Policy Considerations Relating to Privatization in the Food Stamp Program

The Food Stamp Act requires that state civil servants make all decisions about individual households’ eligibility for benefits. Throughout the program’s history, state civil service admini­stration has been taken for granted. Last year, however, USDA approved a waiver for Florida to partially privatize administration of the Food Stamp Program in several counties. Now, at least two states are developing plans to contract with private entities to take over substantial parts of the eligibility determination process. Flor­ida has decided it does not want to wait for the results of the experiment USDA approved and is seeking ap­proval for a loosely-defined waiver to allow it to solicit bids for pri­vatizing statewide an un­determined num­ber of functions within the eligi­bil­ity determination process. On June 25, USDA expressed receptivity to Florida’s pro­posal but sought additional clarification. Texas has pro­posed to close more than half of its local offices, largely replacing them with kiosks and call cen­ters that would determine eligibility based on materials received over the tele­phone and internet. Apparently some or all of these call centers might be operated by private contrac­tors. Both states are pur­su­ing similar changes in their Medi­caid programs.

These proposals raise significant issues, both for these two states and for the future of the national Food Stamp Program. Over one million people received food stamps in Florida during an aver­age month of 2003, with total issuance that year of almost one billion dollars. Texas’s average 1.9 million food stamp recipients received $1.9 billion in food stamps last year. This year, the states rank fourth and first in the country, respectively, in food stamp issuance. Private con­trac­­tors long have operated some discrete func­tions for the Food Stamp Program: printing food stamp coupons, designing computer software, operating electronic benefit transfer (EBT) sys­tems, managing employment and training programs, etc. No firm, how­ever, has ever had con­trol of the entire program — or, in particular, the decision about whether particu­lar households receive food stamps. When the Food Stamp Program began to convert from paper coupons to EBT, USDA and states recognized that the conversion would be costly, disruptive, and subject to potentially serious unforeseen problems. Accordingly, they moved cautiously to minimize harm to vulnerable recipient families. This deliberate approach stands in stark contrast to the rapid, large-scale im­plementation that Florida and Texas are proposing.

Some people oppose privatization based on the principle that discretionary government func­tions should be performed by government employees. Others just as fer­vently support pri­vatization as a way of shrinking government or introducing more private-sector efficiencies into public administration. Beyond these basic ideological positions, how­ever, lie a host of practical considerations about how privatization might work. This paper seeks to examine some of those prac­tical issues.

Without a doubt, the current administration of the pro­gram has significant problems. A number of eligible families feel badly treated. Some forego benefits that they need as a result. Improving access is a goal shared by many public officials, emergency food providers, anti-hunger advocates, and others. Some have seen privatization as a possible answer. Care should be taken, however, before regarding privatization as a panacea. Although private firms have made important contributions in performing discreet tasks for the program, it will be very diffi­cult to design contracts that will give contractors incentives to improve service to low-income families while pro­tecting the program’s integrity. Moreover, hurried or indiscriminate pri­vat­iza­tion could inadvertently cause significant problems with program access, integrity, and cost.

This paper first considers the inherent limits on states’ ability to reap the benefits of com­petition when contracting out program management. It then analyzes the challenges states face in making the transition from operating a program directly to letting and supervising contracts for pro­gram manage­ment, the different skills needed for these two functions, and the diffi­culty of crafting effective contracts for such a multi-faceted function as administering the Food Stamp Pro­gram. It attempts to identify the general factors that make a government func­tion a more or less appealing candidate for contracting out. Finally, it suggests some safeguards that might be prudent if a political decision is made to experiment with contracting out food stamp eligibility determinations.

The Limitations of Competition for Program Administration

One of the main advantages commonly cited for privatization is increased competition. The government, it is said, has held a monopoly on performing these functions for too long. In­vit­ing private contractors to bid for the right to provide those services is thought likely to drive innovation that will allow the program to obtain better services for a lower price. The benefits of competition to govern­ment programs should not be taken lightly: in numerous areas, public-pri­vate partnerships have yielded impressive results. For example, the Special Supplemental Nutri­tion Program for Women, Infants, and Chil­dren (WIC), which provides food vouchers and for­mula to almost half of the infants born in the U.S., requires infant formula companies to compete for its business. Savings from this competitive bidding process are estimated to be about $1.5 billion annually, enough to serve more than one-quarter of all low-income women, infants, and children receiving WIC benefits in an average month — over two million people.

Competition for the right to administer a program, however, differs from competition to provide products or specialized services in several important respects. These differences may undermine the value of that competition significantly.

• First, selecting winning bidders will depend upon a great deal of speculation. At the time the competition takes place, private bidders generally lack the pre­sent capacity to offer those services. No company rents offices and hires eligi­bility wor­kers on the chance that it will be selected to run a program. Thus, the govern­ment agency acting on the bids must essentially speculate about what sort of infra­­structure the agency will be able to build in the future if it wins the bid. That sort of speculation is in­herently error-prone, undermining the value of the com­peti­tion: the value of any im­prove­ments in service and price must be weighed against the risk that the contractor will be unable to perform as promised. Since the government will know the competing bidders’ prices but will only be able to speculate about the service they can provide, its selection of a contractor inevi­tably will be based primarily on price. A bidder that has won the contract by committing itself to operating with the least funding may lack the resources to provide quality service to low-income families.
  
  Many of the notorious problems in military contracting — wea­pons that do not work, cost overruns, etc. — result neither from incompetence at the Defense De­partment nor from the venality of defense contractors. Instead, they result from the inherent dif­ficulty of predicting a new weapon’s capabilities and problems in advance of its con­struction. The Defense Department and NASA may have little prac­tical choice but to endure these problems because the govern­ment does not want to build and operate weapons and spacecraft factories.
  
  Food stamps and other government assistance programs historically have avoided this highly speculative, risky form of contracting. It stands in marked contrast to the more typical approach to contracting in these programs such as WIC’s infant formula pro­curement system. When WIC accepts a manufac­turer’s bid to be the sole pro­vider of infant formula in a state, that company is already producing ac­ceptable product; upon winning the con­tract, it need only ship more of that pro­duct into the state. Thus, while it is relatively clear that picking the lowest bidder among WIC infant for­mula sup­pliers will get the state the best bargain, the same is not true when selec­ting bid­ders to provide admini­stra­tive services: the lowest bidder may not be able to do all that its bid promises (and that the bidder sincerely believes it can do). Once the state selects a bidder, it is likely to need to establish so many addi­tional require­ments that the final contract is quite different from the specifications on which the prospec­tive contractors bid. In the end, the state may have little idea whether one of the unsuccessful bidders might not be able to do a better job of what is actually required.
• Second, programs like food stamps will be at risk for higher costs because their admini­stration is a natural monopoly. The overhead required to maintain facili­ties, develop policies and procedures, train staffs, etc., make it uneconomical for multiple entities — the state and a pri­vate contractor or more than one contractors — to operate eligibility determination systems simultaneously. Indeed, even if they could, it is difficult to see the basis on which competition would take place: surely the program would not want households to select which office to visit based on where they believe they will receive more generous benefits. Thus, the only competition will take place at the initial bidding for the contract; the classic type of on-going competition that one sees between Ford and GM, between ABC and CBS, between Coke and Pepsi, etc., is impossible when the Food Stamp Pro­gram contracts out responsi­bility for eligibility determinations. In much the same way, each state has had to contract with only one company to operate their EBT systems. Once the contract is signed, all competition ceases for the duration of that contract. Poor service by the selected vendor, or the appearance of other ven­dors offering better service or prices, is irrelevant unless the state has grounds to void the contract.
  
  In other natural monopolies, such as the pro­vision of gas or water and sewer ser­vices, public opera­tion or public regulation often have proven necessary to re­strain costs. The Food Stamp Program can in­vite genuine competition at the time it lets contracts, but during the life of those contracts monopoly conditions will exist. During those periods, the system will remain as impervious to market sig­nals as it is in the current, state-run system. Because of the cost and disruption of letting contracts and transferring administrative responsibilities, administrative contracts are likely to run for a number of years, eliminating any com­petition for long stretches of time.
  
  During the life of each contract, the contractor will seek to reduce its costs so that it can maximize its profits. Where federal law provides state options, contractors therefore will prefer whichever one lends itself best to administrative ease. Thus, contractors are unlikely to volunteer to implement optional programs, such as nut­rition educa­tion or job training, that increase their administrative burdens even if those op­tions might benefit recipients. Contractors similarly will have incentives to avoid complex determinations of physical or mental disability or other hardship before sanctioning recipients for failing to comply with program rules.
• Third, and related, whichever contractor is chosen will have great leverage over the state, and great advantages over potential competitors, because the cost of changing program administrators will be very high. Administering a public pro­gram requires sig­nificant, irreducible up-front expendi­tures by each new vendor. These include the costs of hiring staff, months of designing policies and proce­dures for the program and training the new staff in those policies and procedures, the time required to rent and furnish of­fices before they can be put into use, etc. A company having to hire an entire staff of people with the intellectual and social skills to learn and implement complex pro­grams policy in a short period of time is likely to have to pay a premium for that staff — or to compromise severely on qual­­ity — far more than an agency that only needs to replace its occasional losses to re­tirement and attrition. For a contractor to improve the cost or quality of a program’s ad­ministration, it will need to achieve enough new efficiencies to offset these con­sider­able capital costs.
  
  Moreover, these start-up costs will occur each time a new entity takes over ad­mini­­stra­tion of the program. Once a program’s administration has been con­tracted out, there­fore, the contractor will have a substantial advantage over any competing bidder (in­cluding the former public agency, which by then will have had to dismantle its admini­strative infrastructure). This will allow the incumbent contractor to raise its bid and still defeat prospective competitors, whose bids will have to cover these start-up costs. In­deed, private contractors may increase the start-up costs of potential public and private contractors by requiring their more skilled employees to sign contracts promising not to work for competitors in the state for some length of time.
  
  If a contractor providing relatively fungible products, such as WIC infant formula, fails to perform its contractual obligations, the state has other options and so can consider voiding the con­tract and seeking new bids. By contrast, if an administra­tive services vendor is performing badly, the costs and disruption of selecting a new contractor and hav­ing that contractor build up the infrastructure required to operate the program may leave the state with little prac­tical choice but to stay with its existing contractor. This will undercut contractors’ incentives to comply.
  
  For just these reasons, several states have stayed with automation contractors whose systems were causing chaos in their administration of their programs. The states rea­soned that switching contractors would cause considerable disruption and that any new contractor’s system might have comparable problems. The fail­ure of new automation systems caused huge increases in the food stamp quality control (QC) error rates in Florida, Indiana, Michigan, and Los Angeles County, among other places. The disrup­tion, cost, and risk of switching automation sys­tems pales in comparison to that asking a new contractor to set up an entire new eligibility determination infrastructure, with new offices, staff, computers, etc. Thus, states may feel compelled to stay with administra­tive contractors even if their work is far worse than the severely defective automation systems that states have tolerated in the past.
  
  The government’s dependence on a particular contractor may prevent it from reap­ing the full benefit of competition. In particular, the high costs to the govern­ment of shifting from one contractor to another (or rebuilding the infra­structure to administer the pro­gram itself) with make the state ex­tremely vulnerable to con­trac­tors’ pos­sible bank­rupt­cies. Since the contractor will be bidding based on specu­lation about what admini­ster­ing these programs will cost, once it actually rents and furnishes offices, hires staff, etc., it may find that it significantly under­budgeted. In these circumstances, the contrac­tor may threaten to declare bank­ruptcy, leaving the state without the means to operate its programs. If the state judges that threat to be credible, it may have no choice but to increase its pay­ments to the con­trac­tor. Thus, the state effectively must as­sume most of the risk of the inherent un­certainty about the costs of shifting administra­tion to a private contractor: if the con­tract price proves excessive, the contractor keeps the profits, but if it proves insuf­fici­ent, the contractor has lever­age to extract an increase.
  
  Medicaid managed care produced modest short-term savings because plans of­fered low rates to win their initial contracts. Once states had become dependent on them, these plans insisted on higher rates to continue serving Medicaid and the savings disappeared. Similarly, in the early 1990s, vendors of electronic benefit transfer (EBT) services negoti­ated con­­tracts under which they were paid on a per-transaction basis. Since parti­cipation had been rising rapidly, they assumed this would prove highly profitable. When, in fact, participation plummeted after 1994, these contracts proved unpro­fitable. Some con­tractors sug­gested that they might abandon their contracts, put­ting great pressure on the states and USDA to increase the agreed-upon payments.^[1] The leverage of contractors admini­stering key parts of the eligibility determina­tion process would be incomparably greater.
  
  Even the short-term cost savings that Medicaid managed care and EBT brought states came at a high cost to recipient families. Implementation of both systems was chaotic and caused significant hardship in many places. As time passed, con­tractors solved many of their initial problems — but also increased what they charged the states.
• Fourth, the government will have difficulty measuring contractors’ performance and en­suring quality. If an infant formula company delivers expired or otherwise sub­standard formula to WIC, the program’s managers can promptly detect the prob­­lem and assess a penalty. If the problem persists, the government will know and can cancel the con­tract. The enforcement of contracts of this kind is easy be­cause the products the gov­ernment seeks to procure are fairly straightforward to specify.
  
  By contrast, the quality of the administration of the Food Stamp Program is diffi­cult to assess. It entails some relatively objective acts: keeping offices open dur­ing prescribed hours and processing applications and delivering benefits by set deadlines, etc. It in­volves other activities — such as payment accuracy, avoiding improper denials, and claims collection — that can be measured but for which it may be difficult to agree upon appropriate stan­dards. And it includes some efforts that are almost impos­sible to quantify: providing accurate responses to eligibility questions, preventing the application process from discouraging eligible claim­ants, promoting sound nutrition, encouraging in­creased work effort, etc. At best, a contract may specify some specific actions the contractor must take in these areas, but that provides no protection against a contractor whose staff is “going through the motions.” And objective contractual terms are most unlikely to pro­vide mean­ingful guidance on striking the right balance between competing ob­jec­tives, such as payment accuracy and program access.
  
  No program administration, public or private, will excel in all of these areas. As­sess­ing the quality of program administration therefore is very much a matter of sub­jective judg­ment, weighing deficiencies in some areas against strengths in others. These assess­ments typically depend on a wealth of informal communi­ca­tions among various units within the system, a type of communication that con­trac­tors will have strong incentives to control tightly. Even if most observers would regard a particular contractor’s per­formance as deficient, the state may not have objective evidence that is legally suffi­ci­ent to impose penalties or to termi­nate the contract. Absent such evidence, the pros­pect of costly and protracted litiga­tion is likely to deter states from even at­tempting ac­tion against contractors. Although automated systems lend themselves far more readily to objective per­for­mance measures than does overall program ad­ministration, states have rarely if ever recovered damages from automation con­tractors even when sys­tems’ cata­strophic failures caused huge increases in error rates and workload for eli­gibility staff. The difficulty of measur­ing performance under contracts to administer the Food Stamp Program is likely to prevent states from enforcing most quality as­sur­ances the contractors provide. Com­petition thus offers little prospect of improv­ing the quality of program administration or even of allowing states to identify the highest-quality bid­der when letting the contract.
• Finally, contracting could subject the program to massive financial losses, which would damage the program directly and undermine public confidence in its steward­ship of taxpayer funds. No contrac­tor will be able to afford to cover the costs of additional food stamps issued be­cause of serious problems in its admini­stration. Contractors, un­like states, can declare bank­ruptcy. The burden of any significant losses therefore will almost certainly accrue to either state or federal governments. And, in prac­tice, turn­over among state and federal officials and other limitations on the fed­eral-state rela­tionship make it diffi­cult to protect the federal government from severe finan­cial losses if an experiment goes wrong. For example, in the early 1990s, Florida moved rapidly to implement an ambit­ious new system for automat­ing many aspects of the eligibility deter­mination pro­cess. It rapidly implemented the system statewide and ran into severe diffi­cul­ties. The system mishand­led numerous cases, and the state’s food stamp error rate spiked. In 1992 alone, Flor­ida overissued more than $200 mil­lion in food stamps. Although the food stamp quality control (QC) system assessed a correspondingly large penalty, by that time the gover­nor who had rushed through the im­plementa­tion of the flawed automa­tion system had left office. His successor struggled to correct the prob­lem but argued that imposing such a large penalty for something his predecessor had done would hobble his new administration. In the end, USDA re­quired Florida to spend some unmatched state funds to correct its system but did not collect any of the assessed QC penalty. The federal government ab­sorbed all of the $200 mil­lion loss that Florida’s hasty ex­per­iment with automa­tion caused.
  
  Just this year, the Bush Administration forgave a nine-figure QC penalty against Cali­for­nia when Gov. Schwartzenegger argued that he should not have to pay for errors re­sult­ing from a deficient automation system mismanaged by his predeces­sor. The first Bush Admini­stra­tion similarly waived large penalties against states, often in response to pleas from successors of the governors under whom the pen­alties were incurred. Thus, the losses that the federal government can incur from catastrophic failures of state ad­mini­stration in the Food Stamp Program are more than administrations of either party have been willing to extract from the successors of the governors responsible for those failures. Similarly, both Republican and Democratic administrations have declined to require subsequent governors to make good on their predecessors’ commitments that demonstration projects would be cost-neutral to the federal government.
  
  The same thing could happen again if the current privatization proposals go awry. Fis­cal year 2004 is virtually over. It is unclear if Florida could turn over admini­stration to a contractor much before the end of federal fiscal year 2005. Even if it can, USDA will not release the error rate for that year until June 2006. Admini­strative appeals could easily last for six months, by which time Governor Bush will be leaving office under Florida’s term limits. (Texas Governor Perry, al­though not term-limited, also reaches the end of his current term in 2006.) If some­thing goes wrong, any fiscal penalty will have to be paid by his successor. And with Florida issuing over $1 billion in food stamps each year, the cost to the federal govern­ment if the contractor is unable to perform could be huge. (Texas’s annual food stamp issuance exceeds $2 billion.) And replacing or re­habilitat­ing an incom­pet­ent contractor could be far more difficult, expensive, and time-con­sum­ing once the state has dismantled its current eligibility determination infra­struc­­ture than cor­­rect­ing the defective automation system was in Florida in the early 1990s. What­ever the formal terms of the waiver USDA grants, on a project of this scope the states in effect are likely to be gambling with the federal govern­ment’s money.
  
  A fundamental problem with states’ proposals to privatize the administration of the Food Stamp Program is that the states stand to reap substantial gains from any successes but are largely immune to the consequences of any failure. If privatiza­tion reduces administrative costs, states — which pay half of those costs — will reap a financial benefit. If, however, it results in improper issuances of benefits, the federal government — which pays all food stamp benefit costs — and not the state will suffer most of the consequences. In this respect, privatization of pro­gram ad­mini­stration in the Food Stamp Program differs fundamentally from simi­lar pro­posals in TANF, where states have a far higher stake in benefit costs. Bene­fit costs are many times higher than ad­mini­strative costs in the Food Stamp Program, meaning that the potential losses far exceed the potential gains. And of course, if privatization results in the delay or denial of food assistance to eligible needy fami­­lies, benefit costs may decline but at the expense of severe hardship to these families. This imbalance between the potential dangers and advantages they might ex­peri­ence will naturally skew states’ cost-benefit analyses, giving them in­centives to experiment with risky privatization proposals. A state’s proposal in this area there­fore does not deserve the same deference that would be appropriate if the state was going to be the primary one suffering the harm of a failed experi­ment.
  
  The risks are quite real. In states that have privatized administration of TANF, serious disputes have arisen over contractors’ operations of the programs and the propriety of some of the charges to states. Moreover, both Texas and apparently Florida contemplate greatly reducing the role of face-to-face interactions with appli­cants and recipients. The face-to-face interview long has been one of the Food Stamp Program’s primary safeguards against fraud. Direct contact with households also is how states screen applicants for those in need of emergency food stamps, identify persons with disabilities who may need extra help in the application process, and identify other services — job training, nutrition educa­tion, child welfare, etc. — to which the household should be referred. The states have presented no evidence that alternative means will prove equally effective at detecting and deterring fraud — or that they can do so without imposing undue burdens on low-wage working families.

None of this is to suggest that competition has no place in program administration. It does mean, however, that policymakers should not assume that competition will have the same ef­fect in one area of program operations that it does in others. Level-headed analysis of the po­ten­tial risks and bene­fits of any proposal for increased contracting, rather than generic assump­tions about the benefits of competition, will lead to the best use of limited public funds. Any in­vestigation of privatization should proceed slowly, carefully, and with close federal oversight.

Imposing New, Difficult Responsibilities on Government: Designing and Assuring Compliance with Contracts for Program Administration

When the government contracts out administration of important government services, it assumes a fundamentally different role. Where previously the government was required to ad­mini­ster the pro­gram, now it becomes responsible for framing requests for bids, negotiating con­tracts, monitoring per­formance, and enforcing compliance. Both administration and con­tracting are functions that govern­ment sometimes does well and sometimes does poorly. The vast major­ity of government administration and procurement go well and largely unnoticed. But just as some agencies are accused of providing shoddy service to the public, so, too other agen­cies are criticized when they overpay for hammers, toilet seats, and other routine items or when their contractors provide poor service to the public.

The skills required to administer programs and manage contracts are quite different. Good managers must have strong leadership skills and be good judges of people. They must be flexible enough to adapt rapidly to changing circumstances and problems. They must be able to translate poli­cies into instructions their staff can implement. By contrast, letting a sound contract for program ad­mini­stration requires more analytical skills, including meticulous attention to de­tail. Contracting officers must anticipate problems that have yet to arise because they will not have the flexibility to respond to them once the contract is signed. To the extent that they need interpersonal skills, it is as a negotiator rather than as a leader. They must be able to translate policies into contractual specifications that will cause the contractor to act as the agency intends.

Many excellent administrators are ill-equipped to negotiate important contracts. On the other hand, many people skilled at drafting contracts lack the expertise in operating the Food Stamp Pro­gram to anticipate all the contingencies that must be addressed in a contract for the program’s administration. A team is clearly needed, but even if a state agency has the right com­bination of talents to make an ef­fective team, the novelty and enormity of the proposed transfer makes significant problems inevitable: some important aspects of program administration will be taken for granted and some contingencies will not be anticipated. This suggests that any agen­cy proposing to put program administration out for bid should be scrutinized carefully to deter­mine if the personnel it intends to put in charge of contracting have the necessary skills. If it does, a gradual approach to contracting — contracting out some limited functions and for limited parts of the state — is prudent to allow the team to develop the necessary ex­pertise and to limit the consequences of oversights and errors in drafting its first contracts.

States have encountered serious difficulties when trying to make other sharp changes in the way they administer other benefit programs. For example, Medicaid’s conversion to man­aged care required state health departments to switch from working to prevent the overuse of services to trying to ensure that managed care contractors did not cut costs by denying services: in a conventional fee-for-service system, providers’ incentives are to provide as many services as they can, while in managed care the providers’ incentives are reversed. It took about a decade for many states to master the Medicaid managed care contracting process. In the mean time, a host of serious problems ensued. Some con­tractors with inadequate financial planning spent their capitation payments quickly and went bankrupt. This left thousands of Medicaid benefici­aries with no way to receive care until the state was able to scramble around and piece together alternative arrangements, sometimes paying a premium to do so. The doctors and other care pro­viders left unpaid by the bankrupt plans became even less will­ing to treat Medicaid patients than they had been previously, and some moved out of already-underserved low-income communi­ties. Other con­trac­tors apparently con­cluded that the turnover among Medicaid recipients made it cost ineffective to im­munize children; a measles epidemic in Milwaukee resulted in 1989-90. Still other plans became adept at “skimming:” recruiting Medicaid bene­ficiaries likely to require little care while discouraging enrollment by those expected to be un­profitable. Some of the worst prob­lems occurred where states rushed broad scale implementa­tion — as Florida and Texas are pro­pos­ing to do with privatization.

Yet after all of this disruption, Medicaid costs in the states that converted to managed care have continued to rise at rates largely indistinguishable from those in states without man­aged care. Mainstream managed care plans have been unwilling to accept low Medicaid fees, and the small number of Medicaid-oriented plans have had the market power to extract steady rate increases from states. Thus, there are no assurances that the burdens and risks of this sort of massive transformation will ever produce the hoped-for savings.

No one set out to recruit incompetent or irresponsible managed care plans. Nonetheless, states’ experi­ence with fee-for-service care left them ill-equipped to judge when the low-bidding plans could, and could not, be counted on to provide adequate care. In Medicaid managed care, as in food stamp admini­stra­tion, no immediate market correction is available since program re­cipients cannot walk away from un­satisfac­tory contractors.

Similarly, Medicare’s shift from per diem payments to diagnostic related groups (DRGs) for re­imbursing hospital care transformed the risks the program faced in ways that were difficult to anticipate. Previously, with hospitals receiving more funds the longer they kept patients, the program struggled to prompt faster discharges. When DRGs gave hospitals flat payments for each illness without regard to the length of a patient’s hospitalization, some hospitals began dis­charging patients while they were still seriously ill. Here again, the change in providers’ in­cen­tives required entirely different monitoring and enforcement activities. Here, too, some hospitals became adept at skimming: recognizing and recruit­ing patients with relatively inexpen­sive-to-treat forms of conditions covered within a broad DRG. A similar story could be told about Medi­­­­caid’s experience contracting with nursing homes. Over time, the programs adapted. In the short term, the losses in both human and financial terms were substantial.

It is no disparagement of states or USDA to say that they will be unable to anticipate all of the prob­lems that need to be addressed in contracts for administering the Food Stamp Pro­gram. Starting the privatization of program ad­mini­stration with rapid statewide implementation in the states rank­ing first and fifth in food stamp issuances would seem to court excessive risks.

Anticipating all contingencies, or even all important ones, when drafting a request for proposals or a final contract is effectively impossible. A partial parallel can be seen in federal-state relations. The Medicaid statute is, in a sense, a contract between HHS and the states for the program’s operation. States have repeatedly identified flaws in this statute that allowed them to increase the proportion of the program’s costs that the federal government bears beyond that anti­cipated by the statute’s drafters. Time and again Congress has closed loopholes: dispropor­tionate share hospital (DSH) payments, pro­viders’ donations and taxes, intergovernmental trans­fers, skillful use of the upper payment limit (UPL), etc. Time and again, enter­prising states iden­tify others. The states’ behavior, while per­haps an­noying, is in no way illegal or inappropri­ate: they are simply pursuing the full scope of their rights under the law. Governors, budget chiefs, and Medicaid directors are responsible to make the best fiscal arrangement for the people of their states within the law; they owe no duty to the federal government.

In the same way, one need not suppose that contractors hired to take over administration of the Food Stamp Program will inappropriately exploit their positions. Nonetheless, their pri­mary goal is to maximize returns on their shareholders’ invest­ments. Indeed, fail­ing to maxi­mize shareholders’ profits would be unethical and could con­ceiv­ably subject company mana­gers to liability. Just as Medicaid lost large sums of money in the months or years it took Congress to plug loopholes in the federal statute, the Food Stamp Pro­gram can be expected to suffer serious losses as states are forced to await the termi­nation of their contracts — or pay a premium to re­nego­tiate early — to amend portions of their contracts that are found to be flawed.

Similarly, although Congress has amended the WIC program’s statute numerous times over the past three decades, it had not anticipated that stores could increase prices by selling ex­clusively to WIC participants. ^[2] Fortunately, the rapid growth of these WIC-Only stores coin­cided with a reauthorization of the program, and Congress acted to try to rein in their prices. Had the rules for how much WIC paid for foods been locked into a multi-year contract, however, the pro­gram would have been powerless to avoid spiraling losses from these stores’ legal but un­anti­ci­pated activities. And yet despite this legislation, USDA recently amended its budget re­quest to pro­pose that Congress prohibit any new WIC-Only stores from entering the program — appar­ently an in­dication that USDA is uncertain if it can quickly or effectively correct the prob­lem with stores already in the program. If problems arise with contractors administering the Food Stamp Pro­gram, USDA may feel it can do little more than impose corrective requirements on future con­tracts for admini­stration of the program. Addressing problems with vendors al­ready involved with a program can be extremely difficult and time-consuming.

Crafting effective contracts for the administration of the Food Stamp Program is particu­larly difficult because, as noted above, it requires accommodating many different policies that at least par­ti­ally conflict. For example, we want eligibility workers to verify applicants’ statements about their in­come — but not to the point of pestering low-wage workers’ employers, or requiring workers to take so much time off the job, that they are fired, denied promotions, or discouraged from applying for food stamps. States can provide financial incentives for contractors to achieve one goal or another, but achieving the right blend of incentives to induce the contractors to strike the right balance between those objectives, — particularly where some, such as service to low-wage workers, are difficult to measure — may be all but impossible; at best, it will require exten­sive trial and error over the negotiation of several contracts.

Designing contracts that can adapt appropriately to changes in participation also will be very difficult. Serving more households costs more money; if the contract does not increase ad­ministrative costs sufficiently in response to increases in participation — or if it reduces them too little in the event of falling participation — the contractor will have incentive to create barriers to access for eligible families that need food assistance. But if contractors receive too much addi­tional funding for serving more families, they may feel insufficient pressure to maintain program integrity or to support effective job training and similar programs that might help families raise their earnings to exceed the food stamp gross income limits.

Designing contractual terms to handle changes in program rules also would be a chal­lenge. After the contract has been signed, Congress may amend the Food Stamp Act, USDA may change federal food stamp regulations, or the governor or legislature may conclude that some options it has elected no longer serve the best interests of the state and its low-income families. Some of these changes may significantly increase or decrease the cost and burden of administering the program. The state and the contractor may disagree, however, on how much of a difference these changes will make and hence how they should affect the contractual payment rate. States may find that they are unable to adopt beneficial new state options without paying a premium to their contractors and may lobby against otherwise meritorious federal food stamp legislation because they cannot afford the additional costs their contractors will charge to imple­ment it. Where public agencies administer a program, by contrast, no complex negotiations are needed to adopt or implement new program policies.

On any items where contractual requirements are not explicit, contractors will face com­petitive pressures to minimize their investments. Thus, for example, if the contract does not specifically spell out what assistance contractors must provide to people whose disabilities pre­vent them from using automated application systems or people in rural areas without suffici­ent internet access, any contractor that voluntarily builds such accommodations into its budget will be at a disadvantage against bidders that commit only to doing what the terms require.

Crafting reliable systems for monitoring contractors’ performance also poses new and difficult challenges to state administrators. With no experience contracting out administration of the program, even the most conscientious state officials are unlikely to anticipate what data they will need to monitor the contractor properly. Once the contract is signed, the contractor will be under no obligation to honor requests for additional information. One state reported recently that when it asks its EBT con­tractor for information not required under its contract, the contractor commonly responds by quoting a price the state would have to pay to receive that information. If the contractor be­lieves that certain information will reflect badly upon it, the contractor will have every legal right simply to refuse to supply the information. More generally, contractors may withhold informa­tion their contracts do not require them to submit on the grounds that that information is propri­etary. Thus, state officials may have no real idea how the contractor is doing. Even if they re­ceive complaints, they may have no way of knowing how widespread a problem may be.

Contract terms must be particularly clear because litigation with contractors can pro­foundly dis­rupt program administration. If a state terminates a contractor that is performing badly, the contractor is likely to sue for the fees it would receive under the remainder of the con­tract. If the state misjudged in terminating the contract, or if its lawyers perform badly, the state could end up having to pay twice for administering the program during the same period: once to do the work itself (or through another contractor) and again to the dismissed contractor. No state’s budget can readily absorb this kind of blow. Moreover, states effectively will be unable to plan (and to make long-term hires or enter into any new long-term contracts) during the months or years required to resolve the litigation finally. States therefore are likely to be ex­tremely reluc­tant to terminate even the worst contractors — and quite eager to settle any litigation that does arise.

The Types of Activities Most Effectively Contracted out to Private Entities

Some functions are most efficiently performed by the government directly; others are best con­tracted out. The key to sensible program management is to identify which kinds of ac­tivi­ties fall in each category rather than to allow an ideological preference for or against pri­vat­ization force an unwise deci­sion.

One key criteria for determining whether the public or private sectors can most efficiently pro­vide a particular service is which one has the necessary infrastructure. The Food Stamp and WIC Programs rely on private stores to distribute the food they purchase because these stores already have effective food purchasing, distribution, and storage systems; the government would duplicate their in­vestments if it had to build a parallel system for the food assistance programs. Medicare and Medicaid similarly rely primarily upon private hospitals, doctors, and other health care providers because of the inefficiency of developing a duplicate health care delivery system. Most government employees travel on regular commercial airplanes because it would be ineffi­ci­ent for the government to duplicate the airlines’ fleets, route plans, ticketing systems, and other infrastructure. Data processing for states’ food stamp eligibility determination systems similarly is efficiently contracted out because private companies already have invested in the massive ser­vers required to perform that work: a contractor may schedule a batch of computations for the Food Stamp Program in between similar runs for large corporations or other governmental agencies. Early EBT demon­stration pro­jects cost far more to deliver benefits than paper food stamps; once the private sector devel­oped the infrastructure to support electronic purchases in grocery stores to serve credit card cus­tomers, costs to the Food Stamp Program plummeted.

In some cases, neither government nor the private sector may have the infrastructure ready to perform a particular job. In those cases, the government may choose to contract for the work rather than assume the responsibility of building the infrastructure itself. Thus, when the Defense Department or NASA wanted to acquire new types of weapons or spacecraft, neither they nor any private com­panies were equipped to do the necessary work. Because the U.S. government historically has been reluctant to build and operate factories, it chose to contract for that work, essentially paying for private companies to develop the necessary infrastructure.

On the other hand, where the government already has the necessary infrastructure in place, and where private contractors have few other present or future uses for that kind of infra­structure, paying to have the wheel reinvented rarely makes sense. No one would seriously sug­gest that we should aban­don the national parks and contract with private companies to acquire and operate a comparable park system: what would be the point of it? No state is ready to hire private contractors to replace its uni­versity system: the investment necessary to create that system has already been made and does not need to be duplicated.

Although not nearly as conspicuous or imposing as the Everglades or Texas A&M, the infra­struc­ture states have developed to administer the Food Stamp Program also reflects a sub­stantial in­vestment. Selecting, renting, and furnishing offices, developing policies, forms, and computer systems, hiring, training, and supervising staff, and myriad other steps are necessary to establish a system for determining households’ food stamp eligibility. Numerous problems arise in administration that agen­cies’ staff learn to handle by trial and error; that experi­ence, too, rep­resents an investment. A private contractor could eventually reproduce that infra­structure, but why we should want it to do so is unclear. No economies of scale are present: the contractor is unlikely to be able to perform work for other cus­tomers with the food stamp eligi­bility deter­mina­tion infrastructure the way mega-servers handle work for private companies with data pro­cessing needs or check-out stand terminals process non-food stamp cus­tomers’ credit card pur­chases. Any savings are likely to come from reductions in service, such as closing offices. Those reductions, however, can be implemented just as easily by the existing state agencies, without privatization, if they are deemed desirable. Privatization by itself offers no obvious ways of producing significant efficiencies.

Contracting also is more appropriate where the government’s requirements are relatively clear, objective, and stable. Most obviously, the government relies on private vendors for most products it needs rather than building factories of its own. The government also purchases num­er­ous ser­vices that can be concisely and specifically described. Among other things, the Food Stamp Pro­gram long has rented space for its national headquarters from a private landlord. Many state food stamp agencies similarly rent space for their headquarters and local offices. Pri­vate contractors design and print the program’s informational brochures and posters. Other kinds of services men­tioned above — data pro­cessing, air travel, etc. — can be reduced to simple, clear specifica­tions that are amenable to private contracting.

By contrast, governmental functions that require the exercise of judgment to weigh com­peting priorities are difficult to privatize successfully. For example, although many have criti­cized the Federal Emergency Management Administration (FEMA), few would argue that its core functions should be contracted out. Assessing each disaster and determining the nature and extent of help the victims require is an intensely discretionary task. Writing a contract that suc­cessfully captures all of the factors FEMA must consider would be virtually impossible. Even the State Department’s harshest critics generally would not propose that private con­tractors make decisions about this country’s relations with foreign powers.

Similarly, where the way the service is provided needs to adapt to changing circum­stan­ces, a contract that locks in a certain set of specifications for an extended period may not be ideal. This was one of the factors that led the federal government to discontinue its reliance on private contractors for airport security. As new threats emerge, the Department of Homeland Security can adapt security precautions much more quickly by giving orders directly to the air­port staff rather than by negotiating new terms in contracts with private companies. Airport security also requires a delicate balancing of the need to prevent terrorist attacks with the need to avoid the economic damage that would result from making air travel too burdensome. The fed­eral gov­ernment’s direct control of this process allows it to make adjustments more quickly if it finds security vulnerabilities or excessive delays in one or another part of the system.

Administering the Food Stamp Program requires both front-line and more senior staff to make numerous nuanced judgments in response to changing circumstances. As evidence begins to suggest that many households are misunderstanding a particular notice, administrators must decide whether to rewrite that notice, instruct eligibility staff to explain it differently in inter­views, or to simplify the rules that the notice is trying to explain. Eligibility workers must judge whether an applicant’s hesitant an­swers result from embarrassment or from an effort to defraud the program. Aggressively questioning every reticent applicant will drive off many eligible low-wage working families that badly need the pro­gram’s help; ignoring signs of possible prevarica­tion will open the program to abuse. Similarly, when an applicant seems to be having trouble understand­ing the program’s rules, the eligibility worker needs to decide if the problem is a con­fusing explanation, a mental disability that requires an accommodation, or limited English pro­ficiency, which requires an interpreter. The state has no hope of being able to specify how these and countless other delicate judg­ments should be made in the terms of a contract. Instead, all employees of the contractor will be obliged, out of duty to its stockholders, to make decisions based on its bottom line. Although not all state employees may live up to the lofty vision of pub­lic service, at least that vision is the goal to be strived for; employees of a for-profit concern are expected to base their actions on what will best enhance its bottom line.

The acceptability of contracting also depends on tolerance for risk. Transferring respon­sibility to a new entity entails risks both in the transition and over the long term if the contractor proves not to be up to the task. The government no doubt could obtain a great bargain contract­ing out responsibility for transporting the President — what airline would not be willing to absorb substantial losses to be able to advertise that it is the official airline of the President? — but ob­vi­ously no one would consider doing so because the consequences of even a brief error in com­muni­cations could be disastrous. On the other hand, we are comfortable contracting out re­fresh­ment stands in public parks because any failure would cause few serious problems. Decisions about whether to contract out administration of the Food Stamp Program to companies currently lacking the necessary infrastructure therefore depend in part on a judgment about how troubled we would be if those contractors failed and left many eligible low-income families without food assistance for some period of time.

Possible Strategies for Safeguarding Program Integrity and Effectiveness while Exploring Privatization

Privatization need not be an all or nothing proposition. Private contractors already per­form numerous functions for the program, including the design and operation of automation sys­tems, elec­tronic delivery of benefits, research and evaluation, the design of outreach materials, etc. Indeed, the most significant difference between the Food Stamp Program and the com­mod­ity distribution programs that preceded it (as well as today’s TEFAP) is privatization of the dis­tribution of food through com­mer­cial food retailers. That decision sharply reduced program costs and increased recipients’ choices among foods. Thus, the question about whether private contractors would have a role in the Food Stamp Program has never been seriously in doubt. The key questions are, first, how much of the program’s operations will be turned over to private contractors, and, second, which aspects are best turned over to them and which are best kept in the hands of public officials.

If a political decision is reached to explore further privatization of the Food Stamp Pro­gram, a state could try contracting for some kinds of functions to see whether it saves money and improves the program’s opera­tions. State and federal officials should collaborate to develop a detailed list of specifications that should be in any contract to administer the program. Prior to implementation of EBT — a far simpler process — federal and state officials conducted numer­ous studies and tests of technical issues, contract specifications, etc. A state also could try con­tracting for administration in some parts of the state before making a decision about whether, and how, to make statewide changes. The private contractor should open its office first; only when that office is running smoothly should the public office be closed. In automating food stamp benefit delivery, states maintained the infrastructure of paper coupon issuance until their EBT systems were up and running. They should follow a similar approach here. If the state’s initial con­tract specifies that the contractor will take over statewide, its best employ­ees will find jobs with better long-term prospects and the state will have little choice but to pro­ceed with the roll-out even if the contractor’s performance in the initial counties is highly prob­lematic. Just as importantly, the state will not know what issues it needs to address in a contract for statewide administration until it sees the results from a limited pilot. Neither of the plans cur­rently under discussion appears to give the states this opportunity.

A rigorous, in-depth evaluation of any contracting also is essential, both to identify cor­rections needed within the states affected and to help develop expertise to guide any future con­tracting ventures. This is particularly true because of the reduced availability of programmatic data that, as discussed above, is otherwise likely when the state agency is displaced. In addition, strict limits should be im­posed on contractors’ ability to protect as proprietary information, or to destroy, the policies under which they administer the program and the files and data they compile in doing so.

To avoid the sort of loss to the federal or state governments that the failures of automated systems have caused in California, Florida, Michigan, and other states in the past, any con­tractor should be required to be bonded against possible bankruptcy, against the inability to pay any penalties assessed for non-compliance, and against any increase in the food stamp QC error rate. Re­quiring bidders to satisfy bonding companies that they will not produce losses like the $200 million Florida experienced in 1992 will provide the program with additional assurance about their com­pet­ence. Contracts also should include large penalties against contractors that cease to perform before the contract expires; these should be sufficient to compensate the state for the inflated salaries and rents it likely will have to pay to reconstitute its eligibility determina­tion system on short notice.

Privatization ought to be approached with the care appropriate for any dramatic change in the role and responsibilities of government. Contracting problems are inevitable. Beginning on a limited scale, and with an intense evaluation, will help identify those problems while they are still affecting rela­tively small numbers of people. They also will allow the government to negoti­ate broader contracts from a position of strength — while it is still free to decline to work with contractors — rather than with its own infrastructure already disbanded.