Freedom of expression has always ranked high on the American scale of values and fundamental rights. This essay addresses regulation of “commercial speech,” which is defined as speech or messages that propose a commercial transaction. Regulation of commercial advertising occurs in several forms, but it is often controversial. In 1938, the Federal Trade Commission (FTC) was given the authority to regulate “unfair or deceptive” advertising. Congressional hearings were first held in 1939 on proposals to ban radio advertising of alcohol beverages (Russell 1940; U.S. Congress 1939, 1952). Actions by the FTC during 1964-69 led to the 1971 ban of radio and television advertising of cigarettes. In 1997, the distilled spirits industry reversed a six decade-old policy and began using cable television advertising. Numerous groups immediately called for removal of the ads, and Rep. Joseph Kennedy II (D, MA) introduced a “Just Say No” bill that would have banned all alcohol advertisementsfrom the airways. In 1998, the Master Settlement Agreement between that state attorneys general and the tobacco industry put an end to billboard advertising of cigarettes. Do these regulations make any difference for the demand for alcohol or cigarettes? When will an advertising ban increase consumer welfare? What legal standards apply to commercial speech that affect the extent and manner in which governments can restrict advertising?

For many years, the Supreme Court held that the broad powers of government to regulate commerce included the “lesser power” to restrict commercial speech.1 In Valentine (1942), the Court held that the First Amendment does not protect “purely commercial advertising.” This view was applied when the courts upheld the ban of broadcast advertising of cigarettes, 333 F. Supp 582 (1971), affirmed per curiam, 405 U.S. 1000 (1972). However, in the mid-1970s this view began to change as the Court invalidated several state regulations affecting advertising of services and products such as abortion providers and pharmaceutical drugs. In Virginia State Board of Pharmacy (1976), the Court struck down a Virginia law that prohibited the advertising of prices for prescription drugs, and held that the First Amendment protects the right to receive information as well as the right to speak. Responding to the claim that advertising bans improved the public image of pharmacists, Justice Blackmun wrote that “an alternative [exists] to this highly paternalistic approach . . . people will perceive their own best interests if only they are well enough informed, and the best means to that end is to open the channels of communication rather than to close them” (425 U.S. 748, at 770). In support of its change in direction, the Court asserted two main arguments: (1) truthful advertising coveys information that consumers need to make informed choices in a free enterprise economy; and (2) such information is indispensable as to how the economic system should be regulated or governed. In Central Hudson Gas & Electric (1980), the Court refined its approach and laid out a four-prong test for “intermediate” scrutiny of restrictions on commercial speech. First, the message content cannot be misleading and must be concerned with a lawful activity or product. Second, the government’s interest in regulating the speech in question must be substantial. Third, the regulation must directly and materially advance that interest. Fourth, the regulation must be no more extensive than necessary to achieve its goal. That is, there must be a “reasonable fit” between means and ends, with the means narrowly tailored to achieve the desired objective. Applying the third and fourth-prongs, in 44 Liquormart (1996) the Court struck down a Rhode Island law that banned retail price advertising of beverage alcohol. In doing so, the Court made clear that the state’s power to ban alcohol entirely did not include the lesser power to restrict advertising. More recently, in Lorillard Tobacco (2001) the Supreme Court invalidated a state regulation on placement of outdoor and in-store tobacco displays. In summary, Central Hudson requires the use of a “balancing” test to examine censorship of commercial speech. The test weighs the government’s obligations toward freedom of expression with its interest in limiting the content of some advertisements. Reasonable constraints on time, place, and manner are tolerated, and false advertising remains illegal.

This article provides a brief economic history of advertising bans, and uses the basic framework contained in the Central Hudson decision. The first section discusses the economics of advertising and addresses the economic effects that might be expected from regulations that prohibit or restrict advertising. Applying the Central Hudson test, the second section reviews the history and empirical evidence on advertising bans for alcohol beverages. The third section reviews bans of cigarette advertising and discusses the regulatory powers that reside with the Federal Trade Commission as the main government agency with the authority to regulate unfair or deceptive advertising claims.

The Economics of Advertising
Judged by the magnitude of exposures and expenditures, advertising is a vital and important activity. A rule of thumb in the advertising industry is that the average American is exposed to more than 1,000 advertising messages every day, but actively notices fewer than 80 ads. According to Advertising Age (www.adage.com), advertising expenditures in 2002 in all media totaled $237 billion, including $115 billion in 13 measured media. Ads in newspapers accounted for 19.2% of measured spending, followed by network TV (17.3%), magazines (15.6%), spot TV (14.0%), yellow pages (11.9%), and cable/syndicated TV (11.9%). Internet advertising now accounts for about 5.0% of spending. By product category, automobile producers were the largest advertisers ($16 billion of measured media), followed by retailing ($13.5 billion), movies and media ($6 billion), and food, beverages, and candies ($6.0 billion). Beverage alcohol producers ranked 17th ($1.7 billion) and tobacco producers ranked 23rd ($284 million). Among the top 100 advertisers, Anheuser-Busch occupied the 38th spot and Altria Group (which includes Philip Morris) ranked 17th. Total advertising expenditures in 2002 were about 2.3% of U.S. gross domestic product (GDP). Ad spending tends to vary directly with general economy activity as illustrated by spending reductions during the 2000-2001 recession (Wall Street Journal, Aug. 14, 2001; Nov. 28, 2001; Dec. 12, 2001; Apr. 25, 2002). This pro-cyclical feature is contrary to Galbraith’s view that business firms use advertising to control or manage aggregate consumer demand.

National advertising of branded products developed in the early 1900s as increased urbanization and improvements in communication, transportation, and packaging permitted the development of mass markets for branded products (Chandler 1977). In 1900, the advertising-to-GDP ratio was about 3.1% (Simon 1970). The ratio stayed around 3% until 1929, but declined to 2% during the 1930s and has fluctuated around that value since then. The growth of major national industries was associated with increased promotion, although other economic changes often preceded the use of mass media advertising. For example, refrigeration of railroad cars in the late 1870s resulted in national advertising by meat packers in the 1890s (Pope 1983). Around the turn-of-the-century, Sears Roebuck and Montgomery Ward utilized low-cost transportation and mail-order catalogs to develop efficient systems of national distribution of necessities. By 1920 more Americans were living in urban areas than in rural areas. The location of retailers began to change, with a shift first to downtown shopping districts and later to suburban shopping malls. Commercial radio began in 1922, and advertising expenditures grew from $113 million in 1935 to $625 million in 1952. Commercial television was introduced in 1941, but wartime delayed the diffusion of televison. By 1954, half of the households in the U.S. had at least one television set. Expenditures on TV advertising grew rapidly from $454 million in 1952 to $2.5 billion in 1965 (Backman 1968). These changes affected the development of markets — for instance, new products could be introduced more rapidly and the available range of products was enhanced (Borden 1942).

Market Failure: Incomplete and Asymmetric Information
Because it is costly to acquire and process, the information held by buyers and sellers is necessarily incomplete and possibly unequal as well. However, full or “perfect” information is one of the analytical requirements for the proper functioning of competitive markets — so what happens when information is imperfect or unequal? Suppose, for example, that firms charge different prices for identical products, but some consumers (tourists) are ignorant of the dispersion of prices available in the marketplace. For many years, this question was largely ignored by economists, but two contributions sparked a revolution in economic thinking. Stigler (1961) showed that because information is costly to acquire, consumer search for lower prices will be less than complete. As a result, a dispersion of prices can persist and the “law of one price” is violated. The dispersion will be less if the product represents a large expenditure (e.g., autos), since more individual search is supported and suppliers have an extra incentive to promote the product. Because information has public good characteristics, imperfect information provides a rationale for government intervention, but profit-seeking firms also have reasons to reduce search costs through advertising and brand names. Akerlof (1970) took the analysis a step further by focusing on material aspects of a product that are known to the seller, but not by potential buyers. In Akerlof’s “lemons model,” the seller of a used car has private knowledge of defects, but potential buyers have difficulty distinguishing between good used cars (“creampuffs”) and bad used cars (“lemons”). Under these circumstances, Akerlof showed that a market may not exist or only lower-quality products are offered for sale. Hence, asymmetric information can result in market failure, but a reputation for quality can reduce the uncertainty that consumers face due to hidden defects (Akerlof 1970; Richardson 2000; Stigler 1961).

Under some conditions, branding and advertising of products, including targeting of customer groups, can help reduce market imperfections. Because advertising has several purposes or functions, there is always uncertainty regarding its effects. First, advertising may help inform consumers of the existence of products and brands, better inform them about price and quality dimensions, or better match customers and brands (Nelson 1975). Indeed, the basic message in many advertisements is simply that the brand is available. Consumer valuations can reflect a joint product, which is the product itself and the information about it. However, advertising tends to focus on only the positive aspects of a product, and ignores the negatives. In various ways, advertisers sometimes inform consumers that their brand is “less bad” (Calfee 1997b). An advertisement that announces a particular automobile is more crash resistant also is a reminder that all cars are less than perfectly safe. Second, persuasive or “combative” advertising can serve to differentiate one firm’s brand from those of its rivals. As a consequence, a successful advertiser may gain some discretion over the price it charges (“market power”). Furthermore, reactions by rivals may drive industry advertising to excessive levels or beyond the point where net social benefits of advertising are maximized. In other words, excessive advertising may result from the inability of each firm to reduce advertising without similar reductions by its rivals. Because it illustrates a breakdown of desirable coordination, this outcome is an example of the “prisoners’ dilemma game.” Third, the costs of advertising and promotion by existing or incumbent firms can make it more difficult for new firms to enter a market and compete successfully due to an advertising-cost barrier to entry. Investments in customer loyalty or intangible brand equity are largely sunk costs. Smaller incumbents also may be at a disadvantage relative to their larger rivals, and consequently face a “barrier to mobility” within the industry. However, banning advertising can have much the same effect by making it more difficult for smaller firms and entrants to inform customers of the existence of their brands and products. For example, Russian cigarette producers were successful in banning television advertising by new western rivals. Given multiple effects, systematic empirical evidence is needed to help resolve the uncertainties regarding the effects of advertising (Bagwell 2005).

Substantial empirical evidence demonstrates that advertising of prices increases competition and lowers the average market price and variance of prices. Conversely, banning price advertising can have the opposite effect, but consumers might derive information from other sources — such as direct observation and word-of-mouth — or firms can compete more on quality (Kwoka 1984). Bans of price advertising also affect product quality indirectly by making it difficult to inform consumers of price-quality tradeoffs. Products for which empirical evidence demonstrates that advertising reduces the average price include toys, drugs, eyeglasses, optometric services, gasoline, and grocery products. Thus, for relatively homogeneous goods, banning price advertising is expected to increase average prices and make entry more difficult. A partial offset occurs if significant costs of advertising increases product prices.

The effects of a ban of persuasive advertising also are uncertain. In a differentiated product industry, it is possible that advertising expenditures are so large that an advertising ban reduces costs and product prices, thereby offsetting or defeating the purpose of the ban. For products that are well known to consumers (“mature” products), the presumption is that advertising primarily affects brand shares and has little impact on primary demand (Dekimpe and Hanssens 1995; Scherer and Ross 1990). Advertising bans tend to solidify market shares. Furthermore, most advertising bans are less than complete, such as the ban of broadcast advertising of cigarettes. Producers can substitute other media or use other forms of promotion, such as discount coupons, articles of apparel, and event sponsorship. Thus, government limitations on commercial speech for one product or media often lead to additional efforts to limit other promotions. This “slippery slope” effect is illustrated by the Federal Communications Commission’s fairness doctrine for advertising of cigarettes (discussed below).

The Industry Advertising-Sales Response Function
The effect of a given ban on market demand depends importantly on the nature of the relationship between advertising expenditures and aggregate sales. This relationship is referred to as the industry advertising-sales response function. Two questions regarding this function have been debated. First, it is not clear that a well-defined function exists at the industry level, since persuasive advertising primarily affects brand shares. The issue is the spillover, if any, from brand advertising to aggregate (primary) market demand. Two studies of successful brand advertising in the alcohol industry failed to reveal a spillover effect on market demand (Gius 1996; Nelson 2001). Second, if an industry-level response function exists, it should be subject to diminishing marginal returns, but it is unclear where diminishing returns begin (the inflection point) or the magnitude of this effect. Some analysts argue that diminishing returns only begin at high levels of industry advertising, and sharply increasing returns exist at moderate to low levels (Saffer 1993). According to this view, comprehensive bans of advertising will reduce market demand importantly. However, this argument is at odds with empirical evidence for a variety of mature products, which demonstrates diminishing returns over a broad range of outlays (Assmus et al. 1984; Tellis 2004). Simon and Arndt (1980) found that diminishing returns began immediately for a majority of 100-plus products. Furthermore, average advertising elasticities for most mature products are only about 0.1 in magnitude (Sethuraman and Tellis 1991). As a result, limited bans of advertising will not reduce sales of mature products or the effect is likely to be extremely small in magnitude. It is unlikely that elasticities this small could support the third prong of the Central Hudson test.

Suppose that advertising for a particular product convinces some consumers to use Brand X, and this results in more sales of the brand at a higher price. Are consumers better or worse off as a consequence? A shift in consumer preferences toward a fortified brand of breakfast cereal might be described as either a “shift in tastes,” an increase in demand for nutrition, or an increase in joint demand for the cereal and information. Because it concerns individual utility, it is not clear whether a “shift in tastes” reduces or increases consumer satisfaction. Social commentators usually respond that consumers just think they are better off or the demand effect is spurious in nature. Much of the social criticism of advertising is concerned with its pernicious effect on consumer beliefs, tastes, and desires. Vance Packard’s, The Hidden Persuaders (1957), was an early, but possibly misguided, effort along these lines (Rogers 1992). Packard wrote that advertisers can “channel our unthinking habits, our purchasing decisions, and our thought processes by the use of insights gleaned from psychiatry and the social sciences.” Of course, once a “hidden secret” is revealed, such manipulation is less effective in the marketplace for products due to cynicism toward advertisers or outright rejection of the advertising claims.

Dixit and Norman (1978) argued that because profit-maximizing firms tend to over-advertise, small decreases in advertising will raise consumer welfare. In their analysis, this result holds regardless of the change in tastes or what product features are being advertised. Becker and Murphy (1993) responded that advertising is usually a complement to products, so it is unclear that equilibrium prices will always be higher as advertising increases. Further, it does not follow that social welfare is higher without any advertising. Targeting by advertisers also helps to increase the efficiency of advertising and reduces the tendency to waste advertising dollars on uninterested consumers through redundant ads. Nevertheless, this common practice also is criticized by social commentators and regulatory agencies. In summary, the evaluation of advertising bans requires empirical evidence. Much of the evidence on advertising bans is econometric and most of it concerns two products, alcohol beverages and cigarettes.

Advertising Bans: Beverage Alcohol
In an interesting way, the history of alcohol consumption follows the laws of supply and demand. The consumption of ethyl alcohol as a beverage began some 10,000 years ago. Due to the uncertainties of contaminated water supplies in the West, alcohol is believed to have been the most popular and safe daily beverage for centuries (Valle 1998). In the East, boiled water in the form of teas solved the problem of potable beverages. Throughout the Middle Ages, beer and ale were drunk by common folk and wine by the affluent. Following the decline of the Roman Empire, the Catholic Church entered the profitable production of wines. Distillation of alcohol was developed in the Arab world in 700 A.D. and gradually spread to Europe, where distilled spirits were used ineffectively as a cure for plague in the 14th century. During the 17th century, several non-alcohol beverages became popular, including coffee, tea, and cocoa. In the late eighteenth century, religious sentiment turned against alcohol and temperance activity figured prominently in the concerns of the Baptist, Friends, Methodist, Mormon, Presbyterian, and Unitarian churches. It was not until the late nineteenth century that filtration and treatment made safe drinking water supplies more widely available.

During the colonial period, retail alcohol sellers were licensed by states, local courts, or town councils (Byse 1940). Some colonies fixed the number of licenses or bonded the retailer. Fixing of maximum prices by legislatures and the courts encouraged adulteration and misbranding by retailers. In 1829, the state of Maine passed the first local option law and in 1844, the territory of Oregon enacted a general prohibition law. Experimentation with statewide monopoly of the retail sale of alcohol began in 1893 in South Carolina. As early as 1897, federal regulation of labeling was enacted through the Bottling in Bond Act. Following the repeal of Prohibition in 1933, the Federal Alcohol Control Administration was created by executive order (O’Neill 1940). The Administration immediately set about creating “fair trade codes” that governed false and misleading advertising, unfair trade practices, and prices that were “oppressively high or destructively low.” These codes discouraged price and advertising competition, and encouraged shipping expansion by the major midwestern brewers (McGahan 1991). The Administration ceased to function in 1935 when the National Industrial Recovery Act was declared unconstitutional. The passage of Federal Alcohol Administration Act in 1935 created the Federal Alcohol Administration (FAA) within the Treasury Department, which regulated trade practices and enforced the producer permit system required by the Act. In 1939, the FAA was abolished and its duties were transferred to the Alcohol Tax Unit of the Internal Revenue Service (later named the Bureau of Alcohol, Tobacco, and Firearms). The ATF presently administers a broad range of provisions regarding the formulation, labeling, and advertising of alcohol beverages.

Alcohol Advertising: Analytical Methods
Three types of econometric studies examine the effects of advertising on the market demand for beverage alcohol. First, time-series studies examine the relationship between alcohol consumption and annual or quarterly advertising expenditures. Recent examples of such studies include Calfee and Scheraga (1994), Coulson et al. (2001), Duffy (1995, 2001), Lariviere et al. (2000), Lee and Tremblay (1992), and Nelson (1999). All of these studies find that advertising has no effect on total alcohol consumption and small or nonexistent effects on beverage demand (Nelson 2001). This result is not affected by disaggregating advertising to account for different effects by media (Nelson 1999). Second, cross-sectional and panel studies examine the relationship between alcohol consumption and state regulations, such as state bans of billboards. Panel studies combine cross-sectional (e.g., all 50 states) and time-series information (50 states for the period 1980-2000), which alters the amount of variation in the data. Third, cross-national studies examine the relationship between alcohol consumption and advertising bans for a panel of countries. This essay discusses results obtained in the second and third types of studies.

Background: State Regulation of Billboard Advertising
In the United States, the distribution and retail sale of alcohol beverages is regulated by the individual states. The Twenty-First Amendment, passed in 1933, repealed Prohibition and granted the states legal powers over the sale of alcohol, thereby resolving the conflicting interests of “wets” and “drys” (Goff and Anderson 1994; Munger and Schaller 1997; Shipman 1940; Strumpf and Oberholzer-Gee 2000). As a result, alcohol laws vary importantly by state, and these differences represent a natural experiment with regard to the economic effects of regulation. Long-standing differences in state laws potentially affect the organization of the industry and alcohol demand, reflecting incentives that alter or shape individual behaviors. State laws also differ by beverage, suggesting that substitution among beverages is one possible consequence of regulation. For example, state laws for distilled spirits typically are more stringent than similar laws applied to beer and wine. While each state has adopted its own unique regulatory system, several broad categories can be identified. Following repeal, eighteen states adopted public monopoly control of the distribution of distilled spirits. Thirteen of these states operate off-premise retail stores for the sale of spirits, and two states also control retail sales of table wine. In five states, only the wholesale distribution of distilled spirits is controlled. No state has monopolized beer sales, but laws in three states provide for restrictions on private beer sales by alcohol content. In the private license states, an Alcohol Beverage Control (ABC) agency determines the number and type of retail licenses, subject to local wet-dry options. Because monopoly states have broad authority to restrict the marketing of alcohol, the presumption is that total alcohol consumption will be lower in the control states compared to the license states. Monopoly control also raises search costs by restricting outlet numbers, hours of operation, and product variety. Because beer and wine are substitutes or complements for spirits, state monopoly control can increase or decrease total alcohol use, or the net effect may be zero (Benson et al. 1997; Nelson 1990, 2003a).

A second broad experiment includes state regulations banning advertising of alcohol beverages or which restrict the advertising of prices. Following repeal, fourteen states banned billboard advertising of distilled spirits, including seven of the license states. Because the bans have been in existence for many years and change infrequently over time, these regulations provide evidence on the long-term effectiveness of advertising bans. It is often argued that billboards have an important effect on youth behaviors, and this belief has been a basis for municipal ordinances banning billboard advertising of tobacco and alcohol. Given long-standing bans, it might be expected that youth alcohol behaviors will show up as cross-state differences in adult per capita consumption. Indeed, these two variables are highly correlated (Cook and Moore 2000, 2001). Further, fifteen states banned price advertising by retailers using billboards, newspapers, and visible store displays. In general, a ban of price advertising reduces retail competition and increases search costs of consumers. However, these regulations were not intended to advance temperance, but rather were anti-competitive measures obtained by alcohol retailers (McGahan 1995). For example, in 44 Liquormart (1996) the lower court noted that Rhode Island’s ban of price advertising was designed to protect smaller retailers from in-state and out-of-state competition, and was closely monitored by the liquor retailers association. A price advertising ban could reduce alcohol consumption by elevating full prices (search costs plus monetary prices). Because many states banned only price advertising of spirits, substitution among beverages also is a possible outcome.

After peaking in the early 1980s, per capita alcohol consumption is now at roughly the level experienced in the early 1960s. Nationally, the decline in alcohol consumption from 1980 to 2000 was 21.0%. This decline has occurred despite continued high levels of advertising and promotion. At the state-level, the percentage changes in consumption are Illinois, -25.3%; Ohio, -15.5%; and Pennsylvania, -20.5%. Pennsylvania is a state monopoly for spirits and wines and also banned price advertising of alcohol, including beer, prior to 1997. However, the change in per capita consumption in Pennsylvania parallels what has occurred nationally.

Econometric Results: State-Level Studies of Billboard Bans
Seven econometric studies estimate the relationship between state billboard bans and alcohol consumption: Hoadley et al. (1984), Nelson (1990, 2003a), Ornstein and Hanssens (1985), Schweitzer et al. (1983), and Wilkinson (1985, 1987). Two studies used a single year, but the other five employed panel data covering five to 25 years. Two studies estimated demand functions for beer or distilled spirits only, which ignores substitution. None of the studies obtained a statistically significant reduction in total alcohol consumption due to bans of billboards. In several studies, billboard bans increased spirits consumption significantly. A positive effect of a ban is contrary to general expectations, but consistent with various forms of substitution. The study by Nelson (2003a) covered 45 states for the time period 1982-1997. In contrast to earlier studies, Nelson (2003a) focused on substitution among alcohol beverages and the resulting net effect on total ethanol consumption. Several subsamples were examined, including all 45 states, ABC-license states, and two time periods, 1982-1988 and 1989-1997. A number of other variables also were considered, including prices, income, tourism, age demographics, and the minimum drinking age. During both time periods, state billboard bans increased consumption of wine and spirits, and reduced consumption of beer. The net effect on total ethanol consumption was significantly positive during 1982-1988, and insignificant thereafter. During both time periods, bans of price advertising of spirits were associated with lower consumption of spirits, higher consumption of beer, and no effect on wine or total alcohol consumption. The results in this study demonstrate that advertising regulations have different effects by beverage, indicating the importance of substitution. Public policy statements that suggest that limited bans have a singular effect are ignoring market realities. The empirical results in Nelson (2003a) and other studies are consistent with the historic use of billboard bans as a device to suppress competition, with little or no effect on temperance.

Econometric Results: Cross-National Studies of Broadcast Bans
Many Western nations have restrictions on radio and television advertising of alcohol beverages, especially distilled spirits. These controls range from time-of-day restrictions and content guidelines to outright bans of broadcast advertising of all alcohol beverages. Until quite recently, the trend in most countries has been toward stricter rather than more lenient controls. Following repeal, U.S. producers of distilled spirits adopted a voluntary Code of Good Practice that barred radio advertising after 1936 and television advertising after 1948. When this voluntary agreement ended in late 1996, cable television stations began carrying ads for distilled spirits. The major TV networks continued to refuse such commercials. Voluntary or self-regulatory codes also have existed in a number of other countries, including Australia, Belgium, Germany, Italy, and Netherlands. By the end of the 1980s, a number of countries had banned broadcast advertising of spirits, including Austria, Canada, Denmark, Finland, France, Ireland, Norway, Spain, Sweden, and United Kingdom (Brewers Association of Canada 1997).

Advertising Bans: Cigarettes
Prior to 1920, consumption of tobacco in the U.S. was mainly in the form of cigars, pipe tobacco, chewing tobacco, and snuff. It was not until 1923 that cigarette consumption by weight surpassed that of cigars (Forey et al. 2002). Several early developments contributed to the rise of the cigarette (Borden 1942). First, the Bonsak cigarette-making machine was patented in 1880 and perfected in 1884 by James Duke. Second, the federal excise tax on cigarettes, instituted to help pay for the Civil War, was reduced in 1883 from $1.75 to 50 cents a thousand pieces. Third, during World War I, cigarette consumption by soldiers was encouraged by ease of use and low cost. Fourth, the taboo against public smoking by women began to wane, although participation by women remained substantially below that of men. By 1935, about 50% of men smoked compared to only 20% of women. Fifth, advertising has been credited with expanding the market for lighter-blends of tobacco, although evidence in support of this claim is lacking (Tennant 1950). Some early advertising claims were linked to health, such as a 1928 ad for Lucky Strike that emphasized, “No Throat Irritation — No Cough.” During this time, the FTC banned numerous health claims by de-nicotine products and devices, e.g., 10 FTC 465 (1925).

Cigarette advertising has been especially controversial since the early 1950s, reflecting known health risks associated with smoking and the belief that advertising is a causal factor in smoking behaviors. Warning labels on cigarette packages were first proposed in 1955, following new health reports by the American Cancer Society, the British Medical Research Council, and Reader’s Digest (1952). Regulation of cigarette advertising and marketing, especially by the FTC, increased over the years to include content restrictions (1942, 1950-52); advertising guidelines (1955, 1960, 1966); package warning labels (1965, 1970, 1984); product testing and labeling (1967, 1970); public reporting on advertising trends (1964, 1967, 1981); warning messages in advertisements (1970); and advertising bans (1971, 1998). The history of these regulations is discussed below.

Background: Cigarette Prohibition and Early Health Reports
During the 17th century, several of the northern colonies banned public smoking. In 1638, the Plymouth colony passed a law forbidding smoking in the streets and, in 1798, Boston banned the carrying of a lighted pipe or cigar in public. Beginning around 1850, a number of anti-tobacco groups were formed (U.S. Surgeon General 2000), including the American Anti-Tobacco Society in 1849, American Health and Temperance Association (1878), Anti-Cigarette League (1899), Non-Smokers Protective League (1911), and the Department of Narcotics of the Women’s Christian Temperance Union (1883). The WCTU was a force behind the cigarette prohibition movement in Canada and the U.S. During the Progressive Era, fifteen states passed laws prohibiting the sale of cigarettes to adults and another twenty-one states considered such laws (Alston et al. 2002). North Dakota and Iowa were the first states to adopt smoking bans in 1896 and 1897, respectively. In West Virginia, cigarettes were taxed so heavily that they were de facto prohibited. In 1920, Lucy Page Gaston of the WCTU made a bid for the Republican nomination for president on an anti-tobacco platform. However, the movement waned as the laws were largely unenforceable. By 1928, cigarettes were again legal for sale to adults in every state.

As the popularity of cigarette smoking spread, so too did concerns about its health consequences. As a result, the hazards of smoking have long been common knowledge. A number of physicians took early notice of a tobacco-cancer relationship in their patients. In 1912, Isaac Adler published a book on lung cancer that implicated smoking. In 1928, adverse health effects of smoking were reported in the New England Journal of Medicine. A Scientific American report in 1933 tentatively linked cigarette “tars” to lung cancer. Writing in Science in 1938, Raymond Pearl of Johns Hopkins University demonstrated a statistical relationship between smoking and longevity (Pearl 1938). The addictive properties of nicotine were reported in 1942 in the British medical journal, The Lancet. These and other reports attracted little attention from the popular press, although Reader’s Digest (1924, 1941) was an early crusader against smoking. In 1950, three classic scientific papers appeared that linked smoking and lung cancer. Shortly thereafter, major prospective studies began to appear in 1953-54. At this time, the research findings were more widely reported in the popular press (e.g., Time 1953). In 1957, the Public Health Service accepted a causal relationship between smoking and lung cancer (Burney 1959; Joint Report 1957). Between 1950 and 1963, researchers published more than 3,000 articles on the health effects of smoking.

Cigarette Advertising: Analytical Methods
Given the rising concern about the health effects of smoking, it is not surprising that cigarette advertising would come under fire. The ability of advertising to stimulate primary demand is not debated by public health officials, since in their eyes cigarette advertising is inherently deceptive. The econometric evidence is much less clear. Three methods are used to assess the relationship between cigarette consumption and advertising. First, time-series studies examine the relationship between cigarette consumption and annual or quarterly advertising expenditures. These studies have been reviewed several times, including comprehensive surveys by Cameron (1998), Duffy (1996), Lancaster and Lancaster (2003), and Simonich (1991). Most time-series studies find little or no effect of advertising on primary demand for cigarettes. For example, Duffy (1996) concluded that “advertising restrictions (including bans) have had little or no effect upon aggregate consumption of cigarettes.” A meta-analysis by Andrews and Franke (1991) found that the average elasticity of cigarette consumption with respect to advertising expenditure was only 0.142 during 1964-1970, and declined to -0.007 thereafter. Second, cross-national studies examine the relationship between per capita cigarette consumption and advertising bans for a panel of countries. Third, several time-series studies examine the effects of health scares and the 1971 ban of broadcast advertising. This essay discusses results obtained in the second and third types of econometric studies.

Econometric Results: Cross-National Studies of Broadcast Bans
Systematic tests of the effect of advertising bans are provided by four cross-national panel studies that examine annual per capita cigarette consumption among OECD countries: Laugesen and Meads (1991); Stewart (1993); Saffer and Chaloupka (2000); and Nelson (2003b). Results in the first three studies are less than convincing for several reasons. First, advertising bans might be endogenously determined together with cigarette consumption, but earlier studies treated advertising bans as exogenous. In order to avoid the potential bias associated with endogenous regressors, Nelson (2003b) estimated a structural equation for the enabling legislation that restricts advertising. Second, annual data on cigarette consumption contain pronounced negative trends, and the data series in levels are unlikely to be stationary. Nelson (2003b) tested for unit roots and used consumption growth rates (log first-differences) to obtain stationary data series for a sample of 20 OECD countries. Third, the study also tested for structural change in the smoking-advertising relationship. The motivation was based on the following set of observations: by the mid-1960s the risks associated with smoking were well known and cigarette consumption began to decline in most countries. For example, per capita consumption in the United States increased to an all-time high in 1963 and declined modestly until about 1978. Between 1978 and 1995, cigarette consumption in the U.S. declined on average by -2.44% per year. Further, the decline in consumption was accompanied by reductions in smoking prevalence. In the U.S., male smoking prevalence declined from 52% of the population in 1965 to 33% in 1985 and 27% in 1995 (Forey et al. 2002). Smoking also is increasingly concentrated among individuals with lower incomes or lower levels of education (U.S. Public Health Service 1994). Changes in prevalence suggest that the sample of smokers will not be homogeneous over time, which implies that empirical estimates may not be robust across different time periods.

Nelson (2003b) focused on total cigarettes, defined as the sum of manufactured and hand-rolled cigarettes for 1970-1995. Data on cigarette and tobacco consumption were obtained from International Smoking Statistics (Forey et al. 2002). This comprehensive source includes estimates of sales in OECD countries for manufactured cigarettes, hand-rolled cigarettes, and total consumption by weight of all tobacco products. The data series begin around 1948 and extend to 1995. Regulatory information on advertising bans and health warnings were obtained from Health New Zealand’s International Tobacco Control Database and the World Health Organization’s International Digest of Health Legislation. For each country and year, HNZ reports the media in which cigarette advertising are banned. Nine media are covered, including television, radio, cinema, outdoor, newspapers, magazines, shop ads, sponsorships, and indirect advertising such as brand names on non-tobacco products. Based on these data, three dummy variables were defined: TV-RADIO (= 1 if only television and radio are banned, zero otherwise); MODERATE (= 1 if 3 or 4 media are banned); and STRONG (= 1 if 5 or more media are banned). On average, 4 to 5 media were banned in the 1990s compared to only 1 or 2 in the 1970s. Except for Austria, Japan and Spain, all OECD countries by 1995 had enacted moderate or strong bans of cigarette advertising. In 1995, there were 9 countries in the strong category compared to 5 in 1990, 4 in 1985, and only 3 countries in 1980 and earlier. Additional control variables in the study included prices, income, warning labels, unemployment rates, percent filter cigarettes, and demographics.

The results in Nelson (2003b) indicate that cigarette consumption is determined importantly by prices, income, and exogenous country-specific factors. The dummy variables for advertising bans were never significantly negative. The income elasticity was significantly positive and the price elasticity was significantly negative. The price elasticity estimate of -0.39 is identical to the consensus estimate of -0.4 for aggregate data (Chaloupka and Warner 2000). Beginning about 1985, the decline in smoking prevalence resulted in a shift in price and income elasticities. There also was a change in the political climate favoring additional restrictions on advertising that followed rather than caused reductions in smoking and smoking prevalence, which is “reverse causality.” Thus, advertising bans had no demonstrated influence on cigarette demand in the OECD countries, including the U.S. The advertising-response model that motivates past studies is not supported by these results. Data and estimation procedures used in three previous studies are picking-up the substantial declines in consumption that began in the late-1970s, which were unrelated to major changes in advertising restrictions.

Background: Regulation of Cigarettes by the Federal Trade Commission
At the urging of President Wilson, the Federal Trade Commission (FTC) was created by Congress in 1914. The Commission was given the broad mandate to prevent “unfair methods of competition.” From the very beginning, this mandate was interpreted to include false and deceptive advertising, even though advertising per se was not an antitrust issue. Indeed, the first cease-and-desist order issued by the FTC concerned false advertising, 1 FTC 13 (1916). It was the age of the patent medicines and health-claims devices. As early as 1925, FTC orders against false and misleading advertising constituted 75 percent of all orders issued each year. However, in Raladam (1931) the Supreme Court held that false advertising could be prevented only in situations where injury to a competitor could be demonstrated. The Wheeler-Lea Act of 1938 added a prohibition of “unfair or deceptive acts or practices” in or affecting commerce. This amendment broadened Section 5 of the FTC Act to include consumer interests as well as business concerns. The FTC could thereafter proceed against unfair and deceptive methods without regard to alleged effects on competitors.

As an independent regulatory agency, the FTC has rulemaking and adjudicatory authorities (Fritschler and Hoefler 1996). Its rulemaking powers are quasi-legislative, including the authority to hold hearings and trade practice conferences, subpoena witnesses, conduct investigations, and issue industry guidelines and proposals for legislation. Its adjudicatory powers are quasi-judicial, including the authority to issue cease-and-desist orders, consent decrees, injunctions, trade regulation rules, affirmative disclosure and substantiation orders, corrective advertising orders, and advisory opinions. Administrative complaints are adjudicated before an administrative law judge in trial-like proceedings. Rulemaking by the FTC is characterized by broad applicability to all firms in an industry, whereas judicial policy is based on a single case and affects directly only those named in the suit. Of course, once a precedent is established, it may affect other firms in the same situation. Lacking a well-defined constituency, except possibly small business, the FTC’s use of its manifest powers has always been controversial (Clarkson and Muris 1981; Hasin 1987; Miller 1989; Posner 1969, 1973; Stone 1977).

Beginning in 1938, the FTC used its authority to issue “unfair and deceptive” advertising complaints against the major cigarette companies. These actions, known collectively as the “health claims cases,” resulted in consent decrees or cease-and-desist orders involving several major brands during the 1940s and early 1950s. As several cases neared the final judgment phase, in September 1954 the FTC sent a letter to all companies proposing a seven-point list of advertising standards in light of “scientific developments with regard to the [health] effects of cigarette smoking.” A year later, the FTC issued its Cigarette Advertising Guides, which forbade any reference to physical effects of smoking and representations that a brand of cigarette is low in nicotine or tars that “has not been established by a competent scientific proof.” Following several articles in Reader’s Digest, cigarette advertising in 1957-1959 shifted to emphasis on tar and nicotine reduction during the “tar derby.” The FTC initially tolerated these ads if based on tests conducted by Reader’s Digest or Consumer Reports. In 1958, the FTC hosted a two-day conference on tar and nicotine testing, and in 1960 it negotiated a trade practice agreement that “all representations of low or reduced tar or nicotine, whether by filtration or otherwise, will be construed as health claims.” This action was blamed for halting a trend toward increased consumption of lower-tar cigarettes (Calfee 1997a; Neuberger 1963). The FTC vacated this agreement in 1966 when it informed the companies that it would no longer consider advertising that contained “a factual statement of tar and nicotine content” a violation of its Advertising Guides.

On January 11, 1964, the Surgeon General’s Advisory Committee on Smoking and Health issued its famous report on Smoking and Health (U.S. Surgeon General 1964). One week after the report’s release, the FTC initiated proceedings “for promulgation of trade regulation rules regarding unfair and deceptive acts or practices in the advertising and labeling of cigarettes” (notice, 29 Fed Reg 530, January 22, 1964; final rule, 29 Fed Reg 8325, July 2, 1964). The proposed Rule required that all cigarette packages and advertisements disclose prominently the statement, “Caution: Cigarette smoking is dangerous to health [and] may cause death from cancer and other diseases.” Failure to include the warning would be regarded as a violation of the FTC Act. The industry challenged the Rule on grounds that the FTC lacked the statutory authority to issue industry-wide trade rules, absent congressional guidance. The major companies also established their own Cigarette Advertising Code, which prohibited advertising aimed at minors, health-related claims, and celebrity endorsements.

The FTC’s Rule resulted in several congressional bills that culminated in the Federal Cigarette Labeling and Advertising Act of 1965 (P.L. 89-92, effective Jan. 1, 1966). The Labeling Act required each cigarette package to contain the statement, “Caution: Cigarette Smoking May Be Hazardous to Your Health.” According to the Act’s declaration of policy, the warnings were required so that “the public may be adequately informed that cigarette smoking may be hazardous to the health.” The Act also required the FTC to report annually to Congress concerning (a) the effectiveness of cigarette labeling, (b) current practices and methods of cigarette advertising and promotion, and (c) such recommendations for legislation as it may deem appropriate. Beginning in 1967, the FTC commenced its annual reporting to Congress on advertising of cigarettes. It recommended that health warning be extended to advertising and strengthened to conform to its original proposal, and it called for research on less-hazardous cigarettes. These recommendations were repeated in 1968 and 1969, and a recommendation was added that advertising on television and radio should be banned.

Several other important regulatory actions also took place in 1967-1970. First, the FTC established a laboratory to conduct standardized testing of tar and nicotine content for each brand. In November 1967, the FTC commenced public reporting of tar and nicotine levels by brand, together with reports of overall trends in smoking behaviors. Second, in June of 1967, the Federal Communications Commission (FCC) ruled that the “fairness doctrine” was applicable to cigarette advertising, which resulted in numerous free anti-smoking commercials by the American Cancer Society and other groups during July 1967 to December 1970.2 Third, in early 1969 the FCC issued a notice of proposed rulemaking to ban broadcast advertising of cigarettes (34 Fed Reg 1959, Feb. 11, 1969). The proposal was endorsed by the Television Code Review Board of the National Association of Broadcasters, and its enactment was anticipated by some industry observers. Following the FCC’s proposal, the FTC issued a notice of proposed rulemaking (34 Fed Reg 7917, May 20, 1969) to require more forceful statements on packages and extend the warnings to all advertising as a modification of its 1964 Rule in the “absence of contrary congressional direction.” Congress again superseded the FTC’s actions, and passed the Public Health Smoking Act of 1969 (P.L. 91-222, effective Nov. 1, 1970), which banned broadcast advertising after January 1, 1971 and modified the package label to read, “Warning: The Surgeon General Has Determined that Cigarette Smoking Is Dangerous to Your Health.” In 1970, the FTC negotiated agreements with the major companies to (1) disclose tar and nicotine levels in cigarette advertising using the FTC Test Method, and (2) include the health warning in advertising. By 1972, the FTC believed that it had achieved the recommendations in its initial reports to Congress.3

In summary, the FTC has engaged in continuous surveillance of cigarette advertising and marketing practices. Industry-wide regulation began in the early 1940s. As a result, the advertising of cigarettes in the U.S. is more restricted than other lawful consumer products. Some regulations are primarily informational (warning labels), while others affect advertising levels directly (broadcast ban). During a six-decade period, the FTC regulated the overall direction of cigarette marketing, including advertising content and placement, warning labels, and product development. Through its testing program, it has influenced the types of cigarettes produced and consumed. The FTC engaged in continuous monitoring of cigarette advertising practices and prepared in-depth reports on these practices; it held hearings on cigarette testing, advertising, and labeling; and it issued consumer advisories on smoking. Directly or indirectly, the FTC has initiated or influenced promotional and product developments in the cigarette industry. However, it remains to be shown that these actions had an important or noticeable effect on cigarette consumption and/or industry advertising expenditures. Is there empirical evidence that federal regulation has affected aggregate cigarette consumption or advertising? If the answer is negative or the effects are limited in magnitude, it suggests that the Congressional and FTC actions after 1964 did not add materially to information already in the marketplace or these actions were otherwise misguided.4

Smoking prevalence has declined considerably compared to the 1950s and 1960s. Consumption per capita reached an all-time high in 1963 (4,345 cigarettes per capita) and began a steep decline around 1978. By 1985, consumption was below the level experienced in 1947. Cigarette promotion has changed greatly over the years as producers substituted away from traditional advertising media. As reported by the FTC, the category of non-price promotions includes expenditures on point-of-sale displays, promotional allowances, samples, specialty items, public entertainment, direct mail, endorsements and testimonials, internet, and audio-visual ads. The shift away from media advertising reflects the broadcast and billboard bans as well as the controversies that surround advertising of cigarettes. As a result, spending on traditional media now amounts to only $356 million, or about 7% of the total marketing outlay of $5.0 billion. Clearly, regulation has affected the type of promotion, but not the overall expenditure.

Econometric Results: U.S. Time-Series Studies of the 1971 Advertising Ban
Several econometric studies examine the effects of the 1971 broadcast ban on cigarette demand, including Franke (1994), Gallet (1999), Ippolito et al. (1979), Kao and Tremblay (1988), and Simonich (1991). None of these studies found that the 1971 broadcast ban had a noticeable effect on cigarette demand. The studies by Franke and Simonich employed quarterly data on cigarette sales. The study by Ippolito et al. covered an extended time period from 1926 to 1975. The studies by Gallet and Kao and Tremblay employed simultaneous-equations methods, but each study concluded that the broadcast advertising ban did not have a significant effect on cigarette demand. Although health reports in 1953 and 1964 may have reduced the demand for tobacco, the results do not support a negative effect of the 1971 Congressional broadcast ban. By 1964 or earlier, the adverse effects of smoking appear to have been incorporated in consumers’ decisions regarding smoking. Hence, the advertising restrictions did not contribute to consumer information and therefore did not affect cigarette consumption.

Conclusions
The First Amendment protects commercial speech, although the degree of protection afforded is less than political speech. Commercial speech jurisprudence has changed profoundly since Congress passed a flat ban on broadcast advertising of cigarettes in 1971. The courts have recognized the vital need for consumers to be informed about market conditions — an environment that is conducive to operation of competitive markets. The Central Hudson test requires the courts and agencies to balance the benefits and costs of censorship. The third-prong of the test requires that censorship must directly and materially advance a substantial goal. This essay has discussed the difficulty of establishing a material effect of limited and comprehensive bans of alcohol and cigarette advertisements.