Who Pays Sin Taxes?

Sin taxes are excise taxes that are conceived primarily public health tools, not primarily ways to raise general revenue for governmental operations. 

To lesser extent, excise taxes, more generally, can also serve as self-insurance charges/user fees, for example, in using gasoline taxes to pay for roads, in vaccine taxes to pay for vaccine side effect treatment, or in gambling taxes to pay for gambling addiction treatment.

The dollar amount of tax dollars raised by these taxes is fairly modest compared to general taxes in income, sales, and property ownership. But the idea behind sin taxes is that higher taxes on things that are bad for you, but not appropriate to ban entirely, will discourage their consumption.

The impact of about 2% of total income isn't crushing even for the most heavily impacted clusters of households, although since many of these households pay no federal or state income tax, it is still often their greatest federal tax burden.

In the U.S., high taxes on tobacco products have been quite successful in that regard. Only 8% of Americans purchase and smoke tobacco on a regular basis. These taxes have played an important role in greatly reducing smoking rates compared to our international peers, and producing associated health benefits, in concert with other regulatory developments like limiting where it is legal to smoke, increasing the minimum age to purchase tobacco products, and a cultural shift driving by public interest advertising campaigns and growing public health research that has become widely known. But, these taxes are strongly regressive.

In contrast, alcohol taxes have not had nearly the same consumption reducing effects. While 10% of households buy 80% of alcohol, roughly 50% of American households purchase significant amounts of alcohol on  a regular basis (which is far more dry households than I would have expected). Instead, it has operated as a sales tax on a good for which demand in inelastic, and has not had as clearly demonstrable public health benefits. On the other hand, as the study below reveals, the impact of these taxes was mildly progressive in distributional by income group.

Marijuana and gambling taxes, likewise, seem to be functioning more as a way to bribe median voters into supporting legalization to overcome their weak preferences against legalization, rather than reducing consumption or serving sound tax policy. It probably makes sense to reduce them or phase out these taxes once support for legalizing these activities becomes widespread because legalization didn't lead to the parade of horribles that opponents of legalizing marijuana and gambling claimed that it would.

The body text of the new paper notes that:

In New York City, a 1.75L bottle of vodka might sell for as little as $11.99 of which $7.97 is tax; and a $13.00 pack of cigarettes includes $6.86 in taxes. 

To forward these goals, taxes on sin goods have grown in recent years. In 2009 the federal excise tax on a pack of cigarettes increased from $0.39 to $1.01. . . . All but nine states have substantially raised their tobacco taxes in the last two decades, with the median tax on cigarettes more than quadrupling between 2000 and 2021 from $0.34 to $1.78. Meanwhile, tax revenues from alcoholic beverages have grown, due to both rising consumption and state tax rate increases.

In all, combined federal and state taxes on alcohol and tobacco raise nearly $40 billion annually – an amount comparable to the total federal income tax paid by the bottom half of the income distribution.2

2 For state sin tax revenues, please see https://www.taxpolicycenter.org/statistics/state-and-local-tobacco-tax-revenue and https://www.taxpolicycenter.org/statistics/state-and-local-alcohol-tax-revenue. For federal sin tax revenues, see https://www. taxpolicycenter.org/statistics/types-federal-excise-taxes. For income tax statistics, please consult https://www.irs.gov/pub/irs-soi/17in02etr.xls 

The linked sources indicate that for the year 2015, federal excise tax collections were: $10.37 billion on alcohol and $14.45 billion on tobacco. State and local excise tax collections in the year 2015 were: $7.04 billion on alcohol and $18.17 on tobacco. 

The combined total excise taxes in the U.S. for the year 2015 were $17.41 billion on alcohol, and $32.620 billion on tobacco, for a combined $50.03 billion in total alcohol and tobacco excise taxes at the federal, state and local level. 

The body text of the paper continues, stating:

Just 10% of households account for more than 80% of alcoholic beverage purchases by volume, while the bottom half of the distribution nearly abstains from beer, wine and spirits purchases. For cigarettes, 8% of households are responsible for virtually all purchases. . . 

The top 20% of households [ranked by sin taxes paid] pay roughly 90% of all sin taxes, while more than half of households pay virtually no sin taxes. Following the purchase patterns, taxes on cigarettes are more concentrated than those on alcohol beverages. . . . 

[For analysis purposes each person in the large survey sample was assigned to the best fitting of the eight statistical purchase pattern clusters.] These clusters explain 80% of the overall variation in sin tax burden, while demographics alone explain less than 3%. We identify a tiny group of households (2.5% of the population), whom we label Everything and who consume extremely large amounts of nearly all sin goods. . . . They also tend to be disproportionately from the lowest levels of income and education, white, and older. These households bear a striking similarity to those Case and Deaton (2020) describe as most susceptible to “deaths of despair.” A group of non-drinking Smokers are demographically similar and make up an additional 5.5% of the population. These groups pay 68% of existing sin taxes, which average approximately 2% of income[.] . . . 

[A]lcohol taxes appear progressive at most quantiles of purchasers, and around half of households purchase little to no alcohol. High-income households in the top decile spend around $52 per year on [alcohol excise] taxes, while low-income households in the top decile spend only around $32. Households in the top 5% spend more than twice these amounts.

This comes from two sources. First, high-income households purchase more ethanol (around 35% more) than low-income households at nearly all quantiles of the distribution. However, drinks per adult are roughly constant across quantiles, so extra purchases are mostly explained by larger household size. Second, distilled spirits comprise a higher share of drinks for higher-income households, and are taxed at a higher rate per unit of ethanol (or standard drink) than beer.

Cigarette taxes, on the other hand, appear regressive . . . , though the overall correlation between cigarette taxes and income is still quite weak, ≈ −0.06. Regressivity is driven largely by the extensive margin. At lower levels of income, more than 10% of households smoke, while this is not the case at higher income levels. The lowest income households in the top 5% of the cigarette tax distribution spend more than $300 per year on these taxes, while the top 1% of households spend more than $1000 per year. This is consistent with the fact . . . that the top 1.5% of households account for 60% of all cigarette taxes. These heaviest smokers drive both the average burden and the overall regressivity of the tax.

As the table above indicates, about 60% of households are in clusters that consume very little alcohol or tobacco. 

About 32% of households are in clusters that consume moderate or large amounts of alcohol, but not tobacco (about 93% of people who consume moderate or large amounts of alcohol are non-smokers, similar to the share of the population as a whole). 

About 5.5% of households  are in a cluster that consumes little alcohol but consumes tobacco (a 69%  share of tobacco users, which is surprisingly large). 

About 2.5% of households are in a cluster that consumes significant amounts of both alcohol and tobacco.

Put another way, alcohol consumption and tobacco consumption are almost completely statistically independent of each other.

The demographic profiles of the different clusters is interesting as well.

It also is worth noting that more than two-thirds of sin taxes are paid by people who look a lot like Trump supporters who smoke among whites. The body text's characterization of this group as predominantly "white" while technically true, is somewhat misleading and largely just reflects the fact that "white" is the most common racial category in the U.S. 

This group includes mostly proportionate numbers of whites, of blacks and of Hispanics who both smoke and drink. It includes disproportionately smaller numbers of Asians, Hispanics who smoke but don't drink, people under age 35, people with college degrees, high income people, and parents of minors who smoke and drink. 

These excerpts omit speculation in the article on potential future legislation and in particular the potential impact of a sugary beverages tax. A sugary beverages tax, if enacted, however, would also disproportionately impact the group that pays the highest alcohol and tobacco excises taxes as well, although not as starkly.

The paper and its abstract are as follows:

We find that sin good purchases are highly concentrated with 10% of households paying more than 80% of taxes on alcohol and cigarettes. 

Total sin tax burdens are poorly explained by demographics (including income), but are well explained by eight household clusters defined by purchasing patterns. The two most taxed clusters comprise 8% of households, pay 68% of sin taxes, are older, less educated, and lower income. 

Taxes on sugary beverages broaden the tax base but add to the burdens of heavily taxed households. Efforts to increase sin taxes should consider the heavy burdens borne by few households.

Christopher Conlon, Nirupama L. Rao & Yinan Wang, "Who Pays Sin Taxes? Understanding the Overlapping Burdens of Corrective Taxes" NBER Working Paper 29393 (October 2021) DOI 10.3386/w29393.