Tobacco Documents Online

Snloklnb Be111V1or and Policy Conference Series ' The Cigarette Excise Tax Apri ; l 17) 1985 I -1 _ .. . __- Institute for the Study of Smoking Behavior and Policy F tarvard University John F. Kennedy School of Government 79 John F. Kennedy Street Cambridge, Massachusetts 02138 Z H

45 i Foreword ~ In 1983 the federal excise tax on cigarettes jumped from eight to 16 cents I{ per pack. lt had been level at eight since 1951, when the price of a packet of { cigarettes was 19 cents and the tax was 42 percent of that price. By 1982 the price had risen to 75 cents and the tax was down to barely 11 percent of the . price. Over the 32 year period the consumer price index had nearly ! quadrupled. Doubling the tax in 1983 put its real value adjusted for inflation at about half its real value of 1951, and made the tax equal to 18 percent of the 1983 price of 90 cents per pack. The legislation that raised the tax provided for a return to cight cents on October 1, 1985. There are strong signs of Congressional interest in new legislation that would retain the 16 cent tax or even raise it. The motives are several. An obvious one is revenue; the tax generates about $2.4 billion and in the forthcoming years of high deficits, while $2.4 billion isii t much, every bit helps. A second motive, evidenced in sonie of the proposals to earmark part or all of the tax for Medicaid, is to let smokers pay for some of the extra medical costs that their smoking inflicts ou the non-smoking taxpayers and on non-smoking enrollees in health insurance whose premiums usually must cover the extra costs incurred by smokers. And a third motive, increasingly voiced, is the hope that by raising the price of cigarettes a higher tax may induce smokers to smoke less or, better still, induce smokers to quit and non-smokers not to take tup the habit. The timeliness of the issue-the deadline for preventing a 50 percent tax cut being October 1st this year-prompted the Institute for the Study of Smoking 13ehavior and Policy at Harvard's Kennedy Sclwol of Govern- ment to sponsor a conference in Washington last April. It brouglit togrthrr people knowledgeable on excise taxation in general and people knowl- edgeable about cigarettes and smoking. Because cigarette taxation is impor- tant to many state governments, one participant represented that interest. Who Pays? One of the first questions to ask about any excise tax is, Who pays it? A possible answer-namely, the people who smoke cigarettes-might appear too trivial to get us very far. It is not trivial. It may not even be true. In a literal sense the tax is paid by the manufacturer; the "incidence," as econo- mists call it-the actual cost-may fall forward on the consumers of cigarettes or backward on the suppliers of tobacco, or on wages or profit witltin tile manttfacturing companies. In the conference it was generally expected that the effect of the tax would be a corresponding increase in the price of cigarettes. i

s 3 If it is, then while it is obvious that the burden of the tax must fall on rhose who smoke cigarettes and their families, the answer is still not %"acuous because it implies that the tax falls on people who by their 1n1oking may incur medical costs at public expense. This is of interest in the same way that it is of interest that a gasoline tax falls on people who wear out the roads. With most excise taxes the single most important issue in who pays it is tairness. Doc, rhe tax fall equally on rich and poor alike? Does it fall on rich and poor in proportion to their incomes? Does it fall mainly on the rich, or mainly on the poor? There is.a related interest in whether it bears especially "u the elderly, on certain minorities, or on some other population category. Necessity or Luxury? Sometimes another question is asked: is the taxed commodity a luxury or a necessity? By the most common definition a necessity is something that people have extreme difficulty doing without; a luxury is something that they can take or leave. Cigarettes are a paradoxical commodity: they are generally considered not to meet any fundamental need the way food, clothing, shelter, and medical care (to, yet for people who smoke, cigarettes are extraordinarily difficult to give up. There is an irreversibility here. Until one takes up smoking there is no "need," but when one has become a regular smoker it is remarkably difficult to quit. (According to a 1980 survey, more than a third of all smokers had attempted-tmsuccessfully- to) quit during the preceding 12 months; the figure was more than 50 percent for young smokers, aged 21 to 24, both sexes.) The popular notion has always been that it is less fair to tax necessities than to tax luxuries, the idea being that with the luxury one has a choice whether or not to consume the item and pay the tax. Economists have usually taken a different view. They note that two things happen when a commodity is taxed; people both pay the tax on what they continue to r0nsume of the item, and may also reduce the amount they consume or even avoid the tax altogether by giving up that particular item of consumption. Two Burdens Economists point out that there are two "burdens" of taxation here, the financial burden of paying the tax on what one continues to consume, and the burden of escaping the tax by doing without something that one would have preferred to continue consuming. The first of these burdens, the tax actually paid, shows up in the Treasury's revenues; the second burden has been called the "deadweight loss," it being a burden on the consumer but yielding nothing to the Treasury. Economists usually prefer y to tax the items that will generate the least deadweight loss, and to pick ~ items to tax according to the kinds of people-usually measured by their inconus-who consume the item. And this means preferring to tax the i things that people will continue to consume. Here again, cigarettes are different. The "burden" of giving up ciga-; rettes is likely to be of short duration-weeks, months, years perhaps for a~ few-and then the burdens turn into benefits. Most people in this cormtry , who smoke wish they didn't. People who have quit are glad they did. Furthermore, there is probably not much of a "burden" for the people who might become smokers but do not, perhaps because of the higher price of cigarettes. People who simply forgo ever consuming a taxed item because the tax makes the price too high are forgoing an opportunity in a way that yields the Treasury nothing; but if they forgo experimenting with a substance that for many becomes an addiction, the benefit is a lifetime of freedom from smoking. The Poor, The Young, and Other Issues Consumer budget data indicate conclusively that the poor spend a higher proportion of their incomes on cigarettes than people in middle or high income brackets, appreciably more. As a tax, then, the cigarette tax is what economists call "regressive:' But the tax also appears to discourage smoking, and some data indicate that the tax discourages smoking among the poor more than among the well-to-do. The benefits to those who quit, or who are less likely to take up smoking because of the higher tax, can be measured actuarially as several years of added life expectancy. For someone who smokes a pack a day, an eight cent tax increase or decrease amounts to $30 a year. (Those who quit, or avoid taking it up, save $365 per year. ) Two facts deserve emphasis at this point. One is that nearly everybody in the United States who takes up smoking does so at an early age. Nearly all regular smokers became regular smokers by age 21 or 22, most of them several years before that. 'rhe second fact has already been mentioned- half of all young smokers, in any year, seriously try to quit and can't. Tltr sitrgle fttost iiitportrtttt Jittdirtq reported itt tlie cotijerettce pmceedin~s tlieit joNt~u, ntay be this: thegreatest iinrcut r~Jcii~~tretteErrices, and hence ciy~tre~tte t~r~-es, oft smokiny belmarior appears to be on the ynmrgrst qW yrortps. Some other issues deserve attention. An important one is that a large number of young American men, at precisely the age when they may become regular smokers,.are offered tax-free cigarettes at Post Exchange prices on military bases in this country, abroad and on ships at sea. This is a fringe benefit that invites reexanunation. tn

0 ' a 4 Numerous other important topics also deserve attention: Should ciga- rettes be taxed according to their tar and nicotine content? Should tobacco products other than cigarettes, including snuff and chewing tobacco be (axed? Should the proceeds from the cigarette tax be earmarked for a variety of purposes ranging from health education to medical care? What are the likely responses of state cigarette taxes to changes in the federal tax? "hould cigarette taxes be indexed to the consumer price level in the event Of further inflation? All these are important questions. Some of these yuestiotu received attention at the conference; sonie did not. Quite deliberately, I have avoided stating a conclusion or making a recotnmendation in this brief preface. The conference was not convened no promote a tax increase, a tax reduction, or the maintenance of the current tax. If there was a consensus it probably emerges from the discus- •ion reported here; no consensus was made explicit. In any event, legisla- rurs will make up their own minds, no matter what we might recommend. I'he purpose of the conference was to make the best analysis and the best t:actual data that we could discover available to people who must decide. We hope that what is published here will be helpful in that spirit. The conference itself was a lively one. Acknowledgements This conference is the first of a continuing series of conferences intended to examine issues in public policy and behavioral research related to cigarette smoking. The series is sponsored by the Institute for the Study of Smoking Behavior and Policy, a research center dedicated to examining and enhancing the conceptual, analytical and practical linkages between smoking behavior research and policy at all levels. The Institute was established at the John F. Kennedy School of Govern- ment, Harvard University, in April 1984, with the support of the Carnegie Corporation of New York. We are greatly indebted to the Cabot Family Charitable Trust for its support of the overall series and to the Alfred P. Sloan Foundation for providing the funds for this inaugural conference. Tltontas C. Schelling Director The Institnte jor the Study oJSonoking Behavior and Policy John F. Kennedy School oJGoverrnnent Harvnrd Uaiversity v iv

Participants i Robert Batties Senior Program Specialist Office of the Assistant Secretary of Defense (Ilealth Affairs) Washington, DC Frank Cantrel Tax Counsel Senate Finance Committee Washington, DC Philip J. Cook Professor of Public Policy Studies Institute of Policy Sciences and Public Affairs Duke University Durham, NC William Drayton Environmental Safety Washington, DC Ervin S. Duggan Ervin S. Duggan Associates Washington, DC E. Ripley Forbes Special Assistant Subcommittee on Health and the Environment I-louse Energy and Commerce Committee United States Congress Washington, DC Harvey Galper Senior Fellow Brookings Institution Washington, DC Geraldine Gerardi Financial Economist Office of Tax Analysis United States Department of the Treasury Washington, DC Dean Gerstein Study Director Committee on Basic Research in the ' Behavioral and Social Sciences Commission on the Behavioral and Social Sciences antl Education National Research Council Washington, DC Jeffrey Harris, M.D. Associate Professor of Economics Massachusetts Institute of Technology Cambridge, MA Grady Hedgespeth Bureau Chief for Analysis Estimation and Research Massachusetts Department of Revenue Boston, MA Jan L. Hitchcock Research Associate Institute for the Study of Smoking Behavior and Policy Harvard University Cambridge, MA Celia Jaffe Research Assistant Institute for the Study of Smoking Behavior and Policy Harvard University Cambridge, MA Karl Kronebusch Office of Technology Assessment United States Congress Washington, DC Eugene Lewit [)irector of Research and Evaluation New f ersey Medical School Newark, NJ Matthew Myers Staff Director Coalition on Smoking or I lealth Washington, t)C vii

t s s . ? Joseph Pechman Michael A. Stoto Contents 5enior Fellow in Economics Associate Professor of Public Policy lirookings Institution John F. Kennedy School of Foreword ..............................................i Washington, DC Government Harvard University John M. Pinney ..v . . . ements Acknowled Cambridge MA ... .. .. g ......................... Executive Director , Institute for the Stud of Smoking David Sundwall M D. y tiehavior and Policy . , Chief of Staff Participants .......................................... vii I larvard University Senate Health Committee C,uubridge, MA United States Congress Guide to Contents ..................................... xi Washington DC Albert Rees , 1'rrsident Eric Toder One: Commissioned Papers .............................. 1 Alfred P. Sloan Foundation Deputy Assistant Director Excerpts frotn: New York NY Tax Analysis Division , by Eric J. Toder ................ 3 Issues in the Taxation of Cigarettes Congressional Budget Office , Thomas C. Schelling United States Congress Consumption Impacts oja Charkye in the Federal Ciqarette Littauer Professor of Lucius N . Washington DC by Kenneth E. Warner ......................... 16 Excise Tax Political Economy; and Director , , Institute for the Study of Smoking Kenneth E. Warner Behavior and Policy Professor and Chairman Two: The Morning Discussion ........................... 25 Harvard University Department of Health Planning and Cambridge, MA Administration Three: The Afternoon Discussion ........... . . . . . . . . . . . . . 45 Arthur N. Singleton School of Public Health ~ University of Michi an Minorit Chief of Staff g 63 y Ann Arbor MI Appendix ............................................ Committee on Ways and Means , by Eric J. Toder ............... 65 Issnes in the Taxation of Cigarettes United States Congress , Washington, DC Corunmption Inipacts of a Change in the Federal Cigarette Excise Tax, by Kenneth E. Warner ......................... 88 On the Fairness ojCigarette Exicse Taxation, by Jeffrey E. Harris ....1U6 ~;; ix

Dt Guide to Contents Administrative issues, 30, 79-83 Bootlegging: see State taxes Consumption by age, sex, race, income, 29, 35, 76-79 (including Tables 7 and 8), 95 (Table 3), 106-108 (in- cluding Tables I and 2) in response to prices, 29, 32-36, 39 48-50, 72-76, 88-105 Deaths: see Health effects DemanJ: see Consumption Elasticityofdemand:seeConsumption I lealth effects of smoking, 37, 39-40, 42-43, 98-99 Incidence of tax by income, age, race, sex: see Taxes, incidence by income, age, race, sex Lobbies, 53-54 Military services policies, 55-60 tax exemption, 30, 55-57 Nicotine and smoking behavior, 33, 38-39 Prices and consumption: see Consumption recent trends, 36, 70-73 (including Tables 3-5), 90 in response to taxes, 36, 47-48, 93 ~ ltegressivity, 29, 40-43, 76-79 (in- ,' T__( cluding Tables 7 and 8), 84, 106- N 110 , Z Smoking Practices nicotine compensation, 38-39 economizing, 33-34, 48-50 quitting, 37-38, 50 see also: Consumption Smuggling: see State taxes State taxes rates and revenues, 27-29, 66-73 . (including Tables 2, 4, 5, and 6) military exemptions: see Military services interstate smuggling, 30, 55-57, 74-75 relation to federal tax, 31, 36-37, 82-83 Taxes recent trends, 27-28, 66-72 (in- cluding Tables 1, 3, and 5) effect on prices: see Prices current proposals, 51-52 revenues, 27, 67 (Table 1) earmarking,_51-53 incidence by age, income, race, sex, 76-79 (including Tables 7 and 8), 109, (including Tables 3 and 4) Teenagers, 32-33, 35, 39-40, 50, 75-76, 84, 92, 94-98, 100 xl

One Commissioned Papers Participants in the Conference on the Cigarette Excise Tax based their discussion largely on two papers which they had read before the confer- ence-one by Eric Toder of the Congressional Budget Office and one by Kenneth E. Warner of the University of Michigan School of Public Health. This section contains excerpts from those two papers. Their complete texts are published in the Appendix. N 1

. )k Issues in the Taxation of Cigarettes Eric J. Toder Deputy Assistant Director, 7ia.Y Analysis Division, Con,yrrssiottal l3ttdget C)Jf ice .-1 ciVu•ette is the perJrct type of perfect pleasnre. It is e.rqttisite, and it leares oltr tntsatisJied. I l7tat tttorr can one uyattt? -Oscar Wilde, The Picnirr of Doriatt Gray (1891), Chapter 6. I I4trttitr~: The Strryeott Cctteral has rleterrttitted that citarettc sntvkiuq is daqerons to yottr health. -Required statement on package of cigarettes. Introduction ... This paper reviews briefly some of the major tax policy concerns relating to cigarette taxation. Following a review of data on postwar trends in the burden of cigarette taxation in the United States, the paper briefly discusses and evaluates econometric research on the effect of ciga- rette excise taxes on cigarette consumption and reviews evidence on the incidence of cigarette taxes. The final sections of the paper review tax adininistration and enforcement concerns and issues in determining the best level of government at which to impose the tax. Recent Trends in Cigarette Taxation Cigarette excise taxes have been a declining share of both Federal and state excise tax revenues during the postwar period, despite numerous increases in state excise tax rates. Table 1 shows that Federal receipts from cigarette taxation increased in absolute terms from $1.2 billion in fiscal year 1950 to $2.5 billion in 1982, but declined as a share of total revenue from 3.2 percent to 0.4 percent and as a share of the Gross National Product from 0.5 percent to less than 0.1 percent. As a result of the tax increase in Tax Equity and Fiscal ResponsibiliEy Act of 1982, Federal receipts from ciga- rette excise taxes nearly doubled to $4.7 billion in fiscal year 1984, about 0.7 percent of Federal revenues and slightly over 0.1 percent of GNR Federal cigarette excise tax receipts as a share of total revenue and GNP remain below the 1950 level in every year between 1950 and 1975. Table 2 shows that statc cigarette excise tax receipts have grown at a 3 H

I'able 1. Federal *Tax Collections on Cigarettes, Fiscal Years Table 2. State Tax Collections on Cigarettes, Fiscal Years 1950-1984 00 1950-1984 M I Ar igarette Tax Revenues IS millions) . Total Federal Rrvenue (S billions) NP (S billions) Cigarette Tax Revenues as Percent of Total Revenue Cigarette Tax Revenues as Percent of GNP 1,242.8 39.4 265.1 3.15 0.47 1,294.0 51.6 312.8 2.51 0.41 1,474.1 66.2 339.3 2.23 0.43 1,586.8 69.6 361.3 2.28 , 0.44 +i1 1,513.7 69.7 364.2 2.17 0.42 : >;5 1,5(A/.2 65.5 380.6 2.30 0.40 1'66 1,549.0 74.6 411.8 2.08 0.38 1''37 1,610.9 80.0 433.9 2.01 0.37 1''5,i 1,668.2 79.6 443.1 2.10 0.38 1•'i') 1,771.1 79.2 474.4 2.24 0.37 I"IbQ 1,863.6 92.5 497.9 2.01 0.37 I'M 1,923.5 94.4 509.3 2.04 0.38 I'+62 1,956.5 99.7 548.2 1.96 0.36 1963 2,010.5 106.6 578.0 1.89 0.35 1964 1,976.7 112.6 618.2 1.76 0.32 1't65 2,069.7 116.8 659.5 1.77 0.31 I'K>6 2,006.5 130.8 724.1 1.53 0.28 067 2,023.1 148.8 777.3 1.36 0.26 1968 2,066.2 153.0 831.3 1.35 0.25 1')69 2,082.1 186.9 910.6 1.11 0.23 1970 2,036.1 192.8 968.8 1.06 0.21 1971 2,149.5 187.1 1,031.5 1.15 0.21 1')72 2,151.2 207.3 1,128.8 1.04 0.19 1973 2,221.0 230.8 1,252.0 0.96 0.18 1974 2,383.0 263.2 1,379.4 0.91 0.17 1975 2,261.1 279.1 1,479.9 0.81 0.15 1'+76 2,434.8 379.3 2,072.3 0.64 0.12 1977 2,279.2 355.6 1,862.8 0.64 0.12 1'l78 2,374.1 399.7 2,091.3 0.59 0.11 1't79 2,356.1 463.3 2,357.7 0.51 0.10 1980 2,604.4 517.1 2,575.8 0.50 0.10 1981 2,488.2 599.3 2,885.9 0.42 0.0') Nti2 2,496.1 617.8 3,046.0 0.40 0.08 1983 3,4214.4 600.6 3,221.4 0.57 0.11 1984 4,7-19.2 666.5 3,581.1 0.71 0.13 luurcc: The Tobacco Institute, The Tiur Qordru on Tobacco-Historiral Contpi/atiar, vol. 19, 1984, p. 8; Econoniic Repart oJthc President, Washington, D.C., 1985, p. 242; Advisory Contntission on p. 5; Economic Rr port of the President, Washington, D.C., 1985, pp. 242 anJ 318. Intergovernmental Relations, Si,qniJicant Fearnms oJFiscrrl Federalism. 1982-83 Edition, Washington, D.C., January 1984, p. 32. u s c crccnt R P 5tatr as Cigaretteevcn Tax Revenues Revenue GNP of Total as Percent Year (S millions) (S billions) (t billions) Revenue of GNP 1950 413.7 7.9 265.1 5.22 0.16 1951 444.4 8.9 312.8 4.97 0.14 1952 460.3 9!) 339.3 4.67 0.14 1953 477.2 10.6 361.3 4.52 0.13 1954 469.7 11.1 364.2 4.24 0.13 1955 470.2 11.6 380.6 4.05 0.12 1956 532.3 13.4 411.8 3.98 0.13 1957 581.1 14.5 433.9 4.00 0.13 1958 626.8 14.9 443.1 4.20 0.14 1959 706.6 15.8 474.4 4.46 0.15 1960 929.9 18.0 497.9 5.16 0.1') 1961 995.1 19.1 509.3 5.22 t1.=•) 1962 1,085.6 20.6 548.2 5.28 0.'1t 1963 1,132.8 22.1 578.0 5.12 0._U 1964 1,212.3 24.2 618.2 5.1N1 0._'tt 1965 1,327.1 26.1 659.5 5.08 0.20 1966 1,566.6 29.4 724.1 5.33 0.22 1967 1,643.0 31.9 777.3 5.14 0.21 1968 1,915.8 36.4 831.3 5.26 0.23 1969 2,101.8 41.9 910.6 5.01 0.23 1970 2,368.1 48.0 968.8 3.9•1 0.24 1971 2,594.6 51.5 1,031.5 5.03 0.25 1972 2,904.4 59.9 1,128.8 4.85 0.26 1973 3,092.8 68.1 1,252.1) 4.54 0.25 1974 3,225.2 74.2 1,379.4 4.35 0.23 1975 3,284.7 80.2 1,479.9 4.10 0.22 1976 3,4289 89.3 2,072.3 3.84 0.17 1977 3,483.3 101.1 1,862.8 3.45 0.19 1978 3,632.7 113.3 2,091.3 3.21 0.17 1979 3,621.6 124.9 2,357.7 2.90 0.15 1980 3,714.4 137.1 2,575.8 2.71 0.14 1')81 3,850.6 149.7 2,885.9 2.57 0.13 1982 3,922.2 162.7 3,046.0 2.41 0.13 1983 4,117.8 171.0 3,221.4 2.41 0.13 1984 4,233.0 N.A. 3,581.1 N.A. 0.12 Source: The Tobacco lnstitute, The Tav Bnrdrn on Ti~bacco-llistorical CumpiLttiom, vol. 19, 1984, cax N venucs T R 7otal ~ Cigarette i rax ..Igarcuc c .I . 5

t 1 nntch faster rate than Federal receipts over the same period, but have also declined (though only slightly) relative to GNP. State cigarette excise tax receipts increased from $0.4 billion in fiscal year 1950 to $4.2 billion in fiscal 'year 1984. The last two columns of Table 2 show that state cigarette , excise taxes incrrased faster than GNP throughout the late 1950s and 1960s but declined as a share of GNP after 1972 and as a share of total state ret•enues after 1966. Tables 3-5 show how state and Federal cigarette excise taxes have changed in nominal terms, in real terms, and as a percentage of cigarette prices over the same period. Because the Federal government and almost all states impose specific rather than ad tulurem taxes, inflation reduces cigarette taxes in real terms and as a percentage of the price of cigarettes in years when the excise tax rate is unchanged. Table 3 shows that the Federal cigarette excise tax rate was increased only twice during the postward period - from 7 cents per pack in 1950 to 8 cents in 1951, and from 8 cents to 16 cents in TEFRA in 1982. The Federal tax as a percentage of the cigarette price (including taxes) declined from 42.2 percent to 10.7 percent between 1947 and 1982 and now stands at 16.6 percent, about the satne rate as in 1975. In contrast, Table 4 shows that real state tax rates and state taxes as a percentage of the price of cigarettes are higher today than they were in the It ~50s ... Table 5 shows that combined state and Federal cigarette excise taxes declined from 49.9 percent of cigarette prices it11954 to 27.8 percent in 1982, and have since (as of 1984) risen to 31.8 percent largely because of the doubling of the Federal excise tax rate. The combined excise tax rate as a percentage of price, however, remains lower than the.combined rate in effect as recently as 1980. These figures understate the tax burden on cigarettes to sonie degree because they do not include excise taxes imposed by some localities. In addition, most states with general sales taxes include cigarettes in the tax base. General sales taxes represent an additional portion of the sale price of cigarettes claimed by the tax collector, but do not raise the price of cigarettes relative to most other goods. Because, as discussed below, cigarette taxes in high tax states can be Sources: Thc Tobacco Institute, 71tr liirBurden nn 7ii6atto-His)uriral Cntw )iLttiun,..•ol. 19,1984, avoided to some degree by buying froni other states, the dispersion of tax p. 6; U.S. Department of Agriculture (cigarette price data supplied by Ro,ert Miller); Eronolnir rates among states can be important for enforcement and other Federal Repo.r of r1,e President, Washington, o.C.,1985, p. 291. policy issues. Table 6 shows that the number of states taxing cigarettes (including the District of Columbia) has risen from 42 in 1954 to al151 in 1984, but the tax rates are widely divergent, ranging from two cents per . pmrk in North Carolina to 26 cents in Connecticut and Massachusetts. The difference in price per pack between tlte highest and lowest tax states, Table 3. Burden of Federal Cigarette Excise Taxes: 1951-1984 Tax Rate Current $ Tax Rate 1984S - Average Cigarette I'rice Tax Rate 1percentage M ~ Year (centsipack) (cents/pack) (cents, pack) of price) N 1946 7.0 37.5 ~ 1947 7.0 32.6 16.6 42.2 1948 7 0 30 2 17.4 -t0.2 194') . 7.0 . 30.5 17.9 39.1 H 1950 7.0 30.2 18.2 38.5 1951 8.0 32.0 18.9 •12.3 1952 8.0 31.3 19.9 •10.2 1953 80 31.1 20.9 38.3 1954 8.0 30.9 21.2 37.7 1955 8.0 31.0 21.3 37.6 1956 8.0 30.6 21.8 36.7 1957 8.0 29.5 22.4 B.7 1958 8.0 28.7 23.2 3•1.5 1959 8.0 28.5 24.2 33.1 1960 8.0 28.1 24.9 32.1 1961 8.0 27.8 25.1 3t!) 1962 8.0 .5 7 25.4 31.5 63 1 8 .0 .1 2 25.9 30.9 1964 8.0 26.8 26.4 30.3 1965 8.0 26.3 27.7 28.9 1966 8.0 25.6 2').1 27.5 1967 8.0 24.9 30.2 26.5 1968 8.0 23.9 32.1 2•1.9 1969 8 0 22.7 33.9 23.6 1970 8.0 21.4 37.0 21.6 1971 8.0 20.5 38.7 20.7 1972 8.0 19.9 40.9 1').6 1973 8.0 18.7 42.0 1').tl 1974 8.0 16.9 44.1 1K•1 1975 8.0 15.4 47.3 16.9 1976 8.0 14.6 49.3 16.2 1977 8.0 13.7 51.6 15.5 1978 8.0 12.7 54.3 14.7 1979 $8.0 11.4 57.3 14.0 1980 8.0 10.1 62.0 12'J 1981 R8.0 9.1 66.9 12.0 1982 8.0 8.6 74.7 10.7 1983 16.0 16.7 90.1 17.8 198•1 16.0 16.0 96.3 16.6

, a Table 4. Burden of State Cigarette Excise Taxes: 1950-1984 however, has declined in 1984 dollars from 48.7 cents in 1971 to 24 cents in iiar 'lax Rate Current t Icents/pack)a Tax Rate 1984 t (cents/pack) Average Cigarette Price (cents/pack) Tax Rate (percentage of price) 1•>>u 2.4 10.5 18.2 11+s1 2.5 10.2 18.9 1•,52 2.5 9.8 19.9 1•63 2.5 9.8 20.9 1'6a 2.6 10.0 21.2 12.2 1'+55 2.6 10.2 21.3 12.4 1v56 2.9 11.1 21.8 13.4 t')57 3.0 11.2 22.4 13.5 1958 3.1 11.2 23.2 13.4 t'69 3.3 11.9 24.2 13.8 1960 4.1 14.4 24.9 16.5 1961 4.3 14.9 25.1 17.1 1962 4.7 16.0 25.4 18.3 1963 4.7 16.0 25.9 18.2 1'M 5.1 17.0 26.4 19.2 1965 5.3 17.6 27.7 19.3 1966 6.2 19.9 29.1 21.4 1967 6.5 20.1 30.2 21.4 1vi18 7.6 22.6 32.1 23.6 1'169 8.3 23.4 33.9 24.3 1970 9.5 25.5 37.0 25.7 1971 9.9 25.3 38.7 25.5 1972 11.0 27.3 40.9 26.9 1973 11.3 26.4 42.0 26.9 1974 11.1 23.4 44.1 25.1 1975 11.8 22.8 47.3 25.0 1976 11.4 20.7 49.3 23.1 1977 12.0 20.6 51.6 23.3 1978 12.3 19.5 54.3 22.6 1979 12.1 17.3 57.3 21.1 1980 12.6 15.9 62.0 20.4 1981 12.5 14.3 66.9 18.7 1982 12.8 13.8 74.7 17.1 1983 14.3 14.9 90.1 15.9 1984 14.6 14.6 96.3 15.1 1984. As a result, the economic gain from buying cigarettes in low tax states has declined since the early 1970s, though it remains significant. Effects of Taxes on Cigarette Consumption The price of a pack of cigarettes has increased in 1984 dollars from about 78 cents er 1ck in 1950 to about 96 cents er ~tck in 1984-an increase of P p p p` about 25 percent in 34 years-while over the same period the sum of average state and Federal excise taxes in 1984 dollars has declined about 40 cents per pack to about 30 cents per pack. If real excise tax rates had been raised rather than lowered over this period, prices would have risen more and tax policy would have reinforced instead of offsetting other efforts to control smoking. The question is how much difference might this have made. This section provides a brief overview of recent studies of thc efferts of cigarette prices on smokitlg ... In the two years following the enactment of TEFRA, retail cigarette prices increased by about 22 cents per pack, while state and Federal excise taxes rose by only about 10 cents per pack. In the two years preceding TEFRA, prices rose by about 12 cents per pack, while excise taxes were roughly constant. While it is unpossible to draw any conclusions about the effects of the recent Federal tax increase on price without a fuller model explaining the determinants of the recent growth in cigarette prices, the ex erience since 1980 does not contradict the assttm tion that all of the tax P P is paid by consumers. .. Lewit and Coate f nd that cigarette prices affect smoking primarily by reducing the participation rate; the estimated effect on the number of cigarettes per smoker is statistically insignif cant. There are also significant estimated differences in price elasticities among groups; reported price elasticities are much higher for adult males than for adult females and much higher for people aged 20-25 (-0.89) than for other age groups. This Sources: The Tobacco Institute, Tlir TaxBurdrn on Tidxrccu-Hisrorica! Connpilation, vol. 19,1984, finding that price elasticities are higher among younger groups is consis- pp. 6 and 196; U.S. Department of Agriculture; Ecunontic Repnrt ojMe Presidenr. tent with results from a separate study of teenage youths by Lewit, Coate, a. Computed by multiplying average tax rate in the taxing states by the ratio of cigarette sales in the taxing states to total cigarette sales in the United States and Grossman (1981) that reports a price elasticity of -1.2 for the proprn- y sity to smoke among teenagers. Lewit, Coate, and Grossman find that teenagers are also sensitive to "fairness doctrine" variables, in particular to anti-smoking messages that stress the health hazards due to smoking, but smoking increased between 1970 and 1974 because effects of the increased anti-smoking publicity associated with the fairness doctrine were offset by the effects of a decline in real cigarette prices. Based on the research by Lewit and Coate and by Lewit, Coate, and Grossman, it would appear that price clianges oprrale nlairtly by dr'rerrinq t)

ieeur{yers and yattny adtttts jroJJt beginning to smoke. Because smoking tends to be habit-forming, this finding of a high price response among teenagers and young adults suggests that increasing cigarette excise taxes and then either indexing them to price changes or periodically raising them to maintain the higher real tax burdens over time could eventually result in Table 5. Burden of State and Federal Cigarette Excise Taxes: 1950-1984 Tax Rate Current f Tax Rate 1984 f Average Cigarette Tax Rate Price (percentage Year (cents/pack) (cents/pack) (cents/pack) of price) 1'J5u 9.4 40.7 42 18.2 18 9 1951 10.5 .2 . 1952 10.5 41.1 19.9 1953 10.5 40.9 20.9 1954 10.6 40.9 21.2 49.9 1')55 10.6 41.3 21.3 49.9 1956 10.9 41.7 21.8 50.1 1')57 11.0 40.7 22.4 49.3 1958 11.1 39.9 23.2 47.9 1959 11.3 40.4 24.2 46.8 1960 12.1 42.5 24.9 48.6 1961 12.3 42.6 25.1 48.9 1962 12.7 43.4 25.4 49.8 1963 12.7 43.1 25.9 49.0 1964 13.1 43.7 26.4 ' 49.5 1')65 13.3 43.9 27.7 48.2 1966 14.2 45.5 29.1 48.9 1967 14.5 45.0 30.2 47.9 1968 15.6 46.5 32.1 48.5 1969 16.3 46.0 33.9 47.9 1')70 17.5 46.9 37.0 47.4 1971 17.9 45.8 38.7 46.1 1')72 19.0 47.2 40.9 46.4 1973 19.3 45.1 42.0 46.0 1')74 19.1 40.2 44.1 43.3 1975 19.8 38.3 47.3 41.9 1976 1').4 35.3 49.3 39.3 1')77 20.0 34.3 51.6 38.8 1978 20.3 32.2 54.3 37.3 1979 20.1 28.7 57.3 35.0 I'>80 20.6 26.0 62.0 33.3 1981 20.5 23.4 66.9 30.6 P982 20.8 22.4 74.7 27.8 1'J83 30.3 31.6 90.) 33.7 1984 30.6 30.6 96.3 31.8 , much lower smoking rates among the entire public. It also suggests, however, that the full decline in overall smoking rates among the entire population will take many years to occur and that beneficial effects on the hralth of the population will also be delayed to the extent the damage to any individual caused by cigarette smoking is cumulative over many years. In the short run, the major effect of higher cigarette taxes may be to increase taxes paid by current smokers, without significantly reducing either total smoking or its associated health costs. Table 6. Dispersion in Cigarette Excise Taxes Among States ear Number of Taxing States Minimum Rate Current $ (centsi pack) Maximum Minimum Rate Rate Current S 19ri4S (cents/ Icrnts• pack) pack) Maxintum Itate 19tt•I ieents' park) Itange 19,t-1 S (cenn>' pack) 1954 42 0.0 8.0 y~ 0.0 30.9 3t1') 1955 42 0.0 8.0 0.0 31.0 31.0 1956 43 0.0 8.0 0.0 30.0 30.6 1957 43 0.0 8.0 ~ 11.0 2'J.5 29.5 1')58 •t•1 0.0 8.0 0.0 2ri.7 28.7 1959 47 0.0 8.0 0.0 28.5 38.5 1960 48 0.0 8.0 0.0 28.1 28.1 1761 48 0.0 8.0 0.0 27.8 27.8 1962 48 (1.11 8.11 0.0 27.5 27.5 1')63 48 0.0 8.0 ll.O 27.1 27.1 1404. 4') 0.0 8.0 0.0 26.8 26.8 1965 49 0.0 11.0 0.0 36.2 .i6.2 1966 50 0.0 11.0 0.11 35.2 35.3 1967 50 0.0 13.0 OA -ItJ.4 •N1.1 1')68 50 0.0 15.0 0.0 •l4.8 d•1.8 1')6') 51 2.0 16.0 5.7 45.3 39.7 1')70 51 2.0 18.0 5.3 48.1 .12.8 1971 51 2.0 21.0 5.1 53.9 48.7 1972 51 2.0 21.0 5.0 52.1 47.2 1973 51 2.0 21.0 4.7 49.1 aa:l 1974 51 2.0 21.0 •1.2 4•1.2 40.0 1')75 51 2.0 21.0 3.9 40.5 36.7 1')76 51 2.0 21.0 3.6 38.3 3•1.7 1977 51 2.11 21.0 3.4 36.0 32.6 1')78 51 2.0 21.0 3.2 33.1 30.3 1')79 51 2.0 21.1) ?!) 311.1 27.2 1').sU 51 2.0 21.0 2.5 26.5 2•1.11 1981 51 2.0 21.11 2.3 20 21.7 1982 51 2.0 25.0 2.2 26!J '_•1.tt 1983 51 2.0 . 26.0 2.1 27.1 °_5.0 1994 51 2.11 26.0 2.t1 2r1.u 2-1.O Source: The Tobacco Institute, The 7iu tlnrdrn on Til(uial, pp. 145-195. I1

lucidence of Cigarette Excise Taxes ... Assuming the tax is paid by consumers, this section reviews recent evidence on the distribution of consumption of tobacco products by income class.'. The evidence confirms the belief that taxes on tobacco cunsumption are highly regressive; that is, expenditures on tobacco products .ll. a share of income are much greater in lower inconu classes than in higher income classes.2 .. . ... Tobacco consumption declines from 2.9 percent of income for the lowest fifth of the income distribution to 1.7 percent, 1.1 percent, 0.9 percent, and 0.5 percent for successively higher quintiles. As a percentage uf total expenditures, including both frequently purchased and inter- ntittently purchased items, spending on tobacco products increases from 1.3 percent of income in the lowest income quintile to 1.4 percent in the ucond lowest quintile, but then declines to 1.2 percent, 1.0 percent, and 0.7 percent percent in successively higher quintiles. Spending on tobacco products as a percentage of both total income and total expenditures declines more as income increases than does any other consumption category ...including expenditures on food, alcoholic beverages, housing, apparel and services, transportation, health care, entertainment, personal care, reading, education, and the categories labelleci "miscellaneous" and "all other categories:' Administration and Enforcement Issues State cigarette excise taxes are collected from wholesale distributors. (State taxes on manufacturers would, of course, be taxes on cigarettes pro- dttced within a state, not taxes on cigarette consumption by state residents.) 'Strictly speaking, the desired relationship to examine in the incidence of cigarette taxes among income classes is the relationship between cigarette consumption and income. Cigarette consumption, however, accounts for the bulk of spending on tobacco products, so that the relationship between tobacco products and income is quite close to the relationship between cigarette consumption and income. Moreover, other tobacco products (cigarette papers and tubes and cigars) also are subject to federal excise taxes. -"1'he fact that tobacco taxes are regressive by themselves need not mean, however, that an increase in the federal cigarette tax rate must necessarily reduce the progressivity of the entire federal tax system. Since cigarette taxes are a relatively minor revenue source, overall progressivity could be roughly maintained if a cigarette tax increase were combined with provisions to increase the relative tax shares paid by middle- and upper-income people under the personal income tax. Taxation at the retail level would vastly multiply the number of points of collection. S prci jic vs. ~l d Ftilvrertt Trtxation The current Federal tobacco taxes are mostly specific excises on units of well defined products: S8.00 per thousand (16 cents per pack) for small cigarettes; S16.40 per thousand for large cigarettes; t/z cent for each 50 papers for cigarette papers and one cent for each 50 tubes for cigarette tubes (except if they measure more than 61/'z inches in length, in which case every 2~/a inches is regarded as one paper or one tube); and 75 cents per thousand for small cigars (those weighing not more than three pounds per thousand). The one ad nalorenn tax is the tax on large cigars, which is 81h percent of the wholesale price, but not more than $20 per thousand. State cigarette excise taxes are also specific taxes. Before 1976, New Hampshire's excise tax was a fixed percentage of the retail price, but it has since been converted to a per-unit tax ... One advantage of changing to an ad nalurem tax is that revenues would rise to keep pace with inflation-induced increases in cigarette prices while, in contrast, the real revenue yield from the current specific excise taxes declines as the price level rises even if cigarette prices keep pace with inflation. Thus, over time, the excise tax rates must be statutorily adjusted to maintain real yields. One option to correct for the real erosion in cigarette tax revenues, while maintaining the benefits of per-unit taxes, is to index the tax rate to increases in either the general price level or a price index for all cigarettes. gging frum Lott, Tr.,r to Hiylr Tax States Bootle One major concern to state cigarette tax administrators is "bootlegging" of cigarettes across state lines to avoid taxation in high tax states. The large differences in tax rates among states shown in Section 11-ranging from two cents per pack in North Carolina to 26 cents per pack in Massachusetts in 1984-provide a large economic incentive for bootlegging. ... In response to this problem, the Congress in 1978 enacted the Federal Cigarette Contraband Act, prohibiting the transportation, receipt, shipment, possession, distribution, or purchase of more than 60,000 ciga- rettes not bearing the tax indicia of the state in which the cigarettes are found. . . . This law has been very effective in reducing large-scale cigarette smug- gling, although there is some evidence smuggling began to increase again in 1983 ... I ' ~ 13

The enforcement and revenue loss problems resulting from interstate 'I1ipmcnt of cigarettes are one consideration in determining the appropri- .ite Federal role in cigarette taxation ... Issues in State/Federal Relations Onc ctdvantage of collecting cigarette taxes at the Federal level rather than .u tlie state and local level is that Federal taxes involve lower collection and (A•.ision problems. A Federal tax can be collected from a small number of producers and importers and, because the tax rate is uniform throughout the country, there is no incrntive,for cigarette bootlegging. Qn the other hand, separate state and local cigarette taxes allow citizens more freedom 4 choice by allowing a variation in the tax structure among cotnmunities; .ind, currrntly, cigarette taxes are an important revenue source in some states and localities. Concerns about the potential effect on state and local cigarette tax revenues were raised when the Federal government increased the cigarette excise tax in 1982. An increase in the Federal tax causes some reduction in state and local revenues to the extent consumption of cigarettes declines in response to the tax-induced price increase ... One effect of much larger increases in the Federal excise tax rate would be a significant erosion of state tax bases ... On the other hand, a larger Federal share of total national receipts from cigarette taxes can be justified on the argument that the Federal government, tluough medicare, tax . preferences for health insurance, and other programs now bears an increas- ing share of the health care costs attributable to smoking. Nonetheless, state taxes as well as Federal taxes raise cigarette prices and discourage smoking. Alternative policy options that would maintain state involvement in taxation and/or maintain state shares in total cigarette tax receipts include: Assistance to State Enforcement ... Turnback of Federal Revenues to States ... Turnback of the Tax Base to States ... The Reagan Administration's 1982 "New Federalism" proposal included a combination of these last two alternatives ...[but] ... was not acted on by the Congress. I -t Conclusions :.. The evidence reviewed in this paper indicates that the real burden of cigarette taxation has declined significantly in the postwar period, even taking account of the doubling of the Federal excise tax in TEFRA. Many econometric studies show an effect of cigarette prices on consumption, but most studies show demand to be relatively inelastic. More recent research suggests that the effects of tax-induced price increases may be grvatest on teenagers and young adults and may significantly affect the decision to begin smoking. These results imply that the long-run health benefits from higher cigarette taxation may be greater than previously believcd, but also imply that there could be a significant delay before those benefits are fully achieved. On the other hand, recent data shows that cigarette consumption is a much higher share of the income of low income households than of high income households, thereby suggesting that cigarette taxes are much more regressive than almost all other revenue sources. In addition, higher Federal taxes would reduce state cigarette excise tax revenues. A greater reliance on Federal instead of state taxation would, however, reduce enforcement and compliance costs per total dollar of revenue raised and reduce boot- legging from low tax to high tax states. ti 4

. 1 Consumption Impacts of a Change in the Federal Cigarette Excise Tax Kenneth E. Warner Nro/i-ssor and CJtairntun I hT,irtrncnt of!lealth Plannin,q and Adntinistnrtion ticl,ool oJ AtGlic Hcnlth I '»iiersity of ,blichiyart .-l nn Arbor, Michiqvn Introduction Traditionally, legislatures enacted excise taxes to raise revenues or to make a statement about the "morality" of use of the product in question. In the latter instance, the objective was to penalize or discourage the behavior- hence the name "sin tax." In recent years, however, excise taxes on products such as tobacco and alcohol have come to be viewed in a third dimension, one that has largely eclipsed the morality concern: excise taxes can be effective tools of public health policy. Economists have begun to evaluate the consumption impacts of such taxes as ends of inherent health interest, rather than as vehicles to estimate the revenue implications of tax-induced consumption changes ... Furthermore, public health professionals are including excise taxation as an essential element in writings on public health policy to combat smoking and alcohol abuse ... The current debate on the federal cigarette excise tax is timely in the context of both its revenue and consumption-and hence health-impli- cations. Regarding the former, both Congress and the Administration are searching actively for ways to diminish the swollen federal budget deficit. Mile any conceivable excise tax increases cannot be construed as making nwre than a dent in the deficit, for several reasons this tax has taken on ditnensions disproportionate to its possible contribution. For one, the Administration's 1984 deficit-reduction package included several effective tax increases and only one highly visible effective tax decrease-the decision to allow the federal cigarette tax to revert from 16 cents to eight cents per pack on October 1, 1985, as called for in the 1982 TEFRA sunset provision. Second, a cigarette tax can be viewed as a "user fee," a politically attractive bit of nomenclature adopted in a Congressional legislative proposal. The "user fee" notion meshes neatly with the call by the Advisory Council on Social Security to earmark a cigarette tax to offset smoking- I related Medicare expenditures (Rich, 1983). Furthermore, by being labclyd a "user fee," the excise can parade as a non-tax (or quasi-tax) •`revenue enhancer," thereby mitigating the Administratioii s insistence on avoiding tax increases. The political attraction is based too on the fact that fewer than a third of all adults-the smoking population-will bear its burden. The timeliness of interest in cigarette consumption impacts relates spe- cifically to the sunset provision for the current 16 cent tax, which analysts have predicted will cause smoking to increase, and more generally to the objective of the smoking-and-health community to work toward the Surgeon General's goal of a smoke-free generation by the year 2(HO (Koop, 1984). The nature of the consumption impact of a change in the excise tax is the subject of the remainder of this paper. In focusing solely on the consumption impacts, the paper addresses its assigned charge. It should be noted at the outset, however, that one of the major attractions of an increase in the federal excise tax is its ability to simultaneously serve the interests of fiscal and physical health (Warner, 1984). Price Elasticity of Demand for Cigarettes Price elasticity estimation remains a difficult task. To date, no study has successfully addressed the complex issues in smokers' shifting from high to low tar and nicotine (t/n) cigarettes, complicated by the fact of nicotine "regulation" by smokers....Other aspects of smoking behavior, such as price-induced increases in puffing frequency or smoking further down the butt, have escaped attention in all elasticity studies. While such behav- iors may not be relevant to an interest in the revenue implications of a tax or wholesale price change, they are of interest in assessing the health implications of tax-related changes in price. Also of interest to health professionals are differential price responses by income class (Townsend, 1983), t/n level, years of smoking history, size of daily habit, etc. With the exception of the first of these, none of these factors has been studied in elasticity analyses, and the first has been studied only in the context of smoking in Great Britain. An issue further complicating interpretation of elasticities is that there is good theoretical reason to expect an asymmetrical response to price increases and decreases. From survey data, the federal Office on Smoking and Health has concluded that very few people begin to smoke after the age of 21. Hence cigarette price decreases would be unlikely to induce adults to start to smoke (though they might increase smokers' daily consumption), while price increases could encourage some smokers to quit (and others to reduce daily consumption). The implication is that price response might be considerably greater in the instance of price increases than decreases.

• . .i Again, no study has addressed this challenging analytical problem, though it has clear relevance to both economic and health concerns. In particular, it i% central to an assessment of both the revenue and consumption implica- tiuns of a change in the federal cigarette excise tax. l=inall}•, it is important to recognize that the lag between collection of data, analysis, and publication of results inevitably encompasses a period of w%•cral years ... ... Nevertheless, the state of the art experienced a substantial improvement in the earl,v 1980s when Lewit and his colleagues (1981, 1982) produced %tudirs of both adult and teenage price elasticities ... ... For this reason, this paper uses the elasticity estimates by Lewit et al. to estimate the consumption effects to be expected as the result of now- plausible changes in the federal excise tax. ... The total price elasticity estimates ... exhibit a pattern of price responsiveness decreasing with age, as would be expected for three reasons: 11) teens and young adults have smoking habits that arc less well-defined and of shorter duration, implying less habituation or addiction and hence the potential for more price responsiveness; (2) younger people may be more inclined to start smoking as the result of a price decrease than would be older adults, as discussed earlier; (3) on average, younger people will have less disposable income so that price response may include more of an income effect. Worthy of note is that teens' cigarette demand is quite elastic Ahile that of adults is inelastic. One of the most important findings in the estimates of Lewit et al. is that "participation" or prevalence decisions... account for the vast majority of total price response. All of the daily quantity elasticity estimates ... are small and statistically nonsignificant, while all but one of the participation elasticities are significant. The absence of apparent daily consumption response seems counterintuitive. One can imagine a number of adjustment mechanisms that do not involve change in the number of cigarettes smoked -more or fewer puffs per cigarette, smoking further or less far down the cigarette, etc.,-but a change in daily consumption is perhaps the most obvious response. It is possible that anomalies in smokers' reporting of their daily habits on the surveys could disguise a consumption change ... Consumption Impacts of Changes in the Federal Excise Tax In 1984, the weighted average retail price of a pack of cigarettes was 97.8 cents (Tobacco Institute, 1984). If the federal excise tax reverts to eight cents a pack on October 1, as scheduled, and the retail price of cigarettes talls by the same amount, average price will fall by 8.5 percent. If, instead, the tax were to be increased from its current level of 16 cents to 24 cents or 32 cents, average price would rise by 7.9 percent or 15.1 percent, respectively ... From Table 4 we see that ... the currently legislated eight-cent decrease in the excise tax would induce almost 2 million people to smoke who would not do so if the tax were to remain at 16 cents ...Among the 1.9 million are more' than 460,000 teenagers who would begin or continue smoking as a result of the tax decrease. Adding in the most price-responsive adults, those aged 20-25, we find that more than 1 million young people would join the ranks of the smoking population if the tax decrease takes effect. An eight-cent increase would have a quantitatively similar opposite effect. 1.8 million people would be encouraged to quit or not bet;in Table 4. Estimated Changes in Cigarette Stnoking Attributable to Changes in the Federal Cigarette Excise Tax' Age Group 8-cent 8-cent 16-crnt Decrease Increase Increase Change in number of smokers (thousands) 12-17 + 334 - 311 - 5't•1 18-19 + 130 - 121 - 231 20-25 + 608 - 565 -1.o1i0 26-35 + 513 - 478 -')11 36-74 + 3d5 - 321 - 612 75+ + 13 -12 -23 Total +1,943 -1,808 -3,451 Change in aggregate cigarette consumption (u cigerattes, billions) 12-17 + 2.3 -2.1 -1.11 18-19 + 1.0 -0.9 -1.7 20-25 +4.8 -4.5 -8.5 26-35 +a.1 -3.8 -7.2 36-74 + 8.2 - 7.h -1 a.6 75+ + 0.2 -1).2 L-U. TotaP' +20 ., -19.1 -36:1 '"fhc prrcrntagc changrs for 3b 7d yrar olds have been applic.l to the oldest group of adults 05+) as well. Smokers in this age bracket may he more confirmed smokers than vounger adults, pet haps implying less price response, but they are also likely to be poorer on avrrane, implying mnre response to a price change. Note that thete smokers constitute less than 2 percent of the smoking population. For 11t-191•ear-olds, percentage changes midway between those of 12-17 .tnd _'U-25 ?•car-ulds have been used. " The absolute value of each of these totals is about 6 percent greater than the til;ureti calculated by combining the prevalence and daily cnmuntption chanl;es dircctll•. This rewlts I rom the r+tinta- tion procedure for the different elasticities tucd by Lewit et al.

4 4moking, including over 400,000 teenagers and more than half a million ~-uung adults aged 20-25 (and over a million young adults age 20-35). A I Frcent increase in the excise tax, bringing the real value of the tax close to its value in the early 1950s, would encourage almost 3.5 million Americans a0 forgo smoking habits in which they will engage if the tax remains at 16 %'rnts per pack. This figure includes over 800,000 teenagers and almost two million young adults age 20-35. The aggregate annual changes in cigarette consumption are substantial in absolute magnitude-ranging from an increase of over 20 billion cigarettes to a decrease of more than 36 billion-but represent a relatively small proportion of the domestic cigarette market (from 3.4 to 6.2 percent). Discussion Consumption Impacts. The immediately preceding point illustrates a fundamental conclusion of this analysis: the overall relative consumption impact of conceivable tax changes is modest, only on the order of a few percentage points. But the size of the cigarette smoking population and the daily consumption of smokers mean that even modest relative changes become substantial effects in terms of absolute magnitude. This is readily illustrated by the fact that, if our assumptions hold, an eight-cent tax change will alter the size of the smoking population by less than 3.5 percent, but that 3.5 percent represents almost 2 million Americans. The impacts of tax-induced consumption changes are of most immedi- ate importance in the population of middle-aged and older adults, because these are the individuals most prone to experience smoking-related illness. 'Che elasticity estimates produced by Lewit et al. show this group to be the least price-responsive, yet the sheer size of this group means that, under the uperative assumptions, from 330,000 to 630,000 persons will alter their smoking status if the federal tax is changed. In the long run, the toll of smoking is tied to the smoking practices of the youngest generation. There is a widespread consensus that the ultimate conquest of smoking-induced illness can come only from preventing the onset of smoking in the teenage and early adult years. In this regard, the elasticity studies of Lewit et al. and their translation into numbers of smokers are particularly important. Not only do the price responses repre- sent large numbers of young people; they also represent substantial propor- tions. An eight-cent decrease in the federal excise tax would increase tlie ranks of teenage smokers by a tenth. A 16-cent tax increase would diminish the population of teenage smokers by fully 17 percent. The former would lead approximately 460,000 teenagers in the direction of cigarette habits; the latter would lead 820,000 teens away from dependency on cigarettes ... a . It is important to emphasize ... that ... an eight-cent decrease would encourage tens or hundreds of thousands of Americans to smoke who would not otherwise do so; a cax increase of the magnitude considered would encourage many hundreds of thousands, and likely millions, (if Americans to quit smoking and, in the case of youngsters, not to start. Health Implications ... The fact that one lifelong smoker of every three or four dies from a smoking-related illness can be used to produce some "ballpark" estimates of the mortality implications of the contemplated tax changes. For example, if we assume that one of every four tax-induced quitters (or nonstarters) would have died from smoking, and if we adopt the assumptions used to generate Tables 2-4, the eight-cent increase would be credited with averting the smoking-induced premature deaths of 450,000 Americans. The 16-cent increase would avoid 860,000 premature deaths. By contrast, if the eight- cent tax decrease takes effect and the above assumptions hold, over 480,(lt)0 Americans will die prematurely as a result of their tax-induced initiation or continuation of smoking habits. On average, these victims of smoking will die more than two decades earlier than they would have if they had quit smoking or never startcd.t Another perhaps obvious feature of these numbers also deserves empha- sis: the premature deaths averted or produced by a tax change are not realized all at once. The major consumption changes induced by a cigarette price change will occur in the youngest groups of smokers and potential smokers-teenagers and young adults. The numbers ... suggest that fully 80 percent of price-induced changes in smoking prevalence will be found in people 35 years of age or younger; nearly a quarter of all responders will be teenagers. If price response is asymmetrical ... a still larger share of smoking initiation associated with a tax decrease would be found in tile youngest age groups. The import of this distribution is that the majority of the premature deaths that will be associated with a tax decrease, or of the premature deaths avoided as the result of a tax increase, will occur two to four decades into the future. The most immediate mortality implications" relate to 'The author and a colleague have estimated that decreases in snioking prevalence attributable to anti-smoking activities, including excise taxation, had prevented over 20t),00O premature deaths by 1978, with exponentially increasing numbers in the ensuing years. On average; each of the premature deaths averted translated into 23 years of additional life (Warner and Murt, 1983). ,(

•S tax-induced changes in smoking prevalence in middle-aged and older .idults. Given the large numbers of smokers in these age categories, this still translates into a substantial mortality implication for the near-term future; hut it is only a fraction of the totals given above. For example, according to Iablr 4 close to a third of a million Americans aged 36 and older would quit smoking if the federal excise tax were raised eight cents; over 630,000 WOuld quit if the tax were doubled to 32 cents per pack. Given the earlier .isumptions, these figures would translate, respectively, into 83,000 and Ii'1.(N)0 premature deaths that would be averted in the more immediate future, from the year of the tax increase extending into the following two decades. . . The qualitatively important conclusion is that tens of thousands of A mericans will die prematurely if the excise tax falls back to eight cents per I)ack. If, instead, the tax is raised, tens of thousands will lead longer, hcalthier lives than they would otherwise. In both instances, substantial nonfatal illness burdens will be affected as well ... Further Implications About Elasticity Estimates (1) It is quite possible that as Townsend observed in England, lower socioeconomic groups would be more responsive to price changes than would be higher SES groups. Thus a tax-produced cigarette price decrease might cause relatively more poor people to join or remain in the ranks of smokers. Similarly, a tax-linked price increase might induce relatively larger proportions of the poor to forgo smoking. (2) The inelasticity of demand for cigarettes in the U.S. is undoubtedly in part the result of the low price of cigarettes relative to income. If real cigarette price were to increase substantially over time (which would require tax increases well in excess of those considered in this paper), demand elasticities might rise as well. In that circumstance, further price increases or decreases would be expected to have proportionately larger effects on cigarette consumption and hence, ultimately, on the burden of smoking-related illness. Finally, consideration must be given to the effects of inflation on real cigarette price and thus on the prevalence of smoking. During the more than 30 years of its existence, due to inflation the eight-cent federal excise eroded to only 2.5 cents in constant 1951 dollar value. The doubling of the tax in 1983 restored the tax to only about half of its real value in the early 1950s. Similarly, legislating any tax change in 1985, whether an increase or a decrease, will be tantamount to legislating an effective tax decrease in ensuing years, unless provision is made for inflation-compensating tax I I I . increases. This could be accomplished by shifting cigarette taxation to an ed valoretn basis or indexing tbe tax rate to the general price level or a price index for all cigarettes (Toder, 1985). The importance of this is that even though a tax boost in 1985 would discourage smoking, without an infla- tion adjustment the eroding value of the new tax in 1986 and later years would encourage people to start or maintain cigarette habits ... Conclusion An increase in the federal cigarette excise tax is an attractive, effective tool of health policy. Indeed, it is difficult to think of many policy measures that could have a comparable impact on the health of the public. A tax increase, however, is not the public health ideal because the inelasticity of cigarette dem.uid means that tax increases will not eradicate smoking nor even eliminate a large proportion of it, at least for taxes on the order of magnitude discussed here. But it may be precisely that inelasticity that makes a tax increase a viable public policy option at present, because it assures that a tax increase will generate a revenue increase. No one wants to see government fundamentally dependent.T cigarette excise tax revenues, but the situation in America is far from this possibility: the federal excise tax constitutes well under one percent of all federal revenues. In 1985, increasing the federal cigarette excise tax offers several attrac- tions. It promises to increase federal revenues, especially in the short run. It will discourage a large number, if a small percentage, of adults to give up their smoking or not to start. And it will discourage both a large number and a significant percentage of young people frofii starting or continuing to smoke. The legacy of a tax increase would be a significant contribution toward the realization of a smoke-free generation. Tlte legacy of a tax decrease would be tens of thousands of avoidable premature deaths in the coming decades. 21

Two TheMorning , Discussion The conference, which began at 9:30 a.m. on April 17, 1985, took plae:e at the National Academy of Sciences in Washington. Joseph Pechman, Director of Economic Studies at the Brookings Institution, presided. Excerpts from the morning session follow. 1 25

Mr. Pechman: Welcome to the Conference on The Cii;arette [;xcise Tax. My name is Joe Pechman from the Brookings Institution. We have two papers. I would like to keep the summary of the papers within bouncls, but please do as complete a job as possible-say, within 20 minutes; no longer than half an hour. Mr. Toder: I can do it in less. By way of introduction, others have mentioned that the issue of extending the cigarette tax is likely to come up in Congress this year. The federal cigarette tax was kept at eight cents per pack between 1951 and 1982, which is a fairly long time. This was the same period in which there was increased publicity about the hazards of smoking-the Surgeon General's Report, restrictions on cigarette advertising, and the like. While all this was going on, while all these efforts were being made to reduce smoking, the rcwt tax rate imposed by the Federal Government on cigarette consumption was allowed to decline steadily. It's an interesting sidelight on the Federal Government's policy of giving with one hand and taking with the other. The 1982 tax increase was largely due to the budget deficit, which became much worse following 1981. In scraping for ways to pick up some revenue in The Tax Equity and Fiscal Responsibility Act (TEFRA), higher excise taxes on cigarettes, alcohol and other goods- telephone taxes, for example-were considered. The telephone and cigarette taxes passed; alcohol was left alone. In the discussion of '82, there were, of course, references to the health effects of smoking: how smoking was an undesira- ble thing, and so probably a good thing to tax. But 1 would say the largest motivation for the tax was revenue rather than health policy. Cigarette excise taxes have never been a very large fraction of either federal revenue or the Gross National Product. But if you will look at Table 1, you will see that these taxes go from three percent of total federal revenues in 1950 down to four tenths of one percent in '82. That last figure, of course, nearly doubled with the doubling of the tax, as a percentage of GNP. In the states, the experience has been somewhat different. State tax revenues grew, relative to GNP, in the 'S0s and '60s. And cigarette taxes kept up with other state taxes. Since 1970, though, state cigarette taxes have declined relative to GNP and relative to state revenue. Tables 3 to 5 show the tax rates: taxes as a percentage of price. The federal tax was eight cents from'51 through'82, and then jumped to 16 cents. In constant 1984 dollars, though, the tax declined- from 371'z cents to 8.6 cents in 1982. In 1984, it was 16 cents or a little bit less. At the same time, the price of cigarettes went up, as it turned out, slightly more than the Consumer Price Index. The federal tax, as a 27

percentage of the price, went down from 42 percent in 1947 to slightly over 10 percent in '82. Now it is up to 16 percent. . State taxes, as a percentage of price, have increased since the 'SOs, but that increase took place between the mid-'50s and the early 1970s. Since the early 1970s the tax rate, as a percentage of price, has declined. The tax rate has also declined in real terms. Throughout the period however, nominal state taxes have been increasing and the number of states imposing cigarette taxes has also increased; since 1971, all states and the I)istrict of Columbia have taxed cigarettes. Mr. Schelling: Excuse tne, Eric. In the early years, some states didn't have cigarette taxes. Were these states left out of your base, or were they listed as having a zero tax rate? Mr. Toder: In terms of the calculation I did, they are put in as having a zero tax rate. Now, this doesn't show the complete picture of state cigarette taxes. It does leave out some localities-notably New York City, which has a very high tax on cigarettes added onto the state tax; other localities also have excise taxes. In addition, we have some 33 states which include sales of cigarettes in their general sales tax basis. So that is another piece of the price of cigarettes which is claimed by the tax collector. Table 5 shows the cotnbined burden of state and federal cigarette excise taxes. That burden has also declined significantly, although it has increased slightly since 1982. The combined burden, in real terms, is still below the 1978 level. This, of course, says nothing about how much we should tax cigarettes. I am simply pointing out that we are taxing them a lot less than we were. [ iiriatians Brtweett States A final table, Table 6, shows the divergence of taxes between states. The range of taxes was from zero to eight cents in 1954; today it goes from two to 26 cents. In 1984 dollars the difference between highest and lowest states was almost 31 cents in the 'SOs. Now it is down to 24 cents. So the spread has declined a little bit-mainly because the states at the bottom have very low taxes. Mr. Peclunan: Eric, the ones at the bottom-are they just the tobacco- prodttcing states?- Mr. Toder: Mostly. Mr. Pechinan: Suppose you eliminated tobacco-producing states. Is there meich dispersion among the rest of the states then? Mr. Toder: There is a fair amount. I think there are some states, like New hiampshire, which have very low taxes. Mr. Lewit: There are some states like California that tax at ten cents • I I i I I + according to law. New York's tax, the state tax, is not that great. But there is a substantial city tax which makes the tax very high for, I guess, half the state's population. Mr. Hedgespeth: There are only nine states with rates less than 11 cents a pack. EfJects ofthe TiY Mr. Toder: The next part of the paper goes through a very brief review of incidence studies. Since we have Eugene Lewit here, I won't say much about that, other than that, to my knowledge, everyone who has done econometric work has implicitly assumed that the supply of cigarettes is totally elastic. So the effect of any tax would be on consumption-on the price paid by constttners, not the price received by suppliers. In demand studies, there has been a fairly wide range of elasticity estimates. Recent surveys use a range of somewhere between minus.4 and minus .7 as appropriate estimates. Most papers say that based on the evidence, the demand is inelastic. The Lewit and Coate study went further than most, in that it looke d at sample data from individuals. It showed differences among different groups of the population, and differences in the effective price on the quantity of cigarettes smoked and on the decision to smoke or not to smoke, a significant point. Price differentials seem to affect whether or not you smoke rather than the quantity of cigarettes that a person does smoke. And the effect is much greater on younger age groups. On the incidence of taxes, I put together a little bit of data from the Bureau of Labor Statistics. I hoped to have sonie regression data but wasn't able to get that in time. If you look at it carefully; if you compare consumption of cigarettes among different income groups, the higher you go on the income scale, the smaller is the percentage of cigarette consumption relative to income. This means-if you believe the tax falls on consumers-that the cigarette tax is quite regressive. In fact, the decline in cigarette consumption relative to income and total expenditures appears to be much greater for cigarettes than for virtually any other good classified in the tax survey. So cigarette consumption does rise, in absolute terms, a little bit with income, but it tends to fall quite dramatically as a share of income relative to anything else we might tax, including alcohol. This is not to say, of course, that if we have higher cigarettee taxes and use the proceeds for various transfer payments for low income people, that that would not produce a more equal inconm distribution than otherwise ,.4 29

%V()uld be the case. Mr. Pechman: Does any state index? Mr. Todert Not as far as I know. Mr. Hedgespeth: Not as far as I am aware. Mr. Pechntan: What is the political inihibition? Mr. Hedge.;peth: I would say part of it is just the reluctance to be a tax raisrr. In Massachusetts, for example, the cigarette excise that we recently raised was pitched as a revenue protection measure. You could say it was linked in a lagging way to the changing federal excise tax. Talking to compatriots from other states, I find a general reluctance to be known as a tax raiser, even if it is a small tax such as the cigarette tax. t3o(Ok&ittg atid SmtiulittS, Mr. Toder: The next issue is bootlegging. Given the differing tax rates among the states, there is an incentive to buy from low tax states. It takes various forms, the most serious of which in the `70s was organized smug- gling involving criminal elements. There was federal legislation in 1978 which significantly cracked down on and reduced organized smuggling. There are other ways to smuggle: casual smuggling-an individual's buying cigarettes in neighboring states or bringing home a few cartons from a trip to North Carolina. That is really not statistically, nationwide, a major problem. And of course there is not really much you can do about it, other than to keep the state-to-state tax differentials from getting too high. The other problem is the sale of cigarettes through tax-free outlets, particularly in military post exchanges and Indian reservations. This is a source of revenue loss in those states which- Mr. Schelling: Eric, do states tax cigarettes on military reservations? Mr. Hedgespeth: No. Mr. Schelling: Such taxes are preempted by the Federal Government, which says no taxation? Mr. Hedgespeth: That is right. Mr. Schelling: How big is the differential, GrAy, in Massachusetts between the military base price and the outside price? Mr. Hedgespeth: As much as 30 cents. It is not only that they are exempt front our 26-cent state excise tax. There is also a minimum price compo- nent of our law that adds about four to five cents to the retail price of a pack of cigarettes. So you have a potential for a 30-cent spread. Also, as you would imagine, there is probably low overhead on military bases in terms of profit margins. So that is probably at least another nickel a pack. I I I Mr. Toder: The next section deals with federal/state relationships. Since people are talking about raising the federal tax, that, of course, has certain impacts on states, particularly since the demand elasticity facing a single state is going to be higher than the demand elasticity nationally, for example. That is some limit on how much any individual state can tax. In 1982, 1 was asked to do a paper on what the effect of the federal tax increases was on state revenues. I did some quick estimates and came up with a rough calculation that ten percent of the federal revenue gain would be offset by a loss in revenue to the states. This was assuming, of course, that the states did not raise the tax rates, which many did afterwards. For Or A,Yaitcst The Tax Now to conclude: What ought policy to be? There are obviously different ways of looking at that. If you want to discourage smoking- if that's your goal, for health reasons-you probably ought to raise thee tax as high as is politically feasible. On the other hand, if you take the opposite perspective-that people ought to be free to do what they like-you should not impose discrimina- tory taxes on the third of the population that smokes, relative to the rest who don't smoke. The taxation should be based on a more general criteri- on, such as consumption or income. Finally, one can advocate cigarette taxes on some kind of insurance premium notion. Smokers have higher health costs, and thus impose higher financial costs on the rest of tts. In most cases, it's difficult to lower insurance rates-medical insurance and health insurance rates-for non- smokers. So the rest of us are paying for the cost of smokers. But that does not tell us how high the tax should be. I have not as yet, based on what I've seen, gotten a solid numerical guidance as to how high you make the tax on the basis of that criterion. ?lIr. I 1 iiriier 's Presentcttion Mr. Pechman: Thank you very much, Eric. Let's turn directly to Ken Warner. Mr. Warner: First, a little background on this paper that will add color to the discussion. Toni [Schelling] called me just before I was leaving for five weeks in China and said, "Will you write the paper?" I said, "No, I'm about to leave for five weeks in China." Hc said, "Write it on the airplane!" As a matter of fact, because of a snowstorm in Beijing, we were forced to sit on the ground for three'hours in Shanghai. So I did in fact draft the . ,1 11

hare hones of at least the qualitative outline at that time. What I tried to do in the paper overall is use the elasticity estimates that Gene Lewit and his colleagues have developed, both for teenagers and adults, to look particularly at the participation issue: the issue of who ' timokes how much. One of the major contributions that Gene's work has made to the understanding of cigarette price response is to separate out this notion of ' participation and daily consumption. Some problems-such as whether you smoke further down to the butt-are eliminated if in fact the partici- , patiem issue is au accurate one. 'lhr Under-RcporthiY Facor I don't know that 1 even mentioned it in this paper, but Gene mentioned it in his, and I think it's important to keep in mind: people uttder-report their smoking by a substantial percentage, typically on the order of about a third. They under-report, that is, relative to all of the data that we have on consumption based on production and sales. To the extent there are differential rates of under-reporting by income class, or what have you, many analyses based on surveys could introduce some significant biases. The assumption that is typically made is that the under-reporting is unbiased; that is, it's consistent across the board. I find that a nice conven- tion for purposes of continuing along with your estimations, but I don't bclieve it. li)uth and thc' 7hx One tremendously important issue that has never been assessed is the notion that, in fact, if you drop a price, you may have a significant impact on children's behavior. You may cause teenagers to start to smoke when they might not have otherwise-or you may encourage them to continue smoki ng. Logically, you would probably not see a dramatic effect on adult behavior, particuarly older adults whose decisions whether or not to start smoking have been made previously. Conversely, an increase in the tax would affect kids and teenagers. Given the estimates Gene has made, it would have a very profound effect on them. And it might have a substantial effect on adults, in the sense of discouraging them to continue to smoke. A tax increase may just be the straw that breaks-the camel's back for people who want to quit anyway. They're looking for an excuse, and this would give them an excuse. The import of this is the following: A price decrease which would result from a sunset provision might be expected to have less impact in inducing adults to start smoking than I have s t i estimated. But a price increase might have a larger impact on inducing adults to cease smoking. I have used the elasticity figure that Gene and his colleagues have developed, although, as has been noted in the preceding paper, y ou can use higher elasticities. From my reading of the literature, I think that .4 is probably as good as anything we've got. We may have a range between .4 and .7, but I am inclined to believe it's more than .4. In any event, what I've done here is simply take these elasticity estimates, separating out participation and daily consumption for cigarettes-these are on Table 1. Based on studies by Lewit and his colleagites, the first row, ages 12-17, are teenagers, one study. The others come from the adult study, lumped, as you see, into three age groups, the last of which, for obvious reasons, is a bit dissatisfying. There may be substantial differences between people in the younger end of this age group and the older age group, particularly if you look at the period in which these elasticity studies were undertaken and you start looking at women. One thingJeff Harris has pointed out is really fascinating to look at: The peak smoking rates for both cohorts of women were reached within a period of-what was it?-nine or ten years. In other words, because smoking was diffusing at a much later rate among women than men, you see the younger women sort of picking it up in an age pattern quite similar to that of men. But if you go back to the women who are now 40, 50 or 60, many of them were starting to smoke in large percentages. It's not a question of more, it's the relative magnitude of the two responses. For example, you can adjust your smoking behavior to smoke fewer cigarettes and still get the same nicotine dose. And I would expect to see a lot of that, or at least potentially to see a lot of that. So I'm quite surprised by it at this point. 12ich vs. l)t)or Mr. Schelling: Have there been any studies of how poor people smoke comparing it with how rich people smoke? Mr. Warner: Are you trying to separate that out as a conscious variable? Mr. Schelling: If you've got to stop smoking for a moment, and you're rich, you throw it away; if you're poor, you carefully put it out, put it behind your ear and then light it again. Mr. Warner: That would be a good way to get at the phenomenon. Mr. Schelling: My recollection is that the way my generation responded 32 33

to a price increase would have been to abandon filtertips; buy 100 millime- ter cigarettes, cut them in half with a pair of scissors, and smoke twice as many shorter cigarettes. I remember when people carried around cigarette paper so that they could unroll the butts and roll their own. Mr. Pechnian: Tom, that occurred during the 1930s, but it hasn't happened t•cr}• much since then. Mr. Schelling: Because of the rise in real income? Mr. Peclunan: Yes. Mr. Warner: When you look at the Scandinavian countries, in Norway t here's a tremendous amount of rolling your own cigarettes despite the fact that there's not that much income differential. In Sweden, you'll see next to nunc. So these aren't purely economic phenomena. Mr. Drayton: In Sweden it's concentrated in poor neighborhoods; the prices are lower there. Mr. Harris: Part of it might be a problem with the way the data are collected. This is a cross-section study by Gene Lewit and his colleagues. Therefore, we're comparing what individuals report as their cigarette use in relation to the tax rate in the particular state of residence. Most people round off their cigarette use in packs per day and such rounding errors tend to reduce any estimated effect of price on constunp- tion. To give a striking example, suppose that a person smokes 30 cigarettes a day but, as is typical, says "I smoke a pack a day," in response to a survey question. Now: we have a tax increase, and that person reduces his consumption to 20 cigarettes a day. He still says, "I smoke 20 a day:'• That may resolve, or help to resolve, one question: why the cross-section surveys of the kind Gene Lewit has done give a somewhat lower price elasticity than the time series studies. We're not relying on individual self reports of smoking, therefore, maybe for the very low quantity smoker-elasticities are somewhat lower thatl, in fact, they really are. Mr. Pechman: Back to Ken. '17irt't' Sct'11arios Mr. Warner: In Table 2, and in effect throughout the rest of the analysis here, what I've done is look at three possibilities for the general tax. I have, again as Eric noted, adopted the standard assumption of perfectly elastic supply. It's going to be passed on as an eight cent decrease which is the sunset provision; an eight-cent increase which is simply for the sake of parallelism; and a 16-cent increase which was included in part because if you were to increase the tax by about 16 cents, you would be getting back into the ball park of the real value of the federal tax when it was last raised in 1951. So these were benchmark figures. These are all going from 16-from the current tax %vhich was intended to sunset to eil;ht cents which is the first set of columns. So this table simpl}• gives you the percentage change in cigarette consumption. Again, broken down into participation in daily quantities, that would result in changes of these magnitudes to the cigarette excise tax by age group. In Table 3, we estimated numbers of smokers, average daily and rsti- i mated annual consumption by age for 1982. The footnotes explain where i the data come from. : Gene grabbed me last night and said,"Where did you get those j enormously large figures for the daily consumption of teenal;ers?" I was ) shocked when I came up here with some of the figures that were provided to me by the National Center for Health Statistics. It depends on which survey you use. There's a National Institute of Drug Abuse (NII)A) survey, there's the Health Interview Survey, and several others. They end up with radically different figures. So, please consider those teenage consumption figures even more arbi- trary than the adult ones which are under-reported, again, by around a third. _1~ As Jeff mentioned, most people tell you a pack a day but the average is 30 for smokers. ' You also see that, by the way, when you look at total annual consump- tion in the last column, it comes out to 415 billion smoked-compared with the actual total of 600 billion that are sold. We had estimates of smokers from the NIDA survey of high school seniors, also, the Health Interview Survey. Those are all very rough. But, again, the percentages are relatively low and you're talking:about a minority of the smoking population-a very important minority given the hcalth issues in later years-but a minority for purposes of tax effects. A lot of this should be interpreted as qualitative information in a quantitative form. In Table 4, 1 take the ntunbers from the preceding table, and the percentage changes that were estimated on tile basis of tile elasticities for the three plausible tax changes, an eight-cent increase, an eight-cent decrease, a 16-cent increase. Looking first at the change in the number of smokers, the participation or prevalence elasticity; and in the middle, the daily consumption of people who continue to smoke; then finally the change in aggregate cigarette consumption. The most important numbers are at the top part of the table. They lead me to have some reservations about the specific numbers here with the 1.1 35

J eight-cent decrease, which might be snlaller. But if it is smaller, and if the clasticity is.4, that would imply that an increase in taxes should have a more .ubsclntlal effect on deterring smoking, to get more people to quit, since it's not a purely symmetrical response around .4. . I Price Drop? Mr. Forbes: Ken, a question for you. Is it fair for us to assume if the tax .1mps to eight cents, that the industry reduces the price by eight cents? Mr. Warner: That is the assumption here. Mr. Forbes: It seems to me, as a business decision, they're not going to reduce it. They'll leave it as it is and maybe slow future increases'down. But consumers would not see a change in the price. Mr. Warner: The real question is how much monopoly power they will have. In the past I would have been more inclined to support your view; but given the competition with the different prices on cigarette brands now, I think that you'd see a rather substantial decrease in prices once 'such a provision takes effect. First of all, there's the repackaging of cigarettes; now you can buy 25 to a pack for the same price as 20, or you can buy the generic brands. And of course, the new development is the low-price brand in between; it's not a generic, but it's not the standard price either. So there's a fair amount of price competition that we're observing now. I would expect to see that translate into a substantial decrease, whether it will be a full eight cents or not. Mr. Forbes: But you're seeing it not only between giant manufacturers and small manufacturers but the giant ones against themselves. We see Liggett coming out with new generic brands, but we're seeing the majors conic out with generics as well. Mr. Warner: They're trying to differentiate their products. All products do that. Beer-you've got Busch, Budweiser, and so forth. You've got a product line; you compete both with yourself and your competitors. Tile point is, I think there's sufficient competition among the competitors. But I believe that you will see a change that will be quite comparable to whatever the tax change is. That's not to say it will be exactly eight cents. .~nue Replacenient Trxes Mr. Forbes: I think you've got some states coming in with stop-gap Iel;islation. As soon as the federal tax goes down, they go up. Mr. Warner: Absolutely. To the extent that the states have legislation ,pecitically designed to replace a decrease in federal excise tax, that would tend to nlitigate sonle of these consumption impacts. But I would state I again that I am looking just at the eflects ol tlle change in federal taxrs, because there's no way I can tell exactly what's going to happen. Several states have propclsed Mhat you're discussing: a replacement state tax, in effect. In fact, I'm very disturbed from a public health perspective. I'm very disturbed about the notion that the momentum behind that movement might he an excuse to allow the federal tax to fall because there are going to be a handful of states which will ultimately accomplish that tax increase. So you can see this eight-cent decrease as an outer limit, as a boundary. A Tix [Gitlt Health Implications Let me go to the most important point of all: We are not simply talking about revenues, although I do believe that revenue is the basis on which the federal government and most of the states tetld to look at these things. We're talking about a tax change that's going to affect cigarette consunlp- tion and quite clearly is going to have an impact on the numbers of lives lost prematurely. The virtue of trying to quantify sonle of this is to make the qualitative point that we're looking at tens of thousands of premature deaths in one direction or the other. I tried to translate sonle of the consumption changes and the participa- tion changes into nunlbers of people whose premature deaths would either occur or be avoided as a result of these measures. If you take these figures which are changes in participation-in smoking levels-and translate that into an impact on premature deaths, you're looking at about one out of three to one out of four of these people experiencing-or not experienc- ing -premature death as a result of the changes in consumption behavior. We are talking about substantial numbers. I [ ho Quits? Mr. Cook: Another question in trying to estimate the number of lives that are at stake here, is the question about who it is that is quitting or starting as a result of tlle change in the tax. A recent history of smoking suggested that what we've had in the last 30 years was a great reduction in the prevalence of liqdit smoking; that the people quitting over the last three decades have been the people smoking just a few cigarettes each day and thus, presunla- bly, were not at any great health risk as a result. So, if most of the action we get from changes in tax is changes in prevalence, and if that change is the result of llght smokers who quit, then I wonder if there's any logic to this at all? 10 • 37

Mr. Warner: You've got a lot of assumptions~in there which are too Nimple. I don't think we know precisely who it is who has quit. Surveys find the quitters having smoked larger quantities per day than the self- reports of continuing smokers. I don't believe either of those numbers; I think the quitters are exaggerating, and the people who are still smoking .ue underestimating their consumption. We just drni t really know very well what's going on. Mr. Cook: My claim about the history of the prevalence figures, is that accurate? Mr. Lewit: There are two issues. One is that this was quitting that takes place essentially in the absence of an active intervention on the tax or price side. So it was not really a response to the price effect. The second issue is that the nature of cigarettes has changed also. You Wuuld expect that the number of cigarettes for people who continue to smoke would rise, since people have been going to lower and lower nicotine cigarettes. You can't make the assumption that quitting is removing only the marginal people. Mr. Drayton: Gene, don't we have an aggregate figure? There was about a 50 percent reduction in the measure of tar and nicotine. And you see an increase in the number of people smoking. Then it starts leveling off. There's a central assumption that, indeed, you increase the dosage because of the nicotine. But if that were a major factor, you would think the cigarette companies would love it-that they would move as fast as they could to lower nicotine in cigarettes. (:orrrprrrsatory Sruokirg Mr. Schelling: Let's call on Dean Gerstein to describe for us, if he can, the results of the studies about compensation as people move to low tar and nicotine cigarettes. 13i11 Drayton, I noticed in the article you wrote, which was a dozen years ago, that you said there was no evidence that people compensated by smoking more cigarettes, or by smoking differently. I think there have been a number of studies since. Are you pretty well up on it, Dean? Mr. Gerstein: These apparent massive reductions in tar and nicotine index figures are a poor representation of the actual consumption rates. For example, when the plasma nicotine levels of smokers of different apparent tar and nicotine brands is compared, it appears that there is a series of swall adaptations in the number of cigarettes smoked and the way they are smoked. It's possible for someone who is smoking a brand with a low apparent tar and nicotine content actually to be absorbing as much carbon munoxiLle, tar and nicotine as someone who is smoking an apparently much more deadly cigarette! That's, I think, exactly what Ken is getting at. This apparent balance only has a relatively small effect-because one of the adaptations that people make is to smoke more cigarettes. , Mr. Drayton: You can see that with cigars. People don't inhale cigars, by and large, which is an adaptation in the opposite direction. Mr. Gerstein: Except that people who used to smoke cigarettes, and who have switched to cigars, do inhale! Mr. Schelling: The question is whether people do compensate by smoking more cigarettes. My recollection is that most of the studies, as of about four years ago, show that smokers did compensate with more cigarettes, but nowhere near proportionately. Therefore, the thought was that they were probably getting appreciably less tar and nicotine, maybe a little more carbon monoxide and some other products that had to do with the quantity of agricultural product combusted rather than the tar and nicotine. Essentially, it was that the carbon monoxide stayed level or may have gone ttp a little. The tar and nicotine were typically down, but not down in proportion to the measured content. I1/onrerr and Men Mr. Warner: Jttst one or two final points. Gene found very little response to price among women, but substantial response to price among men. He conjectured that this might change over time, as womcn's habits became more like those of the men. If women's smoking begins to mirror that of men's in a more traditional fashion, including their price respon- siveness, then we would be looking at larger elasticities-larger changes in behaviors in response to taxes-than those used-in these figures. It's also important in terms of the health impact to recognize that the most substantial health impact, and probably, the most substantial revenue impact over the longer haul, r.elates to what is happening with kids. Once they get into adulthood, they're not very likely to begin smoking. So if you can inoculate them with a tax at an early stage and get them to age 25, the number who would start to smoke would be much less substantial than the number who started in their teens. A CCrntiorrary Note Mr. Schelling: One cautionary point in all of this is that we are discussing the prevention of premature deaths, the saving of lives, a long way into the 314 39

future. I don't think anybody knows whether coronary heart disease is going to be a major cause of death in the year 2010, or if that will have been taken care of by other means. Therefore, what we're really doing is projecting current disease-specific death rates a long time into the future. I don't know what the prospects are for having early enough diagnosis of lung cancer, or some way to cut lung cancer fatalities early on; but it is not out of the question that, in 30 years, lung cancer might cease to be the threat that it is. Maybe we should recognize that to induce teenagers to quit smoking so that 40 years from now they'll be at less risk for heart disease and lung cancer, is somehow ... supposing that these particular diseases will be as resistant to treatment or prevention as they are now. We're dealing with what you might call the outside limit numerically of preventable deaths. If we had thought that smoking made people more susceptible to polio 35 years ago and a vaccine came along, you'd no longer be taking credit for getting kids to stop smoking on account of polio. Any of these things that prevent deaths in the age group 50-75 take a long time to have effect, if you're going to work on the teenagers. The Older Sittoker What is really disappointing, I think, is that the people most at risk, say, the 45-50 year olds, don't seem to quit much. Because they're the people among whom, if we could get them to quit, we might see a real difference. They're not going to be saved by medical magic. Mr. Warner: Again, because it's terribly critical from the health stand- point: you get small percentages, but given the population of smokers, males in particular, of that age, you're talking still about a large number of people who will be affected by this tax change in one direction or the other. And they are the ones who are most immediately at risk. What Is Regressivity? Mr. Pechman: Okay, Jeff, the regressivity question. Mr. Harris: I have a short comment, to make solely one point: although economists have used various kinds of information to address the question of the regressivity or fairness of a proposed excise tax increase, in fact, the question of the fairness of a tax increase is a political question, not an economic question. Accordingly, when one decides whether or not a tax increase hurts the "little guy," one ought to decide, in political terms and not merely ecunornic term, who the "little guy" is. - t , The conventional procedure, employed by economists, is to assess whether or not particular goods consumption rises or falls as a percentage of iruome among different incoine groups. Since I would like to address this as much as a political question as an economic question, I don't necessarily think that income is the only way to look at the issue of regressivity. I'd like to look at it in a broader sense. I'd like to ask: are certain groups in our society, who arc well-identified politically, more or less favored by a cigarette tax increase? By and large, for example, the elderly population smokes very, very little in comparison to the remainder of the adult population. Therefore, a cigarette tax increase would be a means of raising revenues while sparittq the incomes of the elderly. Second, although there is a common perception that a tax on cigarettes hurts minorities more than the remainder of the population, I don't find that that perception fits the facts. To the extent that one does -a pure analysis of cigarette consumption in relation to income and uses that analysis to address regressivity, I should remind people that even for the lowest income groups, cigarette consump- tion only comprises about two percent of consumption of expenditures on necessities. To focus_ solely on income does not really address the major questions at hand. Mr. Pechman: I am puzzled. I can agree with everything you said, but I learned, and I still state in everything I write, that one measures regressivity by comparing taxable income to total income and by looking at the distribution of the tax burden. Mr. Harris: I raised the question of how to define regressivity for the purposes of advising or counseling decision makers in the public arena. First, I don't know why the relative relation to income is any more or less important than the absolute magnitude of the burden. My numbers suggest that the absolute dollar burden does not decline with income, even if the relative burden does. Second, I don't know why we put the big focus on income solely. No one ever proposed that a cigarette tax was going to redistribute income. I would focus not on whether the cigarette tax redistributes income one way or the other, but what it does to one identified group or another in the political process. Mr. Pechman: With respect to tile first part, I thought that the reason one defines regressivity in the relation between the tax and income, by income levels, is because we make the assumption that there is a relationship between income and ability to pay. It is true there is a lot of different -t 1 4

opinion about that particular point. You might argue that we should calculate regressivity with respect to wealth rather than income, or total consumption rather than income. But on an}• une of those bases-total consumption, income, or wealth-I think that you will find wealthy people pay relatively less cigarette tax, relative to their income, wealth, or consumption, than do people in the lower Inconle classes. I can't understand how you can argue that this is not regressivity. Mr. Galper: If you are concerned about the political problem, the way I think you deal with that is not to claim that the tax is not regressive. Instead Vou should say, "This tax is a regressive element in the system, but it is desirable to raise this tax for a whole bunch of reasons". We have to make sure, liowever, that the whole tax system is taking this into account and has the progressivity that we like. So we should either make offsetting changes elsewhere in the tax system or make effective changes elsewhere on the expenditure side. That is how one deals with the political issue. But to deny that this tax all by itself is regressive, I don't think does deal with that. I I ill Pcvple Sntoke Less? Mr. Cook: I'd like to make two points; both focus on the health effects of cigarette taxation. It seems to me that the point of fairness is what effect au increase in the cigarette tax would have on the poor. One of the impacts that we think will occur will be that the poor will smoke less as a result of that increase in the cigarette tax. That that will improve the health of people. And quite possibly their earnings will increase as a result of improved health. So if we ask, "How well off are the children living in poor families because of the increase in the cigarette tax?" the answer might very well be that they're better off. Why? Because even though perhaps the parents are spending more on cigarettes, they are actually smoking less. They're healthier; their medical expenses are less. So this is a case where we can't be too narrow in our focus. We shouldn't just look at the overall income distribution effects of an increase in the cigarette tax. The second point is that what we're talking about here is small potatoes, compared with the $300 billion plus that we spend on health care every year in this country. If an increase in the cigarette tax reduced the health rare budget by 1 percent, that would be f3 billion or more. And, certainly, a very interesting issue would be the distributional P I impacts of a'S3 billion reduction in health care e xpenditures. That could be pushed further: smokers die young. Now let us ask: is tbat a positive externality or a negative externllity? It could be argued that to extend the lives of more and more people is going to hurt the Medicare system and the Social Security system by raising their costs. it could be argued that maybe these smokers are doing us a terrific benefit by dying young! Maybe we should not tax cigarettes, but subsidize them, from that point of view. I say all this because there are some huge numbers out there that seem to me to be much more important than the question of a few bucks for cigarette taxation. Mr. Rees: I'd like to make it more specific: the same Congress that is going to be considering extending the cigarette tax is also going to be consider- ing several tax reform proposals. All of them have the characteristic that they greatly increase the exemptions, or the tax entry point, for low income people. That effect on the taxation of the lowest income group just swamps the effects of a cigarette tax on the poor. Pol itics and Ecoiromics Mr. Harris: Let me reiterate. One of the things that Eric's numbers show is that the total burden is small potatoes compared to other issues. Second, however, I feel that one of the goals of an economist should be to inform the political decision-making process. The political process does not measure the worth of a person solely on the basis of income. I think minorities and elderly people in this country, to the extent that they have a favored or an appreciated status, are not valued solely accord- ing to their wealth. Therefore, we are now to inform decision-makers that minorities don't smoke in the aggregate as much as one would think they do. And, second, that old people smoke very little. From that point of view, an increase in the cigarette tax is nowhere near as unfair, politically defined as, say, making social security benefits subject to income taxation. When you use the word "regressivity" as if it's any more than a technical term, it's to somehow impose a value decision, make a value judgment on the the tax, which I don't think is appropriate. •13

Three The Afternoon Discussion i Professor Thomas Schelling, director of the Institute for the Study of Smoking Behavior and Policy, began the afternoon discussion with a question: How did the nation's cigarette manufacturers respond to the 1982 cigarette tax increase? He cited evidence that the companies, separ.ately and as a group, cushioned the impact of the tax-or made its effect less abrupt-by raising cigarette prices gradually in the months before the tax went into effect. 45

Mr. Schelling: Ripley Forbes described what happened in 19H2-83. Thcre has been kind of a concerted anticipation and then a staged increase, maybe to reduce the abruptness of the increase. It seems to me if they were able, concertedly, to cushion the impact of an eight-cent increase, they might be even better at cushioning a price decrease, when they'd probably far rather keep the money and run than play on an inelastic demand curve. Did the Cigarette Companies Maximize Profits? Mr. Myers: I think you actually understated when you said there was a concerted effort to cushion. There was a concerted effort to more than cushion. There was a concerted effort to take advantage of that excise tax increase to dramatically increase prices across the board. And they clid it in a way that would provide them increased profits per pack so far in excess of any decrease in consumption that their overall profits would go up. Look at what they did: the tax went into effect January 1. In late summer of the previous year, you find announcements of monthly price increases. So that between late summer of the year before the tax increase and the summer of the following year, the4were approximately eight different price increases, totalling approximately 25 percent of the total cost of a pack of cigarettes. ' A Iterord Year You also find this: when you look at their financial statements and profit and loss statements, purely under cigarette operations, you find that despite the decrease in consumption in '83 it was a record-breakint; year across the board for all major cigarette manufacturers. In tcrms of profit, it was exceeded only by the record-breaking year recently reported for '84. So you may be understating the level. Mr. Stoto: The whole industry went up? Mr. Myers: What you find is various companies taking advantage. It's a massive industry, but there are very few meaningful players. This is an industry in which two companies have fully two-thirds of the market, with a third company having around 12 percent more. So that if you put the three of them together, you're up to about 80 percent of the market. It doesn't take overt collusion to have concerted action. Mr. Stoto: There arc other things going on. For example, if people know the price is going up on January 1 by eight cents or 10 cents, there is a certain amount of anticipatory buying going on. Mr. Myers: You saw a lot of that. They took advantage of that very 1 47

dramatically, because in the late months of '82, you find very heavy ~rvholesale buying, as well as heavy retail buying, dramatically increased inve ntories. Since the companies have already built in the tax by that point, .ill that extra buy ing that takes place then to avoid the tax becomes pure profit. Mr. Galper: The only question is, who's going to get it? Is it going to the manufacturers or the wholesalers? /31auiiiiw the Feds Mr. Myers: The other thing they are allowed to do by taking that tack, from a pure public relations standpoint, is that the public is already psychologically geared up for an increase and are blaming the Federal Government for it, blaming somebody else for it. So the companies are able to raise the price and say generally it's the excise tax-and not have the consumer angry at them. There was still some impact on consumption, but the companies stayed way ahead of the game. Mr. Harris: What happened to the pattern of wholesale prices after the tax, that is, during the year the tax was in effect? Mr. Myers: As I say, what you saw is a progression of about four to five months before and four to five months after. Mr. Pechman: Your describing suggests that they haven't milked the public dry yet. They could ride tip the demand curve much more than they have. Mr. Myers: I think that's probably right. I think the answer to that is that that would be shortsighted. They might make maximized profits in the short run. But over the long run, they won't. And I think that's what is, in fact, going on. How Do Smokers Respond to Increases in Price? Mr. Peclunan: Could we return to one set of questions that at least I haven't heard addressed? Basically, all of the questions relating to what are the elements of the elasticities: Smaller or larger butt? Cheaper brands? Carton rather than pack purchases? Things like that. Gene, do you have any information on that? Mr. Lewit: They are important and interesting questions. I don't know that we have a lot of information. Some measurement studies have gone on in Europe, particularly in Germany, where you have had either duty increases or cigarette tax increases of one kind or another. People have gone around and measured butt length and found out that butt lengths are shorter, which would imply that there is some kind of compensation. The data that we have calculated on the consumption end, in ternns of gross numbers, doesn't indicate that we can find evidence of compensation. First of all, smoking fewer c.igarettes more intensely is on one level a reasonable compensatory behavior. On the other level, there are lots of attributes of smoking that this doesn't really take account of. You would ask yourself if that is the behavior that you are going to see in response to a price increase, why do people even smoke low-tar and -nicotine cigarettes anyway if they realize that they're going to have to smoke more to receive the san:e dose? Presumably, there are a lot of other attributes of cigarettes ~~ that have been chosen, when you are choosing a particular brand at a: particular price, and they can't be captured by the work we've done. Mr. Warner: I don't think that people rationally or consciously choose to inhale deeper or puff more frequently on a cigarette. That is a compensa- tory mechanism that occurs quite subconsciously. So they can do some compensation, but they may not plot it out. * Mr. Lewit: To the extent there is a subconscious compensation, they are not going to result in changes in consumption patterns at the store. The kind of changes that Tom was talking about earlier, where you crush out the cigarette, save it and smoke it later is because they've become more expensive-those are conscious efforts to economize on cigarettes. The kind of compensation that you're talking about, where you switch to a lower tar and nicotine cigarette and smoke it more vigorously, is not necessarily a conscious effort. That is not going to show up in the response to a price increase. I i itys of * Contitcitsntiti~j Mr. Schelling: Is it possible that these various responses, including not using cigarette machines but going to the supermarket, are behaviors substantially confined to people who are quite poor? It may be that people who are quite poor never converted to low tar and nicotine cigarettes anyhow. Therefore, when we look at what they do with low tar and nicotine cigarettes, we are discussing a class of people from whom we wouldn't expect these responses of an impecunious person to a rising price of something that only for the poor is really enough to worry about, in terius of average daily expenditure. Therefore, what we would really have to do is look and see how they smoke, let us say, in any low income class-inner city high schools compared with suburban high schools, say; or how the welfare mother in the tenement smokes compared with the upper middle class suburban housewife; people who are substantially unemployed compared with well-employed truck drivers. It may be that you really have to be pretty poor to want to save a cigarette you've already 1 is 1 49

lit or to shop around. So I can't believe that economizing doesn't occur to the people for .vhom the price increase is significant. What we don't know is whether generic cigarettes are so much the thing with the poor that either they are already buying nothing but generic cigarettes, or if the price goes up, chat's the first place they'll turn. If you are really hooked on cigarettes, it strikes me that giving up rather than walking a quarter of a mile to find c heaper cigarettes will be a remarkable feat, inasmuch as most people can't give up. They'd get out of bed and walk two miles for a cigarette, even if they have resolved to quit. I am just curious who would quit rather than economize. ' I t iuttirr T~ ) ( )uit Mr. Myers: A point that I think we're missing, too, is that all this doesn't happen in a vacuum. We're dealing with a world in which every survey teaches us that most of the people who smoke don't want to be smoking; that they would like to quit. I suspect even teenagers are picking it up now, and have ambivalent feelings. The excise tax increase itself may provide a catalyst to do something which they otherwise think they ought to do. When you were smoking, Tom, as you stated this morning, you weren't really that concerned about quitting for all those health-related reasons. Now, people knoue they're doing something they shouldn't be doing. They have a variety of motivations and reasons to quit, rather than play this s;ame-especially if the game is not a pleasant one to play. Mr. Lewit: People will consciously buy cigarettes in ways that are more expensive because they perceive those ways as being regulatory. By this I mean people will buy cigarettes by the pack as opposed to by the carton, which is much more expensive in terms of mon4tary outlay and even in terms of time. Why? Because every time you want a cigarette, you have to go and purchase a pack of cigarettes as opposed to having a stock on hand. It's much more expensive. But they do this, if you ask them, because they perceive it as being a regulatory mechanism. They don't want to have cigarettes too cheaply and, typically, too cheaply means that the time cost, not only the monetary cost, is low. But if they have theni too cheaply, they will consume more. Now, maybe they're really fooling themselves, but there are lots of mechanisms whereby people actually impose higher prices on themselves than they might otherwise. An Outline of Pending Tax Proposals Mr. Myers: To give you a two-second notion of what's going on, we've passed out a list of the cigarette cxcise tax bills which have been introduced in Congress already this year. Before you read it, it may be useful for you to realize that their sponsors run an extraordinary political gamut-from ultraconservative to ultra- liberal, Republican and Democrat. This is typical of how cigarette-related issues tend to go politically: You have to go throw out traditional notions of political allies, because they don't work on this issue. This is one of the few issues where Ted Kennedy and Orrin Hatch would sign a joint letter, or sponsor a joint bill, or even sit next to each other during a hearing! What you find here is a series of bills which were introduced early. They're listed tip at the top. By and large, it had been our expectation that most of the 16-cent bills, as [ call them, would do nothing more than repeal the Sunset. But as you can see here, that is not the case. Eannarking There is an enormous move among those members of Congress who have been most activist to date on the excise tax to look at this as a mechanism for raising money for particular purposes. As you can see here, the most popular proposed beneficiary has been the Medicare Trust Fund because of concerns over its long-term health. What we're finding right now is that people who are most interested in excise taxes, and cigarette excise taxes in particular, are also concerned about a variety of other health-related issues. They feel very strongly that the excise tax is a perfect way of funding those. Mr. Stoto: Can I ask what precisely it means to earmark? Mr. Myers: It means that the specific money that comes in as a result of that tax, would not go into the general treasury for general purposes. It would, instead, be targeted directly for a particular purpose, into the Medicare Trust Fund or the Hospital Insurance Trust Fund, and therefore be available solely for the use of that trust fund. Mr. Pechman: We have to remember that the alternative to doing sume- ilting like that is either increase payroll taxes or reduce benefits. If you earmark the eight cents, for example, for the Medicare Trust Fund, what you're saying is that we don't want any benefit cuts or tax increases to that extent. z ;n 1 51

. . Mr. Rees: And that addresses the regressivity issue to the extent that payroll taxes are a regressive tax. ; Mr. Myers: That is exactly right. It is those people who have long-term concerns about health policy, and how the poor and elderly in this country are paying for their health care, who are the strongest advocates of the earmarking of the funds for Medicare and other health related issues. Mr. Myers: This issue has been a way for some people concerned about Nledicare and health to speak out on both issues at one time. Early on it was partially a way to offset what they thought was going to be the opposition of the senior citizens lobby to an increased tax on the regressivity issue. Fortunately, that's something that's turned around because AARP- the American Association of Retired Persons-has joined a coalition that we put together to support a 32-cent tax. So the feeling that the senior citizen lobby would be against on this issue, because of the notion of its being regressive, has not materialized. Now, the senior citizen groups are, interestingly, joining us in increasing numbers. The second set of figures down there represents the doubling of the excise tax to 32 cents. You will notice that there is not a single 32-cent bill there that would send all the money into general revenues. They're all earmarking bills. ,Sharitiy Ff'ith The States There are no bills currently that talk about earmarking this money for revenue sharing with the states. But that is a popular idea, and the fact that there is no specific bill on it doesn't mean that it isn't going to be seriously considered if earmarking is seriously considered. Within the Congress, the tax is perceived as doing different things for different people. Those on Ways and Means and Finance see it as a revenue-generating measure; those who are talking about earmarking tend to be concerned about the health related issues. Mr. Hedgespeth: As a matter of perspective, just about every state earmarks its cigarette tax. Mr. Pechrnan: It does? For what purpose? Mr. Hedgespeth: Highway funds, support of the arts. It runs the gamut. 'f'lie llaq ~rrs if Earmarkitly Mr. Rees: Earmarking for something that has infinite revenue-absorbing capacity, like Medicare or Medicaid, doesti t bother me. It's just a substitute for some other kind of tax. On the other hand, when you earmark for narrow special purposes, then you get into serious trouble because you end tip doing one of two things-either overspending, spending foolishly, or creating an argument for cutting the tax back again at a later time. I presume that the people who are interested in the cigarette tax as a way of promoting health would not want to see that happen to the cigarette tax. That would be a powerful argument qqainst narrow earmarking. Earmarking for Medicare, since its costs are so vastly in excess of any revenue we could conceivably raise through cigarette taxes, does not present that problem. Mr. Cook: I'd like to say, when we were talking to the Advisory Council on Social Security, it struck me that earmarking it to the Medicare Trust Fund changes the politics of the tax system completely so that we would never again have a 30-year period where the tax was neglected by Congress. You would have this hungry mouth called Medicare. Mr. Pechman: Al, I must say you are beginning to break down my inhibitions against earmarking. That's a very telling point. It is true that if you don't earmark, the tax could conceivably go down; whereas, if you gave it to Medicare.. . Mr. Toder: What you have is a double-earmarking situation. It's only because when Medicare is tied to a particular tax, in a sense, you can think of it not only in terms of cigarette policy, but in terms of the whole question of financing Medicare. Mr. Rees: It begins to get general revenues into the financing of Medicare, which is something a lot of people have argued for for a long time. [3iiildittY Coalitivtts Mr. Peclunan: Why isn't the aging lobby joining the anti-smoking lobby? This is a natural alignment. Why don't they get together? Mr. Myers: They are together on this one. The reason AARP and the other senior citizens groups are now on this side is because the health lobby went to them and made the case. It said, "This is something you ought to be pushing for. It's in your interest in the long run. You people really don't get hurt because the number of your people who smoke is very, very small." The one lobby that is absolutely absent, of course, is the labor lobby, whom we can't seem to get to with a ten-foot pole at this point. Mr. Pechman: Why? Are they worried about their union members who smoke? Mr. Myers: Labor has been about the most difficult organization to reach on the overall cigarette issue across the board, no matter what the particu- lar issue. This issue has been interesting because back when we started 12 1 5 3

working on the excise tax issue again last fall, we set out to find out who clsc would be interested in working with the basic health lobby. We found that we were able to bring some poverty groups in, much to our surprise, like the Children's Defense Fund. We have brought some religious groups in who have never fought very strongly on this; the United Methodist Church, for example, has endorsed a 32-cent excise tax and agreed to begin working for it. So this is an issue that crosses funny boundaries. We have an ad hoc group of about 38 national organizations, representing about 30 million people, who have endorsed the 32-cent excise tax. .1 Sitiokclcss Generatimi? Mr. Schelling: I think, regarding the teenage kids, you can make an argummnt like this: the reason most teenage kids take up smoking is because other teenage kids are already smoking. If you could ever achieve one smokeless generation, so that every kid that went to junior high school saw nobody smoking, and went on to high school and saw nobody smoking, they would be deprived of nothing they wanted to do- because there'd be no particular reason to smoke. If there's enough smoking going on, the desire to smoke is socially contagious. Smoking is what I call a communi- cable disease -in the literal sense. Therefore, the concern is that everybody smoking generates a need to smoke in other kids. If we can suppress it far enough, smoking may die out, become an extinct behavior, not missed by those who come along later- because they wouldn't know what they were missing. And, in our judgment, they would have discovered by the age of 19 that they were hooked with a habit they didn't want. Suppose, after they f inished high school they asked, "Did I miss anything?" The answer would be, "You missed it for a few years but, by now, you're at the age where you know what you're missing and you're glad you missed it." So there may be a strong case that because smoking is indeed a conta- gious behavior, the suppression of it is in the interest of those whose smoking recruits others to smoke. I think there's one conclusion we might come to-to get away from the fact that we want the money from the tax proceeds. Suppose we let the tax lapse on October first and it goes back to eight cents, and suppose the cigarette companies succeed in holding the price up almost to the present price? Therefore, they're getting the $2.4 billion that we were hoping for. Do we say, "Well, at least this is second best. The worst would have been to have the price go down:' Mr. Pechman: That's persuasive to me. Mr. Rees: I can't believe that's going to happen, Tom. It seenis to me, as a public relations matter, the cigarette companies coiddti t afford that behavior. Mr. Schelling: Even if we urged them? (Laughter.) Mr. Warner: Write it into the legislation. (Laughter.) The Smuggling Problem Mr. Pecluuan: Grady, let's get to the smuggling issue. For the uninitiated like us, why is it so difficult to prevent smuggling? Mr. Hedgespeth: What we've found is that smuggling has taken on a different tone within society. It has become less of a wholesale phenome- non and more of a retail phenomenon. Casual smuggling tends to be our major problem. It has two causes: the disparity of tax rates across the states, but also tax-exempt sales. The difficulty when you start talking about excise tax increases is the potential for a real increase in both tax avoidance and evasion. HeiiiT Mililctry Smokiiig? The per capita consumption for individuals in Massachusetts is relatively low, around 154 packs per person for the adult population. But when you look at consumption on military bases for active duty personnel, it is 1,540 packs! Obviously, that number has some problems; it doesn't include dependents and retirees. But even when you throw in the active and dependent populations, sonie of whom obviously don t make their pur- chases on military bases, you get 350 packs per person. That is anywhere from double to 10 times the consumption rate of the general population. So obviously, we are concerned=especially considering that one of the papers today showed 31 percent of tax evasion being from sales that should not have occurred. From the state's perspective, even if the ability of the states to levy taxes on military installations isn't passed, there should be a cooperative effort of base commanders and the Department of Dcfense to charge the prevailing rate for cigarettes-to discourage this kind of military post purchasing. The profits to be made from that could go to subsidizing more attractive goods than cigarette consumption. It does seem strange that in this day and age, there should be a Depart- ment of Defense policy that encourages cigarette smoking among our active duty personnel. It seems that it would be in th,s best interest of ~1 1 55

tideral and Defense Department policy to follow along with the states. Mr. Pechmau: w th respect to the consumption of nontaxed cigarettes by civilians-is it the large-scale smuggling that accounts for most of it? Or does the purchasing at retail volume where you go across the state line account for a significant amount? Mr. Hedgespeth: We actually think it is split fairly evenly, about 60 percent smuggling by evasion across the borders, 40 percent on the military hasrs. 1'enrry-.-I nte Srtnuggliny Mr. Pechman: With respect to the 60 percent, what are the types? 7here are criminals who literally take cigarettes, truckloads of cigarettes into the state, on the one hand; whereas, you or I just crossing the border cuuld buy sotue cigarettes in New Hampshire. Which is the more predominant? Mr. Hedgespeth: Definitely the casual smuggler. Mr. Pechman: Not the criminal? Mr. Hedgespeth: No, not since the real price differentials declined significantly , and the federal enforcement actions started. Our enforce- ment people believe that the strengthening of enforcement and the legal remedies did a lot to eliminate the North Carolina-Virginia-Massachusetts connection. Our current enforcement efforts have uncovered small to medium-sized smuggling between New Hampshire and Massachsuetts, but nowhere near on the scale of what it used to be. So for us, our current problem-and you could say it is a manageable one-is with military installations. If there is a more intractable problem, it is public knowledge about the effects of casual smuggling. Obviously, there is not very much we are ever going to be able to do about that until we get parity between the states' tax rates, and that is not a likely outcome. Mr. Myers: Could the state government have gone to the military and said, "You really cause a hassle for us? Can't you just charge a comparable price?" Have the Governor of the State of Massachusetts, and the other governors who have military bases, gone either to the Department of Defense-or Congress, or somebody-and said, "This is a problem we have got to stop?" Mr. Hedgespeth: We haven't. I think it has been more of an activity of the western states that have large Indian reservations. I think they have approached that problem. In Massachusetts we haven't-yet. Mr. Lewit: It is my impression that there is an ongoing debate about the whole issue of commissaries, including the fact that there is no sales tax, and there are lots of other potentials for evasive activity on military bases and reservations. Sonic states have come up with a relatively workable plan where the tax is imposed on the reservation, then returned to the Indian nation ou a per capita basis, or sonic other basis, so that the differential disappears. Of course, then, sales on the reservations also go down. Mr. Peclunan: Bob, could you tell us about the attitude of the military toward this issue? Mr. Battjes: I certainly couldn't speak for the military. There are certainly differences of opinion, and how the policy is eventually going to get shaped is questionable right now. On the one hand, you have people concerned from a personnel point of view about the benefits of the military, who feel that commissary and exchange discounts are part of the compensation package that the military receives. So if you begin altering the price structure, you are affecting benefits. Also, there is the point of view that commissaries and exchanges provide merchandise that is desired by their public, and as such, cigarettes are an item in demand and should be available. On the other side are people concerned about health. They're concerned about the impact of low-priced cigarettes on inducing, continuing, or escalating levels of smoking. And then you have a whole range of points of view in between. From the health perspective, we are attempting, through a number of avenues, to have sonie impact on smoking within the military. We sec it as a significant health problem-one which needs to be addressed. As part of that, the issue of sales, particularly within commissaries, is something that we are looking at. How that will shake out ultimately is unknown. I emphasize commissaries because that is where the really low-priced sales occur, as opposed to exchanges where prices are not that much lower than in civilian stores. Mr. Hedgespeth: How amenable do you think a base commander would be to a delegation from the state, sent by the governor or revenue commis- sioner, to take action in this area? Mr. Battjes: I would suspect that the approach needs to be taken more at the national level. Mr. Myers: Is there anything to prevent the military from doing what they do on Indian reservations? If you raise the price of cigarettes in the commissary and you could agree that that extra increment will simply go back to funding the commissary in general. This would allow you to reduce prices elsewhere. Thatway, the benefit package would not really be altered. 56 : 1 ~ 57

Mr. Hedgespeth: That would be exactly why I would say that wouldn't be a good solution. Mr. Myers: Is there something that would prevent the military from doing that other than inertia? Mr. Pechman: You said that was being done on Indian reservations? Mr. Lewit: In Minnesota and a couple of Midwestern states. It is not fi•deral policy; the states have reached an agreement. The states do not have t.ixing powers on the reservations, so they do not collect taxes on the roervations. So they have an agreement whereby the taxes essentially are <oIlected on the sales on the reservations. Then they are rebated to the tribal organization, whatever it happens to be, to support their activities. I don't know the exact form it takes because it varies among the different .tates, but they have reached these agreements. Srnoking and the Military Services Mr. Pinney: I was jttst going to ask Bob Battjes if the Defense Department was also looking at the consumption of cigarettes by overseas forces, which is a category that includes something on the order of 500 million packs a year. Mr. Battjes: That would be part of the total picture. Our basic thrust is trying to alter the social acceptabilitiy of cigarettes. Smoking is, we hope, moving toward being seen as more of a negative behavior. Mr. Pechman: But what action is being taken by the military to achieve that? Mr. Battjes: That is in the process now of being formulated in terms of a program of health promotion, which would include greater emphasis on smoking among other things; making smoking cessation programs availa- ble and encouraging people to give up smoking. Mr. Schelling: I would think it would be hard to get the U. S. Army terribly interested in smoking cessation, inasmuch as smoking won't affect the health of these people during the time they are in military service. Mr. Myers: The Veterans Administration ought to care. Mr. Schelling: There are a lot of people who believe that smoking is helpful in coping with stress; that it is an adaptive behavior on the part of people who have to work long hours, stay alert on guard duty; go out on maneuvers, and so forth. Noncommissioned officers have an extremely high prevalence of smoking. Among commissioned officers, it is diminishing very, very rapidly. But it might be difficult to get highly motivated efforts on tnililtary bases to get soldiers to give up smoking inasmuch as there is nothing in it for anybody in command of the base or in command of the troops at any level. Highly motivated efforts require somebody who is concerned w ith the long-run public health of the United States. "11te EJfrct oit Preparedness Mr. Harris: I am not sure it is really directly relevant to this conference, but there is obviously a direct effect on the national preparedness, based on the enhanced incidence of disabling respiratory infections and days of disability; on work loss, absenteeism, and on the higher risk of small airway disease in the lung which reduces your exercise tolerance-or, to put it in cruder terms, puts you in poorer shape. I can't quantify those, but I would say there is a short-term, itnmcdiate issue for the military. Mr. Schelling: Do you think it is observable to them? Mr. Peclunan: Certainly, the absenteeism. Absenteeism increases and the short-term bonuses would be noticeable, wouldn't it? Mr. Schelling: There is a high correlation in that age group between smoking, drinking, various drugs. So it mayi~e difficult to sort out the separate effect of smoking. I should think that the case could be made that there is a syndrome that includes drugs and cigarettes-as there presumably is in that age group, especially overseas, just as there may be here in the high schools. Then, if you argue that cigarette smoking and various illicit drugs are a syndrome, maybe you can bring the interest in there. The question is, Jeff, whether you could demonstratG to them that it really does affect the performance. Mr. Galper: If they think that if cutting back on cigarettes is going to lead to other types of substance abuse, they might even be less interested. Militciry Health Proutotion Mr. Pinney: Joe, I was on a panel- to get Bob Battjcs off the hook, I think it is fair to say that the Department of Defense and the individual services, have all taken cognizance of the problem. Their health promotion efforts- and that was the purpose of the panel I was on, to review their health promotion efforts-are varied and mixed. But they all take cognizance of smoking as a problem. The question I think that is being formulated in Defense is to what extent there are going to be any more systematic health promotion efforts to help people quit smoking. The problem that is not being addressed as tenaciously, perhaps, is the 4 i18 ~ -`~9

question that this conference is about: is there a price elasticity of demand concern that they ought to have? They are bringing so many young people, who appear to be more price sensitive, into a situation in which they are given cigarettes that are untaxed from the point of view of federal excise, and also sold at discount rates. My guess is that commissary store cigarettes are about 30 to 40 cents a pack. If we send them overseas, we are giving young people an even bigger break in that they sit around being bored, establishing their habit fairly firmly during those crucial years. t don't think there is any problem about the Defense Department's helping them quit. But the real problem is their supplying very cheap cigarettes to that particular age group. Mr. Hedgespeth: There is another factor that ought to be considered. With the possibility of reduced federal revenue sharing, you are going to find states looking at this tax expenditure in their budgets as a very fertile ground to recoup some of that revenue. It is only a matter of time before the decision is going to be made as to who gets the consumer surplus. Is it going to be the enlisted serviceman's, the Department of Defense's to reallocate elsewhere, or is it going to be the states'? I think the Department of Defense ought to act quickly. Because if they don't do something to ease the subsidy, or to move the subsidy into some other area, then states are going to move in a concerted way to get the taxation power on military bases. They still may-even if Defense acts. Some Closing Words Mr. Pechman: Well, we are approaching the end of our agenda. It might be worthwhile to now throw the floor open for any final comments and judgments from the group. Ms. Hitchcock: Could I just make an observation about the thinking about how changes in taxation policy affect individual behavior? We have been in two streams. One is that people may be trying to compensate and change their smoking behavioc And that raised issues in terms of the rationale and effectiveness of the effect of the tax relative to health. Another one was looking at the economic aspects to see less consumption and buying of cigarettes, how that effect came out more in terms of being a smoker or not being a smoker than in terms of the number of cigarettes smoked. This made me think that part of the issue in excise tax philosophy has to be the presentation of the change. In fact, it may be too complicated to try to see if someone cuts down on the ntumGer of cigarettes. But if you could see that peole are quitting because of whatever messages they are getting from the government or the society at large, then that might be one area. Whether they smoke or not seems to be more clear-cut in terms of a measurement. Mr. Pechman: I must say that I have learned a lot of things that I didn't know before I came to this conference. I was not aware, for example, how much the Congress has simply neglected, tried to avoid the issue of the eight-cent cigarette tax. I was certainly interested to hear what Grady had to say about the federal-state relationship. Also, Al, I would like to think about earmarking some more. I hadn't thought that in this area there might be a better rationale for earmarking in relation to Medicare. Any other comments on what people have learned, or not learned, as the result of today? Mr. Rees: I have relearned something. That is that small conferences are more productive than big ones. Mr. Schelling: Was this a small or a big one? Mr. Pechman: This is a small one. And it was very good. Thank you very much, everyone. (Whereupon, at 4:10 p.m., the conference was adjourned.) (1u 1 6 1

I Appendix This section contains the full text of: Issues in the Taxation of Cigarettes, by Eric J. Toder, Congressional Budget Office. Consumption Impacts of a Change in the Federal Cigarette Excise Tax, by Kenneth E. Warner, University of Michigan. and an additional comment: On the Fairness of Cigarette Excise Taxation, by f effrey E. Harris, Massachusetts Institute of Technology. (11

Issues in the Taxation of Cigarettes 04 Eric J. Toder i ~ Deputy Assistant Director, Tax Analysis Division, Co»gressitmal I3uilYrt Office. r_~ N A cigarette is the perfect type of perfect pleasure. It is exqttisite, atul it leaves one unsatisfied. [i4rot more catt otte nunt? -Oscar Wilde, The Picture of Dorian Crny (1891), Chapter 6. t I'rtrttitt,q: The Suqeott General has deterntittetd that ci,yarette stnokit~q is dangerous to your health. -Required statement on package of cigarettes. Introduction Consumption of those commodities regarded as vices by some, recreation by others, and necessities by still others-such as tobacco and alcohol- have long been objects of the desires of tax collectors around the world. In the United States, alcohol taxes were an early target of tax protesters, but cigarettes were not taxed until the Civil War. Subsequently, Federal excise tax rates on cigarettes were changed frequently, with major increases usually occurring during wartime, but remained at a constant rate of 8 cents per pack between 1951 and the enactment of a temporary doubling of the tax rate in the Tax Equity and Fiscal Responsibility Act (TEFRA) in 1982. The first state to impose an excise tax on cigarettes was Iowa in 1921. The number of states taxing cigarette sales gradually increased until, by 1969, all states and the District of Columbia imposed cigarette excise taxes. Prior to TEFRA, the average state excise tax rate was higher than the Federal rate, and it is currently only slightly lower. Proponents of higher cigarette taxation point to the adverse effect of smoking on health and note that many of the health care costs attributable to smoking-related diseases are paid by non-smokers in the form of higher Federal taxes and private health insurance pretniums.t In addition, it is The author is grateful to Rosemarie Nielsen for help in compiling the data, to participants in the Conference for useful comments, and to Linda Brockman and Shirley Hornbuckle for typing the manuscript. The views in this paper are those of the author alone. 69

vuinted out that the real burden of cigarette taxation has declined during the postwar period, despite the increases in specific tax rates imposed by Staes. Opponents of higher cigarette taxation argue that the tax discrimi- n.ues against those people with the same ability to pay who prefer to spend ncL ir income on cigarettes rather than on other goods and scrvices and is highly regressive. In addition to differences about the extent to which L i, garette consumption should be taxed, there is also dispute about the .tppropriate relative roles of Federal and state taxation. Despite the rising concern and publicity about the adverse health effects .~f smoking in the postwar period, and the consequent introduction of .tnti-smoking Federal policies such as restriction's on cigarette advertising, t he Federal excise tax on cigarettes was allowed to decline steadily in real terms throughout the period. The increased interest in cigarette taxation during the consideration of TEFRA in 1982 was prompted mostly by the eulergence of huge prospective Federal deficits in the wake of the 1981 tax cuts. The continued concern about Federal deficits, projected to remain at a.3 percent of GNP in 1990,2 makes the extension of tlle TEFRA tax increase beyond its scheduled expiration date of October 1, 1985 an attractive option for many in the Congress.3 For this reason, analysis of the role of cigarette excise taxes in the Federal tax system and of the effects of such taxes on smoking levels, health, Federal health program costs, and income distribution are particularly appropriate at this time. Tllis paper reviews briefly some of the, major tax policy concerns relating to cigarette taxation. Following a review of data on postwar trends in the burden of cigarette taxation in the United States, the paper briefly discusses and evaluates econometric research on the effect of ciga- rette excise taxes on cigarette consumption and reviews evidence on the incidence of cigarette taxes. The final sections of the paper review tax adnlinistration and enforcement concerns and issues in determining the best level of government at which to impose the tax. Recent Trends in Cigarette Taxation Cigarette excise taxes have been a declining share of both Federal and state excise tax revenues during the postwar period, despite numerous increases in state excise tax rates. Table I shows that Federal receipts from cigarette taxation increased in absolute terms from $1.2 billion in fiscal year 1950 to S2.5 billion in 1982, but declined as a share of total revenue from 3.2 percent to 0.4 percent and as a share of GNP from 0.5 percent to less than l).1 percent. As a result of the tax increase in TEFRA, Federal receipts from cigarette excise taxes nearly doubled to $4.7 billion in fiscal year 1984, .)bout 0.7 percent of Federal revenues and slightly over 0.1 percent of GNP. i Federal cigarette excise tax receipts as a share of total revenue and GNP remain below levels in every year between 1950 and 1975. Table 2 shows that state cigarette excise tax receipts have grown at a much faster rate than Federal receipts over the same period, but have also ~ Table 1. Federal Tax Collections on Cigarettes, Fiscal Years M 1950-1984 r..I ear igarette Tax Revenues t$ millions) Total Federal Revenue (S billions) NP (S billions). Cigarette Tax Revenues as Percent of Total Revenue Cigarette 7ax Revenues as Percrut ot GNl' 1950 1,242.8 39.4 265.1 3.15 0:n 1951 1,294.0 51.6 312.8 2.51 0.•11 1952 1,474.1 66.2 339.3 2.23 0.43 1953 1,586.8 69.6 361.3 2.28 0:14 1954 7 513 1 69.7 364.2 2.17 0.42 1955 . , 1,504.2 65.5 380.6 2.30 0.40 1956 1,549.0 74.6 411.8 2.08 0.38 1957 1,610.9 80.0 433.9 2.01 0.37 1958 1,668.2 79.6 443.1 2.11) 0.38 1959 1,771.1 79.2 474.4 2.24 0.37 1761) 1,863.6 92.5 497.9 2.01 0.37 1961 1,923.5 94.4 509.3 2.04 0.38 1962 1,956.5 99.7 548.2 1.96 0.36 1963 2,010.5 106.6 578.0 1.89 0.35 1964 1,976.7 112.6 618.2 1.76 0.32 1965 2,069.7 116.8 659.5 1.77 u.31 1966 2,006.5 130.8 724.1 1.53 0?8 1967 2,023.1 148.8 777.3 1.36 u'6 1968 2,066.2 153.0 831.3 1.35 u.25 196') 2,082.1 186.9 910.6 1.11 0.23 1970 2,036.1 192.8 968.8 1.06 0.21 1971 2,149.5 187.1 1,031.5 1.15 0.21 1972 2,151.2 207.3 1,128.8 1.()4 0.1'1 1973 2,221.0 2311.8 1,252.0 0.96 0.18 1974 2,383.0 263.2 1,379.4 0.')1 0.17 1975 2,261.1 279.1 1,479!9 0.81 0.15 1976 2,434.8 379.3 2,072.3 0.64 0.12 1977 22,279.2 355.6 1,862.8 0.64 0.12 1978 2,374.1 399.7 2,091.3 0.59 0.11 1979 2,356.1 463.3 2,357.7 0.51 0.10 1')80 2,604.4 517.1 2,575.8 0.50 0. 10 1981 2,488.2 599.3 2,885!1 0.42 u.O) 1982 2,496.1 617.8 3,046.0 0.41) 0.08 1983 3,424:1 6(x).6 3,2_'1:1 0.57 0.11 1984 . 4,749.2 666.5 3,581.1 (1.71 0.13 Source: The Tobacco Institute, Tlir 7itx 13urdrn urt 7iibato-Histariial Curu'+ilutiort, vol. 19, 1'18•1, p. 5; [iionotuii Rt'pnr! af tlle Presultvtt, Washington, I).C., 1985, F~p. 2a2 ant1318. N ,,t, 1 67

. declined (though only slightly) relative to GNP. State excise tax receipts increased from $0.4 billion in fiscal year 1950 to $4.2 billion in fiscal year 19ti-1. The last two columns of Table 2 show that state cigarette excise taxes increased faster than GNP throughout the late 1950s and 1960s, but Table 2. State Tax Collections on Cigarettes, Fiscal Years 1950-1984 '.•ar Total Cigarette State Tax Revenues Revenue (S millions) (S billions) NP (Z billions) Cigarette Tax Revenues , as Percent of Total Revenue Cigarette Tax Revenues as Percent of GNP I ',50 413.7 • 7.9 265.1 5.22 0.16 1'61 444.4 8.9 312.8 4.97 0.14 1';52 460.3 9.9 339.3 4.67 0.14 1U53 477.2 10.6 361.3 4.52 0.13 1954 469.7 11.1 364.2 4.24 0.13 1955 470.2 11.6 380.6 4.05 0.12 1 956 532.3 13.4 411.8 3.98 0 13 1957 581.1 14.5 433.9 4.00 , . 0.13 1958 626.8 14.9 443.1 4.20 0.14 1959 706.6 15.8 474.4 4.46 0.15 1960 929.9 18.0 497.9 5.16 0.19 1961 995.1 19.1 509.3 5.22 0.20 1962 1,085.6 20.6 548.2 5.28 0.20 I')63 1,132.8 22.1 578.0 5.12 0.20 1964 1,212.3 24.2 618.2 5.00 0.20 1965 1,327.1 26.1 659.5 5.08 0.20 1't66 1,566.6 29.4 724.1 5.33 0.22 1'+67 1,643.0 31.9 777.3 5.14 0.21 Or*i 1,915.8 36.4 831.3 5.26 0.23 1469 2,101.8 41.9 910.6 5.01 0.23 1970 2,368.1 48.0 968.8 4.94 0.24 1'+71 2,594.6 51.5 1,031.5 5.03 0.25 1't72 2,904.4 59.9 1,128.8 4.85 0.26 1,03 3,092.8 68.1 1,252.0 4.54 0.25 1'+74 3,225.2 74.2 1,379.4 4.35 0.23 1975 3,284.7 80.2 1,479.9 4.10 0.22 1'176 3,428.9 89.3 2,072.3 3.84 0.17 1977 3,483.3 101.1 1,862.8 3.45 0.19 1978 3,632.7 113.3 2,091.3 3.21 0.17 1'»'1 3,621.6 124.9 2,357.7 2.90 0.15 1't;stl 3,714.4 137.1 2,575.8 2.71 0.14 1981 3,850.6 149.7 2,885.9 2.57 0.13 1982 3,922.2 162.7 3,046.0 2.41 0.13 I'rg3 4,117.8 171.0 3,221.4 2.41 0.13 I't4a 4,233.0 N.A. 3,581.1 N.A. 0.12 i„urce: The Tobacco Institute, The TLv Dardert ia,t Ti,bacco-Histvrical Contpilatintt, vol. 19, 1984, i•. ;s; Ecauo,+tic Riport,J the 1'resulrnt, Washington, D.C., 1985, p. 242; Advisory Commission on Iua•rgovernmental Rclations, SigniftatnrFeateresoJFiscal Fcderalisrn, 1982-83Edition, Washington, I t.C., January 1984, p. 32. declined as a share of GNP after 1972 and as a share of total state revenues after 1966. Tables 3-5 show how state and'Federal cigarette excise taxes have changed in nominal terms, in real terms, and as a percentage of cigarette prices over the same period. Because the Federal government and almost all states impose specific rather than ad naloretn taxes, inflation reduces cigarette taxes in real terms and as a percentage of the price of cigarettes in years when the excise tax rate is unchanged. Table 3 shows that the Federal cigarette excise tax rate was increased only twice during the postwar period-from 7 cents per pack in 1950 to 8 cents in 1951, and from 8 cents to 16 cents in TEFRA in 1982. This translates into a decline in the tax rate measured in 1984 dollars4 from 37.5 cents per pack in 1946 to 8.6 cents in 1982, followed by an increase to 16 cents in 1984. The Federal tax as a percentage of the cigarette price (including taxes) declined from 42.2 percent to 10.7 percent between 1947 and 1982 and now stands at 16.6 percent, about the same rate as in 1975. In contrast, Table 4 shows that real state tax rates and state taxes as a percentage of the price of cigarettes are higher today than they were in the 1950s. On average, state tax rates increased from 12.2 percent of cigarette prices in 1954 to 26.9 percent in 1975, but have subsequently declined to 15.1 percent in 1984. Table 5 shows that combined state and Federal cigarette excise taxes declined from 49.9 percent of cigarette prices in 1954 to 27.8 percent in 1982, and have since (as of 1984) risen to 31.8 percent largely because of the doubling of the Federal excise tax rate. Tlle combined excise tax rate as a percentage of price, however, remains lower than the combined rate in effect as recently as 1980. These figures understate the tax burden on cigarettes to some degree because they do not include excise taxes imposed by some localities. In addition, most states with general sales taxes include cigarettes in the tax base. General sales taxes represent an additional portion of the sale price of cigarettes claimed by the tax collector, but do not raise the price of cigarettes relative to most other goods: Because, as discussed below, cigarette taxes in high tax states can be avoided to some degree by buying from other states, the dispersion of tax rates among states can be important for enforcement and other Federal policy issues. Table 6 shows that the number of states taxing cigarettes (including the District of Columbia) has risen from 42 in 1954 to all 51 in 1984, but the tax rates are widely divergent, ranging from 2 cents per pack in North Carolina to 26 cents in Connecticut and Massachusetts. The difference in price per pack between the highest and lowest tax states, however, has declined in 1984 dollars from t 19 1 0

Table 3. Burden of Federal Cigarette Excise Taxes: 1951-1984 rar Tax Ratc Tax Rate (:urrrnt S 1')84 E ~ccnts• pack) (centsrpack) Average Cigarette Price (cents/pack) Tax Rate (percentage of price) 1910 7.0 37.5 I't•17 7.0 32.6 16.6 42.2 1'+-Itt 7.0 30.2 174. 40.2 I',•Pt 7.0 30.5 17.9 39.1 1',in 7A 30.2 18.2 38.5 1't51 8.0 32.0 18.9 42.3 1't52 8.0 31.3 19.9 40.2 I'63 8.0 31.1 20.9 38.3 I45•1 8.0 30.9 21.2 37.7 I't;5 8.0 31.0 21.3 37.6 O56 8.0 30.6 21.8 36.7 1957 8.0 29.5 22.4 35.7 193H 8.1/ 28.7 23.2 34.5 1'+i't 8.0 28.5 24.2 33.1 I'tau 8.0 28.1 24.9 32.1 1'/61 8.0 27.8 25.1 31.9 1902 8.0 27.5 25.4 31.5 1903 8.0 27.1 25.9 30.9 1'16-1 8.0 26.8 26.4 30.3 1963 8.0 26.3 27.7 28.9 1966 8.0 • 25.6 29.1 27.5 1967 8.0 24.9 30.2 26.5 1'/68 8.0 23.9 32.1 24.9 1969 8.0 22.7 33.9 23.6 1970 8.0 21.4 37.0 21.6 1971 88.0 20.5 38.7 20.7 1972 8.0 19.9 40.9 19.6 I,'173 8.0 18.7 42.0 19.0 1`t7•1 8.0 16.9 44.1 18.1 1'173 8.0 15.4 47.3 16.9 1't76 8.0 14.6 49.3 16.2 1977 8.0 13.7 51.6 15.5 1978 8.0 12.7 54.3 14.7 1'I'9 8.0 11.4 57.3 14.0 1980 8.0 10.1 62.0 12.9 1981 8.0 9.1 66.9 12.0 Iva2 8.0 8.6 74.7 10.7 1943 16.0 16.7 90.1 17.8 1'184 16.0 16.0 96.3 16.6 Sources: The Tobacco Institute, 77te 7iir BurJevt an 7iiGuccu-Historical ComPilarion, vol. 19,1984, p. 6; U.S. Department of Agriculture (cigarette price data supplied by Robert Miller); Ecunnnaic /67+orr of the Presitkwn, Washington, D.C., 1985, p. 291. -n Table 4. Burden of State Cigarette Excise Taxes: 1950-1984 Average Tax Rate Tax Rate Cigarette Tax Rate Current t 198•1 S Price (percentage Year (ccnts/pack)a (cents/pack) (cents/pack) of pncc) I 1950 2.4 10.5 18.2 ~ 1951 2.5 10.2 18.9 I t 1')52 2.5 9.8 19.9 ` 1953 2.5 9.8 20.9 1954 2.6 10.1) 21.2 12' j 1955 2.6 10.2 21.3 12•a 1956 2.9 11.1 21.8 13:1 1957 3.0 11.2 22.4 13.5 1958 3.1 11.2 23.2 13.4 1959 3.3 11.9 24.2 13.8 1960 4.1 14.4 2d!) 16.5 1961 4.3 14.9 25.1 17.1 1962 4.7 16.0 ' 25:1 18.3 1963 4.7 16.0 25.9 18.2 1964 5.1 17.0 26.4 19.2 1965 5.3 17.6 27.7 19.3 1966 6.2 19.9 29.1 21.4 1967 6.5 20.1 30.2 21.4 1968 7.6 22.6 -AN • 32.1 23.6 1969 8.3 23.4 ` 33.9 2•1.3 1970 9.5 25.5 37.0 25.7 1971 9.9 25.3 • 38.7 25.5 1972 11.0 27.3 40.9 26.9 1973 11.3 26.4 42.0 26.9 1974 11.1 23.4 44.1 25.1 1975 11.8 22.8 47.3 25.0 1976 11.4 20.7 49.3 23.1 1977 12.0 20.6 51.6 23.3 1978 12.3 19.5 54.3 22.6 1979 12.1 17.3 57.3 21.1 1980 13.6 15.9 62.0 . 20:1 1981 12.5 14.3 66.9 18.7 1982 12.8 13.8 74.7 17.1 1983 14.3 14.9 90.1 . 15.9 1984 14.6 14.6 96.3 15.1 Sources: The Tobacco Institute, T6e'Gtx Bwdrn nn Tilbacao -Histnriatl GunpiLuiun, vol. 19, 1984. Ppp. 6 and 196; U.S. Department of Agriculture; Eu,notnic Rerort aJ thr Prrsidcnt. a. Computed by multiplying average tax rate in the taxing states by the ratio of cigarette sales in the taxing states to total cigarette sales in the United States. 71

48.7 cents in 1971 to 24 cents in 1984. As a result, the economic gain from buying cigarettes in low tax states has declined since the early 1970s, though it remains significant. Effects of Taxes on Cigarette Consumption The price of a pack of cigarettes has increased in 1984 dollars from about 78 cents per pack in 1950 to about 96 cents per pack in 1984-an increase of Table 5. Burden of State and Federal Cigarette Excise Taxes: 1950-1984 ear Tax Rate Current S (cents/pack) Tax Rate 1984 S (cents/pack) Average Cigarette Price (cents/pack) Tax Rate (pereenta e of price) 1950 9.4 403 18.2 1951 10.5 42.2 18.9 1952 10.5 41.1 19.9 1953 10.5 40.9 20.9 1954 10.6 40.9 21.2 49.9 1955 10.6 41.3 21.3 49.9 1956 10.9 41.7 21.8 50.1 1957 11.0 40.7 22.4 49.3 1958 11.1 39.9 23.2 47.9 1959 11.3 40.4 24.2 46.8 1960 12.1 42.5 24.9 48.6 1961 12.3 42.6 25.1 48.9 1962 12.7 43.4 25.4 49.8 1963 12.7 43.1 25.9 49.0 1964 13.1 43.7 26.4 49.5 1965 13.3 43.9 27.7 48.2 1966 14.2 45.5 29.1 48.9 1967 14.5 45.0 30.2 47.9 1968 15.6 46.5 32.1 48.5 1969 16.3 46.0 33.9 47.9 1970 17.5 46.9 37.0 47.4 1971 17.9 45.8 38.7 46.1 1972 19.0 47.2 40.9 46.4 1973 19.3 45.1 42.0 46.0 1974 19.1 40.2 44.1 43.3 1975 19.8 38.3 47.3 41.9 1976 19.4 35.3 49.3 39.3 1977 20.0 34.3 51.6 38.8 1978 20.3 32.2 54.3 37.3 1979 20.1 28.7 57.3 35.0 1980 20.6 26.0 62.0 33.3 1981 20.5 23.4 56.9 30.6 1')82 20.8 22.4 74.7 27.8 1983 30.3 31.6 90.1 33.7 1')84 30.6 30.6 96.3 31.8 72 1 about 25 percent in 34 years-while over the same period the sum of average state and Federal excise taxes in 1984 dollars has declined about 40 cents per pack to about 30 cents•per pack. If real excise tax rates had been raised rather than lowered over this period, prices would have risen more and tax policy would have reinforced instead of offsetting other efforts to control smoking. The question is how much difference might this have made. This section provides a brief overview of recent studies of the effects of cigarette prices on smoking. The studies on cigarette demand reviewed in this section all assume that the supply of cigarettes is perfectly elastic, and that therefore all differences Table 6. Dispersion in Cigarette Excise Taxes Among States Minimum Maximum Minimum Maxinunn Number Rate Rate Rate Ratc Range of Current S Current S 1984 S 1984 198•1 S Taxing (eents/ (eents/ (cents! (cents/ (cents/ Year States pack) pack) pack) pack) pack) 1954 42 0.0 8.0 0.0 30.9 30.9 1955 42 0.0 8.0 0.0 31.0 31.0 1956 43 0.0 8.0 0.0 30.6 30.6 1957 43 0.0 8.0 0.0 29.5 29.5 1958 44 0.0 8.0 0.028.7 28.7 1959 47 0.0 8.0 0.0 28.5 28.5 1960 48 0.0 8.0 0.0 28.1 28.1 1961 48 0.0 8.0 0.0 27.8 27.8 1962 48 0.0 8.0 0.0 27.5 27.5 1963 48 0.0 8.0 0.0 27.1 27.1 1964 49 0.0 8.0 0.0 26.8 26.8 1965 49 0.0 11.0 0.0 36.2 36.2 1966 50 0.0 11.0 0.0 35.2 35.2 1967 50 0.0 13.0 0.0 40.4 d0.4 1968 50 0.0 15.0 0.0 44.8 44.8 1969 51 2.0 16.0 5.7 •15.3 39.7 1970 51 2.0 18.0 5.3 •18.1 42.8 1971 51 2.0 21.0 5.1 53.9 48.7 1972 51 2.0 21.0 5.0 52.1 47.2 1973 51 2.0 21.0 4.7 49.1 44.4 1974 51 2.0 21.0 4.2 44.2 40.0 1975 51 2.0 21.0 3.9 40.5 36.7 1976 51 2.0 21.0 3.6 38.3 34.7 1977 51 2.0 21.0 3.4 36.0 32.6 1978 51 2.0 21.0 3.2 33.•1 30.3 1979 51 2.0 21.0 2.9 10 1 27.2 1980 51 2.0 21.0 2.5 26.5 24.0 1981 51 2.0 21.0 2.3 2•1.0 21.7 1982 51 2.0 25.0 2.2 26.9 2-1.8 1983 51 2.0 26.0 2.1 27.1 25.0 1984 51 2.0 . 26.0 2.0 26.0 2•1.1) Source: The Tobacco Institute, 77te Tax Burden un Ti~bacal, pp. 145-195. 73

in quantities consumed either over time, among states, or among individu- als represent movements along a demand curve. This is equivalent to assuming that prices per pack increase dollar for dollar with increases in excise tax rates per pack. Barzel (1976) examined differences in cigarette prices over time and among states and estimated that prices rise $1.065 for cvery dollar increase in excise taxes, with the estimated coefficient %ignificantly different from unity. Barzel attributes this result, however, to .1 feature of unit rather than ad valvreon excise taxes; the impact of a unit tax can be avoided in part by changing the nature of the product. By this interpretation, one reaction to an increase in the tax rate per cigarette is to improve the quality of the cigarette by, for example, increasing its tobacco content because additional "quality" per unit is not taxed. Barzel's expla- nation is consistent with an assumption that the price per quality-adjusted cigarette received by producers is unaffected by the level of tax in the long run and that differences in "quality-adjusted" prices among states are fully explained by differences in taxes and, to a lesser degree, shipping costs from major producing states (Kentucky, North Carolina, and Virginia.) In the two years following the enactment of TEFRA, retail cigarette prices increased by about 22 cents per pack, while stue and Federal excise taxes rose by only about 10 cents per pack. In the 2 years preceding TEFRA, prices rose by about 12 cents per pack, while excise taxes were roughly constant. While it is impossible to draw any conclusions about the effects of the recent Federal tax increase on price without a fuller model explaining the determinants of the recent growth in cigarette prices, the experience since 1980 does not contradict the assumption that all of the tax is paid by consumers. The demand for cigarettes has generally been analyzed by econometric equations that estimate differences in cigarette consumption over time, among states, or among individuals as a function of the price of cigarettes, income, and demographic variables. There are problems of interpretation in both time series and cross-section studies. In time series studies, the estimates of both price and income elasticities are sensitive to the specifica- tion of dummy variables representing publicity on the harmful effects of smoking (such as the 1964 Surgeon General's Report) and variables representing Federal policies to limit cigarette advertising. In addition, time series estimates are not stable because of multicollinearity among price, income, and other time trended variables and may represent short run instead of long run responses to price changes. Cross-section estimates of price elasticity may be biased upwards to the extent that they are based kin data ou tax-paid sales, since some tax-paid sales in low tax (and low price states) result in use by consumers in high-tax states. Lewit and Coate (1982) report that studies completed since 1970 usin}; U.S. data have yielded price elasticity estimates ranging from -0.4 to -1.3.5 Miller (1982) reviews a recent survey that reports a price elasticity of -0.7 as the midpoint of recent studiesb and notes that the Tobacco Institute uses the same figure for its analyses of cigarette tax impacts. More recently, Blaine (1983) has reported price and income elasticities of -0.68 and 0.39, respec- tively, based on a regression analysis of cigarette consumption in the United States between 1945 and 1982. In a more detailed analysis of differences in cigarette consumption among a sample of individuals from a 1976 Health Interview Survey, Lewit and Coate (1982) estunate a slightly lower overall price elasticity of -0.42. The study gives a much more detailed breakdown of the smoking 1 I~i response than in other studies. It corrects for the bias in cross-section studies resulting from bootlegging in two ways: first, because the unit of observation is consumption reported by individuals rather than tax-paid sales, and second, by removing from the sample those households within 20 miles of states with lower prices. This eliminates error in the measure of consumption and partially corrects for errors in the price measure that result when households purchase cigarettes outside of their own localities. Lewit and Coate find that cigarette prices affect smoking primarily by reducing the participation rate; the estimated effect on the number of cigarettes per smoker is statistically insignificant. There are also significant estimated differences in price elasticities among groups; reported price elasticities are much higher for adult males than for adult females and much higher for people aged 20-25 (-0.89) than for other age groups. This finding that price elasticities are higher among younger groups is consistent with results from a separate study of teenage youths by Lewit, Coate, and Grossman (1981) that reports a price elasticity of -1.2 for the propensity to smoke among teenagers. Lewit, Coate, and Grossman find that teenagers are also sensitive to "fairness doctrine" variables, in particular to anti- smoking messages that stress the health hazards due to smoking, but smoking increased between 1970 and 1974 because effects of the increased anti-smoking publicity associated with the fairness doctrine were offset by the effects of a decline in real cigarette prices. Although there are a range of price elasticities, the evidence on balance suggests overall price elasticities between -0.4 and -0.7, a fairly narrow range. The price elasticity with respect to a tax increase in any one state is greater than the price elasticity with respect to a uniform increase in all state taxes or an increase in the Federal tax because of opportunities for residents of high tax states to purchase cigarettes taxed in low tax states. Based on the research by Lewis and Coate and by Lewis, Coate, and Grossman, it would '•I 1 75

.lppear that price changes operate mainly by deterring teenagers and ti'oung adults from beginning to smoke. Because smoking tends to be habit forming, this finding of a high price response among teenagers and young adults suggests that increasing cigarette excise taxes and then either indexing them to price changes or periodically raising them to maintain the higher real tax burdens over time could eventually result in much lower smoking rates among the entire public. It also suggests, however, that the full drcline in overall smoking rates among the entire population will take m.my years to occur and that beneficial effects on the health of the population will also be delayed to the extent the damage to any individual c aused by cigarette smoking is cumulative over many years. In the short run, the major effect of higher cigarette taxes may be to increase taxes paid by current smokers, without significantly reducing either total smoking or its associated health costs. Incidence of Cigarette Excise Taxes As noted above, most available econometric analyses of the market for cigarettes have assumed that the supply of cigarettes is elastic and that any changes in tax rates are fully passed through to consumers of cigarettes. In the short run, some of the incidence of the tax may be borne by owners of tobacco-producing land and by investors in cigarette manufacturing. Because short-run demand is quite inelastic, however, it is likely that even in the short run most of the impact of changes in the tax is borne by consumers. Assuming the tax is paid by consumers, this section reviews recent evidence on the distribution of consumption of tobacco products by income class.7The evidence confirms the belief that taxes on tobacco consumption are highly regressive; that is, expenditures on tobacco products :)s a share of income are much greater in lower income classes than in higher income classes 8 Tables 7 and 8 show data on the distribution of tobacco consumption by income class based on the most recently available information from the Survey of Consumer Expenditures (CES) conducted by the U.S. Depart- ment of Labor, Bureau of Labor Statistics (BLS). Table 7 presents data on the distribution of consumption of various commodities by income class based on the BLS diary survey, a survey of frequently purchased goods by ronsttluers over a 2-week period. Estimated annual expenditures on tobacco products as a percentage of income on tobacco products declines from 2.3 percent for the lowest quintile of the income distribution to 1.4 percent, I.1 percent, 0.7 percent, and 0.4 percent for successively higher income quintiles. As a percentage of expenditures on frequently purchased goods, expenditures on tobacco products increase from 2.9 percent of expendi- tures for the lowest quintile to 3.2 percent for the middle quintile, but then decline to 2.5 percent for the second highest quintile and 2.1 percent for the highest quintile. Table 8 presents data from the BLS Overview Survey for 1980-81, a survey that asks households to estimate their total expenditures on differ- ent products for an entire year. The results from this data source are even more striking. Tobacco consumption declines from 2.9 percent of income for the lowest fifth of the income distribution to 1.7 percent, 1.1 percent, Table 7. Expenditures on Selected Goods and Services by Income Quintile: 1981 Diary Survey Item Lowest Second Third Fourth Iiighest Income Quintile Income Before Taxes (annual) 3,688 9,837 16,724 25,360 4d,7fi0 Expenditures ( week(y ) Food, total S 28.15 f 39.96 S 54.26 S 66.62 S 85.30 Alcoholic beverages 2.02 3.65 5.61 5.97 10.10 Tobacco products 1.63 2.60 3.56 3.27 3.58 Personal care products 2.29 2.45 4.22 4.63 7.10 Nonprescription drugs 1.17 1.33 2.09 1.96 2.33 Housekeeping supplies 2.13 3.17 •t.18 5.86 52.94 Enrrg); total 18.31 28.48 36.46 43.29 52.94 Tota' 55.70 81.64 110.38 132.0) 168.8•1 Expenditures ('%) of income) Food, total 36.69% 21.12% 16.8T'6 1.1.6(i'u 9!)1% Alcoholic beverages 2.85 1.93 1.74 1.43 1.17 Tobacco products 2.30 1.37 1.11 0.67 0.42 Personal care products 3.23 1.30 1.31 0.95 u.82 Nonprescription drugs 1.65 0.70 0.65 0.40 0.27 Housekeeping supplies 3.00 1.68 1.30 1.20 n.87 Energ~;total 25.82 15.05 11.34 8.88 6.15 Total " 78.54 43.16 34.32 27.19 19.62 Expenditures M of total for selccted categories) Food, total 50.54'x) 48.95% 49.16% 50 2•PX, ' 50.52% Alcoholic beverages 3.63 4.47 5.08 5.26 5.98 9obacco products 2.93 3.18 3.23 2.47 2.12 Personal care products 4.11 3.00 3.82 3.49 4.21 Nonprescription drugs 2.10 1.63 1.89 1:18 1.38 I lousekeeping supplies 3.82 3.88 3.79 4.42 a.•1•1 y, total Enrr~~ 32.87 34.88 33.03 32.65 31.36 , Tutal 100.00 100.u11 1000) 1000) 101t.(N) Source: U.S. 17epartment of Labor, Bureau of Labor Statistics, l:anrotncr li.eycuJinur Sunt7^. Diary Sturi7; 198U•81, September 1983, p. S•1. 'r, 1 77

Q.9 percent, and 0.5 percent for successively higher quintiles. As a percent- age of total expenditures, including both frequently purchased and intermittently purchased items, spending on tobacco products increases from 1.3 percent of income in the lowest income quintile to 1.4 percent in the second lowest quintile, but then declines to 1.2 percent, 1.0 percent, and 0.7 percent in successively higher quintiles. Spending on tobacco products as a percentage of both total income and total expenditures declines more as income increases than does any other consumption category Table 8. Expenditures on Selected Goods and Services by Income Quintile: 1980-1981 Interview Survey Itrm Lowest Sccond Third Fourth Highest Income Quintile Income Before Taxes 3.473 9,791 16,80') 25,128 44,616 Total Expenditures 7,852 .11,570 15,736 20,714 30,563 Food 1,820 2,452 3,028 3,737 4,949 Alcoholic beverages 129 221 281 329 460 11111 5ing 2,682 3,605 4,448 5.810 8,516 Apparel and services 396 569 810 1,075 1,851 Transportation 1,251 2,278 3,377 4,461 6,050 I Iealtlt care 476 595 700 807 1,066 Entrrtainment 263 440 679 916 1,535 I'ersottal care 77 111 139 178 272 R,.1ding 55 77 110 139 206 Education 170 84 116 200 498 Tobacco 101 164 191 215 217 Miscellaneous 1110 141 232 293 482 t:.tsh contributions 161 273 429 562 1,209 Personal insurance and pe nsions 170 559 1,195 1,993 3,241 L'xpenditures less itenu (if saving' 7,682 11,011 14,541 18,721 27,322 Expenditures 1`:L of inconte) Total Expenditures 226.09% 118.17'b 93.62% 82.4316 68.50"o Fand 52.40 25.04 18.01 14.87 11.11 Alcoholic beverages 3.71 2.26 1.67 1.31 1A3 I luusing 77.22 36.82 26.46 23.12 19.09 Apparel and services 11.40 5.81 4.82 4.28 4.15 Trans >ortation 36.02 23.27 20.09 17.75 13.56 I (~•altlt care 13.71 018 4.16 3.21 2.39 Entcrtainment 7.57 4.49 4.04 3.65 3.44 1'rrsonal care 2.22 1.13 0.83 0.71 0.61 Iteading 1.58 0.79 0.65 0.55 0.46 li'lucation 4.89 0.86 0.69 0.80 1.12 1 itbacco 2.91 1.68 . 1.14 0.86 0.49 \1i+crllancotts 2.88 1.44 1.38 1.17 1.t1Y t:.tsh contributions 4.64 2.79 2.55 2.24 2.71 listed in Table 8, including expenditures on food, alcaholic beverages, housing, apparel and services, transportation, health care, entertainment, personal care, reading, education, and the categories labelled "miscellane- ous" and "all other categories:' Administration and Enforcement Issues This section briefly examines administration and enforcement issues related to cigarette taxation. Three topics are covered: the best level for collection of the tax, the issue of specific vs. ad nlnrrnl taxation, and the problem of "bootlegging" from low tax to high tax states. Lel,cl of Collecrion Most federal excise taxes are collected from manufacturers and importers. These include excise taxes on tobacco, alcohol, and gasoline, but not taxes Table 8 (Continued) Item Lowest Second Third Fourth Ilighest Expenditures (% of total) Income Quintile Personal insurance and pensions 4.89% 5.71% 7.11% 7.93% 7.20%, Expcnditures less itcros of saving' 221.19 112.46 86.51 74.50 61.24 Food 23.18 21.19 19.2•1 18.U•1 16.23 Alcoholic beverages 1.64 1.91 1.7') 1.5') 1.51 1lousin~~, 34.16 31.16 28.27 28.05 27.86 Apparcl and services 5.04 • 4.92 5.15 5.19 6.06 80 19 Transp ortation 15.93 19.69 21.46 21.54 . I lealth care 6.06 5.14 •1.•15 3!)0 3:/9 Entertainment 3.35 3.80 •1.31 4:12 5.02 Personal care 0.98 0.96 0.88 0.86 0.89 Reading 0.70 0.67 0.70 11.67 0.67 Education 2.17 0.73 0.7a 0.97 1.63 Tobacco 1.29 1.42 1.21 1.04 0.71 Miscellaneous 1.27 1.22 1.47 1.41 - 1.58 Cash contributions 2.05 2.36 2.73 2.71 3.96 Prrsonalinsuranee and pensions 2.17 4.83 7.5') 9.62 10.60 Expenditures less items of saving' 97.83 95.17 92.-11 'N/.38 89.10 Source: U.S. Department of Labor, Bureau of Labor Statistics, "Consunucr Exprnditure Survey: Results from the 1980-81 Interview." News Iteleaur, Decwmber 19, 19H-1,'lable 1. a. Refers to expenditures other than for personal insurance and pensions. -8 4 79

i un diesel fuel, which is collected from retailers and users. (For the gasoline tax, a wholesale distributor may elect to be treated as a producer for tax purposes.) The advantage of collecting taxes at the manufacturer level is that it reduces administrative and compliance costs by minimizing the number of points of collection. Moreover, retail excise taxes on specific goods, as opposed to general sales taxes, create significant paperwork per dollar of revenue raised because small retailers who sell, but do not specialize in the taxed good, must keep separate records to identify the portion of their sales that is taxable at each separate excise tax rate. Most Federal retail excise taxes were eliminated in 1965. Those that remain are justified for particular reasons.9 State cigarette excise taxes are collected from wholesale distributors. (State taxes on manufacturers would, of course, be taxes on cigarettes produced within a state, not taxes on cigarette consumption by state residents.) Taxation at the retail level would vastly multiply the number of points of collection. .Siecif ic vs. Ad Valorem Tnxatinn The current law Federal tobacco taxes are mostly specific excises on units of well defined products: $8.00 per thousand (16 cents per pack) for small cigarettes; $16.40 per thousand for large cigarettes; yz cent for each 50 papers for cigarette papers and 1 cent for each 50 tubes for cigarette tubes (except if they measure more than 6~i inches in length, in which case every 2Y4 inches is regarded as one paper or one tube); and 75 cents per thousand for small cigars (those weighing not more than 3 pounds per thousand). The one ad ttiilorent tax is the tax on large cigars, which is 8h percent of the wholesale price, but not more than $20 per thousand. State cigarette excise taxes are also specific taxes. Before 1976, New Hampshire's excise tax was a fixed percentage of the retail price, but it has since been converted to a per-unit tax. The advantage of specific taxes is that they do not give rise to valuation problems in situations where manufacturers and/or wholesalers operate their own retail distribution outlets. Specific taxes can be less appropriate when there are large quality differences that cannot be readily measured by a small number of well-defined physical characteristics; in that case, specific taxes can distort choices among product characteristics and lead to lower effective rates for variants of the product with less quality per measured unit. Quality differentials do not appear to be a serious enough problem in the taxation of cigarettes to warrant sacrificing the administrative advan- tage of per-unit taxes. One advantage of changing to an ad mluretrt tax is that revenues would rise to keep pace with inflation-induced increases in cigarette prices while, in contrast, the real revenue yield from the current specific excise taxes declines as the price level rises. even if cigarette prices keep pace with' inflation. Thus, over time, the excise tax rates must be statutorily adjusted' to maintain real yields. One option to correct for the real erosion in' cigarette tax revenues, while maintaining the benefits of per-unit taxes, is; to index the tax rate to increases in either the general price level or a price index for all cigarettes. BootkqghtY front Low `Tii.m to Ni~h T» States One major concern to state cigarette tax administrators is "bootlegging" of ; cigarettes across state lines to avoid taxation in high tax states. The large differences in tax rates among states shown earlier in this paper-ranging from 2 cents per pack in North Carolina to 26 cents per pack in Massachu- setts in 1984-provide a large economic incentive for bootlegging. The Advisory Commission on Intergovernmental Relations (ACIR) has conducted two major studies of cigarette bootlegging in recent years. (See Advisory Commission on Intergovernmental Relations (1977; 1985).) ACIR reported an "epidemic" of organized cigarette smuggling and illegal diver- sion of cigarettes from the legal distribution system in the mid-1970s. In response to this problem, the Congress in 1978 enacted the Federal Ciga- rette Contraband Act, prohibiting the transportation, receipt, shipment, possession, distribution, or purchase of more than 60,000 cigarettes not bearing the tax indicia of the state in which the cigarettes are found. ACIR's most recent study finds that this law has been very effective in reducing large-scale cigarette smuggling, although there is some evidence smuggling began to increase again in 1983. The ACIR Report lists 5 varieties of tax evasion activities: casual ciga- rette smuggling (individuals buying cigarettes in neighboring, lower-taxed states); organized or commercial cigarette smuggling (transportation of cigarettes between states for profit); mail-order purchase of cigarettes (made illegal in 1949, but Post Office rethinking its enforcement role); purchase of cigarettes through tax-free outlets (international ports of entry, military PXs, and Indian reservations); and diversion within the legal distribution system. They 6nd that large-scale organized smuggling has declined dramatically since the mid-1970s due largely to the Federal Cigarette Contraband Act and that the largest remaining source of revenue loss to state and local governments is due to the tax exemption of sales on Indian reservations and. military bases. Increases in state taxes and the widening of real differentials between low tax and high tax states that occurred between 1981 and 1983, however, have increased incentives for 40 , 1 81

smuggling and many states have reduced resources devoted to enforcing state cigarette tax laws. The enforcement and revenue loss problems resulting from interstate shipment of cigarettes are one consideration in determining the appropri- ate Federal role in cigarette taxation. The next section examines alternative Federal policy options in light of state and local concerns. Issues in State/Federal Relations One advantage of collecting cigarette taxes at the Federal level rather than at the state and local level is that Federal taxes involve lower collection and evasion problems. A Federal tax can be collected from a small number of producers and importers and, because the tax rate is uniform throughout the country, there is no incentive for cigarette bootlegging. On the other hand, separate state and local cigarette taxes allow citizens more freedom of choice by allowing a variation in the tax structure among communities and, currently, cigarette taxes are an important revenue source in sonic states and localities. Concerns about the potential effect on state and local cigarette tax revenues were raised when the Federal governnient increased the cigarette excise tax in 1982. An increase in the Federal tax causes some reduction in state and local revenues to the extent consumption of cigarettes declines in response to the tax-induced price increase. Using a price elasticity of -0.4 tior total cigarette consumption and a higher relative price elasticity of-1.0 for a state's share of national tax-paid sales, Toder (1982) estimated that the 8 cent per pack excise tax increase in TEFRA would reduce state revenues by $176 million at 1984 levels if state tax rates remained unchanged-an amount equal to about 7 percent of the increase in Federal excise tax revenues ($2.45 billion) and about 12 percent of the increase in Federal revenues, net of income tax offsets ($1.48 billion).10 One effect of much larger increases in the Federal excise tax rate would be a significant erosion of state tax bases. As noted above, Federal taxes, compared to state and local taxes, are less expensive to collect and more difficult to evade, but give less opportunity for citizens of different states to decide separately how much they want to tax cigarette consumption. In addition, a larger Federal share of total national receipts from cigarette taxes can be justified on the argument that the' Federal government, through Medicare, tax preferences for health insurance, and other programs now bears an increasing share of the health care costs attributable to smoking. Nonetheless, state taxes as well as Federal taxes raise cigarette prices and ; discourage smoking. Alternative policy options that would maintain state involvement in taxation and/or maintain state shares in total cigarette tax receipts include: Assistance to State Enforcement. Additional aid to enforcement could allow states to capture more of the revenue base by further raising their excise taxes, while attempting to minimize the bootlegging problem. Turnback of Federal Revenues to States. Alternatively, the Federal government might raise the excise tax, but allocate additional revenues to states and localities according to sonic pre-determined formUla. This would improve the efficiency of tax collection without lowering state revenues, but would reduce the ability of citizens of different states to make inde- pendent decisions on how much they want cigarette consumption to be taxed. One version of this approach would be a proposal to reduce a state's share of Federal cigarette revenues to the extent it imposes its own state tax; this revenue distribution formula would be equivalent to mandating a uniform state excise tax to be collected at the Federal level and disbursed to the states. Turnback of the Tax Base to States. Alternatively, the Federal excise tax could be eliminated entirely, leaving the entire tax base to the states. The resulting reduction in Federal revenues could be financed by reducing Federal fiscal assistance to state and local governments. The Reagan Administration's 1982 Federalism proposal included a coni- bination of these last two alternatives. The Administration proposed to triple the Federal excise taxes on tobacco (among other excise taxes) for three years and to distribute the revenues from the excise-taxes to the states in fiscal years 1983-1985 in proportion to state population. After fiscal year 1985, the Federal excise taxes on tobacco (and other excises) would have been eliminated, allowing the states to recover the revenues if they chose to do so by increasing their own excise taxes. In effect, the Administration was proposing, after a transition period, to turn back the tax bases, not the revenues, to the states, while at the same time reducing Federal involvenunt in certain domestic programs.- The Administration's Federalism proposal was not acted on by the Congress. Conclusions This paper has reviewed some of the major issues.. in the taxation of cigarettes. A more thorough examination of these issues is particularly timely this year, given the scheduled expiration of the increase in Federal ti? 83

cigarette excise taxes enacted in TEFRA and the Congressional interest in extending the TEFRA tax rates or increasing rates further as one way of reducing the Federal deficit. Increased taxation of cigarettes has been advocated by some as a way of reducing health costs attributable to smoking. Proponents of increases in the excise tax also note that the real burden of cigarette taxes has declined over time. Opponents of higher cigarette taxation have pointed to the unfairness of taxing cigarette users more heavily than other taxpayers with the same income and to the regressivity of cigarette taxes. The evidence reviewed in this paper indicates that the real burden of cigarette taxation has declined significantly in the postwar period, even taking account of the doubling of the Federal excise tax in TEFRA. Many econometric studies show an effect of cigarette prices on consumption, but nwst studies show demand to be relatively inelastic. More recent research suggests that the effects of tax-induced price increases may be greatest on teenagers and young adults and may significantly affect the decision to begin smoking. These results imply that the long-run health benefits from higher cigarette taxation may be greater than previously believed, but also imply that there could be a significant delay before those benefits are fully achieved. On the other hand, recent data shows that cigarette consumption is a much higher share of the income of low income households than of high income households, thereby suggesting that cigarette taxes are much more regressive than almost all other revenue sources. In addition, higher Federal taxes would reduce state cigarette excise tax revenues. A greater reliance ou Federal instead of state taxation would, however, reduce enforcement and compliance costs per total dollar of revenue raised and reduce boot- legging from low tax to high tax states. i Notes 1. See, for example, Warner (1984). 2. See Congressional Budget Office (1985). 3. An extension of the 16 cent per pack cigarette excise tax through 1990 would increase federal revenues by about $1.7 billion per year for the rest of tlte decade, compared to a baseline that assumes the tax will expire as scheduled on October 1, 1985. This estimated revenue increase is net of reduced federal income taxes; the increase in excise tax receipts is about $2.3 billion. Under the assumption that GNP will be unaffected by changes in the excise tax rate, factor incomes must decline by the amount of the increased excise tax, thereby reducing income tax revenues. 4. as measured by changes in the CPI between 1950 and 1984. 5. See, for example, Schreider, Klein, and Murphy (1981), Ippolito, Murphy, and Sant (1979), Hamilton (1974), and Miller (1975). 6. This finding is based on a survey by Shuffett (1982). 7. Strictly speaking, the desired relationship to examine the incidence of 11 cigarette taxes among income classes is the relationship between cigarette consumption and income. Cigarette consumption, however, accounts for the bulk of spending on tobacco products, so that the relationship between tobacco products' consumption and income is quite close to the relation- ship between cigarette consumption and income. Moreover, other tobacco products (cigarette papers and tubes and cigars) also are subject to federal excise taxes. 8. The fact that tobacco taxes are regressive by themselves need not mean, however, that an increase in the federal cigarette tax rate must necessarily reduce the progressivity of the entire federal tax system. Since cigarette taxes are a relatively minor revenue source, overall progressivity could be roughly maintained if a cigarette tax increase were combined with provisions to increase the relative tax shares paid by middle- and upper-income people under the personal income tax. 9. For example, the diesel fuel tax cannot be collected at the manufac- turer level because taxed diesel fuel used by trucks and automobiles is physically indistinguishable from untaxed home heating oil. (It has been proposed, however, to collect the tax from refiners and to dye taxed fuel, so it can be identified downstream as tax-paid when used for taxable purposes.) 4 ' ~ K~

10. This calculation does not take account of reductions in state income taxes (which would further reduce state revenues) or of any tax-induced Shifts in income between tobacco-producing and other states. The -0.4 elasticity is at the low range of estimates (see Section 111), but may overstate the short-run effect, according to the findings of Lewit and Coate. References Advisory Commission on Intergovernmental Relations, Cigarette Bootleg- giny: A State and Federal Responsibility, Washington, D.C., 1977. Advisory Commission on Intergovernmental Relations, Cigarette Tax Evasion: A Second Look, Washington, D.C., 1985. Barzel, Y., "An Alternative Approach to the Analysis of Taxation,"Journal ,!jPolitical Ecortomy, December, 1976. l3laine, Thomas W., "Problems in the Estimation and Interpretation of Demand Equations: The Case of Cigarettes," unpublished paper, Univer- sity of Kentucky, 1983. Congressional Budget Office, Reducing the Deficit: Spending and Revernie Options, 1985 Annual Report, Part ll, February, 1985. 1Iamilton, James L., "The Demand for Cigarettes: Advertising, the Health Scare, and the Cigarette Advertising Ban," Reviewof Econontics and Statis- tics, 1974. Ippolito, R.A., R.D. Murphy, and D. Sant, Constaner Responses to Cigarette Health Infonnation, Staff Report, Bureau of Economics, Federal Trade Commission, 1979. Lewit, Eugene M., and Douglas Coate, "The Potential for Using Excise Taxes to Reduce Smoking," Journal of Health Econoniics, 1982. Lewit, Eugene, M., Douglas Coate, and Michael Grossman, "The Effects of Government Regulation on Teenage Smoking," Journal of Law and Econo+nics, December, 1981. Miller, Robert I-l., "Cigarettes: Consumption Situation and Outlook," National Tobacco Tax Association, Proceedings of the Fifty-Sixth Annual 1Neeting, 1982. Miller, Robert H., "Pricing Out Tobacco: Price as a Factor in Cigarette Consumption," talk presented at 3rd World Conference on Stnoking and Health, New York, N.Y., June 3,1975. Schneider, Lynne, Benjamin Klein, and Kevin M. Murphy, "Governmen- tal Regulation of Cigarette Health Information,"Journal of Laiv and Eco- nonlics, December, 1981. I Shuffett, D. Milton, "Estimated Impacts of Proposed Increase in Federal Taxeson Cigarettes,"I)epartnient of Agricultural Economics, University of Kentucky, July 12, 1982. Tobacco Institute, The TiiY Burden on Ti~bacco: Historical Contpilation, Volume 19, 1984. Toder, Eric J., "The Effect of the Federal Cigarette Tax Increase on State Tax Revenues,"NationalTobaccoTaxAssociation,Proceedin,gsoJihrFifty- Sitdi Annual n1eetin,y, 1982. U.S. Department of Labor, Bureau of Labor Statistics, Consunier Expeluli- ture Snrtey: Diary Survey, 1980-81, September, 1983. - U.S. Department of Labor, Bureau of Labor Statistics, "Consumer Expend- iture Survey: Results from the 1980-81 Interview," press release, December 19, 1984. Warner, Kenneth E., "Cigarette Taxation: Doing Good by Doing Well," Journal of Public Health Policy, September, 1984. Ei7

Consumption Impacts of a Change in the Federal Cigarette Excise Tax Kennetli E. Warner /'rojessor and Chairnian, !)eparhnent of Health Planninq and Adniinistration, Sclwof uf Public Health, [ lniversity of Michigan .-l nn ArGor, Michigan Introduction Traditionally, legislatures enacted excise taxes to raise revenues or to make a statement about the "morality" of use of the product in question. In the latter instance, the objective was to penalize or discourage the behavior- hence the name "sin tax." In recent years; however, excise taxes on products such as tobacco and alcohol have come to be viewed in a third dimension, one that has largely eclipsed the morality concern: excise taxes can be effective tools of public health policy. Economista have begun to evaluate the consumption impacts of such taxes as ends of inherent health interest, rather than as vehicles to estimate the revenue implications of tax-induced consumption changes (Cook, 1982; Cook and Tauchen, 1982; Harris, 1982; Laughhutin and Lyon, 1971; Lewit and Coate, 1982; Lewit, Coate, and Grossman, 1981; Warner, 1982, 1984). Furthermore, public health professionals are including excise taxation as an essential element in writings on public health policy to combat smoking and alcohol abuse (Beauchamp, 1976; Bonnie, 1978; de Lint, 1980; Ernster et al., 1985; Jacobson, 1981)t. The current debate on the federal cigarette excise tax is timely in the context of both its revenue and consumption-and hence health- implications. Regarding the former, both Congress and the Administra- tion are searching actively for ways to diminish the swollen federal budget deficit. While any conceivable excise tax increase cannot be construed as making more than a dent in the deficit, for several reasons this tax has taken on dimensions disproportionate to its possible contribution. For one, the Administration's 1984 deficit-reduction package included several effective tax increases and only one highly visible effective tax decrease-the decision The author is indebted to the Conference participants for helpful suggestions on an earlier draft, and particularly to Thomas Schelling. to allow the federal cigarette tax to revert from 16 cents to 8 cents per pack on October 1, 1985, as called for in the 1982 TEFRA sunset provision. Second, a cigarette tax can be viewed as a "user fee," a politically attractive bit of nomenclature adopted in a Congressional legislative proposal. The "user fee" notion meshes neatly with the call by the Advisory Council on Social Security to earmark a cigarette tax to offset smoking-related Medicare expenditures (Rich, 1983). Furthermore, by being labeled a "user fee," the excise can parade as a non-tax (or quasi-tax) "revenue enhancer," thereby mitigating the Administration's insistence on avoiding tax increases. The political attraction is based too on the fact that fewer than a third of all adults-the smoking population-will bear its burden. The timeliness of interest in cigarette consumption impacts relates spe- cifically to the sunset provision for the current 16-cent tax, which analysts have predicted will cause smoking to increase, and more generally to the objective of the smoking-and-health community to work toward the Surgeon General's goal of a smoke-free generation by the year 2000 (Koop, 1984). The nature of the consumption impact of a change in the excise tax is the subject of the remainder of this paper. In focusing solely on the consumption impacts, the paper addresses its assigned charge. It should be noted at the outset, however, that one of the major attractions of an increase in the federal excise tax is its ability to simultaneously serve the interests of fiscal and physical health (Warner, 1984). Price Elasticity of Demand for Cigarettes The basis for estimating the cigarette consumption impacts of a change in the federal excise tax is an analysis of the price elasticity of demand for cigarettes. Over the past several decades, numerous economists have under- taken empirical studies of the elasticity, relying on both time series and cross sectional data (Blaine, 1983; Fujii, 1980; Lewit and Coate, 1982; Lewit, Coate and Grossman, 1981; Lyon and Simon, 1968; Lyon and Spruill, 1977; Miller, 1975; Sackrin, 1962). Studies dating since 1970 have found price elasticity estimates ranging from-0.4 to -1.3 (Lewit and Coate, 1982). According to Miller (1982) (as cited in Toder, 1985), a price elasticity of -0.7 represents the midpoint of recent studies and is the figure used by the Tobacco Institute in its analyses of the impacts of cigarette taxes. Price elasticity estimation remains a difficult task. To date, no study has successfully addressed the complex issues in smokers' shifting from high to low tar and nicotine (t/n) cigarettes, complicated by tlu fact of nicotine regulation (Benowitz et al.;1983; Folsom et al., 1984; Gerstein and Levison, 1982; Kozlowski et al., 1980; Russell et al., 1980). In virtually all studies, cigarettes have been treated as a homogeneous product. The exception is 0 4H 1 81)

s ;w Itmited investigation of differential tar and nicotine taxes (Drayton, 1972; I larris, 1980). Other aspects of smoking behavior, such as price-induced mrreases in puffing frequency or smoking further down the butt, have .•,caped attention in all elasticity studies. While such behaviors may not be rcIrvant to an interest in the revenue implications of a tax or wholesale price change, they are of interest in assessing the health implications of tax-related changes in price. Also of interest to health professionals are ,lifferential price responses by income class (Townsend, 1983), t/n level, rears of smoking history, size of daily habit, etc. With the exception Ot the first of these, none of these factors has been studied in elasticity .tnalyses, and the first has been studied only in the context of smoking In Great Britain. An issue further complicating interpretation of elasticities is that there is ~;0od theoretical reason to expect an asymmetrical response to price increases .tnd decreases. From survey data, the federal Office on Smoking and I lealth has concluded that very few people begin to smoke after the age of 21. Hence cigarette price decreases would be unlikely to induce adults to start to smoke (though they might increase smokers' daily consumption), w(lile price increases could encourage some smokers to quit (and others to reduce daily consumption). The implication is that price response might be considerably greater in the instance of price increases than decreases. Again, no study has addressed this challenging analytical problem, though i t has clear relevance to both economic and health concerns. In particular, it is central to an assessment of both the revenue and consumption implica- tions of a change in the federal cigarette excise tax. Finally, it is unportant to recognize that the lag between collection of data, analysis, and publication of results inevitably encompasses a period of several years. While this is a problem in virtually all empirical social science research, it is particularly germane when the enviromnent for the behavior in question is rapidly changing. In the context of smoking, the composition of the smoking population has changed continuously in the past three decades (in terms of age distribution, education, sex mix, etc.), "o that elasticities based on a population of smokers in 1979, for example, may not reflect the behavior of the population in 1985. Furthermore, the cigarette market itself is evolving quickly, with a new trend toward price segmentation (from low-priced generic brand cigarettes to high-priced designer" cigarettes) introducing a factor which might alter elasticities. These considerations emphasize the limited analytical state of the art of cigarette price elasticity estimation. Nevertheless, the state of the art experienced a substantial improvement in the early 1980s when Lewit and his colleagues (1981, 1982) produced studies of both adult and teenage t I . price elasticities. These analyses took into account many of the biases that afflicted previous studies, including multicollinearity among cigarette price, income, and smoking trend variables in time series analyses and the effects of bootlegging on sales data in cross sectional studies. -I'he work by Lewit and his colleagues also has the virtue of exploring differences in elasticities by age and sex. In addition, Lewit and his colleagues introduced the useful innovation of examining separate participation (prevalence) and daily consumption elasticities. The two studies by Lewit et al. represent the best evidence on elasticities to date, and their distinguishing prevalence/quantity elasticities permits a more refined assessment of the likely consumption impacts of a change in the federal cigarette excise tax. For this reason, this paper uses the elasticity estimates by Lewit et al. to estimate the consumption effects to be expected as the result of non-plausible changes in the federal excise tax. Specifically, we will use their estimates to examine the consequences of permitting the tax to fall to 8 cents, as scheduled, and increasing the tax to 24 cents or 32 cents. Thirty-two cents-a doubling of [he current tax-is approxi- mately the level that would be required simply to maintain the real value of the tax at its value in 1951, the last time the tax was permanently increased. Twenty-four cents represents an intermediate increase equal in magnitude, though opposite in direction, to the return to 8 cents scheduled for October 1, 1985. The salient estimates from the work of Lewit et al. are presented in Table 1. Not presented are the researchers' finding that elasticities were much greater (in absolute value) for adult men than for women, a finding Table 1. Cigarette Demand Price Elasticities Calculated by Lewit et al. (1981, 1982) Elasticities Abe Group Total (1) Participation' (2) Quantity per day per sntoker (3) 12-17 -1.40' -1?0• -o.z5` 20-25 -0.8'Y' -0.7-1' -(t?o 26-35 -o.d7• -0.44' -0.04 3(r7a -0.45' -0.15 -tt.15 All adults (20-74) -0.42' -0.26' -0.11) ' Significant at p • .05 ' I'revalence clasticity. tieFkcts the decision of whether or not to sntoke at all. ~' Implied front total and participation elasticiks. - 'ill ' ~ t)1

'hey did not attenipt to explain. They do hypothesize, however, that •ti s price responsiveness may approach that of males as female smoking p.ttterns are becoming quite similar to those of men. Below we discuss the 1mplications of using the non-sex-specific elasticities. The overall adult price elasticity of -0.42 falls at the lower end of the i ange of recent estimates. Use of this estimate, rather than a more average i i9ure, reflects our opinion that Lewit and Coate's (1982) work was better Icsigned than that of earlier studies. Failure to control for bootlegging, for •xample, would bias upward elasticity estimates based otl state cigarette .tlrs data. In addition, use of a lower elasticity estimate assures that 'stinlates of tax-related consumption effects will be conservative. Larger .•Iasticities would translate into larger consumption impacts. The total price elasticity estimates (column (1)) exhibit a pattern of price i-rsponsiveness decreasing with age, as would be expected for three reasons: I) teens and young adults have smoking habits that are less well-defined .1nd of shorter duration, implying less habituation or addiction and hence the potential for more price responsiveness; (2) younger people may be more inclined to start smoking as the result of a price decrease than would he older adults, as discussed earlier; (3) on average, younger people will have less disposable income so that price response may includemore of an income effect. Worthy of note is that teens' cigarette demand is quite elastic %vIlile that of adults is inelastic. One of the most important findings in the estimates of Lewit et al. is that "participation" or prevalence decisions (column (2)) account for the vast majority of total price response. All of the daily quantity elasticity esti- mates (colurtui (3)) are small and statistically nonsignificant, while all but ,Me of the participation elasticities are significant. The absence of apparent daily consumption response seems counterintuitive. One can imagine a number of adjustment mechanisms that do not involve change in the mimber of cigarettes smoked-more or fewer puffs per cigarette, smoking tttrtller or less far down the cigarette, etc.,-but a change in daily con- .umption is perhaps the most obvious response. It is possible that anomalies in smokers'reporting of their daily habits on the surveys could disguise a consumption change. The essential point is that the absence of daily COnsumption effects may not reflect reality and in any case does not imply .tn absence of price response by continuing smokers. For purposes of this 11aper, however, the nonsignificance of the daily quantity elasticities will leatl us to examine only participation and total consumption effects of price changes. For a discussion of the specific methods and limitations of the elasticity <rudics, readers should consult the original papers. Consumption Impacts of Changes in the Federal Excise Tax In 1984, the weighted average retail price of a pack of cigarettes was 97.8 cents (Tobacco Institute, 1984). If the federal excise tax reverts to 8 cents a pack on October 1, as scheduled, and the retail price of cigarettes falls by the same amount, average price will fall by 8.5 percent. If, instead, the tax were to be increased from its current level of 16 ci:nts to 24 cents or 32 cents, average price would rise by 7.9 percent or 15.1 percent, respec- tively.2-' Applying the elasticity figures from columns (1) and (2) of Table 1 implies that the expected percentage changes in consumption would be those presented in Table 2. Two caveats are essential at this point. One relates to the assumption that price will change by the full and exact amount of the tax change, the assumption employed in virtually all cigarette demand elasticity studies that supply is perfectly elastic. The one study to address this issue suggests that retail prices have risen by slightly more than taxes (Barzel, 1976). Some observers conjecture, however, that pricing responses to tax increases and decreases may differ, with tax decreases not producing commensurate decreases in price. This conjecture suggests collusive behavior, a possibility in an oligopoly consisting of half a dozen firms, two of which control two-thirds of the entire domestic market. If this model is valid, it sug- gests that the legislated decrease in the federal tax would not produce an eight-cent decrease in retail price, mitigating increases in cigarette consumption. The second caveat relates to the issue of asymmetrical consumption price response discussed above. If it is true that response to a price decrease is less elastic than response to a price increase, then symmetrical application of the elasticities from Table 1 will bias upward the estimate of consump- Table 2. Expected Percentage Changes in Cigarette Consumption Resulting from Changes in the Federal Cigarette Excise Tax 8-cent Decrease 8-centlncrease 16-centlnereau Age Group Total Partici- pation Total Partici- pation Total Partici- pation 12-17 11.') 10.2 -11.1 -9.5 -21.1 -1H.1 20-25 7.6 6.3 -7.0 -5.9 -13A -11.2 26-35 4.0 3.7 -3.7 -3.5 -7.1 -6.6 36-74 3.8 1.3. -3.6 -1.2 -6.8 -2.3 All adults (20-74) 3.6 2.2 -3.3 -2.1 -6.3 -3.9 13 1 93

non increases associated with a price decrease and downward the estimate ,t consumption decreases associated with a price increase. In terms of f,tble 2, this means that the consumption percentage change estimates for l)e 8-cent decrease would be too large, while those for the 16- and 24-cent ncreases would be too small (in absolute value). We will return to this possibility below. For now, however, we will 1•SUme that a tax change will be reflected in a commensurate change in •tail price and that price response is symmetrical and accurately represented 1k• the elasticities estimated by Lewit and his colleagues. In this case, the 1;;ttres in Table 2 represent expected percentage changes in consumption. Table 3 presents estimates of the numbers of smokers-in each age .ttcgory studied by Lewit et al., as well as older teenagers (18-19) and older -L'nior citizens (75+). Also included in the Table are estimates of average Ltily consumption in each group and the group's to~tal annual consump- i.~n. Combining these with the percentage changes in Table 2 produces :Stimates of the quantitative changes in numbers of smokers and each age roup's annual consumption. These estimates are presented in Table 4, irranged to look at each effect (i.e., participation and total consumption) n:ross the three alternative tax changes. Estimates for 18-19 year-olds ind the most elderly adults (75+) are calculated as described in the i ~,otnote to Table 4. From Table 4 we see that if the operative assumptions held, the currently ?egislated eight-cent decrease in the excise tax would induce almost 2 million people to smoke who would not do so if the tax were to remain at 16 cents. This number includes both people initiating smoking habits and Cantinuing smokers who would have quit absent the economic prod to continue. Among the 1.9 million are more than 460,000 teenagers who %Vuuld begin or continue smoking as a result of the tax decrease. Adding in t lie most price-responsive adults, those aged 20-25, we find that more than I tnillion young people would join the ranks of the smoking population if the tax decrease takes effect. An eight-cent tax increase would have a quantitatively similar opposite : tfect. 1.8 million people would be encouraged to quit or not begin •mokuig, including over 400,000 teenagers and more than half a million young adults age 20-25 (and over a million young adults age 20-35). A 16-cent increase in the excise tax, bringing the real value of the tax close to its value in the early 1950s, would encourage almost 3.5 million Ameri- : ans to forego smoking habits in which they will engage if the tax remains -it 16 cents per pack. This figure includes over 800,000 teenagers and almost two million young adults age 20-35. The aggregate annual changes in cigarette consumption are substanial in Table 3. Estimated Numbers of Smokers, Average Daily and Annual Consumption by Age, 1982 Age Group Number Smokers' (1) (Percentage) Average I)aily Con- sumption of Smokers (# cigarettes) (2) Total Annual Con- sumption of Group (z cigarcttes, billiun.)' (3) 12-17 3.27'),cNw (14.7) 16A 19.1 18-1') 1,577,IN)0 (18.7) 17.5 10.1 , 20-25 '1,665,000 (36.8) 18.0 63.5 26-35 13,722,000 (35.1) 20.3 101.7 36-74 26.963,000 (31.8) 21.8 214.5 75+ 1.014,000 ( 9.5) 17.3 6.4 Total 56,220,000 (29.3) 20.2 415.3 ' Age group population figures were taken from 1982 estimates by the Bureau of the Census. Smoking participation data, provided by the Office on Smoking and Health, DI It ls, were from: a 1982 NIDA household survey (12-17 year-olds) in which smoking was defined as dail}• use of ci arettes during the preceding 30 da s; the 1984 NIDA survey of high school seniors' drug use (c garette use rate, defined as above, applied to 18-19 year-olds); and the 198311ealth Interview Survey (H1S) (adults) in which current smokers were defined as those currently smoking ciga- rettes and having a lifetime consumption of 100 or more, cigarettes. Use of the NIDA data may produce a small underestimate of smoking by 18-19 year-'"ds, as high school seniors are une year younger and may have lower rates of cigarette smoking than high school drop-outs. Age groupings of the HIS data and the studies by Lewit et al. differ slighdy. The I Ils rate for 20-24 y car-olds was used for 20-25 here. Similarly the I lIS 25-34 rate was used for 26-35. HIS broke down rates for over-35 into 35-44, 45-64, and 65+. In the present Table, a rate somewhat higher than the 65+ rate was used for 65-74 year-olds and a lower rate for 75+, the two yielding the I IIS rate for the entire 65+ group. Each of these adjustments introduces potential errors of such small magnitude as to be inconsequential for the largely qualitative purposes of this paper. "Average daily consumption figures for teenagers were estitnated from data in Exhibits A-I and 16 in National Institute of Education (1979) and data supplied by the National Center for I Icalth Statistics (NCHS) from the 1983 NIDA study of high school seniors' drug use. For adults, the figures were estimated from the Census and the 1983 HIS, supplemenicd by data provided by NC1IS from a 1980 telephone poll of the HIS. The latter yielded detailed averages of daily consumption, while the 1983 HIS data were available only in categories (fewer than 15 cigarettes, 15-24, 25+). As with the estimation of numbers of smokers in column (1), estimation of average daily consumption in age groupings corresponding to those used by Lewit et al. required a nutnber of assumptions and tnterpolations. In particular, the teenage daily use estimates might better be considered educated "guesstimates:" All of these figures, however, are quite consistent with earlier estimates of daily consumption as reported on surveys. It is important to keep in utind, as noted in footnote c below, that self-reports of daily consumption fall well below objective measures of consumption. As is discussed in the text; the statistical nonsignificance of the daily cuusumption elasticities makes examination of tax effects on the daily use variable largely a qualitative exercise anyway. ' Equals (1) x (2) x 365. The total estimated annual consumption is only slightly over two-thirds of the nearly 601) billion cigarettes U.S. smokers actually consume each year. This is consistent with the observation that Americans tend to underreport their levels of cigarette consumption (Warner, 1978). ' I ~ 95

.thsolute magnitude-ranging from an increase of over 20 billion ciga- wttcs to a decrease of more than 36 billion-but represent a relatively small I,roportion of the domestic cigarette market (from 3.4 to 6.2 percent). Discussion Consumption Impacts. The immediately preceding point illustrates a tundmental conclusion of this analysis: the overall relative consumption umpact of conceivable tax changes is modest, only on the order of a few percentage points. But the size of the cigarette smoking population and the .1.1ily consumption of smokers mean that even modest relative changes hecome substantial effects in terms of absolute magnitude. This is readily illustrated by the fact that, if our assumptions hold, an eight-cent tax Cable 4. Estimated Changes in Cigarette Smoking Attributable to Changes in the Federal Cigarette Excise Tax' .1~e Group 8-cent 8-cent 16-cent Decrease Increase Increase Change in number of smokers (thousands) 1'-17 + 334 - 311 - 594 + 130 - 121 - 231 + 608 - 565 -1.080 35 +513 -478 -911 + 345 - 321 - 612 -~+ + 13 - 12 - 23 I ~~tal +1,943 -1,808 -3,451 Change in aggregate cigarette consumption (# cigarettes, billions) 1'-17 +2.3 -2.1 -4.0 1i-t') +1.0 -0.9 -1.7 '11-25 +4.8 -4.5 -8.5 '6-35 + 4.1 - 3.8 - 7.2 lli-7a +8.2 -7.6 -14.6 7; } +0.2 -0.2 ~ lutal" +20.6 -19.1 -36.4 1•he percentage changes for 36-74 year-olds have been applied to the oldest group of adults (75+) •~rell. Smokers in this age bracket may be more confirmed smokers than younger adults, perhaps 1itplying less price response, but they are also likely to be poorer on average, implying more -.punse to a price change. Note that these smokers constitute less than 2 percent of the smoking ;-qpulation. For 18-19 year-olds, percentage changes midway between those of 12-17 and 20-25 : ar-olds have been used. I'be absolute value of each o f these totals is about 6 percent greater than the figures calculated by .1mbining the prevalence and daily consumption changes directly. This results from the estima- .. tit procedure for the different rlasticities used by Lewit et al. i ' change will alter the size of the smoking population by less than 3.5 percent, but that 3.5 percent represents almost 2 million Americans. The impacts of tax-induced consumption changes are of most immedi- ate importance in the population of middle-aged and older adults, because these are the individuals most prone to experience smoking-related illness. The elasticity estimates produced by Lewit et al. show this group to be the least price-responsive, yet the shcer size of this group means that, under the operative assumptions, from 330,000 to 630,000 will alter their smoking status if the federal tax is changed. In the long run, the toll of smoking is tied to the smoking practices of the youngest generation. There is a widespread consensus that the ultimate conquest of smoking-induced illness can come only froni preventing the onset of smoking in the teenage and early adult years. In this regard, the elasticity studies of Lewit et al. and their translation into numbers of smokers are particularly important. Not only do the price responses repre- sent large numbers of young people; they also represent substantial propor- tions. An eight-cent decrease in the federal excise tax would increase the ranks of teenage smokers by a tenth. A 16-cent tax increase would diminish the population of teenage smokers by fully 17 percent. The former would lead approximately 460,000 teenagers in the direction of cigarette habits; the latter would lead 820,000 teens away from dependency on cigarettes. As discussed above, it is possible that the assumptions underlying the estimates in Tables 2-4 do not hold. Specifically, smokers' responses to price changes may not be symmetrical and the retail price of cigarettes may not adjust by an amount precisely equal to a federal tax change. In particular, it seems plausible that smoking prevalence by adults is less sensitive propor- tionately to a price decrease than to an increase, and that prices are sticky downward; that is, retail prices per pack of cigarettes might not fall by a full eight cents if the federal tax drops from 16 to 8 cents. If either or both of these conditions held, the consumption impacts estimated in Table 4 would be altered as follows: •(f retail price fully reflected a tax change but adult participation response were asymmetrical, the 8-cent tax decrease would encourage a smaller number of adults to start, resume, or continue smoking; the 8- and 16-cent increases would encourage larger numbers of adults to quit. As there is no a priori reason to assume that smoking by teenagers would be asymmetrical in a given direction, no obvious adjustment of teen response would be called for. • If participation response were symnutrical but price was sticky down- ward, the estimated increases in smoking prevalence would have to be 1 97

reduced to the extent that the 8-cent tax decrease was not passed on to "msunurs. This would apply uniformly to teens and adults. - If price was sticky downward and adult participation was asymmetrical, I he increases in smoking prevalence associated with the 8-cent tax decrease in fable 4 would clearly overrepresent the consumption impact for both trens and adults and might have to be reduced substantially. Conversely, rlie Table 4 decreases in smoking associated with tax increases would have io be increased for adults. Lack of substantial empirical evidence on either of these issues makes it :mpossible to translate the directional adjustments indicated above into ,prcific numbers. It is important to emphasize, however, that except in the :ase of no drop in retail price following a tax decrease, the qualitative t indings of the analysis would hold: an 8-cent tax decrease would encour- .Ige tens or hundreds of thousands of Americans to smoke who would not I irherwise do so; a tax increase of the magnitude considered would encour- Age many hundreds of thousands, and likely millions, of Americans to quit "moking and, in the case of youngsters, not to start. Health Implications. The ultimate importance of tax-induced changes in cigarette consumption lies in their health consequences. No formal .issessment of the health impact has been undertaken for purposes of this paper. However, the fact that one lifelong smoker of every three or four dies from a smoking-related illness can be used to produce some "ballpark" estimates of the mortality implications of the contemplated tax changes. for example, if we assume that one of every four tax induced quitters (or nonstarters) would have died from smoking, and if we adopt the .usumptions used to generate Tables 2-4, the eight-cent increase would be credited with averting the smoking-induced premature deaths of 450,000 Americans. The 16-cent increase would avoid 860,000 premature deaths. Ry contrast, if the eight-cent tax decrease takes effect and the above .Issumptions hold, over 480,000 Americans will die prematurely as a result [If their tax induced initiation or continuation of smoking habits. On .Wrrage, these victims of smoking will die more than two decades earlier rhan they would have if they had quit smoking or never started.4 As with the consumption estimates, these numbers are sensitive to •everal assumptions, including those relating to pricing and symmetry of price response. Also important is recognition that medical practice and Wchnology may improve in the future to the point that many now-fatal I;cart and lung diseases and cancers will become curable. To the extent that t his occurs, the potential mortality savings of a tax increase or the deaths resulting from a tax decrease will be diminished. Morbidity implications are less obvious, as avoidance of death as a medical outcome can translate into increases in experienced illness and disability. Another perhaps obvious feature of these numbers also deserves emphasis: the premature deaths averted or produced by a tax change are not realized all at once. The major consumption changes induced by a cigarette price change will occur in the youngest groups of smokers and potential smokers-teenagers and young adults. The numbers in Table 4 suggest that fully 80 percent of price-induced changes in smoking prevalence will be found in people 35 years of age or younger; nearly a quarter of all responders will be teenagers. If price responses are asymmetrical, as discussed earlier, a still larger share of smoking initiation associated with a tax decrease would be found in the youngest age groups. The import of this distribution is that the majority of the premature deaths that will be associated with a tax decrease, or of the premature deaths avoided as the result of a tax increase, will occur two to four decades into the future. The most immediate mortality implications relate to tax-induced changes in smoking prevalence in middle aged and older adults. Given the large numbers of smokers in these age categories, this still translates into a substantial mortality implication for the near-term future; but it is only a fraction of the totals given above. For example, according to Table 4 close to a third of a million Americans aged 36 and older would quit smoking if the federal excise tax were raised 8 cents; over 630,000 would quit if the tax were doubled to 32 cents per pack. Given the earlier assumptions, these figures would translate, respectively, into 83,000 and 159,000 premature deaths that would be averted in the moree immediate future, from the year of the tax increase extending into the following two decades. It is important to emphasize that all of the numbers in this section are intended to be illustrative only, indicative of the order of magnitude of the health benefit that would follow a tax increase, or of the death toll that would result from a tax decrease. The qualitatively important conclusion is that tens of thousands of Americans will die prematurely if the excise tax falls back to 8 cents per pack. If, instead, the tax is raised, tens of thousands will lead longer, healthier lives than they will otherwise. In both instances, substantial nonfatal illness burdens will be affected as well. Further Implications About Elasticity Estimates All the available evidence suggests that women's smoking has not been as price responsive as that of inen. As noted earlier, Lewit and his col- leagues found price response among women to be much smaller and W)

. , •ratistically nonsignificant. In the present paper, the decision was made to a,V the overall elasticities (i.e., not differentiated by sex) in part because he author believes that the apparent nonresponse of women, if real, may :~r an historical quirk, reflective of an era in which smoking's popularity 1nwng women was growing rapidly. If this is correct, a new study, using I't ,5 data, well might find women's price response approaching that of .nrn. Two implications deserve emphasis: I ) If the apparent difference in price response is real and persists today, the .rnoking prevalence and mortality changes estimated in this paper would lrrive mostly from changes in smoking behavior by men. • ?) If smoking by women has become more price responsive in recent years, the figures presented in this paper likely underestimate the amount of smoking change and hence the mortality implications that would result from tax changes. The estimates by Lewit et al. of inen's price elasticities .1lone are greater than the overall elasticities used here. Increases in women's price responsiveness, without decreases in men's, would lead to higher overall elasticities and hence greater behavioral and health impacts. Of Cuurse, it is possible that male smokers in 1985 are less price-responsive than niale smokers of a decade earlier, partially offsetting the hypothesized increase in price response by women. Price elasticity can be a function of the price level, particularly in its relation to income levels. When price is high relative to a country's or a group's income, a component of price response is income effect; that is, an increasing cigarette price can make smokers aware (consciously or other- Wise) of its impact on their disposable income and hence more price .ensitive. This phenomenon helps to explain why cigarette demand price cIasticity is much higher, compared with that of the jJ.S., in a country like England in which price is higher and income lower. It also explains liownsend's (1983) finding that the absolute value of price elasticity is inversely related to social class in England. A similar explanation applies to rhe conclusion of Lewit and his colleagues that American teenagers have liiglter price elasticities than do adults, and that younger adults are more responsive than older adults.5 Two additional implications derive from cunsideration of income effects: I) It is quite possible that, as Townsend observed in England, lower •()cioeconomic groups would be more responsive to price changes than WOuld be higher SES groups. Thus a tax-produced cigarette price decrease .nigbt cause relatively more poor people to join or remain in the ranks of •nlokers. Similarly, a tax-linked price increase might induce relatively l.irger proportions of the poor to forego smoking. (2) The inelasticity of demand for cigarettes in the U.S. is undoubtedly in part the result of low price of cigarettes relative to income. If real cigarette price were to increase substantially over time (which would require tax increases well in excess of those considered in this paper), demand elasticities might rise as well. In that circumstance, further price increases or decreases would be expected to have proportionately larger effects on cigarette consumption and hence, ultimately, on the burden of smoking-related illness. Finally, consideration must be given to the effects of inflation on real cigarette price and thus on the prevalence of smoking. During the more than 30 years of its existence, due to inflation the 8-cent federal excise eroded to only 2.5 cents in constant 1951 dollar value. The doubling of the tax in 1983 restored the tax to only about half of its real value in the early 1950s. Similarly, legislating any tax change in 1985, whether an increase or a decrease, will be tantamount to legislating an effective tax decrease in ensuing years, unless provision is made for inflation-compensating tax increases. This could be accomplished by shifting cigarette taxation to an ad valvrem basis or indexing the tax rate to the general price level or a price index for all cigarettes (Toder, 1985). The importance of this is that even though a tax boost in 1985 would discourage smoking, without an infla- tion adjustment the eroding value of the new tax in 1986 and later years would encourage people to start or maintain cigarette habits. This is not merely an academic consideration, since tax is a substantial percentage of retail cigarette price. At the present time, federal and state excise taxes constitute 32 percent of average retail price nationwide. Even with the doubling of the federal tax in 1983, this percentage repre- sents an historically low figure. During the two decades from 1954 through 1973, the tax share of retail price was never less than 46.6 percent. An effective moratorium on state tax increases during the next decade (W.rrner, 1981) caused the tax share to fall annually to a low of 26.8 percent in 1982. The current federal tax constitutes 16 percent of retail price. Its predeces- sor, the 8-cent tax, accounted for a larger proportion in all years to 1976 (Tobacco Institute, 1984, Table 13). Conclusion An increase in the federal cigarette excise tax is an attractive, effective tool of health policy. Indeed, it is difficult to think of many policy measures that could have a comparable impact on the health of the public. A tax increase, however, is not the public health ideal because the inelastic- ity of cigarette demand means that the tax increase will not eradicate ~ I W ' 101

•moking nor even eliminate a large proportion of it, at least for taxes on the Wtler of magnitude discussed here. But it may be precisely that inelasticity -hat makes a tax increase a viable public policy option at present, because it „sures that a tax increase will generate a revenue increase. No one wants to cc government fundamentally dependent on cigarette excise tax revenues, ')ut the situation in America is far from this possibility: the federal excise ix constitutes well under one percent of all federal revenues. In 1985, increasing the federal cigarette excise tax offers several attrac- n0m. It promises to increase federal revenues, especially in the short run. It ~~ ill discourage a large number, if a small percentage, of adults to give up heir smoking or not to start. And it will discourage both a large number und a significant percentage of young people from starting or continuing I I smoke. The legacy of a tax increase would be a significant contribution itiward the realization of a smoke-free generation. The legacy of a tax Ircrease would be tens of thousands of avoidable premature deaths in the ,:Oming decades. Notes 1. The interest in excise taxation reflects a broader emerging public health interest in the use of economic incentives to affect behaviors related to health. Health professionals and health benefits managers in business arc exploring a wide range of employment-based incentives (wage bonuses, lottery prizes), insurance incentives (deductibles and copayments, differ- ential premiums and benefits), and tax incentives (deductibility of expend- itures on weihress programs) to encourage health-enhancing changes in such areas as diet, exercise, use of seat belts, drug use, and smoking (Warner and Murt, 1984). 2. The denominator is calculated at the mean of the pre- and post-tax change prices. 3. Obviously we are employing an assumption that all other things remain equal. Changes in cigarette production and distribution costs and in state and local excise taxes will also affect cigarette price. Here we are concerned exclusively with the consumption impact that will result from a federal tax change. 4. The author and a colleague have estimated that decreases in smoking prevalence attributable to antismoking activities, including excise taxation, had prevented over 200,000 premature_deaths by 1978, with exponentially increasing numbers in the ensuing years. On average, each of the prema- ture deaths averted translated into 23 years of additional life (Warner and Murt, 1983). 5. The income effect is not intended to serve as the entire explanation of the differences in elasticities. Young people's smoking habits are less well- established than those of their seniors, a factor which almost certainly contributes to the greater price responsiveness of the former. References Barzel Y. An alternative approach to the analysis of taxation. J. Polit. Ecun. 1976; 84:1177-1198. Beauchamp DE. Exploring new ethics for public health: developing a fair alcohol policy. J. Health Politics, Policy and Laiv 1976; 1:338-354. Benowitz NL, Hall SM, Herning RI, Jacobs P 1ll, Jones R1; Osman AL. Smokers of low-yield cigarettes do not consume less nicotine. N. Er~Yl. J. lbl ed 1983; 309:139-142. Blaine TW. Problems in the estimation and interpretation of demand equations: the case of cigarettes. Working paper, University of Kentucky, 1983. 103

R liunnie RJ. Discouraging unhealthy personal choices: reflections on new i lirections in substance abuse policy. J. Drug Issues 1978; 8:199-219. Cook PJ, Alcohol taxes as a public health measure. Br. J. Addiction 1982; 77:245-250. c:uok PJ, Tauchen G. The effect of liquor taxes on heavy drinking. Bell J. Econ. 1982; 13:379-390. le Lint JE. Alcohl control policy as a strategy of prevention. J. Pub. Health Policy 1980; 1:41-49. I)rayton W, Jr. The tar and nicotine tax: pursuing public policy through tax incentives. Yale Latv J. 1972; 81:1487-1516. 1: rnster V, Holbrook J, Lewit E, Pertschuk M, Steinfeld J,- Warner K, W helan E. Public Policy on Smoking and Health: Toward a Smoke-Free Generation by the Year 200U. Report of a work group prepared for the Subcommittee on Smoking, American Heart Association, 1985 (draft). Folsom AR, Pechacek TF, de Gaudemaris R, Luepker RV, Jacobs DR, Gillutn RF. Consumption of 'low-yield' cigarettes: its frequency and relationship to serum thiocyanate. Am. J. PublicHealth 1984; 74:564-568. Fujii ET. The demand for cigarettes: further empirical evidence and its implications for public policy. Applied Econ. 1980; 12:479-489. Gerstein DR, Levison PK, eds. Reduced tar and nicotine cigarettes: smoking behavior and health. Washington, D.C.: National Academy Press, 1982. I larris JE. Taxing tar and nicotine. Ant. Ecort. Reu 1980; 70:300-311. I larris JE. Increasing the federal excise tax on cigarettes. J. Health Econ. 1982; 1: 117-120. Jacobson M. Grappling with alcohol abuse: when will we take a stand? Nutrition Action October 1981; 6-7. Koop CE. Julia M. Jones Lecture. Presented to the Annual Meeting of the American Lung Association, Miami Beach, May 20,1984. Kozlowski LT, Frecker RC, Khouw V, Pope MA. The misuse of `less- hazardotts' cigarettes and its detection: hole-blocking of ventilated filters. Ant. J. Public Health 1980; 70:1202-1203. [.aughhunn DJ, Lyon HL. The feasibility of tax induced price increases as a deterrent to cigarette consumption. J. Bus. Admin. 1971; 3:27-35. Lewit EM, Coate D. The potential for using excise taxes to reduce smoking. J. Health Econ. 1982; 1:121-145. Lewit EM, Coate D, Grossman M. The effects of government regulation on teenage smoking. J. Law and Econ. 1981; 24:545-569. Lyon HL, Simon JL. Price elasticity of the demand for cigarettes in the United States. Ant. J. Ag. Ecou. 1968; 50:888-895. Lyon HL, Spruill ML. A temporal cross-section analysis of cigarette price elasticity in the United States. Working Paper, Univ of Houston, 1977. Miller RH. Cigarettes: consumption situation and outlook. Proceedings of the 56th Annual Meeting of the National Tobacco Tax Association, 1982. -Miller RH. Pricing Out Tobacco: Price as a Factor in Cigarette Con- sumption. U.S. Department of Agriculture, Economic Research Service, 1975. National Institute of Education. Teenage Smoking: Lnmediate and Long Term Patterns. Washington: U.S. Government Printing Office, 079. Rich S. Advisors back alcohol, tobacco tax boost. II'ashington Post, Nov. 5, 1983, p. A5. Russell MAH, Jarvis M, Iyer R, Feyerabend C. Relation of nicotine yield of cigarettes to blood nicotine concentrations in smokers. Br. Med. J. 1980; 280:972-976. Sackrin SM. Factors affecting the demand for cigarettes. Ag. Econ. Res. 1962; 14:81-88. Tobacco Institute. The Tax Burden on Tobacco, vol. 19. Washington: Tl, 1984. Toder EJ. Issues in the Taxation of Cigarettes. Presented at the Conference on Tobacco Excise Taxes, Harvard Institute for the Study of Smoking Behavior and Policy, Washington, D.C., April 17, 1985. Townsend J. Cigarette Tax and Social Class Patterns of Smoking. Presented at the Fifth World Conference on Smoking and Health, Winnipeg, July 1983. Warner KE. Possible increases in the underreporting of cigarette con- sumption. J. Ant. Statistical Association 1978; 73:314-318. Warner KE. Statelegislation on smoking and health: a comparison of two policies. Policy Sciences 1981; 13:139-152. Warner KE. The federal cigarette excise tax. Pp. 160-167 in National Conference on Smokiny or Health-Developing a Bluepriiu for Action. New York: American Cancer Society, 1982. Warner KE. Cigarette taxation: doing good by doing well. J. Pub. Healtlt Policy 1984; 5:312-319. Warner KE. Murt HA. Premature deaths avoided by the antismoking campaign. Ant. J. Public Health 1983; 73:672-677. Warner KE. Economic incentives for health. Annual Ren Pub. Health 984; 5:107-133. Iii1 i 105

s On the Fairness of Cigarette Excise Taxation Jeffrey E. Harris 11a•sacltusrtts Institute ojTi~clnrology 1 xcise taxes on cigarettes, some have argued, are unfair because cigarette Inoking is more prevalent among lower income groups. Any further nrrease in such taxes, it is claimed, would just "hurt the little guy:" In tlus note, I ask whether the cla'tmed unfairness of cigarette taxes fits the acts. I find that the very lowest income Americans actually smoke less than niddle income Americans. In particular, older Americans, who make up a .izcable fraction of the low income group, have much lower smoking r.ttcs. Moreover, when I consider not only the proportion of smokers, but Ilso the number of cigarettes smoked per day, I find the burden of a cigarette tax hike to be no greater for blacks than for whites. My discussion is predicated on the assumption that manufacturers and retailers would largely pass on any new tax to cigarette smokers. If, to the <ontrary, a tax increase were not reflected in a higher retail price, then the burden of the tax hike might fall on the stockholders of cigarette manufac- lurers, the owners of domestic tobacco allotments, and others in the chain A If cigarette production and distribution. (For a more complete discussion, Nre Harris 1982.) The analysis here is based on unpublished data from surveys conducted hy the U.S. Government. In particular, fromJuly 1978 through December 1979 and July through December, 1980, the U.S. National Center for I Iealth Statistics appended a Cigarette Smoking Supplement to its contin- uing Health Interview Survey. Details of the Health Interview Survey, a .rratified, household-based, face-to-face interview sample that is repre- sentative of the U.S. noninstitutionalized civilian population, arc reported elsewhere (National Center for Health Statistics 1977,1979; Harris 1983). During 1979-1980, the Health Interview survey attempted to contact :iU,875 persons aged 17 or over concerning their cigarette smoking prac- rices. Of those contacted 47,286 (93%) reported their age and current cigarette smoking practices. The results below derive from the tabulated responses of this group. Table 1 shows the percentage of current cigarette smokers by income Itescarch supported by Public Health Service Grants DA-00072 and DA-02620. r and age. As the right-most column shows, among all adults, the percentage of smokers in the group earning less than $5,000 annually is equal to the percentage of smokers in the group earning 525,000 or more. The highest prevalence of cigarette use is in the middle income groups. Further, as the last row shows, cigarette use is relatively infrequent in the older popula- tion. While elderly persons represented almost 14% of the total sample, they made up 34% of the lowest income group (not shown in Table 1). A genuine inverse relation between income and smoking prevalence is found only in the middle age range. Table 2 shows the percentage of current cigarette smokers by income and race. Neither group shows a genuine inverse relation between income and the percentage of smokers. Among blacks, there appears to be virtu- ally no incotne-related gradient in the prevalence of cigarette use. Table 2 shows that a slightly higher percentage of black adults smoke cigarettes. But data on the proportion of smokers do not tell tlle whole story. We need to know the number of cigarettes smoked per day by current smokers in each group. Accordingly, I made the following computation. First, for each age- race-income category, I calculated the averoge reported number of ciga- rettes smoked per day among current smokers. Second, since current smokers are known to underreport the actual amount smoked, I multiplied the reported sample averages by 150%. (From national cigarette consump- tion data, I estimate that the average smoker actually consumed about 31.4 cigarettes daily in 1979. By contrast, the sample mean reported smoking frequency was 20.9 cigarettes daily. My use of the ratio 31.4/20.9 = 150% assumes that the underreporting was uniform among the subgroups.) Having estimated both the percentage of current smokers and the daily Table 1. Percentage of Regular Cigarette Smokers in Relation to Family Income and Age, 1978-80. Age (Years) Annual Family Income (Dollars) .17-30 31-65 •65 /111 avs 17+ <5,(xw 38 42 14 31 5,INN1-9,999 41 41 16 35 11),(Hlil-LI,')'l9 36 3`) I8 36 15,IHN)-2•1,99') 32 36 15 3•1 25,(NX)+ 28 33 17 31 All inconres' 34 36 16 33 Source: U.S. National Center for tk:dth 5tatistics, Hraltlt Interview Survey, unpubhshed data. a. includes those with unreported incomes, who represented 8 percent of the total sample. 111(, " 1 107

• .• •moking frequencies among current smokers, I then estimated the possible impact of an 8 cent per pack (that is, a 0.4 cent per cigarette) increase in the retail price of cigarettes. If such an increase had no effect on cigarette use, dhen the annual dollar burden per capita would equal: prrccntagc of number of cigarettes SO.004 per 365 "i~arette smokers x smoked daily among x cigarette x days per current smo[cers year tiince a price increase might deter some cigarette use, such a computation yields the maximum additional expenditure on cigarettes resulting from the specified tax increase. (In making such a calculation, I finesse the ~lttrstion of the relative price sensitivities of different age-race-income ,rroups. The artificial assumption of zero price elasticities permits me, for pttrposes of exposition, to assess the pure redistributive effects of the tax.) The results are shown in Tables 3 and 4. Overall, an 8 cents per pack tax increase would impose a maximum per capita burden of $15 annually. As ,hown in the top panel of Table 3, among all races, the per capita burden of tlte tax tends to show a positive (rather than an inverse) relation to income. 1=ttrther, as the bottom panel of Table 3 shows, the annual dollar burden per black person would be only about three-quarters of that per white person. As Table 4 shows, the annual dollar burden per middle aged person would be three times that per elderly person. The bottom panel of Table 3 and the right-most column of Table 4 express the computed dollar burdens as percentages of reported annual income. As a proportion of income, the estimated burden of the tax Table 2. Percentage of Regular Cigarette Smokers Among Adults in Relation to Family Income and Age, 1978-80. Race .\nnual Family Income (Dollars) White Black All Races' ; iN10 30 35 31 ;,INNI-9,999 34 35 35 lu!dtll-14;)99 36 36 36 1idHNl-2a;)99 34 36 34 _3,IR10+ 31 35 31 :\II incomesb 33 35 33 iource: U.S. National Center for Health Statutics, Health Interview Survey, unpublished data. .1, lucludes other races and those with unreported race. b. Includes those with unreported income. increase is virtually identical across races. Moreover, a tax increase would impose a much smaller proportionate burden on elderly incomes. I have focused on an 8 cent per pack price increase because the federal cigarette excise tax was raised by exactly that amount in 1983. With total ' federal, state and local tax receipts equal to about 31 cents per pack ; Table 3. Maximum Annual Dollar Burden per Capita of an 8-Cent ! Per Pack Tax Increase: by Family Income and Race.' Race Annual Family Inconre (Dollars) White Race Black All Itaccs' 45,000 13 11 12 5,000-9,999 16 10 15 10,030-14,999 17 12 16 15,000-24,999 16 12 16 25,000+ 15 12 15 % Annual Race i/year` lncomej White 15 0.13 Black 11 0.14 All Racesb 15 0.13 Source: U.S. National Center for Health Statistics, Health Interview Survey, unpublished data. a. See text for details of computation. b. lncludes other races and those with unreported race. c. lncludes those with unreported income. d. Excludes those with unreported income. Table 4. Maximum Annual Dollar Burden per Capita of an 8-Cent Per Pack Tax Increase: by Age.' Age Group S/year 17-30 years 14 31-65 years 18 66+ years 6 all ages 15 '%, Annual Incomet, 0.13 0.14 0.08 0.13 Source: U.S. National Center for Health Statistics, Health Interview Survey, unpublished data. a. See text for details of computation. b. Excludes those with unreported income. 111H ( 109

0 I U.S.D.A., 1985, Tables 31 and 1), the dollar figures (and the proportions uf income) could be scaled up by a factor of 31/8 to reflect the total dollar tax burden. Still, the relative magnitudes in Tables 3 and 4 would remain the same. How do we assess the fairness of a tax increase? Among economists, the conventional wisdom is to assess tax regressivity: that is, to check whether low income persons pay a greater proportion of their income than high income groups. The focus on tax regressivity, I would argue, is too narrow here. If the "little guys' are the elderly, then the data show that the cigarette tax in fact hits the "big guys." In this respect, a cigarette tax increase is fairer than, say, subjecting Social Security benefits to income taxation. Moreover, the evidence demonstrates that the average black person would in fact pay less (not more, as some would suppose) than the iverage white person. In this respect, the cigarette tax is fairer than cutbacks k)n government transfer programs that are targeted to minorities. I References Harris, J.E. (1982), "Increasing the Federal Excise Tax on Cigarettes," Journal of Health Econw,tics, T: 117-120. Harris, J.E. (1983), "Cigarette Smoking Among Successive Birth Cohorts ! of Men and Women in the United States During 1900-80,"Jonrnnl of the National Cancer Institute, 71: 473-479. National Center for Health Statistics (1979), "Changes in Cigarette Smoking , ' Practices Among Adults. United States 1978," Advance Data froin Vital and Health Statistics, no. 52. Harris, J.E. (1977), "Current Estimates From the Health Interview Survey," ; Vital and Health Statistics, Series 10, No. 119. U.S. Department of Agriculture, Economic Research Service (1985), Tobacco. Outlook and Situation Report, TS-191. III i