Smoking, Insurance, and Social Cost
by Robert E. McCormick, Robert Tollison, and Richard E. Wagner

Robert E. McCormick is a professor of economics at Clemson University. Robert Tollison, and Richard E Wagner.

Strong political resistance to new taxation has forced supporters of higher taxes to seek alternative justifications for the taking of more money from the public. That incentive, combined with the desire by many policy makers to control the behavior of others, has breathed new life into what used to be called sin taxes that traditionally targeted alcoholic beverages and tobacco products. But those taxes are now called "corrective taxes." They supposedly compensate for the social costs that private behavior purportedly imposes on the public at large. If successful, the current "corrective tax" attack on tobacco will likely next target other public policy debates, for example, concerning alcohol use, diet, or other forms of behavior said to have social costs. Thus, exposing the errors of the attack could head off similar assaults in the future.

Tobacco has been the prime target of corrective tax legislation. Currently, each pack of cigarettes sold in America carries a tax burden that averages a little over 50¢. Twenty four cents is taken by the federal government and the remainder by state and local governments. A pack of cigarettes has an average retail price of around $1.50. Taxes raise the price of cigarettes by some 50 percent.

While the tax is high by anyone’s reckoning, some critics argue that it is still too low to cover all the social costs of smoking, principally higher medical bills. They want the levy increased to $2.00 or $3.00 per pack. Fortunately, newer studies argue that prevailing levels of tobacco taxation are about right, or perhaps even a little too high. Those studies distinguish between the costs that smokers might impose on themselves and the costs that smokers might impose on nonsmokers, with only the latter cost possibly justifying corrective taxation.

But the new approach can be taken even further. Social costs in fact consist of two components; true external costs and wealth transfers between smokers and nonsmokers. Making that distinction suggests that smokers are currently subsidizing nonsmokers. Further, both the external costs and the transfers are due to the pooling of the medical and retirement costs of smokers and nonsmokers, and barriers that prevent insurers from assigning higher costs to those who produce those costs. But in any case, higher taxes will not eliminate the current distortions in the system. They will themselves produce external costs that destroy wealth, harming smokers and nonsmokers alike.

 

External Costs And Corrective Taxation

In two widely cited reports, the Office of Technology Assessment (OTA) of the U.S. Congress surveyed a variety of independently-produced estimates of the alleged social costs of cigarette smoking. In its 1985 report, OTA found as a best-estimate that smoking resulted in $65 billion of such costs, which worked out to $2.17 per pack of cigarettes sold. In its 1993 report, containing data through 1990, OTA pegged the social cost of smoking to be $68 billion, or $2.59 per pack.

It is now widely recognized in scholarly circles that those kinds of figures confound personal and social or external costs. The bulk of the costs represented in the literature surveyed by OTA are personal costs borne by smokers, such as lost earnings caused by illnesses and earlier death. The newer scholarship finds that the external costs of smoking are positive, but generally less than present levels of tobacco taxation.

Willard G. Manning, et al., for example, in a 1989 study published in the Journal of the American Medical Association found that cigarettes are presently overtaxed and alcohol undertaxed, as compared with those tax rates that would offset the external costs associated with consumption of those products. Kip Viscusi also has published extensively on the tobacco issue. [see page 29]

There are of two types of alleged external costs associated with smoking. First, smokers may incur higher medical costs than nonsmokers in the short run, due to negative effects of smoking on health. Second, smokers have a shorter life expectancy than nonsmokers, and so make fewer claims on retirement programs.

How should present and future costs be compared? Future costs should be brought up to present value. When one lends money, one is giving up the immediate use of that money for a specified period of time. Since money that can be used immediately generally is valued higher than money that can be used in the future, one expects a larger payback in the future, that is, interest on one’s investments. Similarly, when one obtains current benefits through borrowing, one expects to pay more in the future to cover principal plus interest charges. Thus, one must "discount" future costs and benefits when comparing them with current ones. A zero discount rate would mean that current and future costs and benefits are valued equally. A discount rate of 50 percent, however, would mean that very little value is placed on future benefits and little account is taken of future costs.

The Manning estimates on the net costs of smoking use discount rates of 0 percent, 5 percent, and 10 percent. Viscusi’s estimates use 0 percent, 3 percent, and 5 percent. The new literature finds that at a discount rate of 4 percent or less smokers subsidize nonsmokers, and that above 4 percent the reverse is the case. We will use the 0 percent and 5 percent rates in our discussion since they span the range at which the change occurs.

 

Preliminary Smoking Cost Model

When considering whether smokers are subsidized by nonsmokers, the standard scenario holds that, on average, a smoker will incur higher claims for medical costs initially compared to nonsmokers. But in later years, nonsmokers, who die younger, will make fewer claims under the social security programs. Whether the lifetime combined cost attributed to a smoker exceeds that attributed to an average nonsmoker depends on the rate of discount.

The actual calculation of costs is quite complex. To clarify the issues involved, consider a simple, hypothetical example set forth in Table 1. The table describes an economy comprised of five individuals, two smokers and three nonsmokers. Smokers live two periods and nonsmokers live three periods. In the first period, a smoker and a nonsmoker enter the population. The smoker incurs twice as many medical expenses as the nonsmoker, both in the initial period and in the next period. At the end of the second period the smokers die, leaving only the nonsmokers to enter the final, retirement period. The steady state population of that economy is five, two smokers and three nonsmokers, with four individuals of working age supporting one retiree.

The annual cost of medical expenses averaged over a lifetime is taken here to be $200 for nonsmokers. Annual retirement costs are set at $300. While those are hypothetical numbers, they give some indication of the approximate relationship of different costs. This vastly simplified model captures the salient features of the various claims about the external cost of smoking. The higher medical costs incurred by smokers in the first two periods reflect their assumed poorer health. Zero retirement costs reflect the smoker’s shorter average life expectancy. As compared with nonsmokers, smokers impose differentially higher costs in the first two periods while they are a source of saving in the third period. To show clearly the possible subsidy situation, three extreme discount rates, 20 percent, 30 percent and 40 percent, are used.

At a 30 percent discount rate, the present value of medical care plus retirement is essentially the same for smokers and nonsmokers. For medical care alone, the present value of the cost to smokers is $135.64 higher ($544.38-$408.74) than the cost to nonsmokers. For retirement, however, smokers are $136.55 less costly than nonsmokers in present value terms.

With a discount rate of 20 percent, smokers are a source of net benefit to nonsmokers. Smokers incur a higher present value of lifetime medical care costs than nonsmokers by $131.94 ($611.11-$479.17). That higher cost, however, is more than offset by the higher present value of lifetime retirement benefits for nonsmokers ($173.61) than for smokers. The present value of medical plus retirement costs means that smokers subsidize nonsmokers by $41.67 ($173.61-$131.94). That aggregate differential can be placed on a per pack basis by dividing by the number of cigarettes smoked. If, for example, two hundred total packs of cigarettes are smoked annually by the two smokers, smokers provide a net benefit of 21¢ per pack for nonsmokers ($41.67/200).

With a discount rate of 40 percent, the situation is reversed and smokers impose a burden on nonsmokers. With respect to the present value of the lifetime cost of medical care, smokers are $135.57 more costly to serve than nonsmokers ($489.80-$354.23). Nonsmokers have net lower retirement expenses of $109.33. Thus, on balance smokers are more costly by $26.24. Under the assumption that two hundred packs are smoked annually, the excess cost of smokers amounts to 13¢ for each pack smoked.

 

Viscusi’s Cost Per Pack

That simple scheme in Table 1 captures the central structure of the estimates generated by the new scholarship on the economics of smoking. Kip Viscusi has made careful estimates of the actual costs and benefits per pack of cigarettes, broken down in terms of various factors and based on empirical research on actual costs.

Viscusi, for example, adjusts for the changes in tar content of cigarettes overtime. Low tar cigarettes are thought to produce fewer health problems than high tar ones. And tar levels have fallen by a factor of three or four over the postwar period. Where Manning et al. treated all cigarettes as equivalent in measuring consumption, Viscusi constructed estimates of consumption that weighted cigarettes smoked in different years by their tar content. That procedure yields estimates of external costs that are less than if tar content is held constant.

Table 2 shows Viscusi’s estimates of the cost per pack of cigarettes smoked. Using a zero rate of discount, Viscusi finds that each pack smoked generates a cost of -$1.63, that is, value is transferred from smokers to nonsmokers. That result is similar to the illustration in Table 1 with a 20 percent rate of discount, where the present value of the lower retirement costs of smokers exceeded the higher present value of their medical care costs. When expressed on a per pack basis, each pack reduces the claims of smokers on retirement pensions by $2.89. When the future savings from unmade social security payments are weighted equally with current health-related payments, as a zero rate of discount does, smokers are a tremendous source of benefit to nonsmokers, according to the new literature on the economics of smoking.

When a 5 percent rate of discount is used, the situation is reversed because the present value of those social security savings decline much more sharply than does the present value of medical care costs. What was a lower present value of lifetime costs for smokers of $1.63 per pack with a zero rate of discount becomes a higher present value of 32¢ per pack with a 5 percent rate of discount.

But in evaluating those costs, it is important to note that excise tax collections have not been taken into account. The excess of Medicare costs, over social security savings, expressed in present value terms, i.e., 32¢, is equal to around 60 percent of the prevailing rate of cigarette taxation in the United States of around 50¢ per pack. Jane G. Gravelle and Dennis Zimmerman in a 1994 Congressional Research Service report estimated that a 10 percent rate of discount applied to the framework of Manning et al., updated to 1993, would yield an excess smoker cost of 42¢ per pack. Thus, even with a 10 percent discount rate, the estimated excess cost of smoking would still be less than current levels of taxation.

The estimates by Viscusi, Manning, and others, point to a crossover at a discount rate of about 4 percent when smokers go from providing benefits to nonsmokers to actually costing nonsmokers. Those scholars of the new economics of smoking contend that it is that estimated transfer that represents the external cost of smoking which can be internalized through a corrective tax of the same amount. But even that formulation exaggerates the external cost of smoking, because it fails to distinguish transfers from external costs and, thus, overstates the amount of taxation necessary to internalize the externality.

 

The Problem Of Insurance Pooling

To understand the cigarette subsidy problem, it is useful to know what kind of arrangement does not result in subsidies. There are two simple scenarios under which no issues of external costs or income transfers would arise. The first is when all services, medical and retirement, are privately provided in full. The second would require the segregation of smokers and nonsmokers within the medical and retirement insurance programs. Thus, there would be one social security program for smokers and another one for nonsmokers. Under such an arrangement, there would be no way for smokers to impose external costs on nonsmokers, nor any possibility for income transfers between the two groups. If health care were costlier to deliver to smokers than to nonsmokers, there would be some combination of higher price and lower coverage chosen by smokers as compared with nonsmokers. In any case, smokers would pay for their higher costs of medical care. If the ratio of retirement years to working years were lower for smokers than for nonsmokers, smokers would pay a lower real price for retirement coverage than would nonsmokers, with some combination of lower social security taxes while working and higher benefits while retired. In any case, there would be neither subsidization nor external cost as between smokers and nonsmokers.

Possible issues of subsidization and external cost arise only when smokers and nonsmokers are pooled and common pricing is practiced. With respect to the hypothetical data illustrated in Table 1, under either market pricing or collective segregation, smokers would pay $400 annually for their medical care and nonsmokers would pay $233.33. Under common pricing, however, the five members of the society would each pay $300. Common pricing thus subsidizes each of the two smokers by $100 per year, and finances those subsidies by added charges of $66.67 on each of the three nonsmokers.

The situation is reversed, of course, for retirement. From Table 1, the two smokers, who receive no benefits, would pay $60 apiece towards the $300 annual cost of retirement. Each of the three nonsmokers thus gains $40 ($60 × 2 smokers/3 nonsmokers). When medical care and retirement are combined, each nonsmoker loses $26.67 ($66.67 subsidy to smokers -$40 subsidy from smokers). Each smoker receives a subsidy of $40 annually ($100 subsidy from nonsmokers -$60 to nonsmokers). If each smoker smokes one hundred packs of cigarettes annually, the $80 aggregate subsidy distributed over the two hundred packs of cigarettes would be 40¢ per pack. The magnitude derived from that simple illustration is close to that derived from Viscusi’s careful calculations, and somewhat higher than those derived earlier by Manning et al.

The subsidies from pooling insurance, however, are not external costs. They are transfers. An external cost will exist only to the extent that taxes or transfers change the conduct of smokers or nonsmokers. Any claim of external cost must be based on such changes in conduct that would effect the overall wealth in the economy. For instance, if smokers subsidize retirement for nonsmokers, nonsmokers might substitute leisure for work, that is, retire earlier. There would be less wealth created in society because individuals who might otherwise work more years would be out of the labor force. Consider the case of complete inelasticity of demand, in which no matter how high the price, smokers continue to use the same amount of cigarettes. In such a case, the transfer from nonsmokers to smokers would involve no external cost because no one changes her behavior as a result of the transfer.

If insurance pooling forces nonsmokers to subsidize the medical bills of smokers, this would not be an external cost. But if, as a result, smokers were to smoke more, and, as a result, they were to incur even higher medical bills, this would be an external cost. To discover the magnitude of external costs, it is necessary to examine the market adjustments that are likely to result because of the subsidy created by pooling and common pricing.

 

The External Cost Of Smoking

Suppose the claim is accepted that pooling lowers the price of cigarettes by 30¢ per pack from what would be required for full-cost pricing. Any resulting external cost stems from such things as increased smoking resulting from the subsidy. Many estimates have been made of the elasticity of demand for cigarettes. Most estimates, for example, those of Gary S. Becker and Kevin M. Murphy, range between -0.5 and -1.0. The former means that for every 1 percent rise in price, demand goes down by 0.5 percent. The latter means that a 1 percent price hike would result in a 1 percent drop in demand. We shall illustrate our argument for both elasticity estimates, as this should provide a reasonable estimate of the plausible range over which the external cost of smoking might vary.

For the sake of argument, assume that the real total costs of smoking, including medical and other costs, is around $3 per pack of cigarettes, based on work surveyed by the Office of Technology Assessment.

Figure 1 illustrates estimates of the external cost of smoking in a manner that is consistent with the ideas developed in the new economics of smoking, while separating pecuniary transfers from external costs. The full price of cigarettes, assumed to be at least $3 per pack, is denoted by PF, a price at which QF of cigarettes would be purchased. But a 30¢ subsidy brings the price down to $2.70 per pack, denoted here by Ps. The result of $2.70 would thus involve $1.50 that smokers pay to vendors, 50¢ of which is collected in tax, and $1.20 that is the per pack equivalent of the other personal costs that smokers bear. In the presence of that subsidized price, consumers purchase Qs cigarettes, or approximately 30 billion packs annually.

From that point of departure, it is simple to calculate the external cost created by the subsidy; in Figure 1, the triangle ABC. That is approximately half (0.5) of the product of the fall in price, PF-Ps, (30¢) and the increase in quantity, Qs-QF, resulting from the subsidy created by insurance. Nonsmokers subsidize smoker by $9 billion, as shown by rectangle FABCDE (30 billion parks × 30¢ per pack subsidy), but smokers only receive $8.6 billion in benefits from the subsidy, the figure FACDE (27.3 billion cigarettes × 30¢ per pack subsidy + [1/2 × 2.7 billion cigarettes] × 30¢ per pack ). The triangle ABC, representing $405 million, is wasted. It is the external cost of smoker subsidies.

If the subsidy reduces the full price of cigarettes to smokers by 10 percent, it will lead to somewhere between a 5 percent and a 10 percent increase in the number of cigarettes smoked, depending on the elasticity of demand. If the elasticity of demand is -0.5, QF will be 28.5 billion packs (30 billion packs/1.05). If elasticity is -0.1, then consumption will be 27.3 billion packs (30 billion packs/1.1). In any case, the calculation of external cost is straightforward. The change in the number of packs smoked due to the subsidy ranges between 1.5 billion and 2.7 billion packs. Hence, the external cost associated with pooling and common pricing lies between $225 million (0.5 × 1.5 billion packs × 30¢) and $405 million. (0.5 × 2.7 billion packs × 30¢). With 30 billion packs of cigarettes being smoked annually, the external cost of smoking would seem to be somewhere between three-fourths of a penny and 1.4¢ per pack.

A tax of 30¢ per pack would raise the full price of cigarettes from Ps ($2.70) to PF ($3.00) in Figure 1, and this would internalize the externality by reducing consumption to QF. The external cost of smoking, however, does not average 30¢ per pack. It averages about a penny a pack. The difference between the two magnitudes indicates the amount of wealth transfer from nonsmokers to smokers. But such a tax would not only eliminate the subsidy for smokers, it also would reduce the level of overall economic activity. That so-called dead-weight loss (triangle ABC) is between $115 million and $225 million is lost.

If smokers and nonsmokers could interact and trade freely, they would avoid the deadweight losses created by subsidies and taxes. Smokers and nonsmokers might be willing to cover their own expenses and split the difference in the dead weight loss.

 

Conclusion

Two clear conclusions stand out. First, the issue of nonsmokers subsidizing smokers only arises because medical and retirement insurance pools the two groups and does not differentiate between them when assigning costs or distributing benefits. Those are transfer costs. And second, taxing smokers might remove their subsidy and real externalities, but it does not help nonsmokers; it results in an actual loss of wealth to the economy.

It was assumed for the sake of argument that health insurers cannot separate smokers and nonsmokers in current programs. But insurers in other markets separate risk classes by all manner of characteristics, including age, gender, health background, and job category. The fact that there are not differential rates for smokers could mean that there are no externalities worth internalizing. Smokers and nonsmokers may be charged common prices for health insurance, in contrast to the prevalence of differential pricing for life insurance, because there is no sound actuarial basis for seeking to place smokers and nonsmokers into distinct categories. And if there are actuarial differences between smokers and nonsmokers, as is assumed in the new economics of smoking, the external costs those differences entail could be internalized for about a penny a pack. Or, there might be externalities and good reason to treat smokers and nonsmokers differently in medical and retirement programs. That suggests that government regulation of insurance markets prevents such internalizing efforts. Thus it is a political, not and economic defect, that keeps smokers and nonsmokers pooled in government medical and retirement systems. Politics is the cause of the subsidy.

Regulation, 1997, Vol. 20, No. 3. Published four times a year by the Cato Institute. Copyright 1997 Cato Institute. Regulation is available on the World Wide Web and readers may subscribe on-line. Subscription rates are $18 per year for individuals and $28 per year for libraries and institutions. Please send changes of address and subscription correspondence to: Circulation Department, Regulation, 1000 Massachusetts Avenue N.W., Washington, D.C. 20001, or call 202-842-0200.

ISSN 0147-0590 INTERNET subscriptions@cato.org WEB SITE http://www.cato.org