The median from this table is the number that made it into the table at the top.

Obesity

America has, variously, a crisis, an epidemic (Wang and Beydoun 2007), and an apocalypse of obesity. Regardless of the descriptor, (Finkelstein et al. 2009) finds that the obese spend almost $1,500 more per year on medical services and that Medicare and Medicaid account for at least half of this increased spending³. “We estimate that a 1-unit increase in BMI for every adult in the United States would increase annual public medical expenditures (i.e., direct medical costs) by $38.7 billion; an average marginal cost of $175 per year per unit of BMI for each adult in the United States.” (Parks et al. 2012)

This is probably the right moment to bring up (again) that Pigouvian taxes are not a good response to all externalities (Fleischer 2015). In particular, a one point increases in BMI is very harmful to the health of some while it’s healthful for others. Thus, a tax which penalizes all weight gain equally could be called a Procrustean Pigouvian tax. It’s also worth pointing out that the optimal weight from a public health perspective isn’t settled (Afzal et al. 2016).

Alcohol

Coming next in our countdown is a classic sin tax—alcohol. There’s yet more double-counting between this and driving because the biggest alcohol-related externality is drunk driving. With a 5% discount rate, the cost breakdown is as follows:

(One very dubious part of (Manning et al. 1989) is that it counts harms to family members as internal rather than external.)

Because “nearly all external costs are caused by 10% of the drinking population who consume one-third to one-half of all alcohol sold” (Cnossen 2007) a targeted tax on excess drinking as (Manning et al. 1989) calculates would be the first best option. Alas, it’s far from obvious how this targeting would be implemented. We adjust the tax to look at total alcohol consumption then. (Manning et al. 1989) reports that 40% of alcohol consumption is in excess of two drinks per day (their definition of ‘excess drinking’). Multiplying “Total net costs” by 0.4 gets us the final cost of $1.10 per ounce ($37 per liter) which appears in the top table. ((Harwood 2000) gives a broadly similar estimate of $62 per liter.)

Agriculture

I know we’re not supposed to speak ill of that most sacred American—the farmer—but modern intensive agriculture creates lots of external harms in the process of food production. Broadly, these harms can be categorized as harms to natural capital and harms to human health. They total up to $34 billion and are broken down in more detail in the following table:

The figures above are likely to underestimate the true size of external harms in agriculture:

• some costs are known to be substantial underestimates (for example, acute and chronic pesticide poisoning of humans, monitoring costs, eutrophication of reservoirs and restoration of all hedgerow losses);
• some costs currently cannot be calculated (for example, dredging to maintain navigable water, flood defences, marine eutrophication and poisoning of domestic pets);
• the costs of returning the environment or human health to pristine conditions were not calculated;
• treatment and prevention costs may be underestimates of how much people might be willing to pay to see positive externalities created;
• the data do not account for time lags between the cause of a problem and its expression as a cost (i.e. some processes long since stopped may still be causing costs; some current practices may not yet have caused costs);
• this study did not include the externalities arising from transporting food from farms to manufacturers, processors, retailers and finally to consumers.

(Pretty et al. 2001)

Municipal waste

As the malodor makes obvious when passing by a garbage dump, municipal solid waste isn’t entirely inert. It leaches into water and off-gases. These environmental harms are external to the initial disposal.

Unfortunately, as we get further down the list, finding good estimates of these costs gets harder and harder. (Repetto et al. 1992) estimates non-market costs of $135/ton of municipal solid waste in high-cost areas and $81/ton in low-cost areas, but doesn’t really show the work that leads up to this conclusion. The table below has a more detailed breakdown but the numbers are focused on Puerto Rico. Regardless, the final estimate of $77 in the table up top is the average of the average estimates from both sources (that is, $108/ton from (Repetto et al. 1992) and $46/ton from (Miranda and Hale 1999)).

Smoking

We finally reach the other classic sin tax—smoking. Surprisingly, there seems to be fairly widespread belief that cigarette taxes are, if anything, too high to be justified on purely Pigouvian grounds. (Manning et al. 1989) is the seminal paper in this area and it answers its titular question in the affirmative—smokers do pay their way. The basic mechanism is that while “nonsmokers subsidize smokers’ medical care and group life insurance, smokers subsidize nonsmokers’ pensions and nursing home payments” (Manning et al. 1989). The breakdown showing this follows:

Even at a 5% discount rate, which produces the highest estimate of external harms (other discount rates actually suggest that cigarette consumption has net positive externalities), $0.54 per pack is less than the current federal cigarette tax of $1.01 per pack.

All that said, there are several acknowledged omissions in these estimates. The big one is environmental tobacco smoke (ETS; secondhand smoke). At the time of the studies (and perhaps still?), the precise health effects of ETS were unknown so the costs weren’t included in the above estimates. (Chaloupka and Warner 2000) explains:

The potential role of ETS costs in reevaluating the net negative externalities associated with smoking is seen by considering the following figures. Manning et al. (1991) noted that inclusion of the costs of 2,400 lung cancers from ETS (a fairly conservative estimate of this toll (Environmental Protection Agency, 1992) ) as external costs would add approximately 19 cents per pack in external costs (updated to 1994 dollars). In addition, inclusion of the costs of neonatal care for smoking-related low- birth-weight babies would add 3 cents to the total, while including fetal deaths attributable to smoking would add yet another 19 cents. Deaths from smoking-related fires would add a further 9 cents. The ETS costs would skyrocket if one included the estimated 30,000-60,000 heart disease deaths recently associated with ETS (Glantz and Parmley, 1995), adding perhaps 70 cents to the total social costs per pack. Similarly, inclusion of the smoking-induced respiratory tract infections and cases of aggravated asthma in children (Environmental Protection Agency, 1992) would boost the total further, as would inclusion of the long-term developmental disabilities in smoking-related low-birth-weight babies (Hay, 1991). All told, the social costs per pack could easily mount toward several dollars if all of the health hazards associated with ETS are real, many are treated as external to the basic consuming unit, and if all or even a significant fraction of the associated costs are included.

Antibiotics

The majority of antibiotics used in the US are used for animal agriculture (it’s those beloved farmers again)—18.4 million pounds in agriculture compared to 7.0 million pounds in human medicine. Regardless, antibiotic use leads to antibiotic resistance. This externality should be priced at “between 29 and 287€ per kilogram active substance or between 9 and 86% of the average price of commonly used antibiotics” (Vågsholm and Höjgård 2010). Using the 2010 exchange rate EUR/USD exchange rate of 0.75 and adjusting for inflation, we find that this works out to $20–$201/lb of active substance. Picking the middle of the range produces the estimate of $110/lb used in the top table. The total cost calculated is in broad agreement with the estimated national cost of between $100 million and $30 million reported in (Rudholm 2002).

Debt

Somewhat non-obviously, borrowing also has external costs. “A borrower who has one more dollar of liquid net worth when the economy experiences a bust relaxes not only his private borrowing constraint but also the borrowing constraints of all other insiders. Not internalizing this spillover effect, the insider takes on too much debt during good times.” (Jeanne and Korinek 2010)

On a percentage basis, this externality is small and procyclical—0.56 percent of outstanding debt for small and medium enterprises in a boom and 0.48 percent of outstanding debt for households in a boom (Jeanne and Korinek 2010). It only amounts to a sizable cost because the quantities of debt involved are so enormous. American households took on $556 billion of new debt in the year up through Q3 of 2018. SME outstanding debt totaled 5.34 trillion in 2009 (Jeanne and Korinek 2010) (I couldn’t find info on the current rate of SME debt increase.).

Guns

I couldn’t find any direct estimates of the externalities of firearms so I threw together my own estimate. The key number I sought was the number of accidental firearm killings of another person. I ruled out self-killings as internalities (Which is not strictly correct. If a child accidentally kills themselves with their parent’s gun, that’s an externality.). I ruled out intentional killings on the grounds that some unknown fraction of these would still occur even in the absence of guns. Clearly though, the resulting estimate is a conceptual lower bound.

(Xu et al. 2016) records 505 deaths from accidental discharge. (Hemenway, Barber, and Miller 2010) reports that in half of such fatalities, the victim was shot by another person.

In 2016, 16 million guns entered circulation in the U.S. (base on 11.5 million manufactured in the U.S., 376 thousand exported and 5.1 million imported).

Using an even $9 million as the value of a statistical life, we find that the estimated cost of 252 deaths is around $2.26 billion. We divide by the 16 million guns entering circulation to get $134 per gun sale as an order of magnitude estimate of the appropriate Pigouvian tax.

Soda

Even though the soda tax is a ‘popular’ policy, I couldn’t find any serious attempts to estimate the optimal Pigouvian tax for soda. Any attempt would be complicated by the fact that the marginal cost of soda consumption varies widely but systematically across consumers (Fleischer 2015)—it would be much higher for those at risk of type 2 diabetes, for example.

Gambling

The three categories of external costs highlighted in (Clotfelter 2005) are:

Problem gamblers

Problems gamblers constitute ~2% of the adult population. They “driving themselves into bankruptcy, their families into hardship, and their personal relationships into ruin”. Their are inevitably social costs which accompany these problems.

Undermines government credibility

“Government, through its schools, laws, and public pronouncements, advocates certain beliefs, including the value of productive work. To the extent that the sponsorship and, particularly, the promotion of gambling undercut that traditional message, legalized gambling creates a second category of external costs.”

Organized crime

Lotteries don’t seem to do much to encourage organized crime but other forms of legalized gambling may.

Finance

The three categories of external costs highlighted in (Masur and Posner 2015) are:

Speculation

“A transfer of money does not generate social value unless it is part of a transaction that reduces risk or otherwise enables people to spread receipts of money across times or states of the world in a way that advances their interests.”

Races

“When new information about [current] events […] comes into existence, market participants will scramble to be the first to trade on the information. […] This activity is socially wasteful. […] [T]he broader market will not benefit at all if information about terrorism, as embodied in market prices, reaches them a nanosecond quicker than it otherwise does.”

Debt

“Short-term debt creates a negative externality. When a depositor or other short-term lender withdraws money, it increases the probability that the borrower will not have enough funds to pay other lenders when the loans are due or demanded. Those lenders will not be able to recover in full because of bankruptcy.”