PB 4

RAISING REVENUE BY TAXING ACTIVITIES WITH SOCIAL COSTS*** JOSEPH J. CORDES,** ERIC M. NICHOLSON* AND FRANK J. SAMMARTINO* Introduction T HE desire to raise additional federal revenue without raising income taxes has sparked interest in altemative reve- nue sources, including higher excise taxes on alcohol and tobacco, and new taxes on pollution. An important attraction of these taxes is that, compared to other taxes, they are perceived to have less harmful, per- haps even beneficial effects on how mar- kets allocate resources among competing ends.

Taxes on specific goods or activities are generally viewed as interfering with choices that consumers and producers make in the marketplace. This is the point of departure for the extensive literature on optimal taxation that examines how to minimize the economic costs caused by such interference.' There are, however, also instances in which govemment interference with mar- ket choices may he beneficial. For exam- ple, if goods impose costs or provide ben- efits to society that are not fully reflected in the market price, too many or too few resources will be devoted to satisfying de- mands for such goods. To the extent that smoking, abusive drinking, and pollution impose costs on the rest of society, taxing these activities may affect market choices differently than other taxes. Unlike in- come taxes or other excise taxes, which drive a wedge between private and social costs and returns, taxing tobacco, alcohol, and pollution may bring private and so- cial costs and returns into closer align- ment by causing market prices to reflect more fully the external costs of smoking, drinking, and pollution.

Yet, identifying the presence of exter- nal costs is only the first step in deciding •U.S.

Congressional Budget Office, Washington, DC 20515.

**U.S.

Congressional Budget Office, Washington, DC 20515 and George Washington University, Wash- ington, DC 20052.

how, and indeed whether, to tax activities that impose social costs. The extensive economic literature on consumption and production externalities concludes only that if activities impose costs (benefits) on third parties that are not fully reflected in the equilibrium market price, then in many but not all cases some set of taxes (subsidies) could improve the allocation of resources. The literature offers little practical guidance about how to structure taxes to correct externalities. To take this next step, one needs more detailed infor- mation about how certain consumption or production activities impose external so- cial costs, as well as about the size of these external costs. One also needs to know how consumers and producers are likely to re- spond to the imposition of corrective taxes.

This paper surveys some results of cur- rent research on the social costs of smok- ing, abusive drinking, and pollution. In doing so, we hope to provide some back- ground to tax economists, practitioners, and policymakers who may be somewhat unfamiliar with what health economists and other social scientists have to say about the costs of smoking and drinking, and what environmental economists have to say about taxes on pollution.

Taxing Alcohol and Tobacco It is now generally recognized that cig- arette smoking poses severe health risks for smokers, as does abusive alcoholic beverage consumption for drinkers. The 1989 Surgeon General's Report on reduc- ing the health consequences of smoking summarized current medical findings concerning the link between smoking and disease. The report cited cigarette smok- ing as a major cause or a contributing fac- tor of cancer, cardiovascular and respi- ratory disease, and a probable cause of unsuccessful pregnancies and low-birth- weight babies.^ A total of 337,000 deaths 343 344 NATIONAL TAX JOURNAL [Vol. XLIII were attributable to the 10 leading smok- ing-related illnesses in 1985, represent- ing 22 percent of all deaths among men and 11 percent of deaths among women.

A total of 390,000 deaths from all causes were attributable to smoking.

Smoking contributes to fire-related in- jury and property damage. According to the National Fire Protection Association, smoking-related fires—the leading cause of civilian deaths from fires in 1986— caused 1506 civilian deaths, 3,559 civil- ian injuries, and $402 million in property damage.^ The 1986 Surgeon General's report on the health consequences of involuntary smoking concluded that the health of non- smokers is also at risk from smoking. That report cited involuntary smoking as a cause of disease, including lung cancer, Eimong healthy non-smokers.* An estimated 105,000 deaths in 1987 were the result of alcohol-related causes.^ While the greatest medically-related health hazard from chronic alcohol con- sumption is liver disease, alcohol con- sumption is associated with a variety of other diseases and illnesses including cancers, mental disorders, and cardiovas- cular and respiratory diseases. About half of alcohol-related deaths in 1987 were at- tributable to alcohol-related diseases and illnesses.

Alcohol abuse is a major factor in motor vehicle accidents. According to the Na- tional Highway Traffic Safety Adminis- tration, 40 percent of all traffic fatalities in 1987 involved a legally intoxicated participant (either a driver, pedestrian, or bicyclist with a blood alcohol content in excess of .1 percent, the legal level of in- toxication in most jurisdictions). About 10 percent of police-reported motor vehicle crashes also were alcohol-related, with approximately 534,000 people suffering injuries in non-fatal alcohol-related crashes.

Drinkers who cause traffic accidents in- jure many besides themselves. Of the 18,500 traffic fatalities in 1987 involving a legally intoxicated participant, about 37 percent of those who died were passen- gers of vehicles, or drivers and pedes- trians who were not drunk.

Economic Costs of Smoking and Drinking One method for estimating the total economic costs of smoking and abusive al- cohol consumption is to multiply the costs of illnesses and accidents by the attrib- utable risk of tobacco and alcohol con- sumption. The attributable risk is the fraction of total occurrences of a particu- lar illness or accident associated with the specific risk factor of smoking or drink- ing. The total cost associated with a par- ticular illness or accident is usually di- vided into direct and indirect costs. Direct costs refiect immediate outlays such as the expenses of medical treatment for smok- ing and alcohol-related illnesses, property loss from accidents and fires, and special programs such as for highway safety or for victims of fetal alcohol syndrome. In- direct costs are future outlays such as the value of lost productivity due to attrib- utable illnesses and premature death.

Two recent studies provide roughly similar estimates of the economic costs of smoking. Considering only the economic costs of cancer, cardiovascular and respi- ratory diseases attributable to smoking, the Office of Technology Assessment (OTA) estimated total costs of smoking of be- tween $38 billion and $95 billion, with a middle estimate of $65.0 billion in 1985, while Rice, et al. estimated total costs of" $53.7 billion in 1984.^ Based on estimated total consimiption of cigarettes in the U.S.

of about 30 billion packs in 1984 and 1985, the middle OTA estimate and the esti- mate by Rice suggest total costs of be- tween $1.79 and $2.17 per pack.

A 1984 study by the Research Triangle Institute (RTI) calculated that the eco- nomic costs of alcohol abuse were $89.5 billion in 1980 which, the study esti- mated, projected to $116.7 billion in 1983.^ Based on estimated consumption of ap- proximately 500 million gallons of pure alcohol in 1983, this estimate suggests to- tal economic costs of about $1.80 per ounce of alcohol consumed.* External vs. Internal Costs Estimates of the total economic costs of smoking and alcohol abuse are not the es- No.

3] J. J. CORDES, E. M. NICHOLSON AND F. J. SAMMARTINO 345 timates needed to evaluate whether cur- rent taxes on alcohol and tobacco cause smokers and abusive drinkers to face the full social costs of their actions. For this purpose it is necessary to separate total economic costs into those that are paid by smokers and drinkers themselves (inter- nal costs), and costs borne by others (ex- ternal costs). The distinction is impor- tant. If market prices inclusive of tax reflect all relevant social costs, and users are assumed to accurately value the eco- nomic costs of their consumption, ob- served tobacco and alcoholic consumption is neither too high or too low, but instead is the outcome of rational choices made by consumers.

A recent study by Manning et al. di- vided economic costs of smoking and abu- sive alcohol consumption into internal and external costs.* The estimated external costs of smoking cited in the study in- cluded the discounted value of medical care for smoking-related illnesses peiid by group health plans or federal programs, covered sick leave for additional workdays lost by smokers, and the lives of non-smokers and property lost from fires associated with cigarette smoking. These costs were offset by the discounted value of reduced pen- sion and nursing home payments that smokers fail to receive because of pre- mature death, but were increased by the value of lost taxes. External costs of smoking were measured per pack of cig- arettes consumed.

The external costs of abusive alcohol consumption included the discounted value of the lives of non-drinkers lost and prop- erty damage from alcohol-related traffic accidents and fires, alcohol-related costs for the criminal justice system, and the cost of alcohol treatment programs, in ad- dition to medical, sick leave eind pension costs.

Abusive consumption of alcohol was defined in the Manning study as con- sumption in excess of two drinks per day, estimated to be forty percent of total con- sumption.

The estimated external cost of smoking in 1986 dollars ($0.15 per pack) was con- siderably lower than the average com- bined federal and state tax ($0.37 per pack).

The estimated external cost of drinking ($1.19 per ounce of excess con- sumption, or $0.48 per ounce of total al- cohol consumption) was much higher than the average combined tax ($.23 per ounce).

The estimates published in the Man- ning study suggest that current federal and state taxes on alcoholic beverages are less than the external costs of drinking, while current federal and state taxes on cigarettes exceed external costs. One should, however, be cautious about ex- trapolating these results to specific tax rates on either alcohol or tobacco.

Estimating the Economic Costs of Smoking and Drinking. Although the studies of the economic costs of smoking and drinking have been carefully done, all estimates of economic costs are necessar- ily imprecise. Economic costs are likely to be understated because it is difficult to quantify all attributable risks and risks that can be measured may be underre- ported. Economic costs are likely to be ov- erstated for other reasons. The estimates all measure gross rather than net eco- nomic costs. For example, if smoking and alcohol-related morbidity and mortality were reduced, individuals would experi- ence other illnesses and eventually would die from some other cause. The studies also assign all of the difference in mortality to the attributable risk of smoking and drinking even though it is possible that smokers and drinkers differ in important characteristics from non-smokers, such as other health-care behavior.

In order to measure indirect costs, which are a larger portion of economic costs, the studies must also deal with the thorny problem of how to attach monetary values to illness and premature death. Esti- mates of the total economic costs of smok- ing and drinking are based on the value of foregone earnings, which is estimated for both working and non-working indi- viduals. The estimates of internal and ex- ternal costs reported in the Manning study are based on the amount that individuals would be willing to pay to avoid illness or premature death. Either of these methods is subject to problems. Estimates based on foregone earnings are relatively easy to calculate, but have been widely criticized as being a conceptually inappropriate 346 NATIONAL TAX JOURNAL [Vol. XLIII measure of the value of life and limb. Es- timates of willingness to pay to avoid ill- ness and death may be conceptually more appropriate, but much more difficult to calculate.^" Finally, because most indirect costs are incurred in the future, it is also necessary to assume some discount rate to value these costs. A change in the discount rate assigned to future indirect costs can sig- nificantly change the estimated economic costs.

Taxing Alcohol. In determining whether to raise tax rates on alcohol two addi- tional issues need to be considered. The first is the difference between abusive and moderate consumption. The second is the relationship of external costs to the con- sumption of different types of alcoholic beverages, to the setting in which the heverage is consumed, and to the char- acteristics of the consumer.

Abusive vs. Moderate Consumption :

The external costs of drinking result fVom some but not all drinking. Some evi- dence suggests that moderate consump- tion of some alcoholic beverages does not appear to be injurious to health, and may in fact be beneficial." Although moderate drinking does appear to increase the risk of traffic accidents, the risks rise substan- tially with increased consumption.^^ Because it is generally not possible to t£tx only on alcohol consumption that gen- erates external costs, the gains from a re- duction in external costs must be bal- anced against the losses to non-abusive drinkers.'^ These losses include not only the tax payments made by non-abusive consumers, but also the loss in welfare if higher taxes cause non-abusive con- sumers to consume less alcohol.

The relative sensitivity to price of abu- sive and non-abusive consumption is an important determinant of the appropriate rate at which to tax. Higher taxes can be an effective policy to reduce external costs if abusive consumption is sensitive to prices (and if higher taxes are reflected in higher prices). If abusive consumption is not sensitive to price, higher taxes would not effectively reduce external costs and would only penalize non-abusive con- sumers.

There is some evidence that abusive consumption is at least as sensitive to price as moderate consumption. One study has found that the fraction of infrequent users declines less under a price increase com- pared to the fraction of heavy users among teenagers.''* Other research indicates that cirrhosis of the liver, which is positively related to abusive alcohol consumption, is negatively related to state alcohol tax rates.

In a study of 30 states, states which raised liquor taxes had either smaller in- creases or greater reductions in cirrhosis mortality rates.

^^ Type of Consumption Relating the appropriate mix of tax rates among different alcoholic beverages to external costs is also difficult. Although the medical consequences of ethyl alcohol consumption are the same for all bever- ages, current federal excise tax rates per ounce of ethyl alcohol vary by type of bev- erage. The tax on distilled spirits equals about $0.20 per ounce of pure alcohol, while the tax on beer equals about $0.05 per ounce of alcohol and the tax on table wine is about $0.01 per ounce of alcohol.

A 12-ounce can of beer, a 1.4 ounce shot of 80 proof distilled spirits, and a 5-ounce glass of table wine all contain about 0.55 ounces of ethyl alcohol.

Some external costs of drinking vary not only with the quantity of alcohol con- sumed, but also with the type of bever- age, the consumer, and the setting in which the beverage is consumed. Beer, for example, may have higher external costs in some respects than wine or distilled spirits since it is the alcoholic beverage of choice among teenagers and young adults who are responsible for a disproportion- ate share of traffic fatalities. Even here, it is difficult to determine an appropriate tax rate. A higher tax on beer-alcohol could cause youths to consume more wine- or spirits-alcohol.

Taxing Cigarettes. The basic estimates in the Manning study suggest that com- bined federal and state tax rates on cig- arettes exceed the rate needed to inter- No.

3] J. J. CORDES, E. M. NICHOLSON AND F. J. 8AMMARTIN0 347 nalize the externalities of smoking. This conclusion, however, is sensitive to how the different costs of smoking are cate- gorized.

Costs of Smoking Borne by Family Members Though the Manning study treats costs imposed by smokers and heavy drinkers on other family members as internal costs, it recognizes that a different classifica- tion would have raised estimated total external costs. Reclassifying both the costs of passive smoking on other family mem- bers as well as the costs of deaths and in- juries to family members from smoking' related fires as external costs would raise the external cost of smoking to about $0.38 per pack, which would be about equal to the combined federal and state taxes on cigarettes.

Treatment of Taxes, Transfers, and Pensions The Manning study also counts lower tax payments by smokers as an external cost and implicitly treats lower Social Se- curity and pension payments received by smokers as an external benefit of smok- ing. It can, however, be argued that these two effects of smoking are more in the na- ture of income transfers than real exter- nal benefits or costs. Though such trans- fers are relevant to assessing the effects of reducing smoking, they do not repre- sent effects of smoking on the allocation of resources apart from those included in the estimates of the medical and health costs of smoking.

For example, if smokers receive lower Social Security payments, other citizens may appear to enjoy an external "benefit" to the extent they pay lower social secu- rity taxes or are able to receive higher benefits themselves. These benefits, how- ever, are obtained at the expense of smok- ers.

Thus, what may appear to be an ex- ternal "benefit" to nonsmokers is also a cost to smokers. Similarly, lower taxes paid by smokers are not a cost of smoking in addition to the costs of smoking attrib- utable to lower productivity and lower earnings. Lower taxes paid by smokers are simply the way in which this component of the total cost of smoking is distributed among the population.

Not counting lower tax payments by smokers would lower estimated net ex- ternal costs. Not counting the effects of social security on social security and pen- sion payments would have the opposite effect. The overall effect of not counting the effects of smoking on taxes, social se- curity and pensions, would be to increase estimated external costs because smoking is estimated to "save" more in lower so- cial security and pension payments than it costs in taxes.

Should Taxes Be Limited to External Costs?

Even if smokers and drinkers face all relevant costs of their actions, consump- tion of cigarettes and alcoholic beverages could still be too high if users underesti- mate the potential harm to themselves from consumption. If users are fully aware of the risks but are unable to reduce their consumption because of addiction, there might be further justification for govern- ment intervention in the form of taxes.

Alternatives to Taxation Finally, it should be recognized that raising excise taxes on alcohol and to- bacco is not the only way to reduce the economic costs of smoking and alcohol abuse. Stricter enforcement of laws against driving while intoxicated, control of dis- tribution and advertising of alcohol and tobacco products, and educational and ad- vertising campaigns might also reduce external costs.

Limiting Smoking. Current federal ef- forts to limit consumption of tobacco focus primarily on providing information on the adverse health effects of tobacco use, re- quiring health warnings on cigarette products and advertisements, banning to- bacco advertisements in the broadcast media, and restricting tobacco use on gov- ernment worksites and on domestic fiights of U.S. airlines.

In the case of advertising, a next step might be a total ban on cigarette adver- 348 NATIONAL TAX JOURNAL [Vol, XLIII tising. It is, however, uncertain whether such a ban would have the desired effect.

If current cigarette advertising recruits new smokers and encourages more smok- ing among existing smokers, a total ban on cigarette advertising might reduce consumption of cigarettes. If current ad- vertising merely targets brand selections of current smokers, however, a total ban on cigarette advertisements might result in lower prices of cigarettes, by reducing advertising costs of cigarette companies.

Lower prices might result in higher per capita consumption and greater smoking prevalence, particularly among price-sen- sitive teenagers.

Another option used to reduce cigarette consumption is to prevent teenagers from purchasing tobacco. Although 44 states and the District of Columbia imposed re- strictions of sales of tobacco to minors, there is no federal minimum age to pur- chase tobacco. The 1989 Surgeon Gener- al's Report concludes that the degree of restriction is the lowest in many decades, and indeed, "It would be surprising if laws as currently implemented bad mucb ef- fect on tbe initiation of tobacco use by children," Limiting Alcohol Abuse. Mucb of re- cent federal policy to reduce costs asso- ciated with abusive alcobol consumption focuses on drinking and driving, and al- cobol consumption among teenagers. To establish a standard drinking age across all states, tbe Federal Uniform Drinking Age Act of 1984 provided tbat Congress would witbbold Federal bigbway funds from states witb minimum legal drinking ages (MLDA) below 21 years. Currently, all fifty states and tbe District of Colum- bia use 21 years as MLDA, Tbe National Highway Traffic Safety Administration estimates tbat minimum drinking age laws reduce traffic fatalities in affected age groups by 13 percent—saving 1,071 lives in 1987 and a cumulative total of 8,142 lives between 1975 and 1987," In 1989, based on recommendations from a worksbop on drunk driving, tbe Sur- geon Greneral proposed a number of op- tions designed to reduce costs of drinking and driving, Tbese options include auto- matic confiscation of drivers' licenses for driving wbile intoxicated, a reduction in tbe blood alcobol concentration limit from 0,10 percent to 0,04 percent, restriction of certain advertising of alcobolic bever- ages, and expanded use of sobriety cbeck- points,^® Price Incentives other than Taxes.

Insurance policies wbicb distinguisb be- tween users and non-users of alcobol and tobacco migbt be anotber way to reduce external costs by making consumers bear more of tbe costs tbemselves. Federal pol- icies migbt include using non-smoker dis- counts for Medicare and Medicaid, pro- viding incentives for actuarial research to document differences in morbidity and mortality among non-smokers, and pro- viding tax incentives for insurance com- panies to use selective plans.

Non-user discounts migbt not alter con- sumer behavior as mucb as bigber tax rates, bowever. Most smokers begin smoking before tbey begin purchasing in- surance. In addition, smokers cannot sim- ply cut back consumption to qualify for non-smoker discounts; tbey must quit smoking. Non-smoker discounts migbt pose difficult administrative problems for tbe industry, and extensive use of selec- tive plans migbt encourage misrepresen- tation of smoking status.

Taxing Pollution Pollution of tbe environment imposes real tbougb often intangible costs, Tbe nature of tbese costs bas recently been bigbligbted in tbe debate about H,R, 3030, tbe Clean Air Act Amendments of 1990,^^ For example, according to some estimates reducing acid rain from current levels could prevent premature deatbs of as many as 50,000 people a year, improve water quality in acidic lakes and streams in tbe Nortbeast and Midwest, provide clearer skies in 31 states, and prevent damage to nimierous stone buildings. Reducing smog and airborne particles is estimated to pro- vide bealtb benefits ranging from $500 million per year to perbaps as mucb as $10 billion per year. Reducing emissions of toxic cbemicals is estimated to prevent 75 percent of tbe 1,500 to 2,700 of new cases of fatal cancer attributable to ex- No.

3] J. J. CORDES, E. M. NICHOLSON AND F. J. SAMMAETINO 349 posure to such chemicals, reduce the in- cidence of birth defects, chronic respira- tory illness and nonfatal cancers, and reduce risks of major leaks of hazardous chemicals.

The Clean Air Act Amendments are noteworthy from the standpoint of envi- ronmental policy because fees and other price incentives are included in the Act as options to help to control pollution. Eval- uating what role pollution taxes might play in helping to improve environment is not simple. Some basic but useful prin- ciples can be gleaned from the extensive literature on the role of taxes and fees in environmental policy.

Defining the Base of Pollution Taxes The first point is simple. To realize many of the presumed benefits of pollution taxes, it is necessary to tax pollution rather than goods produced and sold by polluters.

Taxing the final output of polluters makes goods that pollute more expensive, which reduces their consumption and pro- duction. For this reason, taxing output is commonly seen to be a direct and simple way of improving environmental quality.

This is not the case.^" Pollution is reduced not only by pro- ducing and consuming smaller amounts of goods that give rise to pollution, but also by reducing the amount of pollution emit- ted per unit of the goods produced and sold.

If output rather than pollution is taxed, the polluter's tax liability depends only on how much is produced, and not on how much pollution is emitted per unit of out- put. As a result, there is no financial in- centive to produce goods in a more envi- ronmentally cost-effective manner. By comparison, if the polluter's tax bill de- pends on the amount of pollution rather than on the amount of output, the pollu- ter will have a financial incentive to spend resource to abate pollution so long as the taxes saved exceed the cost of reducing pollution. Thus, taxing pollution rather than output does more than make goods that pollute more expensive. It also pro- vides incentives to produce output in ways that are environmentally less harmful.

Because taxing final output is fairly blunt and inexact it should not be seen as the preferred instrument of environmen- tal policy. Taxing output may, however, be the policy of choice for pragmatic rea- sons when alternatives that might be su- perior in theory cannot be implemented in practice.

Taxing gasoline is one illustration of this point. There is no doubt that emissions from motor vehicles are a major source of air pollution in the U.S. Taxing gasoline would improve air quality by reducing driving, and, in the long run, by encour- aging the use of more fuel efficient vehi- cles, which lower total emissions. It would not provide a financial incentive to re- duce motor vehicle emissions per gallon of gasoline consumed. Such incentives would be provided by taxes which varied with the amount of emissions rather than gasoline consumption. Such taxes could be levied on new cars, based on their es- timated emissions and on old vehicles in conjunction with an inspection program that measures emissions and recorded an- nual mileage.

Though taxes of this sort may be pref- erable in concept, they may be hard to ad- minister in practice. In this case, gasoline taxes may be a reasonably cost effective means of reducing motor vehicle emis- sions especially when compared to other alternatives such as state II vapor recov- ery—a control used to reduce emissions when filling up a vehicle—or methanol powered vehicles.

Pollution Taxes as an Instrument of Environmental Policy The environmental economics litera- ture also suggests that pollution taxes be used to provide financial incentives that lower the costs of attaining levels of pol- lution mandated by law rather than to provide financial incentives for polluters to reduce pollution to desired levels.

Theoretical Limitations of Price Incentives One reason is conceptual. Even if reg- ulators knew enough about the social costs of pollution to set tax rates based on the 350 NATIONAL TAX JOURNAL [Vol. XLIII environmental harm done by different pollutants, relying exclusively on price incentives might not achieve the desired result.

It is in society's economic interest to re- duce pollution when the costs of reducing pollution are less than the economic costs of the environmental harm avoided. Re- lying entirely on price incentives pre- sumes that this result will be achieved if market prices include the marginal social costs of pollution. This presumption is correct if changing the amount of pollu- tion by a small amount also changes the social costs of pollution. In this case, if regulators know the social costs of pol- lution, all they would need to do is impose a pollution tax equal to the marginal so- cial cost of pollution. Polluters facing such a tax would continually compare the tax with the cost of cutting back on pollution.

An economically rational polluter would have a financial incentive to reduce pol- lution up to the point where the cost of cutting back pollution by one more unit just equaled the tetx.

In this case, what was good for the polluter would also be good for society as a whole because pollution would be cut back only if the marginal so- cial benefits (as embodied in the tax) ex- ceed the marginal social costs.

A number of economists, however, have made a compelling case that pollution by its very nature creates conditions under which price incentives by themselves can- not be relied on to allocate resources ef- ficiently.^' A simple example is the case in which changing the amount of pollu- tion by a small amount has no effect on the social costs of pollution. For example, it may be necessary to cut back pollution by a large amount before any economic benefits are obtained.

In this circumstance, taxing polluters does not ensure that resources are allo- cated efficiently. If polluters are taxed, they will reduce pollution when it is cheaper to cut back than pay the tax. Un- like the case described above, however, what makes financial sense for the pol- luter may not make economic sense for society as a whole. In the jargon of eco- nomics taxing polluters may lead society to a "local" rather than a "global" opti- mum. If there are large costs of getting to the point where cutting back pollution begins to lower its social costs, society could conceivably be better off economi- cally by spending nothing to reduce some forms of pollution.

Practical Limitations Lack of knowledge about the effects and the causes of pollution is cited as a fur- ther barrier to exclusive reliance on price incentives as instruments of pollution control. Though research continues to im- prove our understanding of the causes £ind the consequences of pollution, there is a huge range of uncertainty about the so- cial costs of pollution. This uncertainty is reflected in the wide range of estimates of the benefits of improving environmental quality. To illustrate, estimates from a recent study of the benefits of bringing all areas nationwide into conformity with federal ozone standards range from $51 million to $4.7 billion.^^ The range of es- timated benefits from improving environ- mental quality is as wide as it is because the scientific evidence about the effects of pollution continues to be highly uncer- tain and because there is uncertainty about how to translate the available sci- entific evidence into estimates of the monetary costs of pollution.

A Modified Role for Price Incentives These theoretical and practical limita- tions have prompted several environmen- tal economists to acknowledge that it is unrealistic at this time to substitute pol- lution taxes and other price incentives for legislation that mandates specific levels of environment quality.^^ That is, the mandated goals for improved environ- mental quality contained in legislation like the Clean Air Act and the proposed amendments to it cannot and should not be supplanted by pollution taxes. Price incentives may, however, play a role in achieving legislatively mandated levels of environmental quality.

In effect, this is the role proposed for No.

3] J. J. CORDES, E. M. NICHOLSON AND F. J. SAMMARTINO 351 several of the taxes and fees included in the initial language of th Clean Air Act.

H.R. 3030, as reported by the Committee on Energy and Commerce, amends the Clean Air Act to give EPA authority to impose fees as part of an implementation designed to bring areas into compliance with national air quality standards. In addition, if an area classified as a severe ozone nonattaininent area failed to attain the national ambient air quality standard for ozone by 2010 (assuming the Clean Air Act Amendments are enacted in 1990), the state would be required to collect a tax from major sources of volatile organic compounds (VOCs) of $5,000 per ton of VOCs emitted in excess of 80 percent of a baseline amount. This amount would equal the emissions allowed to be emitted in the attainment year, or if lower, the amount of emissions actually emitted in the year the fee was imposed.

Lowering the Cost of Controlling Pollution Price incentives can play a potentially important role whenever multiple sources contribute to the overall level of emis- sions, and, apart from considerations of cost, it is more or less a matter of indif- ference how some overall target for re- ducing emissions is distributed among the multiple sources. In such cases, the op- erative economic rule would be to distrib- ute the total reduction in emissions so that the marginal cost of reducing emissions by an extra unit was the same at each source.

Current environmental regulations do not distribute reductions in emissions in this manner, but instead impose source- specific standards. The result is that the marginal cost of controlling emissions by an additional unit varies among the dif- ferent sources. This means that the total cost of achieving the legislatively man- dated level of environmental quality is higher than it needs to be. Costs could be lowered by shifting the responsibility for reducing emission from sources with rel- atively high costs of control to those with relatively low costs of control.

Can market incentives help regulate pollution more efficiently in this case?

Perhaps, especially in the longer run.

Suppose, for example that instead of re- quiring proportional reductions at each source, a legislated target for overall pol- lution reduction was enacted, and pollu- ters were required to pay a tax on each unit of effluent discharged. In this case, each polluter would have a financial in- centive to reduce pollution up to the point where the marginal cost of abating an ad- ditional unit just equaled the tax. Be- cause all polluters would face the same tax, the marginal cost of reducing an addi- tional unit would be the same at all sources. By teixing polluters, regulators would know that whatever level of emis- sions was achieved in the aggregate, it was achieved at minimum social cost.

Of course, regulators might not like the initial level of abatement achieved in re- sponse to the tax. In the absence of enough information about the costs of controlling pollution, they would be apt at first to set the tax at the wrong level. The tax rate might be set "too low" in the sense that pollution would be cut back by less than the legislatively tfirgeted amount. The tax rate could also be too high in the sense that pollution would be cut back by more than the legislatively mandated amount.

Proponents of increased reliance on taxes, however, have argued that this shortcoming of pollution taxes could be fixed in the longer run by adjusting the tax rate in response to observations of the level of emissions. Through such an it- erative process, regulators could eventu- ally establish the tax rate at which cost- minimizing responses by different sources of pollution would result in the desired reduction of pollution. Relative to the al- ternative of across-the-board direct regu- lation, fewer resources would be devoted to controlling pollution. In the process tax revenues would be collected on the resid- ual amount of pollution remaining after abatement.

A number of studies have attempted to estimate the difference between mini- mum costs of controlling pollution and the costs that result from current regulatory 352 NATIONAL TAX JOURNAL [Vol.

XLIII practices. These studies suggest that sub- stantial costs could be saved if pollution were controlled more rationally.

One study used engineering data from Du Pont to estimate the costs of reducing hydrocarbon emissions by 85 percent in a variety of ways.

The first alternative con- sidered was source-by-source standards in which emissions from each source, both within a single Du Pont plant, and across Du Pont plants were reduced by 85 per- cent. The estimated total cost of this al- ternative was $105.7 million in 1975 dol- lars.

The second edtemative was plant-by- plant standards in which emissions from each Du Pont plant were reduced by 85 percent, but within each plant, emissions were allowed to vary inversely by source with respect to the costs of controlling emissions. The cost of this alternative was $42.6 million, or 63 percent lower than in the case of source-by-source standards. The final alternative considered was a multi- plant standard which allowed emissions to be traded off among plants subject to the overall constraint that 85 percent of emissions be reduced. The estimated cost of this alternative was $14.6 million.^ The Du Pont study applies to only a single firm, but studies dealing with other forms of pollution have found comparable cost savings. A study of the cost of reduc- ing halocarbon emissions in various ways found that compared to a realistic set of source-by-source controls, using price in- centives to limit emissions would have saved roughly $120 million in costs of control—a savings of 50 percent—while collecting $1.4 billion in emissions charges from polluters.^^ Still other studies that have compared the cost of controlling emissions using direct controls with the least cost solution find cost savings rang- ing from 50 percent to less than 10 per- Estimates such as these seem to sug- gest that pollution taxes can offer a po- tentially powerful one-two punch. They offer a way of reducing the cost to the economy of achieving legislatively man- dated levels of environmental quality and they can raise potentially significant amounts of revenue.

In practice, there are a number of reasons why a more modest agenda for pollution taxes may be in or- der, and why the introduction of such taxes needs to be assessed carefully on a case- by-case basis.

Variation in Environmental Harm at Different Sources One reason for caution is that pollution taxes, (like other market incentives) have somewhat less to offer when regulators are not indifferent for scientific and environ- mental reasons to how reductions in emissions are distributed among different sources.^' In that case, there may or may not be significant gains to achieving leg- islative goals through taxation. For ex- ample, it may not be appropriate to trade off more abatement at a cheaper source against less abatement at a more expen- sive source if regulators believe that emissions from the "more expensive" source are more harmful. Unless a higher tax rate could be imposed on the more harmful source, taxing pollution might be less effective than simply mandating re- ductions at each source.

Uncertain Control of Pollution A second and potentially more serious limitation is that regulators must be pre- pared to live with greater uncertainty about the level of pollution control when they tax polluters than when pollution is controlled at each source. We have noted above that the consequences of such un- certainty may diminish over time as reg- ulators are able to observe how polluters respond to taxes, provided that regulators are able to adjust pollution tax rates to achieve legislatively mandated levels of pollution.

Changing such rates, however, may be easier said than done. If pollution tax rates cannot be changed over time, or can be changed only infrequently, the choice be- tween pollution taxes and direct controls involves comparing the greater certainty of control offered by direct regulations against the lower cost of control offered by pollution teixes. Relying on pollution taxes in place of direct controls would make sense if relatively low value was placed on the certainty of control, but the No.

3] J. J. CORDES, E. M. NICHOLSON AND F. J. SAMMARTINO 353 costs of control were thought to be high.

In cases in which it was thought imper- ative to control emissions by a given amount, the uncertainty associated with taxes would become a more serious lia- bility.

Given the uncertainties associated re- placing source-specific controls with pol- lution taxes, one possible strategy might be to retain current regulations, but sup- plement them with pollution taxes. Indi- vidual polluters would still be required to meet specific targets for reducing pollu- tion, but would also be taxed on their emissions. These taxes would be set at levels that represented regulators' best guesses about the costs of controlling pol- lution at the margin.^* Individual polluters could be expected to respond to such a regime in one of two ways.

A polluter would do nothing if the t£ix rate was less than or equal to the marginal cost of controlling at the level mandated by regulation. In this case, the polluter would pay tax on residual emis- sions, but would otherwise have no incen- tive to change behavior in the short run.

Some polluters, however, might discover that the costs of controlling pollution at the mandated level was less than the cost of paying the tax. Such polluters would have a financial incentive to reduce pol- lution by more than the amount man- dated by regulations.

Because individual polluters would be required to meet the source-specific reg- ulatory standard at a minimum, the type of arrangement just described would guarantee that the overall target for re- duced emissions was achieved. This cer- tainty would, however, come at a price.

So long as source-specific regulations re- mained in effect, polluters would not be free to adjust their levels of pollution abatement so as to equate the marginal costs of abatement. Thus, combining pol- lution taxes with existing regulations in the manner outlined would not lower the cost of attaining the regulatory standard, though it might result in modest further reductions in pollution.

It might, however, provide regulators with more information about the costs of controlling pollution. Over time, this in- formation would allow regulators to set taxes with greater confidence in their ability to move to desired levels of envi- ronmental quality. At such a time, con- sideration could be given to relaxing source-specific controls.

The proposal requiring states to impose a fee of $5,000 per ton on VOCs emitted in excess of 80 percent of a baseline amount in certain areas is an example of a related, though somewhat different ap- proach. Polluters facing this fee in the at- tainment year, could choose to pay it or cut back emissions to 80 percent of the baseline amount. If the cost of reducing pollution over this range was less than $5,000 per ton, firms would have a finan- cial incentive to reduce emissions beyond mandated levels. Thus, by observing the firms' responses, regulators would gain information about the marginal cost of abatement.

Because states would collect the fee, it is not intended as a source of federal rev- enue.

The proposed VOC fee, however, aptly illustrates that there may be times in which the goal of achieving further abatement and the goal of raising tax revenue may not be entirely compatible.

If polluters found it cheaper to reduce emissions beyond the baseline than to pay the tax, no fees would be collected.

Alternative Policy Instruments In thinking about the future role of pol- lution teixes, one should edso recognize that many, though not all, of the benefits as- cribed to pollution taxes may also be ob- tained by modifying the current system of direct controls to allow polluters to trade pollution permits with each other. In ef- fect, current regulations would be the ba- sis for distributing pollution permits "free of charge" to individual polluters which could then be bought and sold.

Under such a scheme, polluters with relatively high costs of controlling pollu- tion would find it profitable to acquire ad- ditional rights to pollute from polluters with relatively low costs of control. Pol- luters with relatively low costs of con- trolling pollution would find the costs of 354 NATIONAL TAX JOURNAL [Vol. XLIII cutting back on pollution to be less than the income they could receive by selling rights to pollute.

The total amount of permits would be fixed to equal the legislatively mandated level of pollution, so that trading of pol- lution permits would not affect the total amount of pollution reduction. It would, however, redistribute abatement among pollution sources with high and low costs of control. As long as trades could be ar- ranged at little or no transactions costs, pollution permits would be traded until each producer faced the same marginal cost of abatement. Thus, like taxes, trade- able permits would lower the cost of achieving legislatively mandated levels of environmental quality. Unlike taxes, tradeable permits would produce a cer- tain level of overall pollution control.

In the short run, tradeable permits ap- pear to provide incentives comparable to those of pollution taxes without the as- sociated uncertainty. In the long run, however, resources would be allocated difTerently if polluters were taxed than if tradeahle permits were initially distrib- uted free of charge. Polluters' profits will be lower if they must pay taxes on all pol- lution than if they were assigned trade- able permits. Thus, taxing polluters cre- ates stronger incentives for resources to exit industries that pollute than does lim- iting pollution with tradeable permits.^* Limiting pollution by means of permits would provide the same long-run incen- tives for entry and exit if permits were sold by the govemment rather than assigned free of charge.^" It would also be equiva- lent in many important respects to taxing pollution. The two main differences would be that revenues from public auctions of pollution permits might be classified dif- ferently than taxes in the budget. The ad- ministrative costs of raising revenues from govemment auctions of pollution permits would also differ from the administrative costs of taxing pollution.

It should be noted that any permit scheme would need to be monitored care- fully to ensure that it did not bave un- intended effects on the ability of firms to compete with each other in different mar- kets.

This issue would be of particular concern if the initial permits were dis- tributed free of charge.

Conclusions Tax economists have been wary about using tax incentives to encourage activi- ties for social purposes. Should tbis war- iness carry over as well to using taxes to discourage activities with sociEil costs? The following broad conclusions can be drawn from the literature surveyed in this pa- per.

First, there is no doubt that both smok- ing and abusive drinking impose costs on parties other than smokers and abusive drinkers. There is also evidence that rais- ing the prices of tobacco and alcohol re- duces smoking and abusive drinking.

Thus, higher taxes on alcohol and higher taxes on cigarettes will reduce the exter- nal costs of smoking and drinking. There is also evidence that at current tax rates, abusive drinkers do not bear the external costs of their actions. This may also be so for smokers. However, tbis conclusion is sensitive to how some of the costs of smoking are categorized.

These findings are consistent with the view that compared to other taxes, excise taxes on alcohol and tobacco may entail smaller deadweight losses, and may even result in efficiency gains if current tax rates do not reflect external costs of smoking and drinking. Enthusiasm for raising excise taxes on alcobol and to- bacco, bowever, should be tempered by the recognition that there may be otber ways of controlling smoking and abusive drink- ing that would not raise revenue. It should also be empbasized that the comparative effects of taxes on economic efficiency is only one of several criteria for evaluating different taxes.

In contrast to taxes on tobacco and al- cohol, which have been around for some time, and which are clearly seen as a di- rect means of reducing smoking and drinking, the role that taxes migbt play in reducing pollution is more speculative and likely to be less direct. For the fore- seeable future, the level of environmental quality is likely to be determined by leg- islation rather than the interplay of mar- No.

3] J. J. CORDES, E. M. NICHOLSON AND F. J. SAMMARTINO 355 ket incentives. Pollution taxes can, how- ever, play a useful role in holding down the costs of maintaining and improving environmental quality. As in the case of alcohol and tobacco taxes, comparable cost- savings may be achieved by means other than pollution taxes.

ENDNOTES ***We gratefully acknowledge helpful comments from Rosemary Marcuss. The views expressed are the authors' alone and should not be attributed to the Congressional Budget Office.

'For a recent discussion see Joel Slemrod, "Optimal Taxation and Optimal Tax Systems," Journal of Eco- nomic Perspectives, 4(1): 157-190, Winter 1990.

''U.S.

Department of Health and Human Services.

Reducing the Heatth Consequences of Smoking: 25 Years of Progress.

A Report of the Surgeon General.

Centers for Disease Control, Center for Chronic Dis- ease Prevention and Health Promotion, Office on Smoking and Health. DHHS Publication No. (CDC) 89-8411:98-99, Prepublication version, January 11, 1989.

Specifically, the report found that cigarette smoking is the m^or cause of lung and laryngeal can- cer, chronic bronchitis and emphysema, a major cause of oral and esophageal cancer, coronary heart disease and cerebrovascular disease (stroke), and a contrib- uting factor to bladder, pancreatic, and renal cancer.

'John R. Hall, Jr. and Alison L. Norton. "The U.S.

Smoking-Material Fire Problem Through 1986," Fire Analysis Research Division, National Fire Protection Association, September, 1988.

••U.S.

Department of Health and Human Services.

The Heatth Consequences of Invotuntary Smoking. A Report of the Surgeon General. U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control. DHHS Publication No.

(CDC) 87-8398, 1986.

V. M. Shultz, D. P. Rice, and D. L. Parker. "Alco- hol-Related Mortality and Years of Potential Life Lost- United State8, 1987." Reported in Morbidity and Mortality Weekly Report, 39(1):173-178, Centers for Disease Control, U.S. Department of Health and Hu- man Services, Public Health Service, March 23,1990.

^Office of Technology Assessment. "Smoking-Re- lated Deaths and Financial Costs," OTA Staff Mem- orandum, Health Program, U.S. Congress, 1985. Dor- othy P. Rice, Thomas A. Hodgson, Peter Sinsheimer, Warren Browner and Andrea N. Kopstein. "The Eco- nomic Costs of the Health Effects of Smoking, 1984." Milbank Quarterly 64(4):489-547, 1986. A recent re- port by the U.S. Department of Health and Human Services estimated total costs of smoking of $52.3 bil- lion in 1985. See: U.S. Department of Health and Hu- man Services, Public Health Service, Center for Dis- ease Control. Smoking and Health, A National Status Report 2nd Edition. DHHS Publication No. (CDC) 87- 8396 (Revised February, 1990).

'Henrick J. Harwood, Diane M. Napolitano, Patri- cia L. Kristiansen, and James J. Collins. "Economic Costs to Society of Alcohol and Drug Abuse and Men- tal Illness, 1980." Research Triangle Park, N.C.: Re- search Triangle Institute, 1984. (PHS Contract No.

ADM 283-83-002).

*An estimated 5,666 million gallons of beer, 528 million gallons of wine and 431 million gallons of dis- tilled spirits were consumed in 1983.

The cost per ounce of alcohol is based on an average alcoholic content of 5.0 percent for beer, 12.9 percent for wines, and 41.4 percent (80 proof) for distilled spirits.

^Villard G. Manning, Emmett B. Keeler, Joseph P.

Newhouse, Elizabeth M. Sloss and Jeffrey Wasser- man. "The Taxes of Sin. Do Smokers and Drinkers Pay Their Way?" Journal of the American Medical Association, 261(ll):1604-1609, March 17, 1989.

'"For a discussion of this issue see Edward M.

Gramiich, A Guide to Benefit-Cost Analysis Prentice Hall, 1990, Ch. 4.

"Most alcoholics who develop liver diseases con- sume a large quantity of alcohol over an extended pe- riod. Some evidence suggests a positive relationship between moderate alcohol consumption and a lower risk of coronary artery disease.

'^Youths (ages 16—21) with a blood-alcohol concen- tration (BAC) between 0.05 and 0.10—corresponding to 1 to 2 drinks over a several hour period, face twice the risk of being in a fatal automobile accident rel- ative to non-drinking youths. Youths with BAC of over 0.10—corresponding to 6 or more drinks over a sev- eral hour period, face 100 times the risk of being in a fatal automobile accident relative to non-drinking youths. See: Charles E. Phelps, "Death and Taxes: An Opportunity for Substitution," Journal of Health Eco- nomics, 7:1-24, 1988.

'*rhomas F. Pogue and Larry G. Sgontz. "Taxing to Control Social Costs: The Case of Alcohol." Amer- ican Economic Review, 79(l):235-243, March 1989.

"Michael Grossman. "Health Benefits of Increases in Alcohol and Cigarette Taxes," National Bureau of Economic Research, Working paper No. 3082. Cam- bridge MA: August 1989.

"Philip J. Cook. "The Effect of Liquor Taxes on Drinking, Cirrhosis, and Auto Fatalities." in Alcohol and Public Policy:

Beyond the Shadow of Prohibition, Mark H. Moore and Dean R. Gerstein eds., Washing- ton, D.C: National Academy Press 1981; and Philip J. Cook and George Tauchen. "The Effect of Liquor Taxes on Heavy Drinking." Bell Journal of Econom- ics, 13(2), Autumn 1982.

'°A similar adjustment could also be made to the estimated external costs of alcohol abuse presented in the Manning study. In this case, because alcohol abuse has a more pronounced effect on productivity the net effect of excluding costs due to taxes paid and trans- fers received by drinkers would be to reduce rather than to increase external costs. The adjustment, how- ever, is small and would not change the basic conclu- sion that the external costs of alcohol abuse exceed current taxes on alcohol.

"U.S.

Department of Transportation, National Highway Traffic and Safety Administration. "1987 Fatality Facts," National Center for Statistics and Analysis, October 1988.

"U.S.

Department of Health and Human Services.

Proceedings of the Surgeon General's Workshop on Drunk Driving. Public Health Service, Office of the Surgeon General. Washington D.C: December 14—16, 1988.

"The remainder of the paragraph draws heavily 356 NATIONAL TAX JOURNAL [Vol. XLIII from the Wall Street Journal, April 5, 1990.

'"See Daniel F. Spulber, "Effluent Regulation and Long-Run Optimality," journal of Environmental Economics and Management, 12:183-116, 1985.

'"For a discussion of this issue see, William Baumol and Wallace Oates, The Theory of Environmental Policy, Cambridge: Cambridge University Press, 1988, Chapter 8, and William W. Baumol and David F.

Bradford, "Detrimental Externalities and Non-Con- vexity of the Production Set," Economica 39:160-176, Mejf 1972.

^AUen J. Krupnick, "Economics and the Ambient Ozone Standard," Resources Washington, D.C: Re- sources for the Future: 9-12, Summer 1988. We are grateful to Fran Sussman in the National Resources and Commerce Division at CBO for providing this ex- ample.

''T'his point was first made in William W. Baumol and Wallace E. Oates, "The Use of Standards and Prices for Protection of the Environment," Swedish Journal of Economics, 73:42-54, March 1971.

The point and some imphcations for the design of environmen- tal policy are developed at more length in Baumol and Oates, The Theory of Environmental Policy, Chapters 11-13.

"Michael T. Maloney and Bruce Yandle, "Bubbles and Efficiency," Regulation, 49-52, May/June 1980.

^A.

Palmer, et al., "Economic Implications of Reg- ulating Chlorofiuorocarbon Emissions from Nonaero- sol Applications (Rand Corporation, Santa Monica, 1980).

For a general discussion see Baumol and Oates, The Theory of Environment Policy, Ch. 13.

™See Baumol and Oates, op. cit.. Chapter 11.

"See Baumol and Oates, The Theory of Environ- mental Policy, Ch. 11, pp. 169-170.

''^For a discussion of some examples, see Congres- sional Budget Office, Reducing the Deficit: Spending and Revenue Options, 432-440, Washington, D.C: U.S.

Government Printing Office, February 1990.

^See Spulber, op. cit. for a rigorous analysis of why failing to tax all pollution fails to provide the needed incentives for entry and exit.

™See Baumol and Oates, op. cit.. Chapter 13, and Spulber, op. cit..