Post by Deanna on Feb 13, 2010 8:37:54 GMT -5
disclaimer: This is for informational purposes only and viewer makes own decisions.
A good discussion needs to involve different opinions and views, not just one.
www.sonoma.edu/users/c/cuellar/Research/FoodStamps.pdf
Do Food Stamps Cause an Over-Consumption of Food?
by
Steven S. Cuellar, Ph.D.
Department of Economics
Sonoma State University
1801 East Cotati Avenue
Rohnert Park CA. 94928
e-mail: Steve.Cuellar@Sonoma.edu
Draft
March 13, 2003
Abstract
This paper analyzes the effects of the federal food stamp program on the consumption of food
and the potential effects the incidence of obesity among food stamp recipients. Obviously, an
effective food stamp program will lead to an increase in the consumption of food. However, if
the food stamp program, because of the fixed quantity nature of its implementation, has led to an
“over-consumption” of food and increased incidence of obesity, then switching from a fixed
quantity subsidy to a cash grant would reduce the incidence of obesity among the poor. To
examine these issues, this paper starts with a theoretical analysis using a standard microeconomic
model of consumer optimization to analyze the effects of a food stamp program on the
consumption of food and the effect that switching from a food stamp program to a program that
provides a cash grant has on food consumption. The paper then proceeds to examine data on the
food stamp program and the incidence of obesity among various groups.
1 World Health Organization (2002).
2 Centers for Disease Control and Prevention, National Center for Health Statistics.
3 World Health Organization (2003).
Obesity is increasingly being recognized as a serious health problem, not just
domestically where the United States leads the world in obesity, but also worldwide.1 According
to the Center for Disease Control, it is estimated that approximately 65 percent of the US adult
population is overweight, while 30 percent of the adult population is considered obese.2 Along
with the increased recognition of obesity as a serious health problem, there has also been an
increased number of studies into its possible causes. For example, a recent study by the World
Health Organization has attributed the increased incidence of obesity worldwide to an increase in
the consumption of refined sugars added to today’s foods.3 In paper recently presented at the
American Heart Association meetings, Pereira et al. (2003) the rise in obesity is attributed to an
increased consumption of fast foods. Chou, Grossman and Saffer (2002) have attributed the
increasing incidence of obesity to the increased rate of women working outside the home which
has resulted in the increased consumption of food prepared outside the home, including fast food
and the decreasing rate of smoking in the US population. Lakdawalla and Philipson (2002)
attribute the rise in obesity to technological change that has simultaneously reduced the price of
food and reduced the amount of physical activity needed for both home and market production.
In addition to the above studies, there have also been attempts to link the federal food stamp
program to the increase in obesity among the poor (Besharov 2002). The public policy
ramifications of this association are clear. If food stamps have led to an increase in the rate of
obesity among the poor, then one solution is to provide cash grants instead of food stamps to
reduce the overindulgence of food by food stamp recipients. This conclusion is reached by
4 There are other programs that provide food to low income families including the school
breakfast and lunch programs and the (WIC) program for women, infants and children.
5 The utility functions have the usual conditions of being monotonic, continuously twice
differentiable and strictly quasi-concave.
employing a well known principle of microeconomic theory, which states that consumers will
weakly prefer a cash grant to an equivalent in kind subsidy. That is, cash grants are Pareto
preferred to food stamps, in that cash will not make any consumers worse off and will make
some better off.
This paper begins with a detailed examination of the micro theoretic foundations
underlying the relationship between food stamps and food consumption. Specifically, this paper
examines how the food stamp program can lead to the over-consumption of food, what effect
switching from a food stamp program to a cash grant will have on food consumption and which
consumers respond to the so called “cash out”.4 The paper then proceeds to examine data on the
food stamp program and the incidence of obesity among various groups by race, gender, income
and education.
To investigate the effects of a food stamp program on the consumption of food, consider
a standard consumer optimization model in which the consumer acts to maximize utility u(x1, x2)
by choosing the combination of two goods, x1 and x2, that maximizes utility subject to the budget
constraint I = x1p1 + x2p2.
(1) Max subject to I = x1p1 + x2 p2, oe i =(1, 2).
xi
U(x1,x2)
where I is an exogenous endowment and p1 and p2 are the prices of x1 and x2 respectively. Define
x1 as food and x2 as a composite commodity consisting of all goods other than food. Assuming
well behaved preferences5, optimization produces a unique solution xi
* = xi (p1, p2, I) for i =(1,2).
Figure 1
This is shown graphically in Figure One. Prior to the introduction of the food stamp program,
the consumer chooses the utility maximizing consumption bundle (x1
0, x2
0). Consider now the
introduction a food stamp program which provides the consumer with fs dollars worth of food
stamps which can only be used to purchase food. The food stamp program allows the consumer
to purchase at least x1
G units of food, where fs = x1
Gp1. The inclusion of the food stamp program
changes the budget constraint to I + x1
Gp1 = x1p1 + x2 p2. The effect of the food stamp program
on the budget constraint is shown graphically in Figure One as a parallel shift in the budget
constraint from AB to ACD. For a consumer who spent more on food prior to the institution of
the food stamp program, the infra-marginal consumer, optimization in the presence of food
stamps results in the utility maximizing demand function x1(p1, p2, I) $x1
G. In this case, the
institution of the food stamp program is equivalent to a pure increase in income, and increases
consumption of all normal goods. The increase in the consumption of food resulting from the
6 Higher quality food would include for example more fruits and vegetables.
institution of the food stamp program is shown in Figure One as the increase from x1
0 to x1
1 while
consumption of other goods increases to from x2
0 to x2
1. It should be noted that, although the
increase in consumption is usually considered an increase in the quantity of food, it could also be
considered an increase in expenditures on food. These are not necessarily the same if the
expenditures result in an increase in the consumption of higher quality food. If for example,
fatty-foods are an inferior good, then the income effect caused by the food stamps would cause a
decrease in consumption of fatty food and an increase in the consumption of healthier foods.6
This is an especially important point as it relates to the problem of obesity, since an increase in
the quality of food consumed may result in a lower incidence of obesity among food stamps
recipients.
Note that the consumer shown in Figure One would be completely indifferent to a cash
grant equivalent to x1
Gp1 dollars. A pure cash grant would have the effect of extending the
budget constraint from the line segment CD to ED, resulting in no effect on the equilibrium
consumption bundle of the consumer. More generally, those whose consumption of food prior to
the institution of the food stamp program is greater than the amount provided by the food stamps
will be indifferent between food stamps and a cash equivalent. These are referred to as inframarginal
consumers where the infra-marginal consumer is one for whom x1
0 > x1
G.
Those who would be made better off by the cash grant are those for whom the pre-food
stamp consumption of food is less than that of the subsidized quantity, x1
0< x1
G. For the supra
marginal consumers, the optimization problem is
(2) subject to I = x1
G
Max p1 + x2 p2, oe i = (1, 2).
xi
U(x G
1 ,x2)
Figure 2
Optimization results in the consumer consuming x1
* = x1
G and x2
* = x2 (x1
G,p2, I-x1
Gp1). This
situation is shown graphically in Figure 2a. In the absence of the food stamp program, this
consumer chooses the equilibrium consumption bundle (x1
0, x2
0) where x1
0< x1
G. Given food
stamps equal to x1
Gp1, this consumer increases consumption to exactly that amount resulting in a
corner solution in which the consumer spends all of his/her income on goods other than food. In
this case the consumer is said to be over-consuming food since they are consuming more food
than they would choose if given a cash equivalent of x1
Gp1. Note that under the food stamp
program, this consumer is made better off, shown graphically in Figure One as an increase in
utility from I0 to I1. However, Pareto improvement is possible, since if this consumer is given a
cash grant, they could increase utility even more by choosing a quantity of food equal to x1
1 < x1
G.
The effect on the budget constraint of switching from a food stamp program to a program that
provides a cash grant is shown in Figure Two as a shift in the budget constraint from CD to ECD.
Given the preferences of the supra-marginal consumer, the cash grant allows the consumer to
reach a higher indifference curve along the EC segment of the budget constraint. Clearly the
consumer represented in Figure 2a has a relative preference for goods other than food. For
comparison, Figure 2b shows a consumer with a relative preference for food, who is indifferent
between the cash grant and the food stamps. The consumer shown in Figure 2b, like the
consumer shown in Figure One, is an infra-marginal consumer.
The results of the theoretical analysis, shown in Figures One and Two, are clear. As long
as food is a normal good, the food stamp program will increase the expenditures of food for all
recipients. The effect on obesity, however, is ambiguous. If low quality foods are a normal
good, then food stamps will result in an increase in the consumption of fat inducing foods thus
contributing to obesity among food stamp recipients. However, if high quality foods are a
normal good and low quality foods are an inferior good, then the effective increase in income
resulting from the food stamps will reduce obesity.
What does the model say about the effect on food consumption of converting the food
stamp program to a pure cash grant? To begin with, it is clear that a cash-out will have no effect
on the infra-marginal consumer. Only those consumers constrained by the food stamp program,
the supra marginal consumer, will reduce the optimal consumption of food when faced with a
cash equivalent. Furthermore, those who reduce food consumption in response to the cash out,
the supra-marginal consumers, are those who consume the least amount of food. Those who
consume the most food, the infra-marginal consumers will not change their food consumption.
Finally, if the proportion of infra-marginal consumers, which includes those with balanced
preferences and those with a relative preference for food, is greater than that of the supramarginal
consumers, then a cash out of the food stamp program will not have a significant effect
Thousands of Food Stamp Participants
Year
1970 1975 1980 1985 1990 1995 2000
1000
10000
20000
30000
Figure 3
Food Stamp Participation 1969-2002 (Thousands)
on the number of people reducing food consumption. Fraker (1990) estimates the proportion of
constrained food stamp recipients to be between 5-15 percent while Whitmore (2002) estimates
the proportion to be approximately 20-30 percent. Clearly, unconstrained or marginal consumers
make up the bulk of food stamp recipients and would therefore result in a small proportion of
people responding to the cash out of the food stamp program. Any subsequent effects on obesity
would be minimal.
An examination of the data on food stamps and obesity further reveals the relationship
between food stamps and obesity among the poor. Consider first the food stamp program.
Although the food stamp program was established in 1964, reliable statistics are only available
beginning in 1969. In 1969 there were nearly 3 million participants in the food stamp program,
while in 2002 there were approximately 19 million. Figure Three shows the average
Monthly Benefits Per Person (Dollars)
Year
PerPerson RPPB
1970 1975 1980 1985 1990 1995 2000
0
20
40
60
80
Figure 4
Nominal and Real Monthly Per Person Food Stamp Benefits 1969-2002 (Dollars)
participation in the food stamp program from 1969-2002. As can be seen from the graph, food
stamp participation reached its peak in 1996 at approximately 25 million prior to the Welfare
Reform Act of 1997. More relevant to the issue of individual food consumption is the average
monthly benefits of those under the food stamp program. The monthly benefit available depends
on a variety of factors including income and family size, and thus varies greatly, reaching a
maximum of approximately $465 per month for a family of four. However, the average monthly
benefit is much lower. Figure Four shows nominal and real average monthly benefits per person
from 1969 to 2002. As can be seen from Figure Four, while nominal monthly benefits have risen
steadily throughout the period, inflation has depleted much of that growth resulting in virtually
no change in real monthly benefits since the mid-1970’s. For example, although the nominal
average monthly benefit was approximately $80 in 2002, the real value has remained
7 Obesity is defined as a Body Mass Index (BMI) greater than or equal to 30, where BMI
is a calculated as person’s body weight in kilograms is divided by the square of his or her height
in meters (i.e., wt/(ht)2.
8 Overweight is defined as a BMI at or above the 95 percentile for each group.
approximately $45 per month for the last twenty five years.
Consider now the prevalence of obesity in the United States.7 Because obesity was not
always considered a disease, data on obesity are incomplete. However, the Center for Disease
Control has collected data on obesity at various intervals starting in 1960. Table One shows
obesity rates by gender, race and age group. For adults over the age of twenty the intervals are
1960-62, 1971-74, 1976-80, 1988-94 and 1999-2000. For children aged 6-11 and adolescents
aged 12-19, the data provided are for those defined as overweight.8
The period for those aged 6-11 years old is the same as that covered for adults except for
the starting period, which is 1963-65. For those aged 12-19, the period covered is the same as for
adults except that the starting period is 1966-70. The data shown in Table 1 are shown
graphically in Figures 5-7.
Table 1
Group 1-Adults Over 20
Year All Male Female
60-62 13.3 10.7 15.7
71-74 14.6 12.2 16.8
76-80 15.1 12.8 17.1
88-94 23.3 20.6 26
99-00 30.9 27.7 34
Group 2-Adults Over 20 by Gender and Race
Year
White
Male
White
Female
Black
Male
Black
Female
60-62 10.5 14.2 14 26.8
71-74 11.8 15.4 16.8 29.7
76-80 12.4 15.4 16.5 31
88-94 20.7 23.3 21.3 39.1
99-00 27.7 30.6 28.8 50.8
Group 3-Children 6-11 by Gender
Year
All
Children
Boys
6-11
Girls
6-11
63-65 4.2 4 4.5
71-74 4 4.3 3.6
76-80 6.5 6.6 6.4
88-94 11.3 11.6 11
99-00 15.3 16 14.5
Group 4-Children 6-11 by Gender and Race
Year
White
Boys 6-11
Black
Boys 6-11
White
Girls 6-11
Black
Girls 6-11
63-65 4.4 1.6 4.5 4.5
71-74 4.1 5.3 3.7 3.3
76-80 6.7 6.7 5.7 11.1
88-94 10.7 12.3 9.8 17
99-00 12 17.1 N/A 22.2
Group 5-Children 12-19 by Gender
Year
All
12-19
Boys
12-19
Girls
12-19
66-67 4.6 4.5 4.7
71-74 6.1 6.1 6.2
76-80 5 4.8 5.3
88-94 10.5 11.3 9.7
99-00 15.5 15.5 15.5
Group 6-Children 12-19 by Gender and Race
Year
White
Boys 12-19
Black
Boys 12-19
White
Girls 12-19
Black
Girls 12-19
66-67 4.7 3.1 4.5 6.4
71-74 5.5 5 6.1 10.1
76-80 4.6 4.8 4.7 10
88-94 11.6 10.7 8.9 16.3
99-00 12.8 20.7 12.4 26.6
Percent
Rates of Obesity for Adults by Race and Gender 1960-2000
Period
Male Female
All
60-62 71-74 76-80 88-94 99-00
10
20
30
40
50
Black
60-62 71-74 76-80 88-94 99-00
10
20
30
40
50
White
60-62 71-74 76-80 88-94 99-00
10
20
30
40
50
Figure 5
Consider first Group 1 shown in Figure 5, which includes all adults twenty and over,
along with a breakdown for adult males and females. Two observations are evident from the
graph of Group 1. The first is that the rate of obesity remained relatively flat until the period
1976-80, after which obesity began to rise precipitously. The second observation is that the rate
of female obesity is persistently above that of males throughout the entire period. When adults
are further broken down by race and gender, the same pattern of obesity emerges. However,
Figure 5 does show that the rate of obesity for black females is significantly above that of white
males and females and black males. According to the data, the rate of obesity for black females
is approximately 15 percentage points greater than that of the general adult population for all
periods observed, although following the same general pattern of growth of the other cohorts in
that group. For example, in the first period observed 1960-62, the rate of obesity among black
Percent
Rates of Obesity for Children Aged 6-11 by Race and Gender 1960-2000
period
Male Female
All
63-65 71-74 76-80 88-94 99-00
10
20
30
Black
63-65 71-74 76-80 88-94 99-00
10
20
30
White
63-65 71-74 76-80 88-94 99-00
10
20
30
Figure 6
females was approximately 27%, compared to about 13% for all adults, while in the most recent
period observed, 1999-2000, the rate of obesity among black females reaches 50% compared to
31% for the adult population as a whole. While obesity has risen for the entire population as a
whole, clearly the rate of obesity among black females has reached an epidemic proportion.
However, the phenomena of black female obesity clearly pre-dates the institution of the food
stamp program by at least a decade.
Figure 6 shows the data for children 6-11 years old. From Figure 6, you can see that the
proportion of those overweight remains relatively flat from the period 1963-65 through 1971-74
and then begins to increase for both male and females. The break down of children by race and
gender, Group 4, shows that the proportion of overweight black females 6-11 years old has risen
much faster than their black male, white male and female counterparts.
Percent
Rates of Obesity for Children Aged 12-19 by Race and Gender 1960-2000
Period
Male Female
All
66-67 71-74 76-80 88-94 99-00
10
20
30
Black
66-67 71-74 76-80 88-94 99-00
10
20
30
White
66-67 71-74 76-80 88-94 99-00
10
20
30
Figure 7
Figure 7 shows percentage of those overweight for Group 5, 12-19 year old adolescents
by gender and race. From Figure 7 you can see that the percentage of those overweight begins to
grow precipitously after the 1976-80 period. An examination of this same group by race and
gender shows a similar pattern but again shows that the percentage of overweight black females
12-19 year olds is again consistently above that of the rest of the cohort.
Reliable data on obesity among food stamp recipients is not currently available, however
data on obesity by income category does exist. Tables 2 and 3 provide data on the incidence of
obesity by race and gender for different income groups. Table 2 shows data on obesity by
income group while Table 3 provides data on the incidence of obesity by income group, where
each income group is measured relative to the poverty level. If food stamps have
Table 2
Rates of Obesity
1988-1991
Black Mexican White
Income Category Male Female Male Female Male Female
Under $10,000 23 53 33 52 33 43
$10,000- under $30,000 31 50 39 51 35 35
$30,000- under $50 34 54 45 43 36 37
Over $50,000 42 49 48 42 31 27
contributed to the rising rate of obesity, then you would expect to observe a higher rate of obesity
among lower income groups.
Table 3
Rates of Obesity
1988-1991
Black Mexican White
Income Category Relative to Poverty Male Female Male Female Male Female
Below the Poverty Level 29.6 50 36.9 50.5 29.8 46.9
At or above the Poverty Level 32.8 49 40.6 44.4 32.3 30.1
Under 131% of the Poverty Level 28.4 51.1 37.2 50.3 33.3 44.7
131-350% of the Poverty Level 33.2 49 41.4 44.4 33.6 31.9
Over 350% of the poverty Level 36.4 46.1 46.1 40 28.7 26.4
The data on from Tables 2 and 3 are shown graphically in Figures 8 and 9.
Figure 8 does not indicate a clear relationship between obesity and income for any of the
groups shown. For Black and Mexican males the incidence of obesity increases as income
increases. For Black females there does not appear to be any correlation between obesity and
income, while for Mexican females the incidence of obesity decreases as income increases.
Percent
Rates of Obesity by Race, Gender and Poverty Status
Income
Male Female
Black
<10,000 <30,000 <50,000 50,000+
25
30
35
40
45
50
55
60
Mexican
<10,000 <30,000 <50,000 50,000+
25
30
35
40
45
50
55
60
White
<10,000 <30,000 <50,000 50,000+
25
30
35
40
45
50
55
60
Figure 8
Similarly, the incidence of obesity for both White males and females decreases as income
increases.
Figure 9 provides a graphical interpretation of the data in Table 3 where income is
measured relative to the poverty level. The relationship between income and obesity becomes
even more muddled when income is measured relative to the poverty level. The incidence of
obesity is relatively constant for each group across all income levels. The only clear relationship
that emerges from Figure 9 is that women have consistently higher rates of obesity across all
racial groups and income levels.
Percent Obese
Income Relative to Poverty Level
Male Female
Black
<Poverty Poverty <131% <350% >350%
25
30
35
40
45
50
55
60
Mexican
<Poverty Poverty <131% <350% >350%
25
30
35
40
45
50
55
60
White
<Poverty Poverty <131% <350% >350%
25
30
35
40
45
50
55
60
Figure 9
Obesity and Poverty by Race and Gender
Table 4 provides the data on the incidence of obesity among various education groups by
race and gender shown graphically in Figure 10. Obesity does appear to decrease as education
increases for Black, Mexican and White women, however the incidence of obesity appears to
increase among Black men and remains relatively constant across all education groups for
Mexican men.
An examination of the available data does not indicate any relationship between obesity
and food stamps. The only clear picture that emerges from the data is that women are afflicted
by obesity at a much higher rate than men across income and education categories. This is
especially true for Black women and to some extent for Mexican women, and occurs for the most
part across income and education levels.
Percent
Rates of Obesity by Race, Gender and Education
Education
Male Female
Black
< HS HS >HS
25
30
35
40
45
50
55
60
Mexican
< HS HS >HS
25
30
35
40
45
50
55
60
White
< HS HS >HS
25
30
35
40
45
50
55
60
Figure 10
Table 4
Rates of Obesity
1988-1991
Black Mexican White
Education Category Male Female Male Female Male Female
Less than 12 years of education 29 55 39 53 39 39
12 years of education 29 51 43 40 36 38
More than 12 years of education 39 49 40 47 30 30
Conclusion
It is clear that obesity has become a national health problem over the last ten to fifteen
years, afflicting approximately thirty percent of the population. The problem of obesity,
9 Trippe (2000).
however, has hit women, most notably minority women where obesity afflicts approximately half
of all Black women, at a disproportionately higher rate than men. The relationship between food
stamps and obesity is much less clear. Analysis using the standard consumer optimization model
shows that food stamps do indeed have the intended effect of increasing expenditures on food.
The effect on obesity, however, is ambiguous since the increase in income resulting from food
stamps may induce a change in the types of food consumed. If low quality foods are a normal
good, then food stamps result in an increase in the consumption of fat inducing foods and result
in an increase in obesity among food stamp recipients. However, if high quality foods are a
normal good and low quality foods are an inferior good, then the effective increase in income
resulting from the food stamps will induce a substitution from lower quality foods to higher
quality foods and reduce obesity. The degree to which low income families switch from low to
high quality foods depends on the nutritional information possessed by low income families. It
may be that low income consumers lack the nutritional education needed to make healthy diet
choices. However, research shows that although low income consumers are not as nutrition
conscious as high income consumers, low income families do choose similar diets to those of
middle income consumers. Furthermore, the same research shows that food stamp recipients
have diets similar to other low income families not receiving food stamps.9 Whether or not food
stamps induce recipients to change the mix of food consumed, food stamps will not necessarily
increase the consumption of food beyond what most recipients would consume if given cash.
Only those consumers with an extreme preference for goods other than food would be induced to
consume more food under the food stamp program than they would under a program that
provided an equivalent cash grant. Since these consumers are those who consume the least
amount of food, it is highly unlikely that food stamps result in an increased incidence of obesity
among these supra-marginal consumers. Consequently, a conversion of the food stamp program
to a cash grant is unlikely to reduce obesity among food stamp recipients. Further doubt is cast
upon the relationship between obesity and food stamps when data on food stamps, obesity and
income are examined. The data fails to indicate a clear relationship between obesity and the
implementation of the food stamp program. Nor does there appear to be any clear relationship
between obesity and income or poverty status. The incidence of obesity among the most effected
groups, namely females and most particularly minority females, predates the beginning of the
food stamp program and occurs across income and education categories.
To be sure, the data examined in this paper are discrete and highly aggregated. Clearly, as better
data becomes available, more research needs to be conducted. Nevertheless, the findings of this
paper do not support, either theoretically or empirically, a causal link between food stamps and
obesity. As a result, a move from a fixed quantity food stamp program to a program that
provides a cash grant is unlikely to have any effect on the very real problem of obesity among the
poor.
REFERENCES
Besharov, Douglas J., “We’re Feeding the Poor as If They’re Starving.” American Enterprise
Institute, December 2002.
Centers For Disease Control and Prevention, National Center for Health Statistics, Hispanic
Health and Nutrition Examination Survey (1982-84), National Health and Nutrition
Examination Survey. Available at
www.cdc.gov/nccdphp/dnpa/obesity/trend/index.htm.
Centers For Disease Control and Prevention, National Center for Health Statistics, National
Health and Nutrition Examination Survey III Phase 1. 1988-1991.
Chou, Shin-Yi, Grossman, Michael and Saffer Henry, “An Economic Analysis of Adult Obesity:
Results from the Behavioral Risk Factor Surveillance System.” National Bureau of
Economic Research Working Paper No. 9247, October 2002.
Lakdwalla, Darius and Philipson, Tomas, “The Growth of Obesity and Technological Change: A
Theoretical and Empirical Investigation.” National Bureau of Economic Research
Working Paper No. 8965, May 2002.
United States Department of Agriculture, Food and Nutrition Services, “National Level Annual
Summary: Participation and Costs, 1969-2002.” Available at
www.fns.usda.gov/pd/fssummar.htm.
United States Department of Agriculture, Food and Nutrition Services, Office of Analysis,
Nutrition and Evaluation, “Characteristics of Food Stamp Households: Fiscal Year 2001,”
FSP-03-CHAR, by Randy Rosso. Project Officer Jerry Genser. Alexandria, VA: 2003.
Available at www.fns.usda.gov/oane/MENU/Published/FSP/Participation.htm
United States Department of Agriculture, Food and Nutrition Services, “Dietary Intake and
Dietary Attitudes Among Food Stamp Participants and Other Low-Income Individuals”
by Carole Trippe. Project Officer Jerry Genser. Alexandria, VA: September 2000.
Available at:
www.fns.usda.gov/oane/MENU/Published/NutritionEducation/NutritionEducation.htm
Whitmore, Diane. “What Are Food Stamps Worth?” Working Paper #468, Princeton University
Industrial Relations Section, July 2002.
A good discussion needs to involve different opinions and views, not just one.
www.sonoma.edu/users/c/cuellar/Research/FoodStamps.pdf
Do Food Stamps Cause an Over-Consumption of Food?
by
Steven S. Cuellar, Ph.D.
Department of Economics
Sonoma State University
1801 East Cotati Avenue
Rohnert Park CA. 94928
e-mail: Steve.Cuellar@Sonoma.edu
Draft
March 13, 2003
Abstract
This paper analyzes the effects of the federal food stamp program on the consumption of food
and the potential effects the incidence of obesity among food stamp recipients. Obviously, an
effective food stamp program will lead to an increase in the consumption of food. However, if
the food stamp program, because of the fixed quantity nature of its implementation, has led to an
“over-consumption” of food and increased incidence of obesity, then switching from a fixed
quantity subsidy to a cash grant would reduce the incidence of obesity among the poor. To
examine these issues, this paper starts with a theoretical analysis using a standard microeconomic
model of consumer optimization to analyze the effects of a food stamp program on the
consumption of food and the effect that switching from a food stamp program to a program that
provides a cash grant has on food consumption. The paper then proceeds to examine data on the
food stamp program and the incidence of obesity among various groups.
1 World Health Organization (2002).
2 Centers for Disease Control and Prevention, National Center for Health Statistics.
3 World Health Organization (2003).
Obesity is increasingly being recognized as a serious health problem, not just
domestically where the United States leads the world in obesity, but also worldwide.1 According
to the Center for Disease Control, it is estimated that approximately 65 percent of the US adult
population is overweight, while 30 percent of the adult population is considered obese.2 Along
with the increased recognition of obesity as a serious health problem, there has also been an
increased number of studies into its possible causes. For example, a recent study by the World
Health Organization has attributed the increased incidence of obesity worldwide to an increase in
the consumption of refined sugars added to today’s foods.3 In paper recently presented at the
American Heart Association meetings, Pereira et al. (2003) the rise in obesity is attributed to an
increased consumption of fast foods. Chou, Grossman and Saffer (2002) have attributed the
increasing incidence of obesity to the increased rate of women working outside the home which
has resulted in the increased consumption of food prepared outside the home, including fast food
and the decreasing rate of smoking in the US population. Lakdawalla and Philipson (2002)
attribute the rise in obesity to technological change that has simultaneously reduced the price of
food and reduced the amount of physical activity needed for both home and market production.
In addition to the above studies, there have also been attempts to link the federal food stamp
program to the increase in obesity among the poor (Besharov 2002). The public policy
ramifications of this association are clear. If food stamps have led to an increase in the rate of
obesity among the poor, then one solution is to provide cash grants instead of food stamps to
reduce the overindulgence of food by food stamp recipients. This conclusion is reached by
4 There are other programs that provide food to low income families including the school
breakfast and lunch programs and the (WIC) program for women, infants and children.
5 The utility functions have the usual conditions of being monotonic, continuously twice
differentiable and strictly quasi-concave.
employing a well known principle of microeconomic theory, which states that consumers will
weakly prefer a cash grant to an equivalent in kind subsidy. That is, cash grants are Pareto
preferred to food stamps, in that cash will not make any consumers worse off and will make
some better off.
This paper begins with a detailed examination of the micro theoretic foundations
underlying the relationship between food stamps and food consumption. Specifically, this paper
examines how the food stamp program can lead to the over-consumption of food, what effect
switching from a food stamp program to a cash grant will have on food consumption and which
consumers respond to the so called “cash out”.4 The paper then proceeds to examine data on the
food stamp program and the incidence of obesity among various groups by race, gender, income
and education.
To investigate the effects of a food stamp program on the consumption of food, consider
a standard consumer optimization model in which the consumer acts to maximize utility u(x1, x2)
by choosing the combination of two goods, x1 and x2, that maximizes utility subject to the budget
constraint I = x1p1 + x2p2.
(1) Max subject to I = x1p1 + x2 p2, oe i =(1, 2).
xi
U(x1,x2)
where I is an exogenous endowment and p1 and p2 are the prices of x1 and x2 respectively. Define
x1 as food and x2 as a composite commodity consisting of all goods other than food. Assuming
well behaved preferences5, optimization produces a unique solution xi
* = xi (p1, p2, I) for i =(1,2).
Figure 1
This is shown graphically in Figure One. Prior to the introduction of the food stamp program,
the consumer chooses the utility maximizing consumption bundle (x1
0, x2
0). Consider now the
introduction a food stamp program which provides the consumer with fs dollars worth of food
stamps which can only be used to purchase food. The food stamp program allows the consumer
to purchase at least x1
G units of food, where fs = x1
Gp1. The inclusion of the food stamp program
changes the budget constraint to I + x1
Gp1 = x1p1 + x2 p2. The effect of the food stamp program
on the budget constraint is shown graphically in Figure One as a parallel shift in the budget
constraint from AB to ACD. For a consumer who spent more on food prior to the institution of
the food stamp program, the infra-marginal consumer, optimization in the presence of food
stamps results in the utility maximizing demand function x1(p1, p2, I) $x1
G. In this case, the
institution of the food stamp program is equivalent to a pure increase in income, and increases
consumption of all normal goods. The increase in the consumption of food resulting from the
6 Higher quality food would include for example more fruits and vegetables.
institution of the food stamp program is shown in Figure One as the increase from x1
0 to x1
1 while
consumption of other goods increases to from x2
0 to x2
1. It should be noted that, although the
increase in consumption is usually considered an increase in the quantity of food, it could also be
considered an increase in expenditures on food. These are not necessarily the same if the
expenditures result in an increase in the consumption of higher quality food. If for example,
fatty-foods are an inferior good, then the income effect caused by the food stamps would cause a
decrease in consumption of fatty food and an increase in the consumption of healthier foods.6
This is an especially important point as it relates to the problem of obesity, since an increase in
the quality of food consumed may result in a lower incidence of obesity among food stamps
recipients.
Note that the consumer shown in Figure One would be completely indifferent to a cash
grant equivalent to x1
Gp1 dollars. A pure cash grant would have the effect of extending the
budget constraint from the line segment CD to ED, resulting in no effect on the equilibrium
consumption bundle of the consumer. More generally, those whose consumption of food prior to
the institution of the food stamp program is greater than the amount provided by the food stamps
will be indifferent between food stamps and a cash equivalent. These are referred to as inframarginal
consumers where the infra-marginal consumer is one for whom x1
0 > x1
G.
Those who would be made better off by the cash grant are those for whom the pre-food
stamp consumption of food is less than that of the subsidized quantity, x1
0< x1
G. For the supra
marginal consumers, the optimization problem is
(2) subject to I = x1
G
Max p1 + x2 p2, oe i = (1, 2).
xi
U(x G
1 ,x2)
Figure 2
Optimization results in the consumer consuming x1
* = x1
G and x2
* = x2 (x1
G,p2, I-x1
Gp1). This
situation is shown graphically in Figure 2a. In the absence of the food stamp program, this
consumer chooses the equilibrium consumption bundle (x1
0, x2
0) where x1
0< x1
G. Given food
stamps equal to x1
Gp1, this consumer increases consumption to exactly that amount resulting in a
corner solution in which the consumer spends all of his/her income on goods other than food. In
this case the consumer is said to be over-consuming food since they are consuming more food
than they would choose if given a cash equivalent of x1
Gp1. Note that under the food stamp
program, this consumer is made better off, shown graphically in Figure One as an increase in
utility from I0 to I1. However, Pareto improvement is possible, since if this consumer is given a
cash grant, they could increase utility even more by choosing a quantity of food equal to x1
1 < x1
G.
The effect on the budget constraint of switching from a food stamp program to a program that
provides a cash grant is shown in Figure Two as a shift in the budget constraint from CD to ECD.
Given the preferences of the supra-marginal consumer, the cash grant allows the consumer to
reach a higher indifference curve along the EC segment of the budget constraint. Clearly the
consumer represented in Figure 2a has a relative preference for goods other than food. For
comparison, Figure 2b shows a consumer with a relative preference for food, who is indifferent
between the cash grant and the food stamps. The consumer shown in Figure 2b, like the
consumer shown in Figure One, is an infra-marginal consumer.
The results of the theoretical analysis, shown in Figures One and Two, are clear. As long
as food is a normal good, the food stamp program will increase the expenditures of food for all
recipients. The effect on obesity, however, is ambiguous. If low quality foods are a normal
good, then food stamps will result in an increase in the consumption of fat inducing foods thus
contributing to obesity among food stamp recipients. However, if high quality foods are a
normal good and low quality foods are an inferior good, then the effective increase in income
resulting from the food stamps will reduce obesity.
What does the model say about the effect on food consumption of converting the food
stamp program to a pure cash grant? To begin with, it is clear that a cash-out will have no effect
on the infra-marginal consumer. Only those consumers constrained by the food stamp program,
the supra marginal consumer, will reduce the optimal consumption of food when faced with a
cash equivalent. Furthermore, those who reduce food consumption in response to the cash out,
the supra-marginal consumers, are those who consume the least amount of food. Those who
consume the most food, the infra-marginal consumers will not change their food consumption.
Finally, if the proportion of infra-marginal consumers, which includes those with balanced
preferences and those with a relative preference for food, is greater than that of the supramarginal
consumers, then a cash out of the food stamp program will not have a significant effect
Thousands of Food Stamp Participants
Year
1970 1975 1980 1985 1990 1995 2000
1000
10000
20000
30000
Figure 3
Food Stamp Participation 1969-2002 (Thousands)
on the number of people reducing food consumption. Fraker (1990) estimates the proportion of
constrained food stamp recipients to be between 5-15 percent while Whitmore (2002) estimates
the proportion to be approximately 20-30 percent. Clearly, unconstrained or marginal consumers
make up the bulk of food stamp recipients and would therefore result in a small proportion of
people responding to the cash out of the food stamp program. Any subsequent effects on obesity
would be minimal.
An examination of the data on food stamps and obesity further reveals the relationship
between food stamps and obesity among the poor. Consider first the food stamp program.
Although the food stamp program was established in 1964, reliable statistics are only available
beginning in 1969. In 1969 there were nearly 3 million participants in the food stamp program,
while in 2002 there were approximately 19 million. Figure Three shows the average
Monthly Benefits Per Person (Dollars)
Year
PerPerson RPPB
1970 1975 1980 1985 1990 1995 2000
0
20
40
60
80
Figure 4
Nominal and Real Monthly Per Person Food Stamp Benefits 1969-2002 (Dollars)
participation in the food stamp program from 1969-2002. As can be seen from the graph, food
stamp participation reached its peak in 1996 at approximately 25 million prior to the Welfare
Reform Act of 1997. More relevant to the issue of individual food consumption is the average
monthly benefits of those under the food stamp program. The monthly benefit available depends
on a variety of factors including income and family size, and thus varies greatly, reaching a
maximum of approximately $465 per month for a family of four. However, the average monthly
benefit is much lower. Figure Four shows nominal and real average monthly benefits per person
from 1969 to 2002. As can be seen from Figure Four, while nominal monthly benefits have risen
steadily throughout the period, inflation has depleted much of that growth resulting in virtually
no change in real monthly benefits since the mid-1970’s. For example, although the nominal
average monthly benefit was approximately $80 in 2002, the real value has remained
7 Obesity is defined as a Body Mass Index (BMI) greater than or equal to 30, where BMI
is a calculated as person’s body weight in kilograms is divided by the square of his or her height
in meters (i.e., wt/(ht)2.
8 Overweight is defined as a BMI at or above the 95 percentile for each group.
approximately $45 per month for the last twenty five years.
Consider now the prevalence of obesity in the United States.7 Because obesity was not
always considered a disease, data on obesity are incomplete. However, the Center for Disease
Control has collected data on obesity at various intervals starting in 1960. Table One shows
obesity rates by gender, race and age group. For adults over the age of twenty the intervals are
1960-62, 1971-74, 1976-80, 1988-94 and 1999-2000. For children aged 6-11 and adolescents
aged 12-19, the data provided are for those defined as overweight.8
The period for those aged 6-11 years old is the same as that covered for adults except for
the starting period, which is 1963-65. For those aged 12-19, the period covered is the same as for
adults except that the starting period is 1966-70. The data shown in Table 1 are shown
graphically in Figures 5-7.
Table 1
Group 1-Adults Over 20
Year All Male Female
60-62 13.3 10.7 15.7
71-74 14.6 12.2 16.8
76-80 15.1 12.8 17.1
88-94 23.3 20.6 26
99-00 30.9 27.7 34
Group 2-Adults Over 20 by Gender and Race
Year
White
Male
White
Female
Black
Male
Black
Female
60-62 10.5 14.2 14 26.8
71-74 11.8 15.4 16.8 29.7
76-80 12.4 15.4 16.5 31
88-94 20.7 23.3 21.3 39.1
99-00 27.7 30.6 28.8 50.8
Group 3-Children 6-11 by Gender
Year
All
Children
Boys
6-11
Girls
6-11
63-65 4.2 4 4.5
71-74 4 4.3 3.6
76-80 6.5 6.6 6.4
88-94 11.3 11.6 11
99-00 15.3 16 14.5
Group 4-Children 6-11 by Gender and Race
Year
White
Boys 6-11
Black
Boys 6-11
White
Girls 6-11
Black
Girls 6-11
63-65 4.4 1.6 4.5 4.5
71-74 4.1 5.3 3.7 3.3
76-80 6.7 6.7 5.7 11.1
88-94 10.7 12.3 9.8 17
99-00 12 17.1 N/A 22.2
Group 5-Children 12-19 by Gender
Year
All
12-19
Boys
12-19
Girls
12-19
66-67 4.6 4.5 4.7
71-74 6.1 6.1 6.2
76-80 5 4.8 5.3
88-94 10.5 11.3 9.7
99-00 15.5 15.5 15.5
Group 6-Children 12-19 by Gender and Race
Year
White
Boys 12-19
Black
Boys 12-19
White
Girls 12-19
Black
Girls 12-19
66-67 4.7 3.1 4.5 6.4
71-74 5.5 5 6.1 10.1
76-80 4.6 4.8 4.7 10
88-94 11.6 10.7 8.9 16.3
99-00 12.8 20.7 12.4 26.6
Percent
Rates of Obesity for Adults by Race and Gender 1960-2000
Period
Male Female
All
60-62 71-74 76-80 88-94 99-00
10
20
30
40
50
Black
60-62 71-74 76-80 88-94 99-00
10
20
30
40
50
White
60-62 71-74 76-80 88-94 99-00
10
20
30
40
50
Figure 5
Consider first Group 1 shown in Figure 5, which includes all adults twenty and over,
along with a breakdown for adult males and females. Two observations are evident from the
graph of Group 1. The first is that the rate of obesity remained relatively flat until the period
1976-80, after which obesity began to rise precipitously. The second observation is that the rate
of female obesity is persistently above that of males throughout the entire period. When adults
are further broken down by race and gender, the same pattern of obesity emerges. However,
Figure 5 does show that the rate of obesity for black females is significantly above that of white
males and females and black males. According to the data, the rate of obesity for black females
is approximately 15 percentage points greater than that of the general adult population for all
periods observed, although following the same general pattern of growth of the other cohorts in
that group. For example, in the first period observed 1960-62, the rate of obesity among black
Percent
Rates of Obesity for Children Aged 6-11 by Race and Gender 1960-2000
period
Male Female
All
63-65 71-74 76-80 88-94 99-00
10
20
30
Black
63-65 71-74 76-80 88-94 99-00
10
20
30
White
63-65 71-74 76-80 88-94 99-00
10
20
30
Figure 6
females was approximately 27%, compared to about 13% for all adults, while in the most recent
period observed, 1999-2000, the rate of obesity among black females reaches 50% compared to
31% for the adult population as a whole. While obesity has risen for the entire population as a
whole, clearly the rate of obesity among black females has reached an epidemic proportion.
However, the phenomena of black female obesity clearly pre-dates the institution of the food
stamp program by at least a decade.
Figure 6 shows the data for children 6-11 years old. From Figure 6, you can see that the
proportion of those overweight remains relatively flat from the period 1963-65 through 1971-74
and then begins to increase for both male and females. The break down of children by race and
gender, Group 4, shows that the proportion of overweight black females 6-11 years old has risen
much faster than their black male, white male and female counterparts.
Percent
Rates of Obesity for Children Aged 12-19 by Race and Gender 1960-2000
Period
Male Female
All
66-67 71-74 76-80 88-94 99-00
10
20
30
Black
66-67 71-74 76-80 88-94 99-00
10
20
30
White
66-67 71-74 76-80 88-94 99-00
10
20
30
Figure 7
Figure 7 shows percentage of those overweight for Group 5, 12-19 year old adolescents
by gender and race. From Figure 7 you can see that the percentage of those overweight begins to
grow precipitously after the 1976-80 period. An examination of this same group by race and
gender shows a similar pattern but again shows that the percentage of overweight black females
12-19 year olds is again consistently above that of the rest of the cohort.
Reliable data on obesity among food stamp recipients is not currently available, however
data on obesity by income category does exist. Tables 2 and 3 provide data on the incidence of
obesity by race and gender for different income groups. Table 2 shows data on obesity by
income group while Table 3 provides data on the incidence of obesity by income group, where
each income group is measured relative to the poverty level. If food stamps have
Table 2
Rates of Obesity
1988-1991
Black Mexican White
Income Category Male Female Male Female Male Female
Under $10,000 23 53 33 52 33 43
$10,000- under $30,000 31 50 39 51 35 35
$30,000- under $50 34 54 45 43 36 37
Over $50,000 42 49 48 42 31 27
contributed to the rising rate of obesity, then you would expect to observe a higher rate of obesity
among lower income groups.
Table 3
Rates of Obesity
1988-1991
Black Mexican White
Income Category Relative to Poverty Male Female Male Female Male Female
Below the Poverty Level 29.6 50 36.9 50.5 29.8 46.9
At or above the Poverty Level 32.8 49 40.6 44.4 32.3 30.1
Under 131% of the Poverty Level 28.4 51.1 37.2 50.3 33.3 44.7
131-350% of the Poverty Level 33.2 49 41.4 44.4 33.6 31.9
Over 350% of the poverty Level 36.4 46.1 46.1 40 28.7 26.4
The data on from Tables 2 and 3 are shown graphically in Figures 8 and 9.
Figure 8 does not indicate a clear relationship between obesity and income for any of the
groups shown. For Black and Mexican males the incidence of obesity increases as income
increases. For Black females there does not appear to be any correlation between obesity and
income, while for Mexican females the incidence of obesity decreases as income increases.
Percent
Rates of Obesity by Race, Gender and Poverty Status
Income
Male Female
Black
<10,000 <30,000 <50,000 50,000+
25
30
35
40
45
50
55
60
Mexican
<10,000 <30,000 <50,000 50,000+
25
30
35
40
45
50
55
60
White
<10,000 <30,000 <50,000 50,000+
25
30
35
40
45
50
55
60
Figure 8
Similarly, the incidence of obesity for both White males and females decreases as income
increases.
Figure 9 provides a graphical interpretation of the data in Table 3 where income is
measured relative to the poverty level. The relationship between income and obesity becomes
even more muddled when income is measured relative to the poverty level. The incidence of
obesity is relatively constant for each group across all income levels. The only clear relationship
that emerges from Figure 9 is that women have consistently higher rates of obesity across all
racial groups and income levels.
Percent Obese
Income Relative to Poverty Level
Male Female
Black
<Poverty Poverty <131% <350% >350%
25
30
35
40
45
50
55
60
Mexican
<Poverty Poverty <131% <350% >350%
25
30
35
40
45
50
55
60
White
<Poverty Poverty <131% <350% >350%
25
30
35
40
45
50
55
60
Figure 9
Obesity and Poverty by Race and Gender
Table 4 provides the data on the incidence of obesity among various education groups by
race and gender shown graphically in Figure 10. Obesity does appear to decrease as education
increases for Black, Mexican and White women, however the incidence of obesity appears to
increase among Black men and remains relatively constant across all education groups for
Mexican men.
An examination of the available data does not indicate any relationship between obesity
and food stamps. The only clear picture that emerges from the data is that women are afflicted
by obesity at a much higher rate than men across income and education categories. This is
especially true for Black women and to some extent for Mexican women, and occurs for the most
part across income and education levels.
Percent
Rates of Obesity by Race, Gender and Education
Education
Male Female
Black
< HS HS >HS
25
30
35
40
45
50
55
60
Mexican
< HS HS >HS
25
30
35
40
45
50
55
60
White
< HS HS >HS
25
30
35
40
45
50
55
60
Figure 10
Table 4
Rates of Obesity
1988-1991
Black Mexican White
Education Category Male Female Male Female Male Female
Less than 12 years of education 29 55 39 53 39 39
12 years of education 29 51 43 40 36 38
More than 12 years of education 39 49 40 47 30 30
Conclusion
It is clear that obesity has become a national health problem over the last ten to fifteen
years, afflicting approximately thirty percent of the population. The problem of obesity,
9 Trippe (2000).
however, has hit women, most notably minority women where obesity afflicts approximately half
of all Black women, at a disproportionately higher rate than men. The relationship between food
stamps and obesity is much less clear. Analysis using the standard consumer optimization model
shows that food stamps do indeed have the intended effect of increasing expenditures on food.
The effect on obesity, however, is ambiguous since the increase in income resulting from food
stamps may induce a change in the types of food consumed. If low quality foods are a normal
good, then food stamps result in an increase in the consumption of fat inducing foods and result
in an increase in obesity among food stamp recipients. However, if high quality foods are a
normal good and low quality foods are an inferior good, then the effective increase in income
resulting from the food stamps will induce a substitution from lower quality foods to higher
quality foods and reduce obesity. The degree to which low income families switch from low to
high quality foods depends on the nutritional information possessed by low income families. It
may be that low income consumers lack the nutritional education needed to make healthy diet
choices. However, research shows that although low income consumers are not as nutrition
conscious as high income consumers, low income families do choose similar diets to those of
middle income consumers. Furthermore, the same research shows that food stamp recipients
have diets similar to other low income families not receiving food stamps.9 Whether or not food
stamps induce recipients to change the mix of food consumed, food stamps will not necessarily
increase the consumption of food beyond what most recipients would consume if given cash.
Only those consumers with an extreme preference for goods other than food would be induced to
consume more food under the food stamp program than they would under a program that
provided an equivalent cash grant. Since these consumers are those who consume the least
amount of food, it is highly unlikely that food stamps result in an increased incidence of obesity
among these supra-marginal consumers. Consequently, a conversion of the food stamp program
to a cash grant is unlikely to reduce obesity among food stamp recipients. Further doubt is cast
upon the relationship between obesity and food stamps when data on food stamps, obesity and
income are examined. The data fails to indicate a clear relationship between obesity and the
implementation of the food stamp program. Nor does there appear to be any clear relationship
between obesity and income or poverty status. The incidence of obesity among the most effected
groups, namely females and most particularly minority females, predates the beginning of the
food stamp program and occurs across income and education categories.
To be sure, the data examined in this paper are discrete and highly aggregated. Clearly, as better
data becomes available, more research needs to be conducted. Nevertheless, the findings of this
paper do not support, either theoretically or empirically, a causal link between food stamps and
obesity. As a result, a move from a fixed quantity food stamp program to a program that
provides a cash grant is unlikely to have any effect on the very real problem of obesity among the
poor.
REFERENCES
Besharov, Douglas J., “We’re Feeding the Poor as If They’re Starving.” American Enterprise
Institute, December 2002.
Centers For Disease Control and Prevention, National Center for Health Statistics, Hispanic
Health and Nutrition Examination Survey (1982-84), National Health and Nutrition
Examination Survey. Available at
www.cdc.gov/nccdphp/dnpa/obesity/trend/index.htm.
Centers For Disease Control and Prevention, National Center for Health Statistics, National
Health and Nutrition Examination Survey III Phase 1. 1988-1991.
Chou, Shin-Yi, Grossman, Michael and Saffer Henry, “An Economic Analysis of Adult Obesity:
Results from the Behavioral Risk Factor Surveillance System.” National Bureau of
Economic Research Working Paper No. 9247, October 2002.
Lakdwalla, Darius and Philipson, Tomas, “The Growth of Obesity and Technological Change: A
Theoretical and Empirical Investigation.” National Bureau of Economic Research
Working Paper No. 8965, May 2002.
United States Department of Agriculture, Food and Nutrition Services, “National Level Annual
Summary: Participation and Costs, 1969-2002.” Available at
www.fns.usda.gov/pd/fssummar.htm.
United States Department of Agriculture, Food and Nutrition Services, Office of Analysis,
Nutrition and Evaluation, “Characteristics of Food Stamp Households: Fiscal Year 2001,”
FSP-03-CHAR, by Randy Rosso. Project Officer Jerry Genser. Alexandria, VA: 2003.
Available at www.fns.usda.gov/oane/MENU/Published/FSP/Participation.htm
United States Department of Agriculture, Food and Nutrition Services, “Dietary Intake and
Dietary Attitudes Among Food Stamp Participants and Other Low-Income Individuals”
by Carole Trippe. Project Officer Jerry Genser. Alexandria, VA: September 2000.
Available at:
www.fns.usda.gov/oane/MENU/Published/NutritionEducation/NutritionEducation.htm
Whitmore, Diane. “What Are Food Stamps Worth?” Working Paper #468, Princeton University
Industrial Relations Section, July 2002.