By Douglas J. Besharov and Peter Germanis

This article originally appeared in The Public Interest, Winter 1999.

WIC WORKS, perhaps better than any other government program in existence," Agriculture Secretary Dan Glickman recently declared. Former Health and Human Services Secretary Louis Sullivan agreed: "The WIC Program results in significant Medicaid savings that far outweigh the program's costs by a ratio of 3 to 1.... That is clearly an overwhelming return on a small national investment." Such statements testify to the extraordinary bipartisan support enjoyed by WIC. Officially known as the "Special Supplemental Nutrition Program for Women, Infants, and Children," WIC is a several billion dollar program, serving millions of women and children. The program's popularity stems from the widespread belief that research studies have proven that WIC "works." But, although some studies suggest real dietary and health improvements, the great benefits cited above apply only to WIC's prenatal program (only a small part of the total program). Even here the evidence suggests much more modest effects than WIC's boosters claim.

Some will argue that this exaggeration is harmless enough and is for a good cause: $ 5 billion a year in additional food and counseling services for low-income children and pregnant and breastfeeding or postpartum women. But overstating WIC's effectiveness prevents an honest appraisal of the program and stands in the way of redesigning the program to increase its impact on those who most need its combination of food packages and counseling.

Anatomy of a program

WIC was established in 1972 as a two-year pilot program partially in response to the 1969 White House Conference on Food, Nutrition, and Health. The conference report argued that nutritional deficiencies among low-income women and children threatened their health and led to higher medical costs. WIC seeks to improve the diets (and, thus, the health) of low-income pregnant, breastfeeding, or postpartum women, as well as their infants and children (up to age five).

Benefits. WIC is a three-part program: (1) vouchers to purchase specific high-nutrition foods to supplement diets, (2) nutritional and health counseling, and (3) referrals to healthcare and social-service providers.

WIC's food "packages" are meant to supplement the recipients' diet, not to meet their entire food or dietary needs. High in protein, calcium, iron, and vitamins A and C, they are designed to provide the nutrients often lacking in the diets of the populations targeted by WIC. Packages typically include iron-fortified infant cereal and formula, juice, milk, cheese, eggs, peanut butter, and beans. Many WIC agencies tailor food packages to meet the nutritional deficiencies of individual WIC clients. For example, if it is known that a WIC participant has high cholesterol, the standard food package may be modified accordingly.

In 1996, the average WIC food package was worth about $ 45 per month, costing the government only $ 31 because of manufacturers' rebates on infant formula. This is across all WIC recipients, however. The total value of the two WIC food packages provided to a postpartum mother and her newborn can exceed $ 100 a month.

Local WIC agencies must spend at least one-sixth of their administrative funds on nutrition education. And WIC recipients must be offered at least two nutrition education sessions each time they are certified. These may be either one-on-one counseling sessions or group classes, designed to teach about the importance of good nutrition and its relationship to good health. Participants are also instructed on how to deal with their own particular nutritional risks and those of their children. Pregnant women are encouraged to breastfeed, for example, unless it's not medically advisable. (In general, these sessions last only about 15 minutes and are voluntary.)

WIC also provides referrals for preventive health care and social services. For example, WIC counselors might send pregnant women to prenatal-care or smoking-education classes, as well as drug- and alcohol-addiction and family-planning services. However, referrals may be a relatively less important aspect of the program because many WIC recipients are already participating in other programs (and may even have been referred from them).

Eligibility. Under federal rules, eligibility for WIC is based on low income and "nutritional risk." Income eligibility is set at family incomes up to 185 percent of the poverty income guidelines ($ 29,693 for a family of four in 1998). (Recipients of welfare, food stamps, and Medicaid automatically meet the program's income criteria.) Nutritional risk is a broad concept including medical conditions like anemia and low weight; a mother's age, history of pregnancy complications, or poor pregnancy outcomes; and inadequate diets.

The process of determining nutritional risk is inexact, if not permissive. According to the Institute of Medicine's Committee on Scientific Evaluation of WIC Nutrition Risk Criteria, some states use "generous" cut-off points and "loosely defined risk criteria." According to the report, there are "serious gaps on the evidence" for some of the risk criteria. And the tools to measure them are unreliable. Moreover, some of the criteria seem to be quite loosely applied, being dependent on the judgment of "street-level bureaucrats" who can qualify someone who is marginally at risk or not at risk at all, as Peter Rossi points out in Feeding the Poor: Assessing Federal Food Aid. Although the extent of these problems is not known, the Department of Agriculture (USDA), which administers WIC, has revised its guidelines to improve the process for evaluating nutritional risk.

Size and scope. WIC is not an entitlement program, unlike other federal nutrition programs such as food stamps. The number of women and children served in a given year is established by the amount Congress appropriates, with states free to add funding.

Always a popular program, WIC has grown rapidly. When it was permanently authorized in 1974, it served only 88,000 women and children, at a cost of $ 10.4 million. Participation increased to nearly 2 million in 1980, at a cost of $ 725 million, and to 4.5 million in 1990, at a cost of $ 2.1 billion. By 1997, participation reached 7.4 million, at a cost of $ 3.7 billion. If one also includes the rebates that infant formula manufacturers are forced to give, the program provides another $ 1.3 billion worth of food, for a total value of about $ 5 billion.

Because of the relatively high income cut-off (almost twice the poverty line), loose interpretations of "nutritional risk," and the fact that the middle class is having relatively fewer children than in the past compared to lower-income Americans, a surprisingly large proportion of Americans receive WIC benefits: nearly half of all infants, one-quarter of all children aged one to five, and the same proportion of pregnant women. (Another 10 percent of American children are eligible for benefits but do not receive them because WIC is "not fully funded.")

WIC "works"?

All of us would like WIC to be as successful as claimed. And it makes common sense that providing food packages and nutritional counseling ought to improve diets. The actual, scientific evidence, however, suggests that WIC's effectiveness is more modest than the three-to-one benefit-cost ratio commonly cited.

From a scientific standpoint, the preferred approach for measuring the effectiveness of WIC - or any social intervention - is a randomized experiment, where individuals eligible for WIC are randomly assigned to a treatment group (which receives WIC) and a control group (which does not). Random assignment assures that members of the treatment group and the control group are comparable, so that any difference in subsequent outcomes could be attributed to program participation. This is a particularly important issue when assessing a program like WIC, where participation is voluntary and some unobserved factor, such as parental motivation, may affect both participation and outcomes - thus creating the danger of "selection bias" in the findings. For example, pregnant women who voluntarily enroll in WIC may be more concerned about their babies than those who do not, and their better birth outcomes may reflect this higher level of concern more than the program's nutritional information or food supplements. Conversely, the bias could be in the opposite direction. If WIC recipients are at greater nutritional risk than nonparticipants, then comparisons to nonparticipants may understate the positive effects of WIC.

However, denying WIC benefits to eligible participants - necessary for a randomized experiment - would raise ethical and practical problems. Hence, even the best WIC studies have been forced to rely on statistical comparisons between those who received WIC benefits and those who did not. This creates unresolvable selection-bias questions in these studies, sharply undermining their findings.

Keeping this in mind, we will review studies of WIC's major target groups: postpartum and breastfeeding mothers, infants, children aged one to five years old, and pregnant women.

Mothers and infants

The only study that we could find on WIC's effects on postpartum mothers - about 11 percent of WIC participants - was conducted by Bette Caan of the School of Public Health, UC Berkeley and her colleagues using California administrative data from the 1980s. They found relatively small effects: For infants born to mothers who received WIC postpartum benefits for five to seven months between their two pregnancies, the mean birth weight was about 4 percent higher than for infants born to women who received postpartum benefits for less than three months or not at all. (Women in both groups received WIC prenatal benefits during both pregnancies.) However, the validity of even this small positive impact is undermined by questions of selection bias: The better functioning mothers presumably sought the WIC benefits for a longer period. Another problem is that about 20 percent of the original sample was dropped due to missing data (which could introduce selection bias if, for example, the dropped cases have systematically different birth outcomes from those that remained).

Some observers have suggested that the free infant formula that WIC provides may reduce breastfeeding, ordinarily considered better for newborns. For, although WIC encourages women to breastfeed, the free infant formula it provides (with a market value of over $ 70 a month) may discourage breastfeeding. In response, beginning in 1993, USDA introduced an enhanced food package for breastfeeding mothers who forego the infant formula food package. Still, as Urban Institute researchers concluded in 1994: "It is not clear whether WIC promotes or hinders breastfeeding."

About 25 percent of WIC participants are infants, and again there is little research evidence of effectiveness. The National WIC Evaluation, conducted in the early 1980s by researchers at the Albert Einstein College of Medicine in New York City and the Research Triangle Institute in North Carolina, is the most comprehensive examination of WIC's impact on infants (as well as children). It found that infants participating in WIC did not have higher caloric intakes than those in the comparison group, although they had higher mean intakes of certain nutrients, especially iron and vitamin C. WIC infants also had lower mean intakes of calcium, reflecting the fact that they were more likely to consume infant formula than whole milk. They were also somewhat more likely to have a regular source of medical care, to be immunized, and not to be anemic than similar low-income children not on WIC.

However, as Peter Rossi concludes, in his comprehensive study of federal food programs, the National WIC Evaluation was "plagued by a number of technical problems concerning the suitability of the comparison group used as well as severe response rate problems for both the WIC and comparison samples. Accordingly, the findings derived cannot be regarded as anything more than suggestive."

The children

Just over half of all WIC participants are children ages one to five, making this the largest target group served by the program. But here, too, the body of research on effectiveness is disappointingly sparse, with a few studies concluding that WIC reduced iron-deficiency among poor children and modestly increased the intake of selected nutrients.

Again, the National WIC Evaluation is one of the leading studies. It found that children participating in WIC, like infants, did not have higher levels of caloric intake than a comparison group, but they did have higher mean intakes of certain nutrients, especially iron and vitamin C. The strongest positive dietary effects were for the most disadvantaged children, including those who were very poor and from female-headed households. WIC children were also more likely to be immunized and to have a regular source of medical care than similar low-income children not on WIC. However, as noted above for infants, these findings should be viewed as nothing more than "suggestive" because of technical problems related to the evaluation.

A more recent study used data from the 1989-1991 Continuing Survey of Food Intake by Individuals to compare preschoolers one to five years old in households with incomes below 130 percent of the poverty level participating in WIC to a comparable group of preschoolers not in WIC. After attempting to control for differences in socioeconomic characteristics and possible selection bias, the study reported that WIC had statistically significant positive effects for 10 of the 15 nutrients examined, with particularly significant increases in the intake of iron and zinc (often lacking in the diets of WIC children). Although the diets of most WIC children already exceeded the recommended daily allowances for many nutrients, these findings suggest significant dietary improvements for particular groups.

Once again, however, this research suffers from the generic weaknesses of comparison groups. Moreover, the considerable differences in the socioeconomic characteristics of WIC participants and the comparison group of eligible nonparticipants make it particularly difficult to judge how successfully the study controlled for these differences.

Ray Yip at the Centers for Disease Control (CDC) and his colleagues took a different approach. Using data from the CDC's Pediatric Nutrition Surveillance System (PNSS), they examined the incidence of anemia among children aged six to sixty months, who were enrolled in public-health programs (primarily WIC). The analysis was limited to six states that consistently participated in the PNSS during the time period examined. Anemia among children six to sixty months of age declined steadily from 7.8 percent in 1975 to 2.9 percent in 1985. Other studies report similar decreases.

The decline in anemia could be due to several factors. First, because the data are limited to children covered by programs such as WIC, this decline could also have reflected a change in the composition of the children covered by the programs. For example, as WIC's funding and enrollment have expanded, the children in the program have probably become less disadvantaged. If the newly eligible children had lower rates of anemia than the earlier WIC recipients, they would have lowered the average rate of anemia, even if the WIC benefit itself had no effect.

In addition, this reduction may be part of a more general downward trend in iron-deficiency anemia, caused by such factors as an increase in breastfeeding, the substitution of iron-fortified formula for unfortified formula and cow's milk, and general nutritional education efforts. WIC may also have contributed indirectly to this trend through its required iron fortification of many foods. As Barbara Devaney, who conducted some of the major WIC studies for Mathematica Policy Research, observes, "Much of these foods that are on the shelves of supermarkets are iron-fortified and affect the diets of nonparticipants as well as program participants." If this is the real explanation, a more direct way of achieving WIC's putatively beneficial effects on anemia would be simply to require manufacturers to fortify infant formula, cereal, and bread.

But this is speculation. No study has successfully isolated WIC's impact from the changing socioeconomic characteristics of recipients, nor from the apparent secular decline in anemia. Moreover, the existing studies are limited to a small number of states not representative of the nation as a whole.

More importantly, the practical significance of these modest findings is unknown. If there had been a socially significant reduction in anemia, one would expect, for example, to see it reflected in a reduction in the behaviors associated with anemia. But, as Devaney comments, "Little is known about the long-term effects of WIC on improving behavioral and cognitive development, outcomes that would presumably result from better iron nutrition status." In sum, the evidence about the effectiveness of almost 90 percent of the WIC program is limited and suggests only small to modest effects.

Pregnant women and newborns

The purported three-to-one-savings calculation, cited by Louis Sullivan and so many others, comes solely from research on one of the smallest parts of WIC, the program for pregnant women (12 percent of participants). In 1992, the U.S. General Accounting Office (GAO) reviewed 17 "methodologically strong" evaluations. It estimated that prenatal WIC participation resulted in a 25 percent reduction in low-birth-weight births (under 2,500 grams) and a 44 percent reduction in very low-birth-weight births (under 1,500 grams). As a result, according to the GAO, each dollar spent on WIC for pregnant women saved $ 3.50 (over an 18-year period) in Medicaid and disability payments, paid by federal, state, and local governments, and in medical-care costs borne by private healthcare providers.

A deeper look at this body of research, however, suggests that the GAO's conclusions should have been considerably more tentative. All but one of the studies reviewed by the GAO were statistical comparisons of WIC participants with nonparticipants, and attempts made to control for other factors that influence birth outcomes varied in their success.

One of the best of these studies was conducted by Devaney and her colleagues. In five states, they matched 1987-88 Medicaid and WIC records and then compared the outcomes of Medicaid recipients receiving WIC services with those not receiving WIC. After applying statistical controls for identifiable demographic and parental-care characteristics that could also affect birth outcomes and Medicaid costs, the researchers found that WIC participants had fewer premature births, and their newborns had 1.6 percent to 3.8 percent higher birth weights (depending on the state). Moreover, their estimates suggest a striking 28 percent decrease in the number of newborns with "low" birth weights (under 2,500 grains), and an even larger, 59 percent decrease in the number of newborns with "very low" birth weights (under 1,500 grams).

Despite these encouraging results, Devaney and other researchers in this field have been forthright in describing the limitations of their findings, most notably because of simultaneity bias and selection bias. Independent of any program effect, the longer a woman is pregnant, the more likely it is that she will enroll in WIC - because she has more opportunities either to learn about the program or to enroll. Thus positive birth outcomes for women who enroll in WIC late in their pregnancy are more likely due to the length of their pregnancy than to the effects of WIC. Failure to account for this "simultaneous" effect exaggerates WIC's impact.

Several studies have tried to address the simultaneity problem. When Devaney and her colleagues, for example, attempted to control for gestational age, the estimated Medicaid savings declined by about 55 percent. Similarly, using national data, Anne Gordon and Lyle Nelson of Mathematica found that accounting for simultaneity bias resulted in substantial reductions in otherwise comparable birth effects. (These corrections may, however, result in an understatement of WIC's impact because they tend to wipe out any positive effects on increasing pregnancy duration.)

Selection bias is also a serious issue in all these studies - as most responsible researchers in the field acknowledge. Two studies made especially extensive efforts to control for selection bias. In Gordon and Nelson's study, WIC's estimated effects on birth weight reversed, becoming "large and negative." The researchers did not consider these findings credible, however, explaining that it "is simply not possible that participation in WIC reduces newborn birth weight or related outcomes." Perhaps. But the consistency and strength of their results strongly suggest that selection bias is, indeed, exaggerating program impacts. Mark Lopez, an economist at the University of Maryland's School of Public Affairs, reviewed their efforts and concluded that the selection-adjusted results "are so consistent in sign, and are statistically significant, the unadjusted WIC results ... are most likely biased up."

Another study that attempted to control for selection bias provides further evidence that WIC's impact may vary by subgroup. Michael Brien of the University of Virginia and Christopher Swann of Mathematica used the same data source as Gordon and Nelson, but a different sample and methodology, to analyze birth outcomes for whites and blacks separately. Their unadjusted results indicate that WIC increased birth weight by 3.8 percent for blacks and 2.7 percent for whites. After correcting for selection bias, however, birth-weight effects increased to 13 percent for blacks and became "not statistically significant" for whites. (They also found that WIC had no statistically significant effect on the incidence of low birth weight or on infant and neonatal mortality.) Of course, race is likely a proxy for various unmeasured factors among participants or programs - for example, marital status at first birth (unwed teen parenthood), extreme poverty, or social disorganization.

Thus the popular claim that "WIC works" is based on research results for only a small portion of the program. It does not adjust for simultaneity and selection bias, and it ignores the more modest effects for the rest of WIC. A more appropriate way to describe WIC's effects is in terms of a range. WIC for pregnant mothers has perhaps zero to substantial impacts on infant mortality, prematurity, and birth weight. But the rest of the program (for infants, children, and postpartum and breastfeeding mothers) has small to modest impacts on anemia and nutrient intake. There is also a glimmer of evidence that WIC's beneficial effects are concentrated among the most needful recipients. These realities should not be surprising, given a whole body of research in related fields documenting the limited impact of nutritional supplements, counseling, and prenatal care on any but the most "severely undernourished."

Policy implications

How should policy makers respond to the likelihood that WIC's effects are modest and probably concentrated among the most disadvantaged recipients? Certainly not by continuing to expand WIC's coverage to progressively less needy families. But, just as certainly, not by simple-mindedly cutting or abandoning the program. The problems that WIC addresses are serious and require attention. In an age when so many government programs for the poor seem to have no effect, and may even make things worse, WIC's possible beneficial effects should not be slighted. (Moreover, in many inner-city neighborhoods, the program's one hundred dollars' worth of food has become an important component of the economic safety net for mothers with newborns.)

Unfortunately, no one really knows how to make WIC more effective. Partly because it has been so politically incorrect to acknowledge the limitations of these research findings, the necessary research and planning have not been conducted. We should systematically explore what might make it stronger, expanding those elements that do and discarding those that do not. Here are the ideas that seem most important to pursue, either under direct statutory grants of programmatic discretion to states or through a case-by-case waiver process similar to that which preceded the passage of the 1996 welfare reform law (and would also require new legislation).

1. Target more WIC resources to the most needful families. About 50 percent of all infant formula sold in the United States is purchased with WIC dollars. One must question why a remedial program like WIC is now provided so broadly. Implicit in current federal rules on the provision of WIC benefits is the judgment that nearly 40 percent of all American children are at "nutritional risk" - and that their mothers need corrective counseling. Since household incomes vary from state to state, in six states as many as 60 percent or more of all newborns receive benefits. That can't be right.

As the research cited above suggests, WIC's positive effects are probably concentrated among its most disadvantaged recipients. Federal rules, however, require that all recipients in the same target group receive basically the same set of WIC benefits, notwithstanding their differences in need. For example, there is a maximum for the amount of food that can be provided. Even though states administer the program, and are presumably in the best position to judge whether various program adjustments are necessary, they cannot make major changes in services to meet the palpably different needs of clients. Why not allow states, at least on an experimental basis, to increase the size of the food package for the most needful. families? More importantly, why not allow states to expand and re-orient WIC's counseling component?

2. Intensify counseling services and make them more directive. To the extent that WIC is successful, an important explanation could be the nutritional and health counseling that clients receive, especially given the relatively small size of the food package for all but newborns. And that may be the result of the directive or authoritative content of WIC's counseling.

A group of projects recently funded by the Centers for Disease Control (CDC) demonstrates how much WIC can change parental behavior, in this case raising immunization rates. In New York City, for example, six volunteer WIC sites were randomly assigned to one of three immunization interventions designed to increase vaccination coverage among preschool children who were eligible for measles vaccination. In two sites, WIC parents were escorted to a clinic in the same location where children could be immunized in an "express lane"; in two other sites, parents were required to return monthly for WIC coupons, rather than every two months, as is the normal schedule, until they had the child immunized; and, in the final two sites, parents were offered a vaccination assessment and education and referral services ordinarily offered in the WIC office. The children at the escort sites were five times more likely to be immunized than children at the referral sites. Similarly, the children at the sites where parents could be required to return each month for their coupons were three times more likely to be immunized. In short, WIC benefits can be used to change recipient behavior for the better.

One possible approach would be to allow exceptions to federal rules that currently prohibit tying WIC benefits to the acceptance of nutritional instruction or engaging in any other responsible behavior. Right now, pregnant smokers cannot even be required to attend smoking-cessation classes, despite the well-known connection between smoking and low birth weight and other adverse birth outcomes. States should be permitted to experiment with such behavior-related rules, in the same way that they can now condition welfare on specific behaviors.

Another possibility would be to test, on a large-scale basis, the nurse home visitor program developed by David Olds at the University of Colorado. His Prenatal and Early Childhood Nurse Home Visitation Program is a highly structured program that utilizes an authoritative or directive approach to counseling low-income, first-time mothers. (Participation is voluntary.) As opposed to interventions that are nonjudgmental, the nurse home visitation projects deliver clear behavioral messages by public-health nurses. As the Washington Post reported, the traditional way of providing family planning says, "If you want to avoid a second baby, here's a condom and how to use it." The authoritative approach says, "You shouldn't have another baby and here are ways to prevent it."

A randomized experiment in Elmira, New York, found that the nurse home visitors achieved a 25 percent drop in smoking by the end of the pregnancy, a 75 percent reduction in premature births among pregnant women who had smoked, and large birth-weight increases for babies born to young teen mothers (nearly 400 grams for mothers aged 14 to 16). In addition, 15-year follow-up findings indicate almost a 31 percent reduction in the subsequent childbearing for low-income, unmarried mothers (1.1 versus 1.6 subsequent births). Verified cases of child abuse and neglect were 79 percent lower, drug and alcohol problems 44 percent lower, arrests among the mothers 69 percent lower, and welfare use 33 percent lower. As a result, the program led to large savings in government spending. (Replications of this study are showing similarly impressive results in Memphis, Tennessee, and Denver, Colorado.)

3. Allow exceptions to federal limits on spending for WIC services. In essence, federal rules now limit "administrative" expenses - which include counseling services - to a fixed amount, averaging about $ 11 per participant per month in 1997. As a result, states are essentially precluded from using WIC funds to provide enhanced (and expensive) services - even to the most needful families.

Redirecting existing WIC funds would come at the cost of not serving many families who would otherwise receive benefits. But that is not as unreasonable an option as it may seem. Just about all eligible pregnant women, infants, and postpartum and breastfeeding mothers are now enrolled in WIC. The beneficiaries of recent expansions have primarily been children ages one to five who are at relatively less nutritional risk. As researchers at the Urban Institute pointed out in 1994, "As the Program expands, much of the new caseload will come from increases in participation by children and postpartum women, the two categories with lowest priority now."

Consider how states have used the savings from their cost-containment programs. As the share of all infant formula purchased by WIC increased, it became apparent that states could use their considerable market power to obtain discounts from formula manufacturers. The rebates they receive now amount to $ 1.3 billion, one-third of actual WIC appropriations.

WIC's funding rules have forced states to use these savings to expand participation, generally to those in lower nutritional risk categories, rather than to improve services. There are legitimate reasons for some limits in a program as large and diverse as WIC. But forcing states to add more and more families to the program is not one of them. Indeed, in so doing, we have lost an opportunity to improve WIC's effectiveness-by directing enhanced services to the most needful.

4. Test alternative service configurations. WIC is a program of the United States Department of Agriculture (USDA), in keeping with its focus on nutrition services and health objectives. But WIC works through state health departments, which in turn fund local health departments, hospitals, public-health clinics, and community health centers.

This elaborate system was conceived in the early 1970s. It went into effect before Medicaid coverage was expanded to include, among others, the same people covered by WIC, and long before Medicaid recipients were universally enrolled in managed-care programs. Thus integrating WIC into Medicaid managed-care systems merits serious consideration. This would avoid costly duplication of administrative services and would make it easier for WIC participants to avail themselves of the services WIC now refers them to, such as prenatal care.

There is, of course, a possible downside to integrating WIC into Medicaid managed-care systems. Managed-care providers, such as HMOs, might eagerly accept adding WIC services to their own programs. Especially if they were to receive additional funding, WIC services could have a positive impact on pregnancy and child outcomes. An Urban Institute study reports that in "at least a couple of states managed care plans and private physicians have already asked about offering WIC services in their offices." We should, nevertheless, be mindful that HMOs have been criticized for neglecting preventative activities. Providers who believe that a substantial portion of their clients will shift to other firms before the long-term costs of failing to provide preventive services are realized will be tempted to skimp on them. Hence, any integration of WIC with managed care should be pursued with caution.

5. Evaluate any programmatic changes. While the foregoing ideas seem consistent with research findings and common sense, they are untried and, therefore, should be carefully evaluated. Fortunately, since they generally involve either maintaining the current level of services to particular clients or enriching them, they could be subject to randomized experiments, the preferred method of evaluation. For example, an experimental design could be used to compare the impact of WIC's current nutritional education and counseling services to a more comprehensive and authoritative intervention - perhaps in the form of a nurse home visitation program.

Myth-making versus reform

Our analysis of the research challenges the conventional wisdom that WIC is a uniquely successful program. In all probability, many recipients benefit from its combination of food packages and counseling services. However, many claims made on its behalf exaggerate its proven effectiveness. We realize that some will label us "enemies of the poor" for our frank treatment of a program that has become sacrosanct to many journalists, policy makers, and politicians. But, unless WIC is assessed honestly, it is difficult to see how a program with such admirable purposes can ever be improved. And wouldn't that hurt poor children even more?

DOUGLAS J. BESHAROV is a resident scholar at the American Enterprise Institute and a professor at the University of Maryland School of Public Affairs. PETER GERMANIS is assistant director of the University of Maryland's Welfare Reform Academy.