some moved to better but less expensive housing; and some moved to worse but less expensive housing. Source. Stephen D. Kennedy & Mireille D. Leger, Final Comprehensive Report of the Freestanding Housing Voucher Demonstration 23 (1990).

44. But even this generalization can prove incorrect. One scholar has

commented:

An oft-cited concern about vouchers is whether there would be enough providers to provide choice for consumers for certain low volume services or in rural areas. For example, in sparsely populated North Dakota, one might anticipate problems ensuring an adequate supply of services. In reality, however, North Dakota officials reported that the supply of services has increased as a result of the consumer demand generated from their reimbursement model of consumer purchasing. This has created opportunities for new providers.

Richard Dougherty with William D. Eggers, Delivering Better Services for the Mentally Ill and Developmentally Disabled: A Consumer Choice Model, 215 PRIVATIZATION CENTER POL'Y STUD. 10 (1996).

MARKETPLACE FOR SOCIAL WELFARE SERWCES

45. As noted above, concerns about the ability to generate a successful market in elementary and secondary education is one reason for the lack of stronger support for school vouchers, and for the popularity of "charter schools," that is, independent schools supported by public funds pursuant to a "charter," but operated by either a public or private group. Students apply to attend the school, and there is usually a rule that prevents the charter school from imposing particular selection criteria. The charters, typically issued by a state or local school board, allow various institutions-such as committees of parents or teachers, local clubs, or even profit-seeking firms-to set up and operate a school. Although the school is subject to the same student learning standards and requirements as schools under public jurisdiction, the managing institution is free to determine its own curriculum and teaching methods. See generally Bruno V. Manno, et al., How Charter Schools are Different: Lessons and Implications From a National Study, 79 PHI DELTA KAPPAN 488-98 (Mar. 1998).

46. Butler, supra note 20, at 120. 47. Hall & Eggers, supra note 17, at 4 n.12. 48. Limiting the value of a voucher can be another form of rationing, but since it is more often used to control costs that are deemed difficult to control, it is discussed below.

49. Hall & Eggers, supra note 17, at 13. 50. Stoesz, supra note 25, at 5. 51. 52. See note 3 for a definition of mediating institutions.

Id. at 6 (emphasis in original).

53. SMITH & LIPsKy, supra note 6, at 204. 54. Joseph S. Dolan, Letter to the Editor, CHRON. OF PHILANTHROPY, Aug.

10, 1993, at 44.

55. SMITH & LIPsKy, supra note 6, at 204.

56. For example, the Urban Institute's National Child Care Survey 1990, based on a nationally representative sample of families with children under age 13, found that the hourly fee for regulated family day care was $1.64 per hour, compared to $1.48 per hour for nonregulated family day care. See BARBARA WILLER ET AL., NAT'L. ASS'N FOR THE EDUC. OF YOUNG CHILDREN, THE DEMAND AND SUPPLY OF CHILD CARE IN 1990 30 (1991).

57. William T. Gormley, Jr., Regulating Mister Rogers' Neighborhood: The Dilemmas of Day Care Regulation, THE BROOKINGS REV., Fall 1990, at 24 [hereinafter Mister Rogers' Neighborhood].

58. 59.

Id. at 25. See WILLIAM T. GORMLEY, EvERYBoDY's CHILDREN: CHILD CARE AS A

PUBLIC PROBLEM 103-04 (1995).

"TOP-DOWN" VS. "BOTtOM-UP" FUNDING?

Box 4

"

Is there already a market for the goods or services involved? Or, can one be created? Conversely, are there barriers to entry for new providers that may preclude the development of true market competition?

" Will the recipients, as a group, be wise consumers of the program's

goods or services?

" Should recipients be allowed to choose service providers or is the

service involuntary or coercive in nature?

" Will the goods or service be rationed, or will they be provided to all

"

who meet specific eligibility criteria? Is close government regulation of the good or service needed? Or can providers be free to offer it as they see fit? Particularly, can mediating structures offer services under the terms of their private values or beliefs?

72. BERGER & NEUHAUS, supra note 2, at 150. 73.

See Butler, supra note 20, at 122 ("Organizations could then receive support with little or no regulation but with the crucial requirement that they justify themselves to the people who part with their own money.").

74.

Id. at 119.

II. MARKET MECHANISMS

75.

See Historical Statistics: 1960-1995, Organisation for Economic Cooperation and Development table 6.6 (Paris, 1997). [hereinafter Historical Statistics]. For 1998 figures, see A World of Taxes, at http://www.oecd.org/daf/fa/statsstats. htm. 76.

Id. For 1998 figures, see Tax burden 2001: International Comparison, at

http: //www.estv. admin. ch/data/sd/e/inter/pdf /fisquo.21. pdf

77.

See Freedom and Fairness Restoration Act of 1997: Hearing on H.A 1040 Before the House Comm. on Ways and Means, 105th Cong. 3 (1997) (Statement of

81.

See Douglas J. Besharov & John C. Weicher, Return the Family to 1954,

WALL ST. J., July 8, 1985.

82. The maximum credit was $400 in 1998 and $500 thereafter (in current dollars). See Taxpayer Relief Act of 1997, Pub. L. No. 105-34, § 101, 111 Stat. 788 (1997) (codified as amended at 26 U.S.C. § 24(a) (1997 .

83. Butler, supra note 20, at 122. 84. The College Board, Trends in Student Aid, available at http://www. (last visited Oct. 26,

collegeboard.org/press/costOl/html/TrendsSAO1.pdf 2001).

85. As discussed below, however, top-down approaches are not the only way to encourage consumption; fee-for-service and voucher systems can work as well.

86. See Economic Report of the President, PUB. PAPERS 32 (1991). 87.

See HOUSE COMM. ON WAYS AND MEANS, 2000 GREEN BOOK 813 tbl. 13-

14 (Comm. Print 2000) [hereinafter 2000 GREEN BOOK].

88. OSBORNE & GAEBLER, supra note 2, at 184.

89.

See 2000 GREEN BOOK, supra note 88, at 813 tbl. 13-14.

90. See OFFICE OF PLANNING, RES. AND EVALUATION, ADMINISTRATION FOR CHILDREN AND FAMILIES, DEP'T OF HEALTH AND HUM. SERV., TEMPORARY AID FOR NEEDY FAMILIES PROGRAM, THIRD ANN. REP. 37, 108-09 (2000). Total TANF spending, however, was over $22 billion and also includes TANF expenditures on work activities, childcare and various other services.

91.

92.

See id. at 108-09.

See 2000 GREEN BOOK, supra note 87, at 156.

93. The cost of food for a three-person family was calculated by subtracting the maximum TANF benefit for a three-person family from the maximum combined TANF and Food Stamp benefit for a three-person family. See 2000 GREEN BOOK, supra note 88, at 388, 390. According to HUD rules, eligible tenants for local public housing are required to pay the higher of (a) 30% of counted income (various deductions are made in determining countable income, e.g., $40 per month per dependent) or (b) 10% of gross income toward the rental rate. See CONG. RES. SERVICE, CASH AND NONCASH BENEFITS FOR PERSON WITH LIMITED INCOME: ELIGIBILITY RULES, RECIPIENT AND EXPENDITURE DATA, FY 1996-1998 116-17 (Comm. Print 1999). Therefore, the cost of a housing subsidy for a three-person family was calculated as the average fairmarket rate (FMR) for two bedroom housing less the average payment toward rent for a family of three, multiplied by the proportion of TANF families in public housing, receiving HUD rent subsidy or other rent subsidy. See Universal Living Wage 2000, Fiscal Year 2001 Final Fair Market Rents for Existing Housing, at http://www.universallivingwage.org/states-fmrtables/pennfmr2001.htm; 2000 GREEN BOOK, supra note 88, at 384 tbl. 7-7; [see ADMINISTRATION FOR CHILDREN AND FAMILIES, U.S. DEP'T OF HEALTH AND HUM. SERVICES, AID TO FAMILIES WITH DEPENDENT CHILDREN: CHARACTERISTICS AND FINANCIAL CIRCUMSTANCES OF AFDC RECIPIENTS, FY 1996 4 table 4 (YEAR). The cost of medical care was calculated as the sum of per capita Medicaid payments for two children and one adult. See 2000 GREEN BOOK, supra note 87, at 915 tbl. 15-16.

94. U.S. CENSUS BUREAU, CURRENT POPULATION REPORTS, P60-209, MONEY

INCOME IN THE UNITED STATES: 1999 40, 46, 49 (2000).

95.

See Allan H. Meltzer and Scott F. Richard, A Positive Theory of the InKind Transfers and the Negative Income Tax, in POLITICAL ECONOMY at 70 (Allan H. Meltzer ed.) ("When choosing the type of redistribution he prefers, the decisive voter reasons as follows: If I allow each recipient to choose the bundle of goods and services that maximize his utility, those below some level of income will decide not to work. By giving income in kind, I can encourage people who are net beneficiaries of the tax-transfer system to make different labor-leisure choices. For example, if I allow the government to distribute only one good, people with relatively low productivity will find that all of their disposable

income is in one consumption good; they are surfeited with the good that the government offers and have none of the other good(s). Since they cannot resell the good they receive, they will be willing to trade leisure for income to purchase the good(s) they do not have. By working they trade leisure, valued at the marginal product of labor, for consumption of a good that has relatively high marginal utility. They will continue to trade leisure for consumption of the good they purchase until they reach a constrained optimum.").

96. Critics of the U.S. Social Security program, for example, contend that it decreases savings and, as a result, capital investment and economic growth. Instead of saving and investing money for their retirement, workers rely on the government-through a payroll tax-to take care of them in their old age. See Martin Feldstein, Privatizing Social Security: The $10 Trillion Opportunity, THE CATO PROJECT ON SOCL SECURITY PRWVATIZATION (Cato. Inst., Washington, D.C.), Jan. 31, 1997, at 6 ("Social Security privatization would increase the economic well-being of future generations by an amount equal to five percent of GDP each year as long as the system lasts. Although the transition to a funded system would involve economic as well as political costs, the net present value of the gain would be enormous-as much as $10-20 trillion.").

97.

See Eugene Steuerle, Uses of the Negative Income Tax Framework, 12

Focus 30-32 (1990).

101.

See 2000 GREEN BOOK, supra note 87, at 106-07, This. 15-13, 15-14.

102. The recent movement toward managed care may make the American Medicaid program more like a voucher system, because recipients, in effect, will have a certificate to select one particular provider of the full range of medical care.

103. SeeJOSEPH P. NEWHOUSE & THE INSURANCE EXPERIMENT GROUP, FREE FOR ALL? LESSONS FROM THE RAND HEALTH INSURANCE EXPERIMENT 40, 79 (1993).

104.

Id. at 40, 79, 339.

Id. at 339.

105. In addition, free care marginally improved both near and far corrected vision .. .and increased the likelihood that a decayed tooth would be filled-an effect found disproportionately among the less well educated. Health of gums was marginally better for those with free care. And serious symptoms were less prevalent on the free plan, especially for those who began the Experiment poor and with serious symptoms. Finally, there appeared to be a beneficial effect on anemia for children. Id.

106. See, e.g.,JOHN HOOD, SOLVING THE MEDICAID PuzzLE: STRATEGIES FOR

STATE ENTITLEMENT REFORM 28 (1997).

107. Government provided membership in a health maintenance organization (HMO) or preferred provider organization (PPO) is often viewed as the equivalent of a voucher, if the individual can choose among the HMOs or PPOs.

108. U.S. Dept. of Agriculture, Food and Nutrition Services, National Level Annual Summary: Participation and Costs, 1969-2000, available at http://www.fns. usda.gov/fsp/Default.htm (last visited Oct. 26, 2001). The state share of administrative expenses was calculated based on the historical ration of federalto-state administrative expenses.

109. Karen Cunnyngham & Lisa Fowler for U.S. Dept. of Agriculture,

Food and Nutrition Service, Office of Analysis, Nutrition, and Evaluation, Characteristics of Food Stamp Households: Fiscal Year 2000 (Advance Report), June 2001, available at http://www.fns.usda.gov/oane/MENU/Published/FSP/FILES/Participation/2000advrpt.pdf (last visited Oct. 26, 2001).

110. Stoesz, supra note 25, at 7. 111.

REBECCA M. BLANK, IT TAES A NATION: A NEW AGENDA FOR FIGHTING

POVERTY 162 (1997).

112. 4 (1998).

See PETER H. Rossi, FEEDING THE POOR: ASSESSING FEDERAL FOOD AID

Box 5

VOUCHER OPTIONS

* Form of scrip (coupons, certificate, or bookkeeping entry " Limitations on the goods or services obtainable * Providers designated (licensed, registered, or otherwise restricted) " Amount (covers full cost of good/service or sliding-scale copayment

based on income or other determinant of need)

" Refundable (in whole or in part) " Number (offered to all who meet specific eligibility criteria, or

rationed)

113. Administrative and practical constraints, however, often limit the number of providers willing to accept child care vouchers. See Douglas J. Besharov & Nazanin Samari, Child Care after Welfare Reform, in THE NEW WORLD OF WELFAuu 467-68 (Rebecca Blank & Ron Haskins eds., 2001).

114. There may be a requirement that the provider accept the voucher as

full payment for the good or service.

115.

See Dep't of Housing and Urb. Dev., Quality Housing and Work Responsibility Act of 1998, 64 Fed. Reg. 8193. For a general discussion of the .shopping incentive," see Citizens' Housing and Planning Association, Housing Guidebook, 1999, available at http://www.chapa.orglhsggdbkadobe/RentalAssistance. PDF (last visited Oct. 4, 2001) ("The voucher program ...[gives] tenants a fixed subsidy equal to the difference between 30% of their income and a payment standard based on the FMR. HUD allows PHAs to set the payment standard but it must be between 80% and 100% of the FMR. Voucher holders are generally permitted to lease units with gross rents in excess of the FMR, but because their subsidy amount is fixed, households that lease higher cost units must bear the entire extra cost. On the other hand, the fixed subsidy meant households leasing a unit that costs less than the payment standard end up paying less than 30% of their income for housing (Congress began phasing out this feature-called a "shopping incentive"-in FY '99).").

116. John C. Weicher, The Voucher/Production Debate, in BUILDING FOUNDATIONS: HOUSING AND FEDERAL POLIcY 266 (Denise DiPasquale & Langley C. Keyes eds., 1990); Telephone Interview with John C. Weicher, Senior Fellow, Hudson Inst. (Apr. 14, 1998) ("Actual program outcomes show that about 24% of voucher recipients rent units for less than 95% of the Fair Market Rent. 19%

rent units for 95% to 100% of the FMR. The others pay more than the FMR, paying the difference out of their own pocket. That's another distinctive feature of the voucher: They can do that.").

117. See, e.g., MICHAEL J. BENNETT, WHEN DREAMS CAME TRUE: THE GI BILL AND THE MAING OF MODERN AMERICA (1996). To receive the voucher, eligible service members and veterans must choose an institution and certify their enrollment. Students can receive their benefits either by a check that is issued to them directly, or they can request that the benefit be deposited directly into their checking or savings account. For a general description of the Montgomery GI Bill, see U.S. DEP'T OF VETERANS AFFAIRS, VETERANS BENEFITS ADMIN., 2001, THE MONTGOMERY GI BILL-ACTIVrE DUTY: SUMMARY OF THE EDUCATIONAL BENEFITS UNDER THE MONTGOMERY GI BILL-ACTIVE DUTY EDUCATIONAL ASSISTANCE PROGRAM CHAPrER 30 OF TITLE 38 U.S. CODE, available at http:// www.gibill.va.gov/education/c3Opam.htm (last visited Oct. 4, 2001).

118.

See U.S. DEPT OF VETERANS AFFAIRS, VETERANS BENEFITS ADMIN. 2001. CHAPTER 30 INCREASED EDUCATIONAL ASSISTANCE ALLOWANCE, available at http:// www.gibill.va.gov/education/News/ch3ORateslOOlOl.htm (last visited Oct. 4, 2001).

CONCLUSION

119. JOHN HOOD, SOLVING THE MEDICAID PUZZLE, STRATEGIES FOR STATE

ENTITLEMENT REFORM 29 (1997).

120. For example, having schools go "out of business" in the middle of the school year may be a consideration against a pure voucher system for elementary and secondary education.

Box 6

WHICH MARKET MECHANISM? * Would a tax reduction allow many people to afford the good or service

without government aid?

* Would cash assistance to help people purchase the good or service

create too large a work disincentive or other behavioral effects, without assuring that recipients would spend the money as intended?

" Should recipients be encouraged to consume the good or service? If so,

to what degree?

" Will recipients tend to substitute non-cash benefits for expenditures

they would ordinarily make, thus undermining the purpose of targeted assistance?

" If vouchers are used, will a secondary or black market for the voucher

be likely to develop?

* Will the amount of aid provided vary so that a voucher for a set

amount of assistance would be inappropriate, and a more flexible feefor-service system needed?

" Should recipients be encouraged to be cost conscious? If so, to what

degree?

" Will the phaseout of non-cash benefits (or a sliding-scale copayment)

create marginal tax rates that discourage work or savings?

" Are recipients able to buy the same or equivalent services at sharply

different prices?