ESSAYS ON ISSUES THE FEDERAL RESERVE BANK MARCH 2011 OF CHICAGO NUMBER 284

Chicag­o Fed Letter

Who would be affected by soda ? by Leslie McGranahan, senior economist, and Diane Whitmore Schanzenbach, associate professor of human development and social policy, Northwestern University, and faculty research fellow, National Bureau of Economic Research

In 2009–10, 17 states considered expanding taxation on sugar-sweetened beverages (SSBs) as a potential source of funds and a means to curb obesity. This article examines the various types of soda proposals, the underlying economic theory, and the anticipated impact of the proposed taxes on different population groups.

In early 2009, proposals for a federal the form of the beverage. Local taxes on tax on SSBs began to be discussed as a soda are a source of additional variation potential source of in tax rates. funds for expanded 1. Spending on SSBs consumed at home Forty-five of the 50 states have a sales health care. Although tax on general merchandise. In most Annual Share of Share of Share of specific legislation was states with sales taxes, food for home spending total food at-home food not proposed in 2009, ($) spending spending spending consumption is taxed at a lower rate the Congressional (------percent ------) than other items. In some states, soda Budget Office esti- for home consumption is treated as gen- All 142.31 0.33 2.37 4.18 mated that a tax of eral merchandise rather than as food By race/ethnicity 3 cents per 12-ounce and is subject to the normal, higher rate.2 Black 111.47 0.35 2.51 4.11 soft drink would White 146.51 0.32 2.36 4.23 Food consumed at restaurants or fast- Hispanic 155.66 0.41 2.57 4.32 generate revenues of food establishments is usually subject to $24 billion over four By educational the regular or a higher restau- attainment years.1 Federal taxation rant meal tax. In most states, vending Less than high school 130.96 0.53 3.22 4.74 of SSBs seems unlikely High school 141.79 0.42 2.89 4.61 machine sales of soda are subject to the Some college 147.02 0.37 2.55 4.46 in the near future. same tax as grocery store sales of soda. College graduate 141.71 0.25 1.90 3.64 However, the debate By poverty status has moved to the state Independent of all these laws, soda Below poverty line 117.83 0.50 2.97 4.42 level, where numer- purchased with benefits from the

Source: Authors’ calculations based on data from the Bureau of Labor Statistics’ 2008 ous states and some Supplemental Nutrition Assistance Consumer Expenditure Survey. localities have either Program (SNAP, formerly called the imposed or are con- Food Stamp program) is not subject to sidering imposing an additional sales the sales tax.3 The Food Security Act of tax or an tax on SSBs. 1985 (Public Law 99-198) declared that a state would not be able to participate in Current taxation of SSBs the Food Stamp program if sales taxes Taxation of SSBs is complex because soda were collected on food purchased with and other sweetened beverages are food stamps. As a result, no state currently purchased from many different outlets, collects sales tax on soda (or any other including grocery stores and restaurants, good) purchased with SNAP benefits. and in two primary forms, fountain sodas Taken together, these regulations mean and sealed containers. Currently, state that the state tax on soda can vary within taxes on soda vary both across states a given state. and within states by point of sale and but it might encourage consumers to 2. Calorie intake, by population group switch to brands with a higher sugar con-

Total daily Calories % calories % tent. Other excise tax proposals would calories (SD) from SSBs (SD) from SSBs (SD) levy taxes based directly on sugar content. All 2,115 (1,005) 121 (213) 5.6 (9.6) Excise taxes also would lead to increased By race/ethnicity prices of fountain and restaurant bev- Black 2,085 (1,056) 123 (190) 6.2 (9.8) erages, because increased excise taxes White 2,142 (998) 124 (227) 5.6 (9.8) Hispanic 2,104 (1,050) 129 (178) 6.0 (8.1) would apply to beverages before they are shipped to restaurants. If states expanded By educational attainment Less than high school 1,958 (1,048) 137 (211) 7.4 (11.4) soda taxation by adding grocery pur- High school 2,110 (1,182) 159 (252) 7.3 (11.3) chases of soda to the tax base, however, Some college 2,157 (970) 131 (216) 5.8 (9.4) College graduate 2,161 (866) 78 (163) 3.3 (6.2) it would not affect restaurant prices, as restaurant purchases are already subject By poverty status Below poverty line 1,994 (1,195) 180 (279) 9.0 (13.3) to the sales tax.

Note: SD indicates standard deviation. Finally, the intersection of proposed soda Source: Authors’ calculations based on data from the 2007–08 National Health and Nutrition Examination Survey (NHANES). taxes and SNAP benefits depends on the form of the tax. An increased sales tax on According to data from the Rudd Center “overconsumption” of the good. In the soda would not apply to these purchases. for Food Policy and Obesity at Yale case of the soda tax, reducing consump- On the other hand, excise taxes that are University, 17 states filed legislation in tion may reduce average body weight and levied earlier in the production process 2009–10 to expand soda taxation; two obesity rates. This, in turn, might reduce would be passed on to SNAP recipients in of those states, Colorado and Washington, health problems related to obesity, such the same way that they are passed on to passed such legislation. With a new as diabetes and heart disease, and could other consumers—by being incorporated legislative session beginning in 2011, reduce health care expenditures.4 into the list price of the drink. states may again begin to expand the Public finance theory argues that it Impact on different population groups taxation of soda. usually does not matter whether a tax is An increase in soda taxes would affect There are two primary forms of increased imposed on the producers or consumers people differently, depending on how taxation. In some states, proposals would of a good, because the producers can pass much soda they purchase, where they buy subject soft drinks to the general sales tax some of the tax on to the consumers. it, whether they purchase it with SNAP and no longer treat them as food items. The incidence of the tax burden is shared benefits, and the form of the tax. In This type of proposal was enacted in depending on the elasticity of demand figure 1, we calculate average annual Colorado. The second type of proposal and supply, regardless of whether pro- spending on SSBs consumed at home would create an excise tax on distribu- ducers or consumers directly remit the (at-home SSBs) for members of various tors of soft drinks. This was enacted in tax. However, in the case of the soda tax, demographic groups, based on data from Washington State, where bottlers needed there are important differences between the Bureau of Labor Statistics’ 2008 to pay a $0.02 tax per 12 ounces of soda. the excise tax (paid by suppliers) and sales Consumer Expenditure Survey (CEX). We This legislation was subsequently repealed. tax (paid by consumers and remitted by also present at-home SSB spending as a retailers) proposals. Specifically, sales share of total spending, as a share of total Economics of soda taxation taxes are ad valorem—in proportion to the food spending, and as a share of total In general, taxes distort function- good’s price. In contrast, the excise tax spending on food consumed at home. ing and reduce economic efficiency. Pro- proposals have been based on volume, ponents of soda taxation argue that it with a tax either per ounce of soda or The CEX is the most comprehensive falls into the realm of a “Pigovian tax.” per teaspoon of added sugar. survey of household expenditure pat- Such a tax may actually increase total terns in the United States. The survey Soda prices tend to be lower on a per economic efficiency in the presence of provides detailed information on drinks ounce basis for larger containers (such from consumption. In other purchased for consumption at home as two-liter bottles or 12-packs) than for words, if there is a negative effect from and more limited data on drinks pur- individual cans or single-serve bottles or 5 consumption of a particular good that chased away from home. We present for generics. As a result, if an ad valorem is not directly felt by consumers, then data on the sum of spending on carbon- tax is levied, then the tax per ounce of when left to their own devices, consumers ated drinks, fruit-flavored noncarbonated beverage will be lower for larger servings will consume more of the good than is drinks, and sports drinks. While diet and generic brands, and consumers may socially optimal. Therefore, taxing such drinks would not be subject to additional switch to these cheaper options. Some a good will not only raise revenue for taxation under some proposals, carbon- excise tax proposals would levy taxes per the government, but may also improve ated drinks are not separated by calorie ounce. This would not give consumers overall social welfare by reducing content in the data. an incentive to switch to larger containers, In figure 1, we show that the average calories from SSBs than those in college that the tax response will differ systemati- household spent $142 on sugar-sweetened graduate households. Similarly, the cally based on consumers’ obesity status. beverages at home in 2008.6 Among the poor get a larger share of their daily cal- eight subpopulations in the data, only ories from SSBs. Across all groups, the Conclusion Hispanics exceeded average spending average share of calories from SSBs is In this article, we have documented by more than $10, while three groups greater than the average share of food consumption patterns for SSBs, in terms spent over $10 less—blacks, those with spending on SSBs (shown in figure 1), of both expenditure and caloric intake. less than a high school education, and indicating that SSBs are a relatively in- In general, less educated and poorer the poor. The next three columns of expensive source of calories.9 population groups concentrate a higher figure 1 show spending on SSBs as a pro- proportion of their total spending on portion of total spending (column 2), Potential impact on public health SSBs and get a larger fraction of their total food spending (column 3), and total A recent study of the effects of current daily caloric intake from these relatively spending on food at home (column 4). soda taxes on obesity finds that higher inexpensive sources. In the absence of Sugar-sweetened beverage consumption taxes do not reduce obesity rates.10 Re- substantial differences in the elasticity of as a share of total spending falls as edu- cent proposals, however, suggest raising demand across groups, this implies that cational attainment increases. The same the tax rates to levels much higher than increased soda taxation would primarily pattern is found for SSB spending as a those that are currently seen and may, affect these disadvantaged groups. share of food spending and at-home food therefore, lead to a different outcome. However, the impact will vary somewhat, spending. We also observe that SSBs depending on the particular form of Two conditions would need to be met in represent a larger share of the spending the tax and whether purchases are made order for an expanded soda tax to affect of the poor than is true for the overall using SNAP benefits. the rates of overweight and obesity. First, population. As a result, a soda tax would increased taxation would need to lead to likely have a disproportionate effect on 1 Congressional Budget Office, 2008,Budget reduced consumption of soda. Recent the less educated and poor groups; these Options Volume 1: Health Care, December, research suggests that soda consumption available at www.cbo.gov/ftpdocs/99xx/ groups are also more likely to benefit is modestly responsive to price changes. doc9925/12-18-HealthOptions.pdf. from the SNAP program. According to a recent study by Andreyeva, 2 See Chicago Fed Letter, No. 284, Supplemental We have shown that SSBs comprise a Long, and Brownell, an increase in soda Figure, “Soda taxes in 2009,” available at non-trivial share of food spending. In price of 10% would decrease consump- www.chicagofed.org/webpages/people/ mcgranahan_leslie.cfm#, under Other figure 2, we provide data on the role SSBs tion by somewhere between 2.7% and Resources. play in individuals’ diets. The data here 8.1%.11 Second, lower soda consumption come from the 2007–08 wave of the would need to result in a reduction in National Health and Nutrition Examination overall caloric consumption. Survey (NHANES),7 and the sample is Charles L. Evans, President ; Daniel G. Sullivan, There are many possible responses to limited to adults aged 20 or older. Calorie Executive Vice President and Director of Research; increased soda taxes. Consumers might Spencer Krane, Senior Vice President and Economic consumption is collected using 24-hour purchase soda that has more sugar or Advisor ; David Marshall, Senior Vice President, financial recall interviews, and we use the first markets group ; Daniel Aaronson, Vice President, calories per ounce. They might also re- day’s dietary recall for those that report microeconomic policy research; Jonas D. M. Fisher, spond by consuming other drinks that Vice President, macroeconomic policy research; Richard multiple days of data. The NHANES Heckinger, Assistant Vice President, markets team; are not subject to the tax or shifting their series of studies is the largest national Anna Paulson, Vice President, finance team; William A. consumption to other items. A recent Testa, Vice President, regional programs, and Economics data source on caloric intake. We con- study that estimated the impact of a 40% Editor ; Helen O’D. Koshy and Han Y. Choi, Editors ; struct the measure of SSBs by combining Rita Molloy and Julia Baker, Production Editors ; tax on all SSBs found that the tax would reported intake of sugared sodas, sweet- Sheila A. Mangler, Editorial Assistant. reduce an individual’s calories from SSBs ened teas, and energy drinks. Overall, Chicago Fed Letter is published by the Economic by about 17 kcal per day and calories Research Department of the Federal Reserve Bank as shown in figure 2, adults on average of Chicago. The views expressed are the authors’ from consumption of all beverages by report consuming 121 calories per day and do not necessarily reflect the views of the 12 kcal per day.12 The study found that Federal Reserve Bank of Chicago or the Federal from SSBs. The standard deviation of some of the drop in SSB consumption Reserve System. calories from SSBs is large, indicating is offset by a higher intake of calories © 2011 Federal Reserve Bank of Chicago that there is substantial variation across Chicago Fed Letter articles may be reproduced in from other beverages, but the authors individuals in intake. On average, SSBs whole or in part, provided the articles are not do not investigate the effects of SSB reproduced or distributed for commercial gain account for 5.6% of total reported calorie and provided the source is appropriately credited. taxation on consumption of food. The intake. Average SSB consumption varies Prior written permission must be obtained for effects on food consumption and, hence, any other reproduction, distribution, republica- only slightly by race/ethnic group, but the effects on total calories, depend on tion, or creation of derivative works of Chicago Fed varies much more widely by educational Letter articles. To request permission, please contact whether calories from food and drinks attainment.8 Individuals in households Helen Koshy, senior editor, at 312-322-5830 or are substitutes or complements. In addi- email [email protected]. Chicago Fed with a high school diploma or less get Letter and other Bank publications are available tion, because elasticity of demand varies at www.chicagofed.org. over twice the proportion of their daily with consumer preferences, it is possible ISSN 0895-0164 3 Note that food stamp benefits may not 8 For consistency with the CEX data, here the 11 Tatiana Andreyeva, Michael W. Long, and be used to purchase food or beverages at household’s education level is defined as Kelly D. Brownell, 2010, “The impact of restaurants or fast-food establishments. the maximum of the household head and food prices on consumption: A systematic 4 Congressional Budget Office, op. cit., p. 192. spouse’s reported educational attainment. review of research on the price elasticity of 9 demand for food,” American Journal of Public 5 The NHANES data include food consumed We do not present data on food consumed Health, February, pp. 216–222. away from home, because these data are away from home, while the CEX only covers 12 presented as totals at a meal and do not food consumed at home, but this is unlikely Eric A. Finkelstein, Chen Zhen, James separate out the items in the meal, with to explain this discrepancy. Our results Nonnemaker, and Jessica E. Todd, 2010, the exception of alcoholic beverages. are generally in line with those of Cynthia “Impact of targeted beverage taxes on higher- Ogden, Molly Lamb, Margaret Carroll, and lower-income households,” Archives of 6 The CEX unit of analysis is a “consumer and Katherine Flegal, 2010, “Obesity and Internal Medicine, Vol. 170, No. 22, December, unit” and is similar to a household. socioeconomic status in adults: United pp. 2028–2034. Similar results are found 7 NHANES is a major program of the National States, 2005–08,” NCHS Data Brief, No. 50, in Travis A. Smith, Biing-Hwan Lin, and Center for Health Statistics (NCHS). NCHS December. Jonq-Ying Lee, 2010, Taxing Caloric Sweetened Beverages: Potential Effects on Beverage is part of the Centers for Disease Control 10 Jason M. Fletcher, David E. Frisvold, and Consumption, Calorie Intake, and Obesity, and Prevention (CDC) and has the respon- Nathan Tefft, 2010, “Can soft drink taxes ERR-100, U.S. Department of Agriculture, sibility for producing vital and health sta- reduce population weight?,” Contemporary Economic Research Service, July. tistics for the nation. Economic Policy, Vol. 28, No. 1, January, pp. 23–35.