Printer Friendly

The emergency food relief system: an empirical study.

One consumer market that has failed to attract much attention from consumer economists is the market for free food. There is evidence that the use of emergency food relief (EFR) in the United States--primarily through soup kitchens and food pantries--grew dramatically during the early 1980s and has continued at high levels (Brown 1987; Phisicians Task Force 1985; U.S. Conference of Mayors 1986). Not only has use continued at high levels, but EFR appears to be used by a much broader cross section of the population than the stereotypical single homeless man (Campbell et al. 1987; Hunger Action Network 1987; McGrath-Morris 1988; Rauschenbach et al. 1990; Weber 1988).

The increased role of soup kitchens and food pantries in providing support to the poor has raised two sets of public policy concerns: (1) why is there a need for an EFR system and (2) does the EFR system adequately meet the demand for its services from recipients? Relevant for the first set of policy concerns are the following questions: Are there holes in the "social safety net" such that needy persons do not qualify for public assistance and must resort to other forms of assistance? Are procedures for obtaining public assistance so onerous to some that they prefer alternative means of support? To what extent do users of EFR have regular sources of income (e.g., from employment, Social Security/pensions, or public assistance), but find their income insufficient to cover living expenses; are these insufficiencies chronic or episodic? Are all users of EFR truly in need of food assistance, or are some simply freeloaders?

Results from surveys of EFR users suggest that eligibility restrictions for public assistance programs; obstacles to using public benefits among the eligible; homelessness; and insufficiency of wages and public assistance benefits all play a role in the demand for EFR (Hunger Action Network 1987; Legal Action Center for the Homeless 1987; Physicians Task Force 1985; Rauschenbach et al. 1990). In addition, observed use of soup kitchens throughout any given month has been shown to be related to the payment schedules of public assistance benefits and food stamps (Thompson et al. 1988). No studies have been found that estimate demand for EFR or that compare those poor persons who use eFR with those who do not.

The second set of policy concerns focuses on the ability of the EFR system to meet demand for its product. Are there areas where needy persons lack access to EFR? If so, what factors affect the ability of the EFR system to meet this demand? Finally, what are the consequences when food assistance is not available? While structure and operation of soup kitchens and food pantries have been described (McGrath-Morris 1988; Weber 1988), the questions posed here have not been fully addressed.

A theoretical model of the EFR system is developed in this paper. From this model, reduced form equations of the supply of soup kitchen and food pantry meals are specified and estimated using data from 57 New York State counties outside of New York Cith (hereafter referred to as upstate New York). These equations provide insights into the performance of the EFR system as well as into aspects of the demand for EFR meals by recipients.

DESCRIPTION OF THE EFR SYSTEM IN NEW YORK STATE

Besides federal government entitlement and food assistance programs (e.g., food stamps, WIC, school breakfast/lunch programs), free food is distributed to needy individuals primarily through soup kitchens, food pantries, and surplus commodity distributions (under the federal Needy Family Food Distribution Program). This study is limited to soup kitchens and food pantries.

Soup kitchens provide prepared meals for individuals. While some are associated with shelters and are restricted to shelter residents, most soup kitchens provide meals on a first-come, first-serve basis. Few have eligibility requirements or limit frequency of use. Soup kitchens are less prevalent than food pantries and exist mostly in urban areas.

Food pantries provide food to individuals for preparation in their homes. Some also issue vouchers that recipients can use to purchase food at local food stores or restaurants. Food pantries may have regularly scheduled hours of operation, although many operate informally, providing food by request. Food pantries provide a quantity of food that varies with family size and available supply. The number of days of food distributed per visit varies considerably, although a four-day supply is typical (Weber 1988). Eligibility requirements and limits on frequency of use are often employed by food pantries. Because food pantries provide food that requires preparation, homeless persons are not prevalent among users. Families are somewhat more prevalent among users of food pantries than among users of soup kitchens (Campbell 1985).

In 1986, an estimated 372,137 persons received food each month from 1,273 New York food pantries. Of these, 959 pantries operated in upstate areas, serving 149,495 persons per month or 40 percent of the statewide total. The 286 soup kitchens statewide served 557,085 meals per month. Most of these operated in New York City. Ninety soup kitchens operated in upstate areas, serving 143,990 meals per month or 26 percent of the statewide total (N.Y. Dept. of Health 1988).

Funds and food used by EFR programs are obtained from multiple sources. Nearly all obtain resources from private donations (N.Y. Dept. of Health 1988). In New York State a network of regional food banks serve EFR programs. Food banks receive unsalable and surplus foods from food companies and stores and sell them to food pantries and soup kitchenson a poundage basis, typically around ten cents per pound. Although operated by private nonprofit organizations, regional food banks in New York State have received some state financial support since 1984. Purchases from food banks are reported by 46 percent and 37 percent of upstate New York soup kitchens and food pantries, respectively (Campbell 1985).

EFR sites are community based. Fundraising operates mostly at the local level. Food banks operate at a regional level. Seven food banks serve the entire state. Corporate donations of surplus and unsalable food to food banks may not be solely motivated by altruism, but also by the desire to minimize costs by taking advantage of federal tax benefits (Birnbaum 1982).

The public sector directly supports the EFr system in several ways. At the federal level, the Job Stimulus Bill of 1993 established a program that channels federal funds to support EFR and emergency shelters through local voluntary agencies, typically the county United Way agency. This program is commonly referred to as the FEMA program after the Federal Emergency Management Administration which administers it. Allocations to each county are made on a formula basis, advised by advisory boards on the state and federal levels composed of representatives of charitable organizations. The Temporary Emergency Food Assistance Program (TEFAP) distributes surplus commodities to localities through local voluntary agencies. (1) Leftover commodities not distributed by these agencies can be given to local food pantries and soup kitchens.

New York State provides support for EFR through the Supplemental Nutritional Assistance Program (SNAP). Funds are distributed to regional food banks on a formula basis. Individual EFR programs then apply to the food bank for assistance to purchase food. (2) Finally, some local governments provide modest levels of support.

A MODEL OF THE EFR SYSTEM

EFR programs can be seen as providing benefits to two populations. Most obvious are the recipients of the food assistance. However, EFR programs, as with other charitable organizations, provide social goods that convey benefits to the general population, and specifically to those who make donations. EFR can be seen as a form of in-kind income transfer provided to (presumably) needy persons that supplements similar activities performed by governmental organizations (e.g., through the Food Stamp Program). The government's advantage over charitable organizations in serving as a conduit for transfers is that it can levy taxes and avoid free rider problems. The existence of EFR programs (and other charitable organizations) however can be largely explained by the existence of a segment of the population which does not control political decisions and whose strong preferences for redistributive activities lead them to voluntarily supplement activities of government by making contributions to charity (Rose-Ackerman 1981; Weisbrod 1975). (3) Contributions to charitable organizations therefore provide benefits to donors (e.g., a good feeling that they have done their part to help the poor). In the case of EFR programs, contributions may take the form of money, goods, or labor.

EFR programs then can be seen as facing two sets of demands. One is the demand for food assistance by recipients. This will be referred to as "recipient demand." The other is the demand for the provision of food assistance to the poor by EFR programs on the part of donors. This will be referred to as "donor demand."

Observations on the operations of EFR programs lead to a general conclusion that recipient demand is nearly always greater than supply of food assistance. Support for this assertion is provided by the widespread use of devices by EFR programs to ration the demand for food assistance by recipients. (4) Consequently, the quantity of food assistance provided by EFR programs is largely determined by their resource base, which is primarily determined by donor demand. (5) A number of factorsare hypothesized to influence the donor demand within a given market area. These include the (1) average income of potential donors, (2) price of giving, (3) relative size of the population of potential donors, (4) average preferences of the population of potential donors, and (5) perceptions of demand for food assistance from EFR programs by the poor.

Weisbrod and Dominquez (1986) were the first to argue that at the market level, private donor demand can be cast in a manner similar to that of other conventional goods and services where demand is a function of income levels and the price of donating. The price of donating to the EFR system can be thought of as the quantity of output produced per marginal dollar donated (or equivalent value of goods or services donated). Thus, donations to those EFR programs whose costs are lower at the margin (due to scale economies or lower input prices) will be regarded as more productive and have a lower price. Furthermore, because contributions to charitable organizations are tax deductible, the average price of nonlabor donations will decrease as average incomes (and subsequently marginal tax rates) increase. (6)

Margolis (1982) and Young (1989) put forth "fair share" models of charitable giving that provide further assistance in specifying a donor demand equation. These models posit that individual decisions regarding the level of charitable giving not only depend on the individual's preference for the level of the public goods being provided, but also an assessment of the individual's fair share contribution toward its provision. Therefore donor demand at the market level is hypothesized to be a functioin of the relative size of the population of potential donors to that of potential recipients. However, the direction of the relationship between the relative size of the donor population and market level donor demand is ambiguous. On one hand, greater numbers of potential donors should increase aggregate donations, ceteris paribus. On the other hand, individuals' perceptions of their fair share contribution will decrease as the number of potential donors increases.

While preferences for donations to EFR programs across market areas are assumed to vary with differences in the demographic characteristics of the population of potential donors, these preferences are also influenced by perceptions of the size of recipient demand in the area. An increase in the demand for food assistance by recipients, for example due to a deep recession, will increase the marginal utility of a contribution (Margolis 1982). Evidence that this occurs is given by the rapid increase in the size of the EFR system nationwide during the early 1980s in response to a deep recession and tightened eligibility requirements for government transfer programs. More recently, the significant increase in contributions to the Red Cross in 1980 for victims of Hurricane Hugo and the San Francisco area earthquake provides a similar example in a different charity market.

Given the rationing devices utilized by EFR programs, recipient demand for EFR is not directly observed. However, perceptions of this demand by potential donors to EFR programs are likely to be influenced by the number of requests for free food received by EFR programs as well as general perceptions of the magnitude of poverty in the area. These perceptions are influenced by news reports as well as the fundraising efforts of EFR program managers. The underlying recipient demand for food assistance from EFR programs is likely to be a function of the size of the poverty population, price levels, the coverage of public assistance programs, demographic characteristics of the poverty population assumed related to feelings of stigma and tastes for this form of charity, and the degree EFR programs are accessible to the target population.

DATA AND METHODS

Between November 1984 and February 1985, Cornell University researchers, under contract with the New York State Department of Health, conducted a census of all soup kitchens and food pantries in the state (Campbell 1985). EFR programs identified were asked about their operations and clientele in a telephone interview. A total of 85 soup kitchens and 882 food pantries were identified in upstate areas. Shelters serving food exclusively to its residents were not included in the EFR census. Soup kitchens associated with shelters that also served nonresidents were included in the census; usage data from these sites included only meals served to nonresidents. Programs in New York City are not included in this analysis because the magnitude of the market for EFR there differs significantly from that in the rest of the state and because of data quality and comparability problems.

The 57 upstate counties are diverse in size, population, urbanicity, and economic base. Upstate counties include several which serve as suburban areas to New York City, seven that include central cities of SMSAs, and many that are largely rural. Data from the EFR census were supplemeted by county level data from the 1980 U.S. Census, New York State agencies, and other sources. As much of these supplementary data were available only at the country level, local market areas for EFR were assumed to conform to county boundaries. Population centers are typically near the middle of counties; therefore the number of users who cross county lines for EFR is probably small. Many EFR users have limited transportation opportunities, so actual market areas may be smaller than the entire county.

Limitations on available data prevent the estimation of structural equations. Instead, a reduced form equation for the market supply of EFR is estimated. Market supply ([Q.sup.s]) is specified as a function of a vector of variables hypothesized to be related to private donor demand ([Q.sup.d]) and variables indicating government support (G):

[Q.sup.s] = [Q.sup.s]([Q.sup.d], G) (1)

In one-half of upstate counties, there are no soup kitchens. Because the dependent variable in the soup kitchen equations is censored, soup kitchen equations are estimated using tobit. (7) All upstate counties have food pantries, so OLS is used in food pantry equations.

Variable Descriptions

Variable descriptions and data sources are summarized in Table 1. Table 2 contains descriptive statistics for variables used in the analysis.

Dependent variables

SKSERVED and FPSERVED are dependent variables. SKSERVED is the sum of the usual number of meals served per week reported by all soup kitchens within a county. FPSERVED is the sum of the usual number of households served per week reported by all food pantries in the county. Neither SKSERVED nor FPSERVED is a perfect measure of the quantity of EFR provided. SKSERVED does not differentiate between the size or quality of meals served by soup kitchens. FPSERVED does not account for variation in size of food packages provided by food pantries.

Price of donations

One variable is included to capture variations in the price of donations. There is a little reason to suppose that the average efficiency of EFR programs across counties varies. Consequently, the price of donations will vary only with the price of inputs. As most labor and capital for preparation of EFR meals are donated, food is the most important input that is purchased. The price of food from conventional sources is not likely to vary considerably across the upstate region. Nor does the price of food from food banks vary significantly across the six food bank regions that serve upstate New York. The greatest variation in food prices faced by EFR sites results from the fact that five counties are not served by food banks. A dummy variable is therefore constructed that indicates whether the county is served by a food bank (FDBK).

Income

Average income levels of the potential donor population in each market area are approximated with county median family income (MEDINC).

[TABULAR DATA OMITTED]

TABLE 2

Descriptive Statistics for Variables Used in Soup Kitchen and Food Pantry Use and Coverage Equations (N = 57)
 Variable Name Mean Standard Deviation
 SKSERVED 565,63 1,287.60
 FPSERVED 618.23 885.78
 SKSPVPOP 19.78 29.16
 FPSPVPOP 40.77 44.82
 PERRURAL 54.93 26.14
 SIZE 36,442.00 63,406.00
 NUMPVRTY 15,911.00 20,363.00
 WELFCOV 0.54 0.17
 FDSTPCOV 0.69 0.17
 FMR 303.07 51.85
 FAMPOOR 0.59 0.953
 SINGPOOR 0.17 0.049
 UNITPOP 10.68 11.98
 RICHPOOR 9.77 3.39
 MEDINC 18,986.00 3,255.20
 FEMA 46,633.00 93,511.00
 SNAP 8,253.40 11,672.00
 FDBK 0.912 0.29


Relative size of the population of potential donors

The relative size of the potential donor population is given by the inverse of the poverty rate, or the number of nonpoor county residents per poor resident (RICHPOOR).

Donor preferences

Donor preferences for EFR are captured by a single variable which is constructed by dividing total annual United Way contributions in the county by the nonpoor population (UNITPOP). Here, it is assumed that preferences for donations to EFR programs (possibly through United Way) are related to preferences for the range of charitable and social service activities typically supported by United Way agencies.

Perceptions of recipient demand

Perceptions of recipient demand by potential donors are assumed to be related to levels of actual recipient demand. As indicated recipient demand is not directly observed. However, a set of variables assumed to be related to levels of recipient demand is included. Several variables related to the size of the poverty population and the depth of their poverty are included. The number of persons living under the poverty line from the 1980 census (NUMPVRTY) is included. (8) Average income among the poverty population is likely to be related to coverage by public assistance and food stamp programs. WELFCOV is defined as the number of AFDC, SSI, and Home Relief (New York State's general assistance program) recipients divided by NUMPVRTY. FDSTPCOV is defined as the average number of food stamp recipients during 1984 divided by NUMBPVRTY. Although food stamps are provided to most public assistance recipients, eligibility extends beyond those who qualify for cash assistance programs. The precise relationship between WELFCOV or FDSTPCOV and average income within the poverty population is somewhat unclear. Poor people who do not participate in income assistance programs include those in households who fall under the poverty line but above eligibility limits for AFDC and Home Relief. They also include those who choose not to participate or are ineligible for reasons other than income. To the extent that the poor judge the stigma costs of receiving public assistance as similar to that of receiving EFR assistance, WELFCOV and FDSTPCOV may also serve as proxies for the average stigma cost that the county poverty population would associate with use of EFR. Consequently, there is no clear expectation as to the sign of coefficients on these variables.

Price data do not exist at the county level for upstate counties. One good for which price data exist is rental housing. The Department of Housing and Urban Development (HUD) constructs fair market rents in order to set subsidy levels under the Section 8 and Housing Voucher programs. The fair market rent for a unit of a given size is set at the 45th percentile of rental units meeting HUD housing quality standards. Lack of prices on other goods and services, including the important substitute to EFR meals--purchased food, is not likely to be terribly serious insofar as prices on most consumption goods do not vary substantially across upstate New York. Variation that does exist is likely to be related to the urbanicity of the county, which is included in the model. (9)

Although the food provided to needy recipients by EFR programs is free, use of EFR programs carry an implicit price associated with feelings of stigma and with the costs associated with travel to and from EFR programs. Two variables are included to capture demographic characteristics of the poverty populations in each county that are assumed to be related to the level of stigma costs that would be by potential recipients of EFR. The three public assistance programs for the poor operating in the state serve somewhat distinct groups. AFDC serves families with children, Home Relief primarily serves single persons and some married couples without children, and SSI serves the elderly and disabled poor. Demographic characteristics of the poor population receiving benefits in each county are captured by the proportion of public assistance populations receiving benefits under AFDC and Home Relief (FAMPOOR and SINGPOOR, respectively). The proportion of SSI recipients is excluded so asa to ensure the model was of full rank. The impact of the prevalence of elderly and disabled persons among the impoverished population is implied by the coefficients on FAMPOOR and SINGPOOR.

Two variables to capture variation in access costs are included. As public transportation opportunities are more limited and distances to the nearest EFR site are likely to be greater in rural areas, the percentage of county population that lives in rural areas is included (PERRURAL). The second variable is the population of the largest town or city in the county (SIZE). Public transportation opportunities generally improve with increased municipality size. In addition, SIZE may also provide a crude measure of the size of a spatially concentrated core of potential EFR users. This may be particularly important for the provision of soup kitchen meals. There is likely to be a minimum number of users below which the average cost of providing soup kitchen meals becomes unreasonably high. (10) Food pantries are able to operate efficiently at a much smaller scale of operation. Interpretation of the coefficients on PERRURAL and SIZE is confounded because urbanicity may also be related to social norms that affect stigma costs and to price levels.

Government support

Two variables are included to capture governmental support for the EFR system. FEM A is measured as fiscal year 1984-1985 county allocations under the FEMA program. Disbursements of funds to county EFR programs under the SNAP program are also included (SNAP).

RESULTS

Table 3 presents tobit and OLS results for soup kitchen and food pantry supply. Tobit coefficients are multiplied by the standard error of the estimate in order to obtain coefficients comparable to the OLS coefficients. Both models account for most of the variation in the supply of EFR. Most variables have coefficients of the same sign across the two equations, although coefficients in the soup kitchen equation are generally more significant. Differences in the results between these two equations may reflect differences between the operation of soup kitchens and food pantries, although it is also likely that FPSERVED is measured with considerably more error than SKSERVED resulting in larger standard errors on coefficients in the food pantry model. (11)

The coefficient on FDBK, the food dummy variable, is negative in the soup kitchen equation and positive in the food pantry equation, although neither coefficient is statistically significant. The expectation was that the sign on this variable would be positive. The only counties not served by food banks are five contiguous counties in the lower Hudson Valley. The negative sign on FDBK in the soup

[TABULAR DATA OMITTED]

kitchen equation suggests the possibility that some regional impact not captured by other variables is being picked up by this variable.

Contrary to expectations, median family income (MEDINC) is negatively associated with use, at the .10 level in both equations. As this variable represents the median incomeof all families in the county, poor and nonpoor alike, it is possible that this is not a good measure for the average income of potential donors to EFR. However, alternative variables, such as the percentage of county families with incomes of $35,000 or greater, are tried. These substitute variables consistently have signs suggesting that donations to EFR are inferior goods. This finding is consistent with the fact that most EFR programs are sponsored by churches; contributions to churches as a percentage of income decline rapidly at higher income levels (Jencks 1987). The coefficient on the ratio of nonpoor to poor persons in the county (RICHPOOR) is positive and statistically significant in the soup kitchen equation. This suggests that recipients of EFR are likely to be better served in counties with lower poverty rates. UNITPOP, the measure of preferences of potential donors for EFR, has a positive but insignificant coefficient in both equations. Variables measuring county FEMA and SNAP allocations consistently have insignificant coefficients.

NUMPVRTY is positively, but insignificantly, associated with the supply of both soup kitchen and food pantry meals. The insignificance of NUMPVRTY is surprising. One possible reason is that NUMPVRTY is highly correlated with another variable omitted from the models. Because NUMPVRTY is nearly perfectly correlated with the value of federal surplus food commodities distributed in each county (r = .98), the latter variable is not included in the models. Surplus food is likely to serve as a substitute for EFR. As a result, the coefficient of NUMPVRTY may reflect both the positive influence of the number of poor on recipient demand and negative influence of surplus commodity distributions on recipient demand. Unfortunately, this hypothesis cannot be tested with available data.

Public assistance coverage (WELFCOV) is positively associated with soup kitchen and food pantry use, although the coefficients are statistically insignificant. The measure of the coverage of the food stamp program, FDSTPCOV, is negatively associated with EFR use, although statistically significant at only the .10 level in the soup kitchen equation. This suggests that food stamps serve as a substitute for EFR use by some recipients. As indicated previously, the signs expected on these variables are ambiguous.

FMR, the indicator of the local price of housing, has a positive coefficient in both equations, although neither coefficient is significant. Coefficients on the two variables included to measure accessibility to EFR sites, PERRURAL and SIZE, have expected signs, although only SIZE is statistically significant. Soup kitchen use appears to be associated with urbanicity. The highly significant positive coefficient on SIZE is consistent with the hypothesis concerning the minimum efficient size needed for these facilities. The coefficient on SIZE may also reflect the fact that larger cities are likely to have multiple soup kitchens, increasing the potential for double counting individual users.

The variables for the composition of the poverty population, FAMPOOR and SINGPOOR, are positively and significantly related to soup kitchen use. Coefficients on these variables in the food pantry

[TABULAR DATA OMITTED]

equations are also positive, but not significant. These results imply that the elderly poor are less likely to use EFR than the nonelderly poor. There are several possible explanations for this result. In most counties SSI benefits are higher than those paid to other public assistance recipients. The elderly may lack transportation opportunities, possibly because they are more likely to live in rural parts of counties. Finally stigma costs associated with use of EFR programs may be greater for the elderly poor than the nonelderly poor.

Explication of the determinants of supply provides only indirect information about factors which explain differences across counties in how well the needy population is served by EFR. In order to address this question, the model is respecified using new dependent variables. SKSPVPOP is defined as SKSERVED divided by NUMPVRTY. Similarly, for food pantries, FPSPVPOP is FPSERVED divided by NUMPVRTY. Independent variables remain unchanged. The tobit and OLS results of these equations are in Table 4.

Generally, signs on coefficients are similar to those found in the supply equations. Not surprisingly, the coefficients on NUMPVRTY become negative. The coefficient on NUMPVRTY is statistically significant in the SKSPVPOP equation. Coefficients in RICHPOOR also are more significant. These results suggest that those counties with more severe poverty problems find their poor less well served by the EFR system. The percentage of county population that lives in rural areas (PERRURAL) is more significant in the soup kitchen equation, while SIZE is less significant in the food pantry equation.

CONCLUSIONS

This paper represents the first effort to model the emergency food relief system, as it operates through soup kitchens and food pantries. The empirical work reported here uses a small sample, estimates only reduced form equations, and is not free of problems of measurement error. Therefore, conclusions should be regarded as tentative. Nevertheless, the equations explain much of the variation among counties in the supply of EFR meals, and the results are generally supportive of the theoretical model. The empirical work also casts some light on the research and policy questions posed at the beginning.

The results do not directly serve to address the specific reasons that individuals use EFR programs as the analysis is at the market level and because it is hypothesized that the supply of EFR in part reflects perception of demand for it by donors. Nevertheless, the estimated equations provide evidence that EFR use is associated with poverty, characteristics of the poverty population, and coverage of public assistance programs.

The estimated equations are more informative about how well the EFR system meets recipient demand than donor preference. A mixed story is seen. On one hand, results suggest that the system does respond to variation in recipient demand. On the other hand, results suggest that performance of the EFR system in satisfying the demand for free food varies across market areas. First, EFR coverage is lower in rural areas than in urban areas. It was expected that this would be the case for soup kitchens because they face larger minimum efficient scales of operation. However, no evidence was found to suggest that the absence of soup kitchens in rural areas is made up by an increased coverage of food pantries. While it is suspected that demand for EFR among the rural poor is lower than among the urban poor due to lower accessibility to and tastes for EFR, research is necessary to determine if, alternatively, EFR systems tend to be less effective in rural areas. Second, supply levels are seen as reflecting characteristics of the potential donor population. This suggests that some areas are less well served than others, presumably because donor demand varies. Finally, lower coverage of EFR programs in areas with high poverty rates suggests that the EFR system's capacity to meet the demand for food assistance has limits. It should also be noted that the results do not allow an evaluation of how well even the best EFR systems are able to satisfy the demand for food assistance, or the degree to which persons who are in need of food assistance fail to demand or obtain it.

A better understanding of this market requires more focused survey information. Future work on the supply side of this market should examine operations and usage at the individual program level. Reasons for use of EFR would be most accurately determined by analysis of individual or household level data collected on a sample of low income persons, representative of users and nonusers of EFR, that is integrated with information on the characteristics of the local EFR network. Such research may provide important new insights to the performance of public assistance and food assistance programs.

REFERENCES

Birnbaum, Jeffrey H. (1982), "Bitter Harvest: Charity That Delivers Surplus Food to Needy Is Split by Accusations," The Wall Street Journal (October 25): 1, 20.

Brown, J. Larry (1987), "Hunger in the U.S.," Scientific American, 256 (February): 37-41.

Campbell, Cathy C. (1985), Joint Report on Emergency Food Relief in New York State, Ithaca, NY: Cornell University and New York State Department of Health.

Campbell, Cathy C., Janet Weber, David Pelletier, and Janice Dodds (1987), "The Development of a Surveillance System to Monitor Emergency Food Relief in New York State," American Journal of Public Health, 7 (October): 1350-1351.

Hansmann, Henry (1980), "The Role of Non-Profit Enterprise," Yale Law Journal, 91 (November): 54-100.

Hunger Action Network of New York State (1987), Food with Dignity, A Study of People Using Food Pantries in New York State, Albany, NY: Author.

Jencks, Christopher (1987), "Who Gives to What?" in The Nonprofit Sector: A Research Handbook, Walter Powell (ed.), New Haven, CT: Yale University Press: 321-339.

Kennedy, Senator Edward (1983), Going Hungry in America, Report to the Senate Committee on Labor and Human Resources, Washington, DC: Author.

Legal Action Center for the Homeless and New York University (1987), Below the Safety Net: A Study of Soup Kitchen Users in New York City, New York: Author.

Margolis, Howard (1982), Selfishness, Altruism, and Rationality, New York: Cambridge University Press.

McGrath-Morris, Patricia (1988), "An Evaluation of the Nutritional Quality of Meals Served in Soup Kitchens in New York State and an Evaluation of the Factors that Determine Quality," unpublished Master's thesis, Division of Nutritional Sciences, Cornell University, Ithaca, NY.

New York Department of Health, Bureau of Nutrition (1988), "New York State Emergency Food Relief Programs and the Supplemental Nutrition Assistance Program," Albany, NY: Author.

New York State FEMA Committee (1986), Report on FEMA 4B State Set Aside Allocations, Albany, NY: Author.

New York State Nutrition Surveillance Project (1987), Surveillance Reference Data Base, Ithaca, NY: Cornell University Division of Nutritional Sciences.

Physicians Task Force on Hunger in America and Harvard University School of Public Health (1985), Hunger in America, The Growing Epidemic, Boston: Harvard University School of Public Health.

Rauschenbach, Barbara S., Edward A. Frongillo, Jr., Frances E. Thompson, Elizabeth Andersen, and Debra A. Spicer (1990), "Dependency on Soup Kitchens in Urban Areas of New York State," American Journal of Public Health, 80(January): 57-60.

Rose-Ackerman, Susan (1981), "Do Government Grants to Charity Reduce Private Donations?" in Nonprofit Firms in a Three Sector Economy, Michelle White (ed.), Washington, DC: The Urban Institute: 95-114.

Thompson, Frances E., Douglas Taren, Elizabeth Andersen, George Casella, Jennifer K. Lambert, Cathy C. Campbell, Edward Frongillo, Jr., and Debra Spicer (1988), "Within Month Variability in Use of Soup Kitchens in New York State," American Journal of Public Health, 78(September): 1298-1301.

U.S. Conference of Mayors (1986), The Continued Growth in Hunger and Homelessness in American Cities: 1986, Washington, DC: Author.

U.S. Department of Housing and Urban Development (1985), "Section 8 Housing Assistance Payment Program: Fair Market Rent Schedules for Existing Housing and Moderate Rehabilitation," Federal Register, 49(130): 27658-27713.

Weber, Janet (1988), "A Description of Food Pantries in New York State and Food Stamp Participation by their Clients," unpublished Master's thesis, Division of Nutritional Sciences, Cornell University, Ithaca, NY.

Weisbrod, Burton (1975), "Toward a Theory of the Voluntary Non-Profit Sector in a Three-Sector Economy," in Altruism, Morality and Economic Theory, Edmund Phelps (ed.), New York: Russell Sage Foundation: 171-195.

Weisbrod, Burton (1988), The Nonprofit Economy, Cambridge, MA: Harvard University Press.

Weisbrod, Burton and Nestor D. Dominquez (1986), "Demand of Collective Goods in Private Nonprofit Markets: Can Fundraising Help Overcome Free-Rider Behavior?" Journal of Public Economics, 30: 83-95.

Young, Dennis (1986), If Not for Profit, for What? Lexington, MA: D.C. Heath.

Young, Douglas (1989), "A Fair Share Model of Public Good Provision," Journal of Economic Behavior and Organization, 11: 137-147.

(1) As of 1989, food under the TEFAP program is distributed to EFR sites through food banks.

(2) The system described is that which existed in 1985-1986. Since that time, the program has been changed to allow EFR sites to use funds for purposes other than the purchase of food. In fiscal year 1985-1986, public funding for EFR programs in all of New York State amounted to the following:

[TABULAR DATA OMITTED]

(3) This theory does not address the question why nonprofit organizations are predominant in fulfilling these demands. One hypothesis is that these public good markets are often characterized by a lack of information and that donors are more likely to trust the motives of nonprofit entrepreneurs than profit-oriented entrepreneurs to satisfy these demands in a manner consistent with the donors' preferences (Hansmann 1980; Weisbrod 1988). Nor does Weisbrod's theory explain why, in the absence of coercion, individuals fail to become free riders and engage in charitable giving or other similar altruistic behaviors. Several have offered theories to explain this behavior (Margolis 1982; Young 1986).

(4) Apart from limits on the number of sites within a given area, individual sites will serve people on a first-come, first-serve basis until available food is exhausted. Other rationing devices typically employed by soup kitchens include limiting the number of mealtimes the site is open and reducing quality or quantity of food provided. Rationing devices used by food pantries include limiting the number and length of times of food distribution, imposing income eligibility standards, limiting the number of times a person can use the facility during a month or year, and reducing the size of the food packages. In addition, they seldom advertise, relying primarily on word-of-mouth or referrals (Weber 1988).

(5) Under differing sets of assumptions, government grants to charities might serve to either increase or decrease private donations (Rose-Ackerman 1981). Inconsistent empirical evidence on this question has been found (Weisbrod 1988). A full explanation of the determinants of government support of the EFR system is beyond this study. However, it is reasonable to assume that private donor demand is influenced by levels of government support and in turn that government policy responds in part to perceptions of whether private donors have the ability to address recipient demand.

(6) Weisbroad and Dominquez (1986) further argued that price will be a function of the percentage of each dollar donated that goes to service provision, rather than fundraising. This will vary across EFR programs, but there is no reason to believe that this will systematically vary across market areas.

(7) To ascertain the appropriateness of tobit, the soup kitchen equations are also estimated using OLS on the subsample of counties with soup kitchens. Results are very similar to the tobit equations reported.

(8) NUMPVRTY suffers as a measure of the number of poor because it is based on conditions five years prior to the EFR census. A more contemporary measure of the number of poor was tried: the number of public assistance recipients in the county in 1984 (NUMWELF). This variable has the disadvantage that the coverage of public assistance programs may vary from county to county. NUMWELF and NUMPVRTY, however, are highly correlated with one another (r = .97) and use of NUMWELF does not alter results substantially.

(9) As the equations estimate the supply of meals from specific types of EFR sites (i.e., soup kitchens or food pantries), relative prices of alternative types of EFR should be entered into the models. Variables indicating availability of food pantry and soup kitchen meals are not included in the soup kitchen and food pantry equations, respectively, because these variables are thought to be endogenous. It is likely that supply decisions by soup kitchen site managers (or potential site organizers) are influenced by the use of food pantries, and vice versa. This may be especially true for organizations which sponsor both kinds of programs. Nearly one-half of upstate soup kitchens are associated with food pantries (Campbell et al. 1987). The value of surplus food commodities distributed in each county is available. However, this variable is highly correlated with NUMPVRTY, and therefore is not included.

(10) Some EFr users, particularly homeless persons, may be attracted to larger towns and cities because of the greater array of services available.

(11) Correlations between independent variables indicate that several pairs are highly correlated (r > .70). However, tests of alternative specifications indicate that problems of multicollinearity in the data do not substantially affect significance levels and have no effect on the interpretation

James D. Reschovsky is a Research Fellow, Agency for Health Care Policy and Research, Rockville, MD. The article was written the author was Assistant Professor, Department of Consumer Economics and Housing, Cornell University, Ithaca, NY. of results.
COPYRIGHT 1991 American Council on Consumer Interests
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Reschovsky, James D.
Publication:Journal of Consumer Affairs
Date:Dec 22, 1991
Words:6787
Previous Article:Impact of married women's employment on individual household member expenditures for clothing.
Next Article:Helping consumers choose a credit card.
Topics:


Related Articles
WFP gets food aid moving.
Local Results of National Hunger Study Released; Definitive Report Examines Who is Seeking and Who is Providing Emergency Food Relief Services.
Study finds hunger in suburbs and rural areas. (Nation).
America's Second Harvest conducting hunger survey.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters