| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
The authors are with the Moore School of Business, University of South Carolina, Columbia.
Correspondence: Requests for reprints should be sent to Glenn W. Harrison, PhD, Department of Economics, Moore School of Business, University of South Carolina, Columbia, SC 29208 (e-mail: harrison{at}moore.badm.sc.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. This study compared the present economic value of the 1998 tobacco settlement with the present economic value of the damages attributable to tobacco.
Methods. The 1987 National Medical Expenditure Survey was used to estimate the smoking attributable fraction (SAF) of medical expenditures. SAFs were then applied to Medicaid and other expenditures.
Results. Settlement payments covered only 40% of Medicaid treatment costs already incurred and only 30% of past and projected future Medicaid costs. Excess medical expenditures for all other payment sources were roughly comparable to those incurred by Medicaid.
Conclusions. Although the settlement may reduce future smoking prevalence rates by limiting the ability of tobacco companies to promote smoking and by raising cigarette prices, euphoria over the huge settlement funds should be balanced by a sober comparison with the even larger damage amounts.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
These settlement amounts exclude $10 billion of other payments made by tobacco companies, as well as roughly $40 billion to be paid to Mississippi, Florida, Texas, and Minnesota. Attorneys' fees, not included in these amounts, will easily run into the tens of billions. Thus, the sum of the payments exceeds $250 billion.
This seems like a huge amount, and it is. But before one declares victory in this battle of the tobacco war, perspective is urgently needed. We consider 3 dimensions of the settlement from the perspective of plaintiff damages experts involved in individual state cases. (Glenn W. Harrison served as an expert and calculated damages for the cases brought by 13 states, while Maribeth Coller used these calculations to determine damages in terms of current dollars for the cases brought by 5 states. Harrison also serves as a damages expert in the case brought by 22 BlueCross BlueShield plans in federal court in New York, and both serve as damages experts for 2 HMOs in Minnesota.)
First, we illustrate the computation of the damages that the settlement is intended to compensate. Second, we compare the settlement amounts with these damages. Finally, we show that comparable damages have been incurred in terms of medical expenditures by private citizens, private insurance companies, worker's compensation, and Medicare. The legalities of collecting damages in each of these cases are complex, and many cases have been summarily dismissed by the courts. Without forming any judgment on the legal issues, we calculate the damages attributable to smoking in all of these cases.
The analysis involved 3 significant features. First, consistent with recent studies,1–5 damages were not restricted to diseases known to be caused by smoking. Our estimates allowed for the possibility that smoking may cause the treatment costs to be higher for any given medical condition, even if smoking did not cause the condition that led to the medical intervention. Second, we took into account the effects of passive smoking by considering the current smoking status of all members of each dwelling unit.
Finally, we estimated a separate model for children that also measured the effects of the biological parents' smoking. Children constitute a substantial fraction of all Medicaid expenditures, and there are well-known clinical links between the smoking history of mothers and fathers and illness in their children. The only other damages calculation in the Medicaid litigation to include excess health expenditures for children was provided by Jeffrey Harris in the Florida case, and it was restricted to damages from specific respiratory conditions.6
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Second, the smoking attributable fraction (SAF) of expenditures was calculated for each individual as follows: SAF = (EXPS – EXPNS)/EXPS, where EXPS is the predicted level of Medicaid expenditures for an individual, given his or her actual smoking history, and EXPNS is the predicted level of Medicaid expenditures if smoking were nonexistent. By equation construction, the SAF for a nonsmoker is zero. Finally, SAFs were applied to annual Medicaid expenditures to obtain the dollar amount of these expenditures attributable to smoking. Such "excess" expenditures constituted the damages that were the focus of the litigation.
We estimated the SAF with data from the NMES undertaken by the US Department of Health and Human Services in 1987, the most recent version of this survey for which complete results were available at the time damages reports were prepared for litigation. The Medical Expenditure Panel Survey of 1996 is now available, but the 1987 NMES represented a better choice for our purpose because it was completed in the middle of the period under consideration (1965–2008).
One advantage of our econometric approach, which is now widely used in the public health literature,1–3,5,7 is that it directly estimates the SAF for medical expenditures rather than inferring it from an SAF for mortality or an SAF for utilization. Another advantage is that it makes no a priori judgments as to which medical conditions involve expenditures that are attributable to smoking, effectively "letting the data speak." Earlier traditions, such as the mortality risk approach, relied entirely on the best available clinical and epidemiological evidence that linked smoking causally to specific medical conditions, such as cancers and respiratory problems. More recent evidence points to a much broader medical impact of smoking.8
The expected levels of expenditures used to calculate the SAF represented the product of 2 terms: the probability that an expenditure will occur and the level of expenditure if it does occur. Decomposing the expected levels of expenditures into these 2 components allows for the possibility that various characteristics have different effects on the probability of incurring an expenditure than they do on the level of expenditure once it occurs. Similar 2-equation "hurdle" specifications9–11 are common in the health economics literature.
In our specification, the first component is determined via a regression equation relating the probability of positive Medicaid expenditures to various individual characteristics. All Medicaid-eligible individuals in the NMES are included in the estimation of this equation, after deletion of some observations owing to missing values. The second component is determined via a regression equation relating the dollar amount of expenditures to individual characteristics, conditional on the individual's incurring positive Medicaid expenditures in 1987. Details of the estimation procedure, data set, and regression results are provided in an appendix available at http://dmsweb.moore.sc.edu/glenn/tobacco/ajph.htm.
Measures of Smoking Intensity
Eight measures of smoking were included among the individual characteristics used in the regressions. Smokers were identified as those who reported having smoked 100 or more cigarettes in their lives, and the model included number of years smoked, past smoking intensity, whether the individual currently smoked, and current smoking intensity. Smoking intensity was measured via typical number of cigarettes smoked per day. In the case of all variables in the regressions that were not dichotomous, we included squared values to capture possible nonlinear effects.
Other Explanatory Variables
We also included variables adjusting for factors other than smoking that might cause the health expenditures of smokers and nonsmokers to differ. Measures of income, education, risk attitude, and health status were included to control for differences in health care use not due to smoking. Measures of health status and risk attitude included body mass index and indicator variables for the following: medical treatment for drug use, medical treatment for alcohol use, pregnancy, regular exercise, regular seat belt use, and consideration of oneself as above average in terms of risk taking. Other variables included age, sex, race, census region, and indicators of whether the individual was married, employed, a veteran, a homeowner, a resident of an inner city, a resident of a suburb, or an English speaker. Measures of income and education were also included.
Passive Smoking
A notable feature of the NMES is that it provides information on all members of each family covered, as well as information on all members of each dwelling unit. Hence, we were able to identify the current smoking activities of everyone in the dwelling unit and thus to attribute some of the expenditures of other individuals living in that dwelling unit to the "passive" effects of smoking.
Children and Teenagers
Separate equations were estimated for adults 18 years or older and for children and teenagers up to the age of 13 years. For the latter, current and past smoking intensity of the biological father and mother were included in addition to the measures of passive smoking just described. Because the NMES contains no information on the smoking behavior of individuals younger than 18 years, we were unable to capture the possible effects of these individuals' smoking on Medicaid expenditures. The SAF for youths aged 14 to 17 years was assumed to be zero. This assumption created a conservative bias in estimated SAF values for children and teenagers. To allow for the possible confounding effect of alcohol abuse among mothers, we included a binary indicator for mothers who were receiving treatment associated with alcohol abuse. Because of smaller sample sizes, we used all medical expenditures rather than simply Medicaid expenditures when estimating the model for children.
Simulating the Absence of Smoking
Given the estimates from the 2 terms described earlier, one involving the probability that an expenditure will occur and the other involving the level of the expenditure if it occurs, we estimated the SAF for each individual in the NMES. To estimate EXPS, the expected level of Medicaid expenditures given the observed smoking history of the individual and all other individuals whose smoking affects expenditures, we used the estimated regression coefficients and the observed smoking history of everyone in the sample. To estimate EXPNS, the expected level of Medicaid expenditures in the absence of smoking, we assumed that all individuals had never smoked and did not currently smoke. By setting all smoking variables to zero, we were able to predict the probability of an expenditure's being incurred as well as the expected expenditure (conditional on expenditures being positive) for each individual had there been no history of smoking.
Disease-Specific Calculations
Earlier studies estimating smokingattributable expenditures often restricted attention to specific diseases or conditions.12 Our broader approach, examining all expenditures viewed as being attributable to smoking, was made possible by the availability of the NMES database. Inclusion of these expenditures was appropriate, because smokers may have higher treatment costs for conditions not caused by smoking. For example, if smoking generates a degree of heightened surgical risk owing to the increased probability of anesthesia complications, then any smoker who undergoes surgery could have higher costs as a result of smoking.
It is possible to determine which damages arise from the disease-specific pathway (smoking causes the disease, which causes the medical intervention, which in turn causes the extra medical cost) and which damages arise from other pathways (smoking causes greater medical care use or cost, even if it did not cause the original medical intervention). We illustrate this breakdown, which played an important role in the litigation by explicitly linking the medical testimony on causality and liability to the testimony on economic damages.
Because the NMES includes codes from the International Classification of Diseases, Ninth Revision (ICD-9), it was possible to identify expenditures by disease code. We were then able to modify the damages calculation to estimate a smoking-disease-specific SAF by including only those expenditures for diseases that could be viewed as directly attributable to smoking, relying on the best epidemiological and clinical evidence. We chose these categories of diseases and ICD-9 codes on the basis of the advice of medical experts who were to testify in the trial taking place in Hawaii. Among adults, these diseases included certain cancers, cardiovascular and cerebrovascular disease, certain respiratory conditions, ulcers, and pregnancy complications; among children and teenagers, the diseases included were limited to asthma, ear infections, and respiratory infections.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Finally, we obtained a weighted SAF for Medicaid by averaging the SAF for adults with the SAF for children and teenagers and then weighting the result by their respective expenditure shares in the NMES. The adult SAF for Medicaid was 16.3%, and the SAF for children and teenagers was 1.7%. Because the adult share of Medicaid expenditures was 85%, the resulting overall weighted SAF for Medicaid was 14.1%. To estimate annual smoking-attributable Medicaid expenditures in dollars, we then multiplied the SAF by total annual Medicaid expenditures reported by the Health Care Financing Authority.
To illustrate the magnitude of diseasespecific expenditures relative to general medical expenditures, we turn to 1 particular state for which these more detailed computations were finalized before the Master Settlement Agreement. The smoking-disease-specific SAF in Hawaii for those patients whose claims were paid by Medicaid was estimated to be 27%, higher than the overall SAF for Hawaii. This result was expected, given that we restricted attention to disease codes for which we had strong a priori beliefs that smoking was a causal factor.
When we applied the smoking-diseasespecific SAF to the Medicaid expenditures of Hawaii associated with the identified ICD-9 codes, we found that disease-specific damages constituted 65% of overall smokingattributable Medicaid expenditures.13 Thus, approximately one third of the overall smokingattributable Medicaid expenditures in Hawaii can be viewed as due to increased treatment costs of conditions not caused by smoking.
Table 1
shows annual smoking-attributable expenditures by Medicaid and by other state and local plans that were covered in the settlement agreement. In terms of historical dollars, the total damages ranged from a low of $0.7 billion in 1965 to a high of $34.8 billion in 1998.
|
Cutler et al.4 used the National Health Interview Survey, and a different methodology, to estimate a Medicaid SAF of 5.6% for Massachusetts. They used a 2-stage regression model for adult utilization and employed nonregression methods to ascertain extra utilization due to low-birthweight babies and nursing home admissions. Their statistical analysis focused on the calculation of a utilization SAF estimating the additional inpatient days, but not the additional treatment costs per day, attributable to smoking. This utilization SAF was then used as a proxy for an expenditure SAF. With the exception of the increased utilization associated with low-birthweight infants, they did not include children in their analyses. They also used a disease-specific approach in their analysis of nursing home utilization.
Aggregate Damages
To evaluate whether the settlement amounts were adequate, we converted all calculated damage amounts to 1998 dollars. We compounded historical damages by using the opportunity cost of the funds to the responsible party.14,15 This rate is the weighted average borrowing rate on corporate bonds of the firms named in the settlement. The final column of Table 1
shows the effects of compounding at the rates applicable to each year.
In addition to past damages, the states also relinquished their right to sue in the future for damages associated with tobacco products. To evaluate the settlement in terms of the damages, it is thus necessary to consider future medical costs. In fact, the damages calculations prepared for the trials in virtually all states included an allowance for future excess health expenditures to be incurred to treat smoking-related illnesses due to past smoking behavior.
Admittedly, quantification of future costs is relatively difficult, because future smoking prevalence is unknown. However, if we assume a continuation of historical trends in smoking and medical expenditures for 10 years into the future (with excess costs set to zero after 10 years), which was the assumption used in much of the litigation, we can estimate future costs with enough precision to evaluate the terms of the Master Settlement Agreement. Because the payments for the settlement of these future costs are guaranteed, we used the US Treasury bill rate in converting values to 1998 dollars.
Once we converted past damages (over the 1965–1998 period) to 1998 dollars and converted future damages (over the 1999–2008 period) to 1998 dollars, we summed these amounts to determine aggregate damages. This calculation yielded past damages of $1033 billion and future damages of $371 billion. Total estimated damages, in terms of 1998 dollars, were $1404 billion.
Comparison of Damages With the Settlement
Before comparing the terms of the settlement with the damages, we take into account 2 adjustment factors that will affect the stream of future payments. The settlement payments will be adjusted upward to reflect inflation but will be adjusted downward if cigarette sales fall. Beginning with payments made in 2000, the Master Settlement Agreement stipulated that all future payments would be adjusted upward annually to reflect inflation. This adjustment will be determined via the change in the consumer price index relative to the January 1999 level. If this change is less than 3% in any year, 3% will be used as the adjustment factor.
The Master Settlement Agreement also includes a provision for future payments to be adjusted to reflect the aggregate number of cigarettes shipped in the United States, with volume changes measured relative to 1997 levels. This provision specifies that payments will increase by 1% for each percentage point increase in volume and decrease by 0.98% for each percentage point decline in volume. Given the new marketing and advertising restrictions placed on the industry, as well as the increase in antismoking campaigns and recent increases in cigarette prices, it is likely that the volume adjustments will result in decreased future payments.
To estimate the present value of the settlement, we made 3 conservative assumptions. First, we assumed that the volume of cigarette sales would remain constant, so that there would be no downward adjustment in payments. Second, we assumed that inflation would be, on average, 3% and adjusted each year's payment upward by 3%. Finally, we used a low real discount rate of 2%. This brought the present value of all payments (including the "up front" payments due in 1998 and in 2000 through 2003, along with the annual payments beginning on April 15, 2000) to $417 billion.
Although the dollar amount of the settlement appears staggering, one should bear in mind that it represents only 40% of past damages and 30% of total damages. Moreover, if the prevalence of smoking continues to decline, the percentage of total damages recovered will be even lower. For example, if we assume that smoking declines, such that the volume of cigarettes shipped is 5% less in 2000 than in 1997, and then further decreases by 1% for each of the next 20 years, the present value of the settlement, with a 2% real discount rate, is reduced to $332 billion, or just 24% of total damages.
Damages Beyond Medicaid
The same statistical methodology can be applied to calculate an SAF applicable to non-Medicaid medical expenditures. Figure 1 shows the SAFs and aggregate damages for each category of expenditure in billions of 1998 dollars. The Medicare SAF is 9.7%, close to the SAF of 9.4% estimated by Zhang et al.16 for Medicare expenditures in 1993. It is apparent that although Medicaid accounts for about one half of total damages, the other categories are substantial. None have been settled as yet; indeed, there is much debate over the liability of tobacco manufacturers for these costs.
|
Overall, we found that 8.3% of total US expenditures on medical care are attributable to smoking. In a review of the literature, Warner et al.17 concluded that the estimates of 6.5% and 14.4% provided by Miller et al.1,2 are plausible lower and upper bounds for the national SAF. If we consider only adult expenditures, our national SAF estimate of 9.0% falls in the middle of this range. Moreover, even though we used a lower SAF than did Miller et al.,2 we included expenditures for nonadults, and thus our estimate of total US medical expenditures attributable to smoking in 1993 is very close to theirs ($74.5 billion vs $71.8 billion).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Several other costs and benefits not calculated here should also be noted. First, 2 offsets are often mentioned: excise taxes on cigarettes and the so-called "death benefit" (cost savings that accrue to Medicaid, and especially Medicare, for smokers who die young).18,19 In the case of Medicaid, it has been argued that savings on nursing home expenditures may offset some of the increased health costs of smokers. The death benefit offset was considered by many of the judges in the state Medicaid cases; most of those judges had decided that it could not be presented at trial. Hence, including this offset would not be appropriate when evaluating settlement amounts in proportion to the damage amounts that the defendants could claim in court.
Furthermore, damage calculations undertaken for litigation purposes need not necessarily include all factors that might be included in cost–benefit analyses undertaken for policymaking purposes. Although excise taxes and possible death benefits may be relevant to determining the overall effects tobacco has on society, the lawsuits were brought not by "society as a whole" but by specified legal entities. For example, it could be argued that the death benefit primarily saves federal expenditures under Medicare and pension payments.19 For this reason, these savings should not be considered to offset the damages incurred by other entities.
Second, no attempt has been made to factor in the devastating psychosocial costs of reduced life spans and increased morbidity borne by smokers and their families, although such costs would probably dwarf the damages reported here. The question of whether smokers knowingly assumed risk continues to be a key point of contention in this litigation. The costs just described are appropriately considered in punitive damage awards, along with a factual determination of whether the plaintiff class assumed some share of the risk.
Although the settlement represents only a partial recovery of the damages from past smoking, it is a landmark victory against the tobacco industry and regarded by many as a favorable settlement in light of the risk and expense inherent in litigating separate suits in each state. It is also important to note that the Master Settlement Agreement includes significant restrictions on the marketing and advertising efforts of the tobacco companies. Such restrictions, including prohibitions on marketing aimed at youth smokers, use of cartoon characters in advertising, and payments by tobacco companies to have products placed in movies or television shows, would probably not have been implemented had the courts ruled on these cases, because they would violate the tobacco industry's rights to free (commercial) speech.
In addition, outdoor advertising and brand sponsorship of events have been severely restricted. The various restrictions carried differing phase-in dates, but all were effective on or before July 1, 1999. Although these elements of the Master Settlement Agreement do nothing to compensate the states for the costs they have already incurred in treating smoking-related illnesses, they should help to reduce smoking prevalence rates and thus medical expenditures due to smoking in the future. To the extent that future damages are reduced as a result of these restrictions, the settlement may be viewed as more favorable than indicated by our evaluation based on payments relative to damages.
However, the possibility of reduced smoking prevalence rates is a double-edged sword in terms of the settlement. The reduction in smoking is good news from a public health standpoint, but it further reduces the extent to which the states will be compensated for the medical costs they have incurred (and will incur) as a result of past smoking. Because of the volume adjustments included in the settlement, as cigarette sales fall, so will the dollars received in the future.
In complex litigation, it is common for settlements to represent "cents on the dollar." Is the tobacco settlement, which we calculate to be 30 cents on the dollar, a good settlement? The answer to this question depends on how we proceed from this point. The extent of future damages is, in large measure, in the hands of public heath officials, who can use settlement funds to attempt to reduce prevalence rates and hence future damages. States are free to impose excise taxes of their own that seek to cover future costs. For that matter, they could also use excise taxes on future smoking to cover some of the uncompensated costs of past excess Medicaid expenditures.
The main conclusion from our analysis is that the medical and public health communities need to understand what the tobacco settlement represents before victory is declared in the Medicaid settlement. We need to realize that the past costs of tobacco have not yet been reimbursed in the United States, that future costs are not reimbursed at all, and that huge settlement sums or excise taxes must be viewed in relation to the even larger damages to be correctly evaluated. We must be diligent in pressing for such an understanding as future lawsuits arise and settlements are agreed upon.
|
Footnotes |
|---|
Accepted for publication June 7, 2001.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
2. Miller LS, Zhang X, Rice DP, Max W. State estimates of total medical expenditures attributable to cigarette smoking, 1993. Public Health Rep.1998;113:140–151.[Medline]
3. Miller LS, Zhang X, Novotny T, Rice DP, Max W. State estimates of Medicaid expenditures attributable to cigarette smoking, fiscal year 1993. Public Health Rep.1998;113:447–458.[Medline]
4. Cutler DM, Epstein AM, Frank RG, et al. How Good a Deal Was the Tobacco Settlement? Assessing Payments to Massachusetts. Cambridge, Mass: National Bureau of Economic Research; 2000. Working paper 7747.
5. Miller VP, Ernst C, Collin F. Smoking-attributable medical care costs in the USA. Soc Sci Med.1999;48:375–391.
6. Harris JE. Estimates of smoking-attributable Medicaid expenditures in Florida. Available at: http://dmsweb.moore.sc.edu/tobacco/FloridaEstimates.pdf. Accessed August 17, 1997.
7. Bartlett JC, Miller LS, Rice DP, Max W. Medical care expenditures attributable to cigarette smoking: United States, 1993. MMWR Morb Mortal Wkly Rep.1994;43:469–472.[Medline]
8. American Council on Science and Health. Cigarettes: What the Warning Label Doesn’t Tell You. Amherst, NY: Prometheus Books; 1997.
9. Greene WH. Econometric Analysis. 3rd ed. Upper Saddle River, NJ: Prentice Hall; 1997.
10. Cragg J. Some statistical models for limited dependent variables with application to the demand for durable goods. Econometrica.1971;39:829–844.
11. Melenberg D, van Soest A. Parametric and semi-parametric modeling of vacation expenditures. J Appl Econometrics.1996;11:59–76.
12. Centers for Disease Control and Prevention. Medical-care expenditures attributable to cigarette smoking during pregnancy—United States, 1995. JAMA.1997;278:2058–2059.[Medline]
13. Harrison GW. Tobacco damages to the state of Hawaii. Available at: http://dmsweb.moore.sc.edu/tobacco/hireport.pdf. Accessed April 1, 2002.
14. Coller M, Harrison GW. Time value and the expert witness: guidance from the tobacco litigation. J Forensic Accounting. 2001;2:145–160.
15. Pattel JM, Weill RL, Wolfson MA. Accumulating damages in litigation: the roles of uncertainty and interest rates. J Legal Studies. June 1982:341–364.
16. Zhang X, Miller L, Max W, Rice DP. Cost of smoking to the Medicare program, 1993. Health Care Financ Rev.1999;20:179–196.[Medline]
17. Warner KE, Hodgson TA, Carroll CE. Medical costs of smoking in the United States: estimates, their validity, and their implications. Tob Control.1999;9:290–300.
18. Viscusi WK. The government composition of the insurance cost of smoking. J Law Economics.1999;62:575–609.
19. Manning WG, Keeler EB, Newhouse JP, Sloss EM, Wasserman J. The taxes of sin: do smokers and drinkers pay their way? JAMA.1989;261:1604–1609.[Abstract]
This article has been cited by other articles:
|
|
 |
|
  W. Max and T. Tsoukalas Economics on trial: the use and abuse of economic methods in third party tobacco litigation Tob. Control, December 1, 2006; 15(suppl_4): iv77 - iv83. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. A. Sloan, E. S. Carlisle, J. R. Rattliff, and J. Trogdon Determinants of States' Allocations of the Master Settlement Agreement Payments Journal of Health Politics Policy and Law, August 1, 2005; 30(4): 643 - 686. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
