In previous articles, I examined two typical features of state and local retail sales taxes that prevent them from functioning as comprehensive taxes on consumption: the taxation of business purchases and the exemption of consumer services. In this article, I examine another feature, the exemption of food sold for home consumption (groceries).
The normative status of this feature is very different from that of the first two features. As I explained in my other articles, the taxation of business purchases and the disparate treatment of consumer goods and services are irrational policies that lack coherent justification. In contrast, the exemption of groceries serves the essential purpose of easing the fiscal burden on those with limited economic means. Economic analysis suggests, however, that this policy is not a desirable way to advance that purpose. I therefore recommend that states move to end their grocery exemptions, in a manner that does not endanger the well-being of those in need.
In this article, I survey the practices of states in this regard and then consider the general question of whether necessities should be exempt from sales tax. I first examine this question from the perspective of optimal tax theory. This theory supports taxing necessities and luxuries at uniform rates under some assumptions, although the rationale for doing so is more subtle than is often thought. Recognizing that optimal tax theory is stylized, I then consider how real-world factors might affect the decision on how to tax these items.
Current State Practice
In a November 2009 study, the Center on Budget and Policy Priorities examined the practices of the 46 jurisdictions (45 states and the District of Columbia) that have general sales taxes. CBPP found that seven states taxed groceries at regular sales tax rates--Alabama, Hawaii, Idaho, Kansas, Mississippi, Oklahoma, and South Dakota. Seven other states taxed groceries at preferential rates--Arkansas, Illinois, Missouri, Tennessee, Utah, Virginia, and West Virginia. The remaining 31 states and the District of Columbia exempted groceries from sales tax. Local sales taxes may be imposed on groceries even when they are exempt at the state level.
In my previous articles, I found a strong correlation, positive 0.75, between the extent to which states taxed business purchases and the extent to which they taxed consumer services, indicating a high degree of consistency regarding the extent to which states want to broaden the tax base. That consistency was striking because good tax policy recommends base broadening in the one case (consumer services) and opposes it in the other (business purchases). In contrast, states' willingness to tax groceries appears to have only a weak relationship to their willingness to tax business purchases or consumer services. I find that the taxation of groceries has a correlation of positive 0.25 with each of the other tax base measures, which is only borderline statistically significant. I previously found that western states tended to tax more business purchases and more consumer services, but that pattern does not sustain here. Although western states are somewhat more likely to tax groceries, the difference is statistically insignificant. I find no significant relationship between a state's taxation of groceries and the fraction of its residents who are elderly or poor, or the level of the general sales tax rate.
A couple patterns do emerge in the data. I find that the taxation of groceries has a statistically significant correlation of negative 0.36 with the fraction of the state's popular vote won by President Obama in the 2008 election, indicating that politically liberal states are less likely to tax groceries. Also, the taxation of groceries has a statistically significant correlation of negative 0.29 with per capita income, indicating that higher-income states are less likely to tax groceries.
The trend over time has been to move away from the taxation of groceries. In a 1998 report, Nicholas Johnson and Iris Lav of CBPP said that 20 states taxed groceries. As noted above, only 14 states now do so. This year, two states have reduced their taxes on groceries. Arkansas adopted legislation in March lowering its already preferential 2 percent rate to 1.5 percent. At about the same time, West Virginia lowered its already preferential 3 percent rate to 2 percent.
There have been some efforts this year in the other direction, seeking to increase sales taxes on groceries, but they have been unsuccessful. In February the Utah Senate passed a bill to end the preferential treatment of groceries while lowering the overall sales tax rate, as had been recommended by the Utah Tax Review Commission in 2009, but the measure stalled in the House.8 Rhode Island Gov. Lincoln Chafee (I) has proposed taxing groceries, which are currently exempt, at a 1 percent rate, but this proposal has not been enacted.
Some grocery purchases enjoy a guaranteed sales tax exemption. Federal law prohibits the imposition of sales tax on food purchased under the food stamp program, which was officially renamed the Supplemental Nutrition Assistance Program in 2008, or under the Women, Infants, and Children (WIC) program. Sales tax on food stamp purchases is prohibited by 7 U.S.C. section 2013(a), which was enacted in 1985, and sales tax on WIC purchases is prohibited by 42 U.S.C. section 1786(c)(4).
Although the sales tax exemption for consumer services is likely to gain importance over time as services become an ever-larger share of consumer budgets, the groceries exemption will likely lose importance as groceries become a smaller share of consumer budgets. The national income and product accounts data compiled by the Bureau of Economic Analysis show that expenditures for food and nonalcoholic beverages purchased for off-premises consumption fell from 18.7 percent of total personal consumption expenditures in 1949 to 13.8 percent in 1969, 8.9 percent in 1989, and 6.6 percent in 2009. Nevertheless, the impact of the groceries exemption on the sales tax base is far from trivial, so the issue merits careful consideration.
As I observed in my earlier articles, tax policy analysts are united in condemning the taxation of business purchases and the exemption of consumer services. Opinion is much more divided about the groceries exemption, however, reflecting the fact that unlike the other two policies, it serves a coherent purpose. In their 1998 CBPP report, Johnson and Lav ultimately adopt a favorable view of the exemption, although noting some of its disadvantages. In contrast, State Tax Notes contributing editor David Brunori recently wrote:
The exemption for groceries distorts the market, benefits rich people, and makes compliance and administration more expensive. It's an inefficient and expensive way of providing tax relief. What states should do is exempt purchases made with federal food stamps. Then we'd know that only the poorest folks are getting relief.
I now turn to an examination of the policy issues, first looking at optimal tax theory and then considering real-world complications not reflected in the theory.
Optimal Tax Theory
Under some assumptions about consumer preferences, optimal tax theory supports the conclusion that all consumer products should be taxed at uniform rates, with no preferences for necessities. Strikingly, that conclusion holds even if policymakers put strong emphasis on income redistribution.
The modern theory of optimal taxation assumes that individuals differ in their ability to earn income, so that income inequality arises in the absence of taxation. To promote redistribution, the government taxes wages and makes transfer payments, even though doing so impairs economic efficiency by discouraging work. Policymakers therefore face a trade-off between redistribution and efficiency. Because the simple model has no saving, all income takes the form of wages, and all wages are consumed. Accordingly, the analysis can be viewed as applying to a consumption tax, such as the sales tax, as well as a wage tax.
The special case in which all consumer products should be taxed at a uniform rate occurs when consumers' choice between the various products is weakly separable from their decisions about how much to work. Under weak separability, an individual with a fixed amount to spend on consumer products would divide that money among the various products in the same way, regardless of how many hours of work are required to earn that money. In intuitive terms, working more or less does not alter the relative attractiveness of the various consumer products because no product has any special relationship to work.
Under weak separability, taxing a necessity creates less work disincentive, compared with revenue raised, than taxing a luxury. The reason is that as people work more and increase their earnings, the share they spend on necessities declines. As a result, the share of the last, marginal, dollar devoted to necessities is always smaller than the average share of all earnings that they devote to purchases. But the fraction of the marginal dollar devoted to necessities determines the size of the work disincentives generated by taxing them, while the average share spent on necessities determines the revenue gain from taxing them. Because the marginal share is smaller than the average share, the taxation of necessities has a smaller effect on work incentives, compared with revenue raised.
A simple story explaining why luxuries and necessities should be taxed at uniform rates is that rate differentials create inefficiency by giving consumers an artificial incentive to shift more of their budgets toward necessities. As we will see, that ultimately turns out to be the critical consideration. But that simple story is not the full story. If it were, uniform taxation should be preferred by decision-makers who put greater stress on efficiency, while those who put greater emphasis on redistribution should prefer to impose lower tax rates on necessities. Yet optimal tax theory yields the quite different conclusion that all decision-makers, regardless of their attitudes toward efficiency and redistribution, should join in supporting uniform taxation.
Two changes must be made to the simple story. One concerns how tax rate differentials affect efficiency while the other accounts for the choices available to policymakers concerning the overall level of taxation.
First, the effects of tax rate differentials on efficiency are different than those set forth in the simple story. Uniform taxation is actually not the most efficient policy. Instead, holding the average level of tax rates fixed at some constant positive level, economic efficiency is actually promoted by taxing necessities at a higher rate than luxuries. To be sure, that rate differential inefficiently distorts choices in favor of luxuries at the expense of necessities. But it also reduces the underlying distortion from work disincentives because, as explained above, taxing necessities causes less work disincentive per dollar of revenue raised.
By itself, this first modification makes it even more mysterious why all policymakers should support uniform taxation. It suggests instead that policymakers with a strong commitment to efficiency should support higher tax rates on necessities rather than uniform tax rates. Meanwhile, policymakers with a strong commitment to redistribution should still take the opposite stance, supporting lower tax rates on necessities to advance redistribution, despite the efficiency loss.
But a second modification must also be introduced. The above analysis considers how tax rate differentials across various products affect efficiency and redistribution when the overall level of tax rates is held constant. But policymakers choose the overall level of tax rates, as well as choosing whether and how to vary tax rates across luxuries and necessities. Although those policymakers who are committed to efficiency could advance their goals by taxing necessities at higher rates than luxuries, they could instead lower the overall level of tax rates and reduce transfer payments. Similarly, although policymakers who want to promote redistribution could advance their goals by taxing necessities at lower rates than luxuries, they could instead raise the overall level of tax rates and increase transfer payments. The relevant issue is how each group of policymakers should choose between those different strategies. In other words, how does taxing different products at different rates compare as a way to adjust the efficiency/redistribution balance, relative to simply changing the overall level of tax rates?
In the case of weak separability, optimal tax theory gives a simple answer: It is always better to change the overall level of tax rates than to impose different rates on different products. Regardless of how they feel about the trade-off between efficiency and redistribution, all policymakers should therefore impose a uniform tax rate on all products. Although taxing luxuries at a higher rate would increase redistribution, that goal could be advanced at lower efficiency cost by raising tax rates on all products and providing larger transfer payments. Therefore, redistributive policymakers have no reason to tax luxuries at a higher rate. Conversely, although taxing necessities at a higher rate would reduce inefficiency, that goal could be advanced with less impairment of redistribution by lowering tax rates on all items and cutting transfer payments. Therefore, efficiency-minded policymakers have no reason to tax necessities at a higher rate.
The intuition behind that result is the following. With weak separability, people devote more of their budget to necessities only if they have lower overall consumption. Giving a tax break to those who spend more of their budget on necessities is unnecessary if policymakers can instead give a break to those with low overall consumption (by increasing transfer payments, financed with higher tax rates). The former strategy has the disadvantage that it inefficiently encourages everyone, rich and poor alike, to spend more of their budget on products that are classified as necessities.
This brings us full circle to where we started. The distinctive problem with a preferential rate for necessities is indeed that it creates an inefficient incentive to consume more necessities. But the full story is more complicated. Efficiency alone would actually call for taxing necessities at higher rates. But if policymakers have properly chosen the overall level of taxation, they view that increased efficiency as offset by the loss of redistribution. Tax rate uniformity is not the most efficient policy: It is instead the policy that best balances efficiency and redistribution, provided that the overall levels of tax rates and transfer payments have been set properly.
The conclusion that it is desirable to tax all products at uniform rates depends on the assumption that consumer preferences exhibit weak separability, so that no product has more of a connection to work than any other. More generally, it is likely that some products are complements to work and some are substitutes for work. Optimal tax theory calls for lower tax rates on the complements to work (tax relief for work-related costs) and higher tax rates on the substitutes for work. Rearranging tax rates in that way reduces the inefficiency from work disincentives without impairing redistribution.
Of course, that raises the question of whether food is a work-related cost and therefore deserving of preferential tax treatment. At first glance, there seems to be a clear link between food and work. After all, someone who eats no food will starve to death, obviously making him unable to work. But the relevant question is not what would happen if food purchases were reduced to zero. Instead, the relevant question is what would happen if tax rates were altered slightly from uniformity to give food a small preference, causing people to slightly increase their food purchases and slightly reduce their purchases of other items. Food is a work-related cost if, and only if, that shift would cause people to work more (or to work more productively). In the contemporary United States, it seems unlikely that a slight increase in food consumption at the expense of other consumer products would have a noticeable effect on work, although a different result might apply in developing countries. In the contemporary United States, then, there appears to be no case for giving food tax relief as a work-related cost.
Optimal tax theory offers valuable guidance for tax policy, but only if it is carefully applied with an eye to its limitations. The theory captures critical aspects of real policy choices, notably the trade-off between efficiency and redistribution, but its stylized nature inevitably omits many features of the real world. The conclusions of the theory must therefore be tempered by a consideration of those omitted features. In this case, the consideration of real-world factors strengthens the argument for uniform taxation in some respects and weakens it in others.
The theory assumes that applying differential rates to necessities and luxuries raises no administrative difficulties. In the theoretical model, the different types of products can be distinguished infallibly at no cost. In practice, making that distinction is quite difficult. States have adopted complex rules that still fail to reliably isolate the types of food consumption that are appropriately viewed as necessities. States' definitions of exempt food and beverage purchases are complex because they seek to deny the exemption to some types of food and beverages that are not viewed as luxuries, such as candy and alcohol, and also seek to screen out restaurant meals. In the pursuit of those objectives, the state statutes make exceptions and then make exceptions to the exceptions. Those complications strengthen the argument against preferential treatment.
The relevant New York statutes, for example, begin by exempting "food, food products, beverages, dietary foods and health supplements, sold for human consumption." The exemption is denied, however, to "candy and confectionery," "fruit drinks which contain less than seventy percent of natural fruit juice," "soft drinks, sodas and beverages such as are ordinarily dispensed at soda fountains or in connection therewith (other than coffee, tea and cocoa)," and "beer, wine or other alcoholic beverages."
New York does tax food and drink "when sold in or by restaurants, taverns or other establishments in this state, or by caterers" for on-premises consumption, including any cover, minimum, or entertainment charges. It also taxes some food and beverages sold by those establishments for off-premises consumption. Tax is imposed if the vendor or any person whose services are arranged for by the vendor "serves or assists in serving, cooks, heats or provides other services with respect to the food or drink" after delivery off the premises. Tax is also imposed on food sold for off-premises consumption if "the sale is made through a vending machine . . . except the sale of drinks in a heated state," although tax is not imposed if the food or beverage is sold through the vending machine at a price of 75 cents or less. Moreover, food and drink sold by restaurants and taverns for off-premises consumption are taxed, unless the food and drink are sold in an unheated state and "are of a type commonly sold for consumption off the premises and in the same form and condition, quantities and packaging, in establishments which are food stores other than those principally engaged in selling foods prepared and ready to be eaten." But sandwiches sold for off-premises consumption are taxed, even if they would otherwise qualify for the exception just stated. The statute goes on to exempt food and drink sold to airlines for consumption while in flight; food and drink, except alcohol, sold to elementary and secondary school students at restaurants and cafeterias located on school premises; and food and drink, except alcohol, sold on university campuses to students pursuant to contractual arrangements under which the students do not pay cash at the time they are served.
Those administrative complications strengthen the case for taxing groceries at the same rates as other products. But one important consideration may cut the other way.
It may not always be clear whether tax rates and other transfer payments will actually be adjusted to achieve a desired balance between efficiency and redistribution. That concern undoubtedly helps explain why support for taxing groceries has been stronger among conservatives than among liberals, as suggested by the statistical results mentioned above.
One approach is to adopt specific measures that are intended to offset the sales tax burden on groceries. Five of the seven states that fully tax groceries--Hawaii, Idaho, Kansas, Oklahoma, and South Dakota--offer a credit or rebate specifically described as an offset to the tax on groceries. Those credits and rebates are not based on the amount of groceries actually purchased by the recipient households but instead are based on income and household size.
As Johnson and Lav found in the 1998 CBPP report, the provisions of the credits tend to be restrictive, a conclusion that remains true today. In Kansas, for example, a household may claim the food tax refund only if the household's adjusted gross income is below $35,000 and the household includes a disabled person, a person 55 or older, or a child 17 or younger. The credit is $90 per person if the household income is below $17,500 or $45 per person if the household income is between $17,500 and $35,000. At a 6.3 percent state sales tax rate, even the larger credit amount effectively rebates the tax on $1,429 of food purchases per year ($119 per month) per person. Johnson and Lav also noted that those credits often fall prey to budgetary constraints. That point also still holds; Idaho, which has been increasing the size of its credit, recently voted to delay the increase scheduled for 2011 because of budgetary difficulties.
In the end, relying on these credits and rebates to offset the distributional impact is an uncertain strategy. But it also appears to be unnecessary. The more general point is that the overall level of redistribution achieved in a state, or a country, is likely to depend on the values of its citizens, not on the existence of a particular provision such as a groceries exemption.
Further, the distributional concerns raised by ending the groceries exemption would be alleviated if that policy were accompanied by ending the exemption of consumer services, including services purchased by well-off households, as discussed in my previous article. Of course, groceries purchased with food stamps would remain exempt, as required by federal law. Also, the move to tax groceries could be delayed until the economy has recovered from its current weakness and could then be phased in over a period of time.
The sales tax exemption for groceries is intended to serve the essential purpose of easing the fiscal burden on poorer households. In the end, however, it is not a well-designed policy to achieve that goal. States should move to apply sales tax to groceries in a deliberative manner that maintains protection for those in need.Alan D. Viard is a resident scholar at AEI.