Summary The Supplemental Nutrition Assistance Program (SNAP) is the largest of the federal government’s 15 nutrition assistance programs operating in the United States, giving approximately 40 million Americans benefits that can be used to purchase food for at-home consumption. Following decades of life as the Food Stamp Program, the program was renamed SNAP in 2008 in legislation stating that: ‘a supplemental nutrition assistance program is herein authorized which will permit low-income households to obtain a more nutritious diet… by increasing food purchasing power for all eligible households who apply for participation.’ A study published recently in the
Siobhan Miller considers the following as important: benefits, food stamps, Public Finance, welfare
This could be interesting, too:
Siobhan Miller writes How do American families spend food benefits?
Emma Hyman writes The incentives of centralized school admissions systems
The Supplemental Nutrition Assistance Program (SNAP) is the largest of the federal government’s 15 nutrition assistance programs operating in the United States, giving approximately 40 million Americans benefits that can be used to purchase food for at-home consumption. Following decades of life as the Food Stamp Program, the program was renamed SNAP in 2008 in legislation stating that: ‘a supplemental nutrition assistance program is herein authorized which will permit low-income households to obtain a more nutritious diet… by increasing food purchasing power for all eligible households who apply for participation.’
A study published recently in the American Economic Review finds that SNAP is successful at increasing household food expenditure. A household receiving, say, a $200 monthly SNAP benefit can be expected to increase its monthly expenditure on groceries by a little over $100. This far exceeds what would be expected for a cash benefit of comparable size, which might increase food spending by only $20 or so. It goes against a core principle of microeconomic theory: that money is ‘fungible’ and how it is spent does not depend on the form in which it is received.
So if households are not abiding by the principle of fungibility, what principles do guide their behavior? The researchers – Justine Hastings and Jesse Shapiro – turn to the hypothesis of ‘mental accounting’, developed by economics Nobel laureate Richard Thaler. The idea is that people put money in different mental accounts depending on their source or intended use. So, for example, a household receiving the SNAP benefit might think of it as ‘food money’ and psychologically earmark it for grocery spending. If that same household got its $200 as cash, it might spread the money across several mental accounts and therefore spend it quite differently.
Formulating a quantitative model of household behavior based on the idea of mental accounting, the researchers find that it matches the main facts that they document about the spending response to SNAP. In addition to the fact that household grocery spending responds more SNAP benefits than to cash income, the model successfully predicts another fact documented in the study: that bargain-seeking behavior (measured by coupon redemption and store-brand purchases) declines following SNAP receipt for grocery purchases but not for non-grocery purchases.
The study concludes that the SNAP program is more successful than economic theory would predict at boosting families’ grocery spending. This makes it especially important to understand how it affects the composition of purchased food, a question the authors are pursuing in work with Ryan Kessler.
The preliminary findings of that research suggest that while SNAP is successful in increasing food expenditures, the success does not translate into meaningful gains in the nutritional quality of purchased foods. These results suggest the value of further innovation in the SNAP program to help families eat healthily.
SNAP (the Supplemental Nutrition Assistance Program, formerly known as the Food Stamps Program) is an important part of the social safety net in the United States. It is the second-largest means-tested program in the country. Preliminary data indicate that in September 2018, 19.4 million households enrolled in SNAP, out of a total of around 127.6 million.
The program gives households a monthly benefit—around $250 per month on average in recent years—in the form of funds available on an Electronic Benefit Transfer card. These funds can be used to buy groceries at a wide range of retailers. One recent estimate is that SNAP accounts for 10% of purchases of food for at-home consumption.
Both the name and design of the program suggest a focus on food, and indeed the underlying legislation emphasizes ‘increasing food purchasing power’ as a key aspect. The US Department of Agriculture has said that SNAP ‘raises food expenditures’ and helps ‘to put food on the table’.
Economists have long emphasized that the program may have less to do with food than appearances would suggest. To understand why, think about a hypothetical household initially spending $350 of its monthly cash income on groceries. Now, imagine that the household starts receiving $200 a month in SNAP benefits. How can the household adjust its monthly spending?
One way is to spend $550 total on groceries—$350 in cash and $200 in SNAP. But another way is to continue to spend $350 total on groceries, with $200 coming from the SNAP benefit and the rest from cash. This would free up $200 in cash that the household formerly spent on groceries and that could now be saved or spent on anything else the household wishes to buy.
Because the SNAP benefit affords this household the option of spending $200 more a month on anything, it functions economically just like a cash benefit. And because it can be spent like cash, traditional economic theory predicts that it will be—that is, that giving the household a $200 SNAP benefit should lead to the same changes in spending as giving the household a $200 cash benefit.
The equivalence between SNAP and cash is an example of the fungibility of money, the microeconomic principle that how money should be spent does not depend on the form in which it is received. This principle implies, for example, that a windfall received through a larger end-of-year bonus at work should be spent in the same way as one received through a higher tax refund or a surprise discount on the purchase of the family car.
SNAP is literally a textbook example of this principle, appearing, for example, in the 2000 edition of Greg Mankiw’s Principles of Microeconomics and the 2004 edition of Edgar Browning and Mark Zupan’s Microeconomics.
If this principle applies to SNAP recipients, then the program should effectively function as a cash assistance program, at least for the large majority of recipients for whom SNAP does not cover the full grocery budget. If it does not, then a basic implication of microeconomic theory fails to hold in an important economic context.
Testing fungibility in this context requires measuring how SNAP recipients’ spending changes as a result of receiving program benefits. This is challenging for several reasons. One is that SNAP recipients are likely to differ from non-recipients in hard-to-measure ways. This means that differences in spending patterns between those receiving SNAP benefits and those not receiving them may reflect underlying differences between households rather than the true effect of the program on spending.
Several recent studies circumvent this challenge in various ways and reach different conclusions, some finding that recipients’ behavior is consistent with the fungibility of money and others finding that SNAP benefits affect food spending by more than an equivalent cash benefit.
In a recent article in the American Economic Review, we approach this question with a novel research dataset. The dataset is a retail panel that includes anonymized information on the purchases of nearly half a million regular customers of a US grocery retail chain over a nearly seven-year period. Over that period, thousands of customers in the panel transition on to SNAP.
We observe a given customer’s spending behavior both before and after entry into the program, making it easier to isolate the effect of SNAP separately from other differences across households in their spending proclivities. We take advantage of this and other features of the dataset to employ research designs that differ from those used in past research.
Based on evidence in our own data and in the large body of research estimating the spending response to income changes, we expect that a dollar of cash income would translate into around $0.10 in additional grocery spending, implying that a $200 benefit would take our hypothetical household’s monthly grocery spending from $350 to $370.
By contrast, we estimate that one dollar in SNAP benefits translates into just over $0.50 in additional grocery spending. So, $200 in SNAP benefits would take our hypothetical household’s spending from $350 to $450 instead of the $370 predicted by the principle of fungibility.
We use economic theory to formulate several statistical tests for whether households act according to the fungibility of money, and we find that they do not.
So if households are not abiding by the principle of fungibility, what principles do guide their behavior? Our data do not allow a definitive answer to this question, but an appealing explanation comes from the burgeoning body of work on psychology and economics.
The hypothesis of mental accounting, developed by Nobel laureate Richard Thaler, holds that households put money in different mental accounts depending on their source or intended use. Our hypothetical household getting SNAP, for example, might think of it as ‘food money’ and psychologically earmark it for grocery spending. If that same household got its $200 as cash, it might spread the money across several mental accounts and therefore spend it quite differently.
This hypothesis finds support in a number of studies, including a field experiment in Morocco on labeled cash transfers, a study of the UK Winter Fuel Payment, and in our own past work on the response to gasoline price fluctuations.
We formulate a quantitative model of behavior based on the idea of mental accounting and find that it can match the main facts that we document about the spending response to SNAP. In particular, in addition to the fact that household grocery spending responds more to SNAP benefits than to cash income, the model successfully predicts another fact that we document: that bargain-seeking behavior (measured by coupon redemption and store-brand purchases) declines following SNAP receipt for grocery purchases but not for non-grocery purchases.
The retail data that we use in our study afford many advantages, including a high level of detail and the ability to observe the same household’s decisions over a long period of time. But these data also have some important limitations.
One is that SNAP participation is measured only indirectly in the data, through a household’s mode of payment for groceries. Another is that, because our data come from a single retail chain, fluctuations in spending can be due either to changes in the household’s total grocery bill or to changes in the allocation of spending across retailers. The restriction to data on regular customers of a single retail chain also means that we do not have a nationally representative sample.
Our research includes detailed analysis of all of these limitations, but ultimately further work will be needed to probe the sensitivity and generality of our findings.
Our findings have a number of potential implications for analysis of the SNAP program. One is that the program is more successful than economic theory would predict at increasing grocery spending. This makes it especially important to understand how it affects the composition of purchased foods, a question that we are pursuing in work in progress with Ryan Kessler.
Another is that incorporating additional elements of psychological realism into our models of program recipients might help economists to develop a better understanding of their behavior and also to predict more accurately the effects of alternative ways of designing or implementing the program.