Tobacco Documents Online

'R " J. To: From: Su bj ect: PHILIP MORRIS U. S. A. INTER - OFFICE C ORRESPONDENCE RICHMOND, VIRGINIA Mr. Harry Daniel Date: January 5, 1982 Myron Johnston Cigarette Price Elasticities and the Implications for Philip Morris In an earlier memo ("Teenage Smoking and the Federal Excise Tax on Cigarettes," September 17, 1981) I reported on a National Bureau of Economic Research (NBER) working paper, "The Effect of Government Regulation on Teenage Smokin.g." In that study the authors used the recently released tapes of the 1966-1970 Health Examination Survey of the National Center for Health Statistics and constructed a rather elegant cross-sectional and longitudinal model of teenage smoking behavior. They concluded that the smoking participation elasticity of teenagers is -1.2 and that the total demand elasticity for teenagers is -1.4. This means that a ten percent increase in the price of cigarettes would lead to a decline of 12 percent in the number of teenagers who would otherwise begin to smoke and a 14 percent decline in the total number of cigarettes demanded by teenagers. I have just finished examining another paper by two of the same authors entitled "The Potential for Using Excise Taxes to Reduce Smoking."* On the basis of the findings of this paper, population data, and data from the Public Health Service (PHS) tapes, the Tracking Study, and the POL National Panel, I believe that cigarette price increases, whether of our own doing or as a result of excise tax increases, will have.less effect on Philip Morris sales than the price elasticities alone would lead one to believe. Details of the paper and the reasons for my relative optimism are described below. The Paper in Brief In this study the authors used the recently released 1976 Health Interview Survey tapes, and, using essentially the same procedure used in the earlier paper, constructed the best model I have seen to date from which to calculate the price and income elasticities of cigarettes. With ten other variables held constant, they calculate the overall price elasticity of cigarettes to be -0.42, which is about what the consensus has long held it to be. Thus, a ten percent increase in the price of cigarettes would lead to a 4.2 percent decline in sales. They found the income elasticity to be 0.08, which means that a ten percent increase in income would lead to slightly less than a one perce~t~ ~~~~ increase in the demand for cigarettes. `- `e ` JAN~' *This paper has subsequently been published as NBER working paper #764. It was supported by a.grant from the National Center for Health Services Research to the New Jersey Medical School. The authors are Eugene Lewit, Office of Primary Health Care Education, New Jersey Medical School, and Douglas Coate of the Department of Economics at Rutgers University and the NBER.

2 This paper is undoubtedly the best work I have seen to date on price and income elasticities of cigarettes, although, in fairness to other researchers (myself included) who have wrestled with this problem, it should be noted that the authors were privy to information not accessible to other investigators: The PHS public use tapes made available to the general public have been purged of all information that would compromise the confidentiality of the data (e.g. name and address). The authors of this paper, however, working as they did under a PHS grant, had the complete file and, on the basis of the addresses, were able to assign a price per pack figure for all respondents. Price elasticity works in two ways. A change in price can cause a change in the number of people who use a product or in the quantity of a product the users consume, or, usually, both. In the case of cigarettes, the effect is primarily in the smoker participation rate (price elasticity of -0.26). The effect on the quantity of cigarettes demanded by smokers is much smaller (price elasticity of -0.10). Thus a ten percent increase in the price of cigarettes would cause a decline of 2.6 percent in the number of people who would have otherwise begun to smoke, and a decline of one percent in the number of cigarettes purchased by smokers. This relative insensitivity of sales to price changes on the part of current smokers is consistent with the POL National Panel data, which show little variation between states in the number of cigarettes smoked per day by smokers. In the case of income elasticity, the situation is exactly the reverse: Changes in income have a greater effect on the consumption of cigarettes by smokers than on the smoker participation rate. Income elasticity for the smoker participation rate is 0.03, and for the quantity smoked by smokers 0.06. This is almost identical to the 0.065 correlation between income and consump- tion that was calculated for smokers on the POL National Roster, and the 0.061 and 0.057 correlations found for smokers from the two PHS tapes. (Memo, Johnston to Zoler, "Average Daily Cigarette Consumption by Delivery Level and Measures of Socio-Economic Status", August 7, 1981.) Thus changes in income have very little effect on the demand for cigarettes among smokers.and even less effect on the number of people who smoke. Price Elasticities by Age and Sex As mentioned in my September 17 memo, many of us have hypothesized that price elasticities may be different for different demographic or socio-economic groups, and the authors, in their earlier paper, did indeed find that the smoking participation elasticity for teenagers was -1.2 and their total demand elasticity to be -1.4. In this later paper the authors investigate three different age groups: 20-25 year-olds, 26-35 year-olds, and persons over age 35. Results for these three age groups and teenagers are as follows:

Table 1 Total Demand Elasticities and Smoker Participation Elasticities by Age Total Demand Elasticity Smoking Participation Elasticity Teenagers -1.44 -1.19 20-25 year-olds -0.89 -0.74 26-35 year-olds -0.47* -0.44* Ages 36 and over -0.45 -0.15* Total -0.42 -0.26 * Not significantly different from zero. Note that the total demand elasticity for the total population is lower than for all of the subgroups (although the total demand elasticity for the 26-35 year-olds is not significantly different from zero). This is because, with increasing disaggregation of the data, the calculations become less precise and the confidence levels broader. Since the elasticities reported for the total population are more accurate than for the age groups, a better measure of the true elasticities can be obtained by adjusting the elasticities shown in Table 1 to the elasticities for the total population. These results are shown in Table 2. Table 2 Total Demand Elasticities and Smoker Participation Elasticities, Ages 20 and over, Adjusted Total Demand Elasticity Smoking Participation Elasticity 20-25 year-olds -0.72 -0.64 26-35 year-olds -0.38* -0.38* Ages 36 and over -0.36 -0.13* Total -0.42 -0.26 * Not significantly different from zero.

As noted in the earlier memo, price increases would affect the demand for cigarettes among teenagers chiefly through the smoker participation elasticity. That is, the ch i ef effect wou l d be to i nf l uence the dec i s i on as to whether or not to begin smoking, rather than to influence the demand for cigarettes among committed smokers. This appears to be true also of the 20-25 year olds, among whom the total demand elasticity is nearly twice that of the total population, and, as with the teenagers, appears to operate through the decision regarding beginning (or continuing) to smoke, since the smoker participation elasticity accounts for a great proportion of the aggregate demand elasticity. Among the 26-35 year-olds the elasticities do not quite reach significance, but what elasticities do exist appear to operate chiefly through the smoker participation elasticities. I find this rather puzzling in view of the fact that the vast majority of smokers begin to smoke before age 20 and an even greater proportion before age 25. Probably the best explanation for the observed greater effect of smoker participation elasticities than of the demand for cigarettes by smokers among both the 20-25 year-olds and the 26-35 year-olds is to be found in the design of the study itself: This, as noted, was a cross-sectional study using 1976 data, and since the state-to-state price differentials had remained relatively constant for a number of years prior to 1976, the smoker participation elasticities of people in their twenties may be more apparent than real. That is, price may have had its impact on the decision to smoke several years prior to the study, when these people were still in their teens. Among people over age 35 the overall demand elasticity is statistically significant, but it is not determinable whether it operates through the smoker participation elasticity (i.e. quitting) or the demand for cigarettes by smokers (i.e. cutting down). It is unfortunate that the authors did not disaggregate this older group still further. I suspect that the elasticities for people aged 36 to about 50 would have been non-significant, and those for the older ages significant, and it would have been useful to know through which mechanism price has an effect. The authors further disaggregated the data to calculate the elasticities by age and sex. These unadjusted data are shown in Table 3. Table 3 Total Demand Elasticities and Smoker Participation Elasticities by Age and Sex Total Demand Smoking Participation Elasticity Elasticity Males 20-25 -1.40 -1.28 26-35 -0.32* ~~ -0.29* 36 and over -0.66 ~ 1°`~ -0.25* Females 20-25 -0.30*.'/- -0.14* 26-35 -0.58* -0.39* 36 and over -0.12* 0.07* Total -0.42 -0.26 * Not significantly different from zero.

5 The demand for cigarettes by females is generally insensitive to price: Not one of the elasticities for females is significantly different from zero. Among the 20-25 year-old males both the total demand elasticity and the smoking participation elasticities are significant, while the demand for cigarettes by smokers (not shown) is not. Thus total demand for cigarettes for this age group appears to have (or to have had) an effect through the decision regarding beginning to smoke. For males 26 to 35 neither measure is significant, but price does seem to have an effect on total demand for males over 35, and among these males the effect is split between the smoking participation rate and the quantity smoked by smokers. While not directly comparable, this is intuitively consistent with the observation in my August 7 memo that income is more highly correlated with consumption for males than for females. The Significance for Philip Morris ~ The conclusion to be drawn is that any increase in the price of cigarettes will have its greatest effect on the young, and, in particular, on young males. Unfortunately, it is among the young that we have our greatest market penetration, and theoretically price increases should affect Philip Morris to a greater extent than the total industry. For a variety of reasons, however, this may not be the case. For one thing, we have a large and increasing share of female smokers, and females, in turn, constitute an increasing proportion of total smokers. Since females, as noted above, are relatively insensitive to price, those of our brands that have this greatest appeal to females (all of the 100's) would be relatively immune to price increases. Since Marlboro has its greatest appeal to the young, and to young males in particular, it would appear to be particularly vulnerable to price increases. I am inclined to believe, however, that it is less vulnerable than these elas- ticities alone would suggest, thanks largely to the changing age composition of the population and to the fact that the Marlboro market has changed. Consider first the changing age composition of the population. As I have noted on many occasions, the number of 15-19 year-olds peaked in 1976 and is now declining at the rate of about three percent per year. Furthermore, as I noted in my report on young smokers ("Young Smokers--Prevalence, Trends, Implications, and Related Demographic Trends", March 31, 1981), the prevalence of teenage smoking has been declining at an accelerating rate. As a result, teenagers, who accounted for about ten percent of all smokers (and seven percent of sales) in 1974, now account for only about 7.5 percent of smokers and five percent of sales. With the continued decline in the number of teenagers and in the teenage smoking prevalence rates, their importance as a market will continue to decline. Thus the elasticities will operate from an already low and declining base. The number of 20-24 year-olds reached its maximum in 1981 and will now begin to decline. As noted above, it is my opinion that the smoker participation elasticity for this group exerted its

effect in years prior to the survey, and is therefore overstated to a certain extent. This group, too, will decline in importance relative to the total smoking population, even in the absence of any effect of price elasticities, as a result of the decline in the number of 20-24 year-olds and the declining prevalence rates of people moving into that age group. This under-25 group is, of course, much more important to Marlboro than to the total industry, but it is less important now than in 1974: Our share of this market held steady while the number of smokers declined. Conversely, Marlboro's share of the 25-44 year-old smokers has been increasing. Between 1974 and 1981 its share of 25-34 year-old smokers increased from 11.1 percent to 14.1 percent while that of 35-44 year-old smokers rose from 5.8 to seven percent. Meanwhile the total population aged 25-44 increased at an average annual rate of 3.0 percent. As a result, even though smoking prevalence rates among people in this age group declined, people 25 to 44 now account for about 45 percent of all Marlboro Red smokers--up from about 37 percent in 1974-- and, since they have a higher average daily consumption than younger smokers, account for about 50 percent of sales. This group, it should be noted, includes the group in which the price elasticity is not significantly different from zero. Furthermore, the number of 25-44 year-olds will grow at an average annual rate of 2.7 percent over the next five years, and the smoking prevalence in this group, which declined from 44.5 percent in 1974 to 39.2 percent in 1978, has begun to level off. It is, however, the changes that have taken place over the past decade in the educational and occupational characteristics of smokers of Marlboro Red that ~6/ ~f, give me the greatest cause for optimism. Ten years ago Marlboro had its highest market share among college-educated white collar workers, and among students. Now, for all age groups, its market share is lowest among the upper white collar There is in fact an workers and highest among the lower blue collar workers , . , almost perfect inverse relationship between occupational level and Marlboro market penetration. Even among the 18-24 year-olds, blue collar workers are about 50 percent more likely to smoke Marlboro Red than are students. It is also the blue collar workers who are most likely to smoke, least likely to quit smoking, and who have the highest average daily cigarette consumption, while the upper white collar workers are least likely to have ever smoked, most likely to quit, and who have, other things held constant, the lowest average daily consumption. ~ There is also, for both sexes and all age groups, an almost perfect inverse relationship-between Marlboro Red share and educational attainment. Smokers who have not graduated from high school are about twice as likely to smoke Marlboro Red as college graduates. As was the case with occupation, so also with education: The higher the level of educational attainment the greater the probability of never starting smoking, the greater the probability of quitting, and the lower the average daily consumption.

Although perhaps not related to elasticities of demand for cigarettes, there is one other factor that should help insulate us from any adverse effects of an increase in excise taxes on cigarettes: the racial composition of smokers of Philip Morris brands. There have, to my knowledge, been no studies that addressed the question of racial differences in cigarette demand elasticities, but whether because of different price or income elasticities, or some social phenomenon, the fact is that, at least since 1974, smoking prevalence among blacks has declined much more sharply than among whites. This is true of both sexes and virtually all age groups, and appears in both cross-sectional and cohort data. (This was reported for teenagers in my May 1981 report on young smokers, and an examination of racial differences among adults based on the PHS data will appear in a forthcoming memo.) At any rate, for whatever reason, the fact that our market is about 95 percent white will help to ameliorate the ill effects of an excise tax increase, or any other form of price increase. Unfortunately, we know nothing about the relative price (or income) elasticities of demand for cigarettes among different occupational groups or levels of education. What we do know about their relative propensities to smoke, coupled with the changes that have occurred in the age composition of the population and in the characteristics of smokers of our brands, leads me to believe that price increases will have less effect on Philip Morris sales than the bare price elasticities alone would lead one to believe. MEJ:f cc: Max Hausermann Leo Meyer Richard Thomson Jon Zoler Tom Goodale Al Udow Douglas Nelson Peter Meyers Colon Rowe