Printer Friendly

Competition and the cigarette TV advertising ban.



Whether advertising increases or decreases competition has been debated for decades. On one hand, Kaldor [1949-50!, Bain [1956!, and others argue that advertising reduces competition by enabling the leading firms in an industry to increase product differentiation. This lowers the existing elasticity of demand for the firm's output and erects barriers to the entry of new firms into the industry. On the other hand, Telser [1964!, Nelson [1975!, and others argue that advertising increases competition by disseminating information about the price and other product attributes more widely among consumers. This increases the demand elasticities facing the existing firms and facilitates the entry of new firms. The results of empirical studies of these alternative theories conflict, and so the debate continues. (1) Empirical studies generally focus on large cross-sectional industry samples and compare various measures of competition (e.g. profits, industrial concentration, or market share stability) between high-advertising consumer-goods industries and low-advertising producer-goods industries. The principal drawback of this approach is that one must control for the many differences among industries in addition to advertising which might affect the proxies for competition.

The present paper takes a different approach to test the same hypotheses. The 1970 ban of cigarette advertising on television can be viewed as a quasi-experiment. I assume that it substantially reduced the efficacy of cigarette advertising, whatever its effects on competition, and test for these effects by comparing industry performance measures before and after the ban. The results, on balance, indicate that the ban reduced competition in the cigarette industry.


In 1970, federal legislation was passed banning the advertising of cigarettes on television and radio, effective 2 January 1971. The ban was the result of increasing evidence assumulated during the previous two decades regarding the adverse effects of smoking on health and was not related to considerations of industry competition. Prior to the ban, the cigarette industry was a heavy user of television advertising (see Table I). From 1963 to 1970, the industry's television (plus radio) advertising-to-sales ratio averaged 7.4 percent. (2) The industry's total advertising-to-sales ratio during this period averaged 9.8 percent, which was among the highest of all industries. (3) Television advertisements accounted for 76 percent of this total. In effect, the ban forced the industry to drop its principal advertising medium virtually overnight.

Of course, the industry could and did substitute advertising in other media for the lost television advertising. However, during the first five years after the ban, increased spending in other media did not offset the elimination of television advertising expenditures. Consequently, the mean total advertising-to-sales ratio fell to 6.7 percent during 1971-75. (4) These advertising substitutes were qualitatively inferior, exacerbating the impact of the spending reductions on efficacy. Advertising intensity did not approach pre-ban levels again until 1976. During 1976-80, the advertising-to-sales ratio averaged 11.0 percent, higher than the mean for the pre-ban years 1963-70. (5) Because the substitutes were inferior to television advertisements, the increased advertising expenditure per dollar of sales does not mean that the market impact of advertising was necessarily greater than for the pre-ban period. (6)

It should also be noted that the ban did not cause a decline in industry sales. It it had, this might have generated its own confounding effects on the performance measures analyzed below. Total cigarette sales rose from about 470 billion units in 1960 to about 533 billion units in 1970 and about 620 billion units in 1980, a 32 percent increase over 1960.


Two previous studies provide some evidence on the competitive impact of the cigarette advertising ban. First, Mitchell and Mulherin [1988! test the hypothesis, developed from the economic theory of regulation, that the ban benefited cigarette firms. (7) Using standard stock market event analysis, they examine monthly returns during a thirty-month window ending in December 1970 for a portfolio containing four of the six major tobacco firms. They find abnormally high, statistically significant, positive returns and attribute this in part to the expectation of reduced competition caused by the ban. Second, Holak and Reddy [1986, 219! study the ban" extrapolate managerial implications" of such public policy actions. Using time series regression analysis, they estimate, inter alia, the ban's impact on a brand loyalty proxy and the price elasticity of demand for a sample of ten major cigarette brands during 1950-79. The implications of their results are mixed. On one hand, they find higher brand loyalty after the ban, a result consistent with the competitive view of advertising. On the other hand, they find higher price elasticities after the ban, which is consistent with the monopoly view.

Other studies of the advertising ban focus on its effectiveness (or lack thereof) in achieving stated public policy goals, such as reducing overall cigarette consumption. (8) They present no evidence on its competitive impact. Calfree [1985! finds that federal restrictions on cigarette advertising content prior to the ban reduced competition among firms in supplying safer cigarettes. However, he does not address the broader competitive questions raised by advertising and discusses the ban only in passing.

The general level of competition in the cigarette industry is analyzed in recent papers by Sullivan [1985! and Ashenfelter and Sullivan [1987!. Both papers examine the 1955-85 period and find that the actual responses of industry prices, sales, and revenues to cigarette excise tax changes are inconsistent with the predictions of the monopoly or perfect cartel model. Neither paper considers the possible effects of advertising, including the television advertising ban, on industry competition. In an earlier paper, Telser [1962! analyzes the effects of cigarette advertising on firm and industry demand. He concludes that advertising was perhaps the principal means of competition among cigarette firms prior to World War II, but that its importance declined during the post-war period (i.e., 1946 to 1959) as product innovation became more important, and as price competition increased after the successful 1946 antitrust case against the major tobacco firms. Schmalensee [1972! also examines the impact of cigarette advertising on demand, in part extending and updating Telser's work and correcting alleged deficiencies therein. Schmalensee's statistical results, as he acknowledges, are weak because of various econometric and data problems. He nevertheless tentatively concludes that "advertising seems unlikely to be a very powerful force in the cigarette industry" [1972, 215!. He does not comment specifically on its competitive impacts.


The premise of this paper is that the ban on television advertising substantially reduced the efficacy of industry advertising, whether its effects be pro- or anti-competitive. I assume firms sell differentiated products (cigarette brands) and that each confronts a downward-sloping demand curve. These firms compete to attract customers away from the others, as well as to retain their old customers and attract new smokers. Reduce (increased) competition among firms means less (more) elastic demand for the individual firm's product and lower (higher) cross-elasticities of demand among firms.

The anti-competitive theory holds that advertising's principal effect is to differentiate products and to create and reinforce brand loyalties, thereby reducing brand switching. Brand and firm demand is made less elastic, and cross-elasticities of demand among brands and firms are reduced by advertising. In contrast, the pro-competitive theory assumes that the principal effect is to increase the information available to consumers regarding the positive attributes of alternative brands, including new brands, thereby increasing brand switching. Brand and firm demand is thus made more elastic, and cross-elasticities of demand among brands and firms are greater than in a world with no advertising. Individual firms, to be sure, advertise to differentiate their brands and increase their sales and profits. However, the simultaneous efforts of these firms to each increase its own market power via advertising results in reduced market power for all, compared to a world with no advertising.

The two competing theories of advertising's impact on competition allow the formation of testable hypotheses using specific measures of industry competition. I select share stability, market-share and industry-concentration levels and trends, profitability, and entry. These have often been used as proxies for competition in studies of advertising and competition, and they can be computed from available data. The use of multiple proxies eliminates the problem of selecting and relying on a single "best" proxy, and, if a consistent pattern emerges across the different measures, allows for more robust results.

I compare the competition proxies for ten years before 1970 (1960-69) and ten years after 1970 (1971-80). The year 1970 is excluded because the drop in expenditures on television advertisements by the cigarette industry in that year (see Table I) might have been related to the passage of the ban. These periods are sufficiently long to provide good estimates of the stability and trend measures. The early period begins late enough (1960) that the competitive impacts (positive or negative) of television's introduction during the 1950s should be fully manifest in the performance measures for that period.

Predications about market-share stability can be derived from each theory's implications

regarding the elasticity of demand facing the firm and cross-elasticities among firms. (9) The market power theory implies that, with television advertising, firm demand is relatively inelastic and cross-elasticities are low. This means that firm share is relatively unresponsive to the market behavior (e.g., price changes) of other firms i.e., firm share is relatively stable. In contrast, the competitive theory predicts relatively elastic firm demand and higher cross-elasticities, implying less stable shares. With the television advertising, more information is available to each consumer about brands not currently consumed. This reduces the ex ante search costs of experimenting with new brands and therefore increases the likelihood of switching.

I use two measures of share stability. The first is the standard deviation of annual market shares for each period. The second is the standard error about a linear market-share time trend (i.e., share regressed on time) during each period. For example, low variation about a trend induced by other factors might be consistent with the market power theory, but would be obscured by the impact of the trend on the standard deviation. Share stability measures are computed at both the firm and brand levels. (10)

Market-share and industry-concentration level and trend predictions can be derived from the market power theory because it argues that advertising-induced brand loyalty and barriers to entry give a relative advantage to established leading firms. Such problems are exacerbated if scale economies in advertising exist. This implies high equilibrium market shares for leading firms (i.e., high industry concentration) relative to those which would exist in a more competitive state, as an Mueller and Hamm [1974!. If the television ban reduces this advantage (i.e., reduces market power), then declining shares for leading firms (i.e., declining concentration) would be expected after the ban as the industry adjusts to a new equilibrium characterized by lower leading-firm shares. (11) In contrast, the competitive theory generates conflicting predictions here. On one hand, simply reversing the market power argument implies increasing large-firm shares and declining small-firm shares after the advertising ban, as in Lynk [1981!. On the other hand, if there are scale economies in advertising, or if advertising promotes economies of scale or scope in manufacturing by expanding the markets of efficient leading firms, as argued by Benham [1972!, then predictions regarding, the ban's effect on leading-firm shares would be the same as under the market power view.

My individual brand and firm market-share trend measure is the coefficient of the independent time variable in a regression of market share on time. I measure market concentration at both the brand and firm level using the Hirfindahl-Hirschman Index (HHI), defined as the sum of the squared market shares of all firms or brands in the market. Concentration trends are determined by the simple linear regression of HHI on time.

The third basic performance measure is profitability. The market power view, as argued by Comanor and Wilson [1967!, maintains that advertising generates increased industry profits. Lower profits are expected after the ban, as advertising-generated monopoly rents are reduced or eliminated by the consequent rise in competition. If, instead, advertising promotes competition, higher profits are expected after the ban as interfirm rivalry is reduced. One might also argue that if advertising does not generate market power and the industry is competitive, economic profits are driven to zero whether television advertising is allowed or not. This implies essentially no change in profit rates.

Industry profitability is measured by the price-cost margin using Census of Manufacturers data for SIC #2111 ("Cigarettes"). Since my test concerns a single industry, the problems with comparing margins among industries noted by Ornstein [1975! and Liebowitz [1982! are avoided.

The last performance measure is entry. A central tenet of the market power theory is that advertising is a barrier to entry, whereas if advertising promotes competition, it will ease market entry, as argued respectively by Bain [1956! and Brozen [1974! among others. Entry is expected to be easier (more difficult) after the ban if advertising promotes monopoly (competition). Since the number and identity of cigarette firms is essentially unchanged from 1960 to 1980 (see below), brand entry only is analyzed. This makes the monopoly prediction somewhat ambiguous. In particular, Schmalensee's [1978! brand proliferation, argument implies that excessive brand entry might be expected prior to the ban as extant firms attempt to create firm entry barriers, and such entry might diminish after the ban. Nevertheless, the prediction under the competitive theory remains unchanged, and increased entry after the ban would conflict with this view.

My first measure of entry is simply a count of the new brands in each period. The second measure is the market share of new brands three years after their introduction, an attempt to account for entry success. Counting a large (small) number of entrants may be misleading if their success rate is low (high). In addition, the market power theory implies that advertising provides large firms with an advantage in introducing new brands. This is tested by comparing the relationship between firm size and brand entry before and after the ban.

The above-mentioned stability in the number and identity of firms in the cigarette industry, which in fact extends back to the 1930s, raises three issues. First, it does not mean that the presence of absence of television advertising had no impact on the ease of entry. It means simply that advertising's impact was not sufficient to actually elicit new firm entry. Since new firm entry necessarily involves new brand entry, ease of brand entry is a proxy for ease of firm entry (subject to the above caveat). Second, such structural stability suggests the presence of other deterrents to entry. I assume that these are constant during the study period. This includes the possible entry-deterring effects of publicity regarding the hazards of smoking and the consequent expected reduction of demand. As Calfee [1985! observes, by the early 1960s, the harmful (e.g., cancer-causing) effects of smoking were widely known. Third, entry or its absence is neither necessary nor sufficient to identify competition or its absence. Rivalry can exist among incumbent firms, and the impact of advertising on such rivalry can be identified using the previously defined non-entry competition proxies.

A final point concerns advertising as capital. If television-generated advertising capital does not depreciate rapidly, its effects might linger for perhaps a few years after the ban, depending on the depreciation rate. (12) For example, the change in firm demand elasticities, which underlies profitability and stability predictions, might occur gradually, rather than (say) in the first year after the ban. (13) Because the preban/post-ban distinction would no longer be sharp, my tests would be less likely to detect changes. Nevertheless, the ten-year post-ban perios should be sufficiently long to contain a non-trivial period following the transition to an insignificant stock of television-generated advertising capital.


The principaldata used in this study are from Maxwell [1986!. He reports annual U.S.sales figures, in physical units (billions of cigarettes), for virtually all cigarette brands sold in the U.S. between 1925 and 1985. The raw data are arranged by brand for each of the six U.S. cigarette manufacturers, with each brand being further broken down by specific brand variations (e.g., Marlboro Lights and Marlboro 100s). To compute brand shares, I aggregate the sales of all variations of the same brand (e.g., Marlboro's share includes all eleven Marlboro brand variations), except when a brand is sold in both filter and non-filter variations. I treat these as separate brands. For example, the Philip Morris (non-filter) share is the sum of Philip Morris (non-filter) Regular and Kings, and does not include Philip Morris filter cigarettes sold briefly during the 1960s. I segregate filters and non-filters because of the strong secular trend away from non-filter cigarettes during the study period.

The brand sample consists os twenty-four brands sold in all years from 1960 to 1980. Table II presents the data used in my analysis of this sample for the 1960-69 and 1970-80 time periods. The first column for each period reports the mean market share, followed by its standard deviation. The sample brands account for 94.4 percent of all cigarette sales during 1960-69, and 83.0 percent during 1971-80. In terms of absolute sales, the sample brands increase from about 477 billion cigarettes sold per year for 1960-69, to about 489 billion for 1971-80. The total share of the eight non-filter brands dropped 23.0 percentage points between the two periods, while the total share of the sixteen filter brands gained 11.6 percentage points. The last two columns for each time period in Table II summarize the results of a simple linear regression of market share on time for each brand. The standard error of the time trend regression estimate is reported, along with the coefficient of the independent time variable and its statistical significance. All eight non-filter brands have statistically significant negative share trends in both periods, while only two of the sixteen filter brands have statistically significant negative trends in both periods.

My firm sample consists of the six firms reported in Maxwell [1986!. They account for at least 99.5 percent of U.S. cigarette sales in each year from 1960 to 1980. Firm shares are calculated using all brands sold by each firm, not just the twenty-four in my brand sample. Table III summarizes the firm data following the format of Table II. American Brands and the Liggett Group show statistically significant share declines during both periods, while Philip Morris has statistically significant share increases. The other firms' shares do not show definite trends over both periods. American and Liggett have the largest non-filter shares during 1960-80, averaging about 65 percent and 31 percent respectively.


Market Share Stability

Table IV presents tests of the difference between means before and after 1970 for the standard deviation of market shares and the standard error of linear market-share time trends for both brands and firms. Between 1960-69 and 1971-80 the mean standard deviation of brand shares drops from 0.82 to 0.50, a difference which is statistically significant at the 5 percent level. Between these periods, the mean standard error of the linear brand-share time trends falls from 0.27 to 0.15. This difference is statistically significant at the 1

percent level. Turning now to the results for firms' shares, the mean standard deviation is essentially unchanged between the two periods (1.62 vs 1.64). However, the mean standard error of the time trends drops from 0.69 to 0.39, a difference which is statistically significant at the 5 percent level. Thus, the evidence suggests that share stability increased after the television ban, supporting the competitive theory of advertising. (14)

Market Share and Concentration Levels

and Trends

The next step is to examine the relation between the individual firm- and brand-share trends and initial shares for each period. Table V presents the results of regressing the coefficient of the simple brand- of firm-share time-trend variables (see Tables II and III) on initial market share and variables which account for the effects of the secular shift toward filter cigarettes during the 1960-80 period. For example, since non-filter brands had relatively high shares in 1960, a negative correlation might be expected between initial shares and trends, independent of advertising's effects. For this reason, brand-share regressions include a non-filter dummy variable (equal to 1 for non-filters and zero otherwise), and firm-share regressions include the average non-filter share of each firm's sales.

Looking now at the regression results, brank-share trends for 1960-69 are negatively correlated (5 percent significance level) with initial share, even after accounting for filter status. This correlation disappears for 1971-80. Thus, small-share brands exhibit relatively better share growth before the television advertising ban, an advantage which disappears after the ban. This evidence is inconsistent with the market power theory. In contrast, initial firm shares and firm-share trends show no correlation in either period, a possible upshot of the small sample size (d.f.=3 for the two-variable equations). This evidence neither supports nor contradicts either theory.

Additional information on market-share trends can be obtained by examining industry-concentration trends. The average firm HHI increases slightly between the two periods (from 2161 to 2229), while the average brand HHI decreases slightly (from 770 to 761), although the difference between either pair of means is not statistically significant. Table VI presents the results of regressing annual brand and firm HHIs on time for the two periods. The brand HHI has a clear negative trend during 1960-69. During 1971-80 the trend is positive although not statistically significant. The firm HHI also have a significant negative trend during 1960-69. During 1971-80, it has a strong positive trend. Thus before the television advertising ban both brand and firm concentration were declining. (15) After the ban, the brand HHI decline disappears, while the firm HHI trend clearly reverses. These results contradict the market power theory.


I turn now to an examination of the industry price-cost margin (PCM) before and after the ban. I define the PCM as follows:

PCM = [VP - (PR + CM + ADV)!/VP

where VP is value of production, PR is payroll, CM is most materials, and ADV is advertising expenditures. (16) Data for VP, PR, and CM are from the Census of Manufacturers and Annual Survey of Manufactures (SIC #2111, "Cigarettes"), while data for ADV are from Table I. Because of missing advertising data, PCMs cannot be calculated prior to 1963. The mean value for the PCM increases from 35.9 percent during 1963-70 to 40.9 percent during 1971-80, a difference which is significant at better than the 1 percent level (t=3.79). Thus, price-cost margins in the cigarette industry are higher after the ban.

The variation in advertising intensity after the ban (see Table I) allows identification of the separate effects of the cessation of television advertisements and overall advertising intensity of the PCM via multiple regression analysis. To this end, I regress the PCM on a ban dummy variable (equal to zero before the ban and one afterwards), the industry advertising-to-sales ratio (ASR), and the real annual percentage GNP growth (GR), a proxy for possible business cycle effects. The resulting OLS estimate is shown in Table VII. (17) The statistically significant negative sign on GR indicates that cigarette industry margins are countercyclical. The coefficient of ASR indicates that the PCM is negatively related to advertising intensity during the sample period. The positive sign on BAN's coefficient indicates that the elimination of television advertising increased the PCM, holding constant advertising intensity. These results are inconsistent with the market power view of advertising, and support instead the competitive theory.


The last step in the analysis is an examination of entry. Since virtually all cigarette sales throughout 1960-80 were made by the same six firms (i.e., no firm entry occurred) the analysis focuses on brand entry. Twenty-one new brands were launched during 1960-69, compared to fourteen during 1970-80. (18) Of these fourteen, thirteen entered during 1975-80, and only on entered during 1971-74 (in 1971). For 1975-80, the entry rate was 2.2 brands per year, about the same as the 2.3 brands per year for 1960-70. To measure entry success, I compute the mean market share of new brands in their third year. For brands entering in 1960-69 this share is 0.60 percent, and is somewhat higher (0.65 percent) for the 1975-80 entrants. The sole 1971-74 entrant (Maverick) lasted but one year. (19) Thus, the ban was followed by a virtual cessation of brand entry which lasted four years, during which advertising intensity was substantially lower than before the ban (see Table I). The renewal of entry in 1975 corresponds to a period of increasing advertising intensity which culminated at levels higher than before the ban. In sum, advertising is associated with brand entry. (20) Implications for firm entry are somewhat ambiguous (see discussion in section IV) however, contrary results would have been inconsistent with the competitive theory of advertising.

The relationship between the firm size and entry is examined using the data in Table VIII, which show the number of new brands and their success (again measured by third-year market share) for each firm ranked by mean market share in each period. If advertising is more advantageous to large firms, this would be reflected in relatively greater success by large firms in bringing new brands to market before the ban. The top two firms account for 40 percent of all new brands in 1960-69 and 43 percent in 1971-80. The correlation between firm market share and average third-year brand share for 1960-69 and 1971-80 is, respectively, +0.26 and +0.51, although neither of these is significantly different from zero (d.f.=4). The mean third-year share for the eight brands introduced by the top two firms in 1960-69 is 0.76 percent, compared to 0.50 percent for the thirteen brands introduced during this period by the next four firms. For 1971-80, the same comparison is, respectively, 0.87 percent (six brands) and 0.40 percent (eight brands). Neither of these differences is statistically significant.

However, this picture changes for the 1971-80 period if two outliners are ignored. Without second-ranked Phillip Morris' Merit (third-year share equals 2.93 percent) and fifth-ranked Lorillard's Golden Lights (third-year share equals 2.38 percent), the mean third-year new-brand share is 0.46 percent for the top two firms (five brands) and 0.12 percent for the remaining firms (seven brands), a difference which is statistically significant at the 5 percent level (t=2.39). In addition, the simple correlation between firm share and the average third-year share of new brands is +0.86, statistically significant at the 1 percent levl (t=3.51). Accordingly, there is weak evidence that a large-firm advantage existed after the ban, rather than before.


The premise of this paper is that the television advertising ban had a non-trivial impact on competition, either adverse or beneficial depending upon the relation between advertising and competition. The ban therefore constitutes a quasi-experiment of the effects of advertising on competition. My tests consist of pre- and post-ban comparisons of (1) firm and brand market-share stability, (2) firm and brand market-share and industry-concentration levels and trends, (3) industry price-cost margins, and (4) new brand entry. The results indicate that, first, brand and firm shares are more stable after the ban, net of trend effects. Second, leading-brand shares and brand and firm concentration were declining before the ban, and are stable or increasing after the ban. Third, the ban appears to have increased profit margins, and these margins are negatively related to advertising intensity during the study period. Finally, new brand entry virtually ceased during the first four years after the ban, resuming only in conjunction with a substantial increase in advertising intensity. Also, brands introduced by leading firms may have been relatively more successful after the ban. On balance these results are inconsistent with the market power theory of advertising. The evidence suggests that, whatever other impacts the television ban has had, its unintended side-effect on cigarette industry competition has likely been negative.

(*1) Associate Professor, Graduate School of Business Administration, University of Colorado at Denver. I have received helpful comments from Rich Foster, Sue Keaveney, and two anonymous referees. Remaining errors aremy own. I would also like to thank the Graduate School of Business Administration, University of Colorado at Denver, for financial support.

(1.) See Scherer [1980!, Comanor and Wilson [1979!, and Eckard [1987 1988! for reviews of these studies.

(2.) Radio advertising was relatively minor, less than 8 percent of television advertising. Henceforth my references to television advertising imply television plus radio advertising.

(3.) The mean advertising-to-sales ratio for a sample of 228 SIC four-digit industries in 1967 was 1.76 percent (Eckard [1987!, Table I). In that year, only seven of these industries had ratios higher than 9.8 percent.

(4.) The difference between mean advertising-to-sales ratios for 1963-70 and 1971-75 is statistically significant at the 1 percent level.

(5.) The difference between mean advertising-to-sales ratios for 1963-70 and 1976-80 is statistically significant at the 1 percent level.

(6.) Schneider et al. [1981! suggest that advertising intensity increased in the late 1970s because of the introduction of new low-tar brands. Indeed, new brand entry also increased in the late 1970s (see below). There appear to be no regulatory changes that would explain the increase in advertising. Also, THE FTC advertising data indicate that the increase was not concentrated in one particular medium, suggesting that the increase was not caused by a particular technological innovation.

(7.) The cigarette firms apparently favored the ban. Doran [1979, 95! reports that the Tobacco Institute Executive Committee, representing the industry, offered to voluntarily end television advertising by September 1970, conditional upon a grant of antitrust immunity. The industry was also concerned about the possible adverse impact on sales of anti-smoking messages aired between 1968 and 1970 under the Fairness Doctrine. These messages would (and did) discontinue after the cessation of cigarette television ads.

(8.) These include Hamilton [1972!, Doran [1979!, Warner [1979!, and Schneider et al [1981!.

(9.) See Eckard [1987! for a more thorough discussion of the relevant theory and a review of previous advertising/competition studies using both firm and brand market-share stability measures as proxies for competition. This approach is generally attributed to Telser [1964!.

(10.) Calfee [1985! notes that cigarette "health scare" advertising during the 1950s increased brand switching. However, he argues that this effect ceased before 1960 as regulations prevented firms from making health-related claims and counterclaims. Accordingly, regulatory changes did not have major exogenous impacts on share stability measures during the 1960s.

(11.) Conflicting evidence on this general point exists. Lynk [1981! argues that the introduction of television advertising reduced concentration and presents evidence that this impact persisted into the early 1960s. In contrast, Mueller and Rogers [1980 1984! argue that the impact was positive and persisted into the late 1970s. Each of these studies uses conventional concentration ratios and examines relatively large (albeit different) samples of industries. See Eckard [1988! for evidence that rising industry concentration may be a poor proxy for alleged adverse consumer welfare impacts of advertising.

(12.) Thomas [1989! estimates that advertising capital depreciates rapidly, basing her conclusion in part on an analysis of cigarette advertising. See also Comanor and Wilson [1979!.

(13.) the main exception to this would be the impact on entry under the competitive theory, i.e., the ban should have an immediate impact here. In fact, new brands denied the entry-facilitating mechanism of television advertising might actually be at greater disadvantage initially because extant brands have undepreciated stocks of television advertising capital. As this stock depreciates, the disadvantage of new entrants declines (but does not disappear). Schneider et al. [1981! use this argument to explain the virtual cessation of new brand entry during the first four years after the ban.

(14.) If the trend toward filtered cigarettes was nonlinear, and if most of this shift occurred during 1960-69 (note that the non-filter brand linear trend coefficients in Table II are all lower in absolute value for 1971-80), then the linear trend-based standard error results could be spurious i.e., the difference between the standard errors computed from the linear trends could be exaggerated. This possibility exists at both the brand and firm levels. To test for this, I recomputed the trends in each period adding a time-squared term. The mean standard erros from this "quadratic trend" for the brands are 0.18 for 1960-69 and 0.09 for 1971-80. For the firms these means are 0.52 and 0.26, respectively. The brand mean difference is statistically significant at the 1 percent level, while the firm mean difference is significant at the 5 percent level. Relative to the linear trend results, the differences between

mean standard errors for brands and firms drop from 0.12 to 0.09 and from 0.30 to 0.26, respectively. Thus, while the linear trend somewhat exaggerates the differences between mean standard errors between the two periods, the differences remain statistically significant.

(15.) Under the competitive theory, this trend could be explained as a continuation during the 1960s of an adjustment to lower industry concentration from the introduction of television advertising in the 1950s, as suggested in Lynk [1981!.

In addition, it could be argued that the secular shift away from non-filter cigarettes (beginning circa 1950) is responsible, as high share non-filter brands continued to decline during the 1960s. But this would not necessarily imply a decline in HHI. The impact on HHI depends on how the consequent filter share increase is distributed among individual brands. For example, if it is concentrated in the top two or three brands, HHI could increase. As it happens, of the top three filter brands during 1960-69, Winston's share increased, Kent's decreased, and Salem's was stable. The 1960 market share (rank) was 11.2 percent (2nd) for Winston, 8.1 percent (4th) for Kent, and 7.5 percent (5th) for Salem.

(16.) Domowitz et al. [1986! suggest adjusting the PCM to remove the spurious effects of inventory fluctuations. However, this is not necessary here because the Census reports value of production for SIC #2111, not value of shipments. Note also that the PCM is defined net of advertising expenditures. Otherwise, as noted by Scherer [1980! and Liebowitz [1982!, given Census variable definitions the PCM would implicitly contain the advertising-to-sales ratio, creating a spurious positive correlation between the two variables.

(17.) Annual percentage GNP growth rares are from the 1987 Economic Report of the President. Industry ASRs are from Table I.

(18.) There was no pre-ban entry rush. Four new brands were introduced in 1970, two in 1969 and one in 1968. Four brands also entered in 1964 and in 1966. The annual entry rate for 1968-70 equaled that for 1960-67.

(19.) The mean third-year share for the four 1970 entrants was 0.34 percent.

(20.) The simple correlation between advertising intensity and the number of new brands per year during 1963-80 is +0.38 (p=.06).


Ashenfelter, Orley and Daniel Sullivan. "Non Parametric Tests of Market Structure: An Application to the Cigarette Industry." The Journal of Industrial Economics, June 1987, 483-98.

Bain, Joe S. Barriers to New Competition. Cambridge, Mass.: Harvard University Press, 1956.

Benham, Lee. "The Effect of Advertising on the Price of Eyeglasses." Journal of Law and Economics, October 1972, 337-52.

Brozen, Yale. "Is Advertising a Barrier to Entry?" in Advertising and Society, edited by Yale Brozen. New York: New York University Press, 1974, 79-109.

Calfee, John E. "Cigarette Advertising, Health Information, and Regulation Before 1970." Working Paper No. 134, Federal Trade Commission, Washington, D.C., 1985.

Comanor, William S. and Thomas A. Wilson. "Advertising, Market Structure and Performance." The Review of Economics and Statistics, November 1967, 423-40.

_____. "The Effect of Advertising on Competition: A Survey." Jounal of Economic Literature, June 1979, 453-76.

Domowitz, Ian, R. Glenn Hubbard, and Bruce C. Petersen. "The Intertemporal Stability of the Concentration-Margins Relationship." The Journal of Industrial Economics, September 1986, 13-34.

Doran, G. The Smoking Paradox. Cambridge, Mass.: Abt Books, 1979.

Eckard, E. Woodrow, Jr. "Advertising, Competition, and Market Share Instability." The Journal of Business, October 1987, 539-52.

_____. "Advertising, Concentration Changes, and Consumer Welfare." The Review of Economics and Statistics, May 1988, 340-3.

Federal Trade Commission. Report to Congress Pursuant to the Federal Cigarette Labeling and Advertising Act-1984. Washington, D.C., 1986.

Hamilton, James L. "The Demand for Cigarettes: Advertising, the Health Scare, and the Cigarette Advertising Ban." The Review of Economics and Statistics, November 1972, 401-11.

Holak, Susan L. and Srinivas K. Reddy. "Effects of a Television and Radio Advertising Ban: A Study of the Cigarette Industry." Journal of Marketing, October 1986, 219-27.

Kaldor, Nicholas. "The Economic Aspects of Advertising." The Review of Economic Studies 18(1), 1949-50, 1-27.

Liebowitz, Stanley J. "What Do Census Price-Cost Margins Measure?" Journal of Law and Economics, October 1982, 231-46.

Lynk, William J. "Information, Advertising, and the Structure of the Market." The Journal of Business, April 1981, 271-303.

Maxwell, John C., Jr. "Historical Sales Trends in the Cigarette Industry: A Statistical Summary Covering 61 Years (1925-85)." Furman Selz Mager Dietz & Birney, Inc., Members New York Stock Exchange, 1986.

Mitchell, Mark L. and J. Harold Mulherin. "Finessing the Political System: The Cigarette Advertising Ban." Southern Economic Journal, April 1988, 855-62.

Mueller, Willard F. and Larry G. Hamm. "Trends in Industrial Market Concentration, 1947 to 1970." The Review of Economics and Statistics, November 1974, 511-20.

Mueller, Willard F. and Richard T. Rogers. "The Role of Advertising in Changing Concentration of Manufacturing Industries." The Review of Economics and Statistics, February 1980, 89-96.

_____. "Changes in Market Concentration of Manufacturing Industries, 1947-77." Review of Industrial Organization, Spring 1984, 1-14.

Nelson, Phillip. "The Economic Consequences of Advertising." The Journal of Business, April 1975, 213-41.

Ornstein, Stanley I. "Emperical Uses of the Price-Cost Margin." The Journal of Industrial Economics, December 1975, 105-17.

Scherer, Frederick M. Industrial Market Structure and Economic Performance, 2nd ed. Boston: Houghton Mifflin Company, 1980.

Schmalensee, Richard. The Economics of Advertising. London: North-Holland Publishing Co., 1972, Chapters 5-6.

_____. "Entry Deterrence in the Ready-to-Eat Breakfast Cereal Industry." The Bell Journal of Economics, Autumn 1978, 305-27.

Schneider, Lynne, Benjamin Klein, and Kevin M. Murphy. "Governmental Regulation of Cigarette Health Information." Journal of Law and Economics, December 1981, 575-612.

Sullivan, Daniel. "Testing Hypotheses about Firm Behavior in the Cigarette Industry." Journal of Political Economy, June 1985, 586-98.

Telser, Lester G. "Advertising and Cigarettes." Journal of Political Economy, October 1962, 471-99.

"Advertising and Competition." Journal of Political Economy, December 1964, 537-58.

Thomas, Lacy G. "Advertising in Consumer Goods Industries: Durability, Economies of Scale, and Heterogeneity." The Journal of Law and Economics, April 1989, 163-94.

Warner, Kenneth E. "Clearing the Airwaves: The Cigarette Ad Ban Revisited." Policy Analysis, Fall 1979, 435-50.
COPYRIGHT 1991 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Eckard, E. Woodrow, Jr.
Publication:Economic Inquiry
Date:Jan 1, 1991
Previous Article:A note on risk aversion and preferences.
Next Article:Firm participation in steel industry lobbying.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters