新闻传播学相关参考资料合集：新闻媒体的竞争

R ESEARCH PAPER S ERIES RESEARCH PAPER NO. 1808 Competing For The Public Through The News Media David P. Baron June 2003 Research Paper No. 1808 COMPETING FOR THE PUBLIC THROUGH THE NEWS MEDIA David P. Baron Stanford University June 2003 Abstract Interest groups seek to influence economic activity through public and private politics. Public politics takes place in the arenas of public institutions, whereas private politics takes place outside public institutions often in the arena of public sentiment. Private politics refers to action by interest groups directed at private parties, as in the case of an activist group launching a campaign against a firm. This paper presents a model of informational competition between an activist and an industry, where each interest group seeks to influence public sentiment and does so by advocating its position through the news media. Citizen consumers make both a private consumption decision and a collective choice on the regulation of a product that has an externality associated with it. In the absence of the news organization the collective choice in not to regulate. The activist and the industry obtain private, hard information on the seriousness of the externality and advocate favorable information and may conceal unfavorable information. The news media can conduct investigative journalism to obtain its own information and based on that information and the information it has received from its sources, provides a news report to the public. Due to its role in society, the media has an incentive tobiasitsreport,andthedirectionofbiasistowardregulation.Itsbiasservesto mitigate both the market failure by decreasing demand and a government failure by leading to regulation. The activist then has an incentive to conceal information unfavorable to its interests, whereas the industry fully reveals its information. Competing for the Public Through the News Media 1 David P. Baron Stanford University I. Introduction Interest groups seek to influence economic activity through public and private politics. Public politics takes place in the arenas of public institutions, whereas private politics takes place outside public institutions often in the arena of public sentiment. Public politics fo- cuses on government policy, such as regulation and tax and expenditure policy. Private politics refers to action by interest groups directed at private parties, as in the case of an activist group launching a campaign against a firm. This can be independent of gov- ernment but generally draws strength from the public, as in the case of a boycott. This paper presents a model in which interest groups compete for public sentiment through the news media to influence the public politics of regulation and private politics through the consumption decisions of citizens. In addition, the paper provides an explanation of media bias in terms of the role of the news media in society. In private politics activist strategies often focus on pressuring firms through the mem- bers of the public both in their roles as consumers and their roles as constituents of public officeholders. In response, opposing interest groups may counteract those strategies through both private and public political strategies. For example, the strategic competition between activists and firms frequently focuses on public sentiment about the activities of the firm. The instrument of that competition is advocacy—communicating information to the public favorable to the interest group’s objectives while remaining silent about information un- favorable to those objectives. This communication can be direct to the public, but it can be more efficient to communicate through the news media, which can be a low-cost means of communicating with the public. In his study of 24 ecological boycotts led by activists Friedman (1999, Ch. 8) found that 22 were directed at the news media. Since interest groups direct their communication strategies to it, the news media has a role in private politics (Baron (2002)(2003b)). The news media can be viewed as a private institution in 1 This research was supported by NSF Grant No. SES-0111729. I would like to thank Bharat Anand, Larry Bartels, and seminar participants at Columbia University, Harvard University, and Princeton University for their comments. 1 whose arena interest groups compete to influence public and private politics. In the 1950s and 1960s this role led commentators to refer to the news media as the fourth branch of government or the “fourth estate.” The news media provides information to the members of the public for their private and collective decisions, and in doing so it has considerable discretion in determining what information is reported. Moreover, it may have its own objectives based on self-interest as well as on principles of journalism as embodied in the profession. The media thus can have a strategic role in influencing public sentiment and hence the outcomes of private and public politics. The media’s reporting also can affect the strategies of its sources, and it may conduct investigative journalism to uncover the concealment of information. The strategic competition considered here is between interest groups, an activist and an industry, and pertains to an externality associated with the consumption of a product. The competition takes the form of communicating information about the seriousness of the externality to the public through the news media. The news media exercises discretion on what it reports to the public, and based on that report, the members of the public individu- ally decide whether to consume the product and collectively decide whether to regulate the externality. The media’s reporting strategy thus can affect both private and public politics, and in equilibrium the media biases its news report to mitigate both a government failure and a market failure. Bias in this case can serve the interests of the public, even though individuals are fully rational and skeptical of the news report. In addition to providing information, the news media may be able to make normative arguments regarding the ex- ternality with the objective of changing the public’s preferences. The change in behavior resulting from such moral suasion can be a substitute for regulation. II. An Example of an Issue and the Competition The issue of corporate average fuel economy (CAFE) standards for light vehicles pro- vides an example of a competition between activists and firms with the news media provid- ing information to the public for their private and collective decisions. The CAFE standard for passenger cars has remained at 27.5 mpg since 1985, and the standard for light trucks, which includes SUVs, has remained at 20.7 mpg since a small increase in 1996. Environ- mental activists and their allies in Congress have sought to increase the standard. The 2 Sierra Club has worked for a standard of 40 mpg for automobiles to be achieved over 10 years. Since SUVs and light trucks represent nearly 50 percent of the market, Senators John Kerry and John McCain sought to replace the two standards with a single standard of35mpgtobeachievedover12years. A principal component of the strategy of the environmental groups has been to em- phasize the seriousness of the global climate change issue and recently to add the national security issue of the reliance on imported oil. In public politics environmentalists sought to increase constituent pressure on representatives. The automakers countered by empha- sizing the revealed preferences of consumers for larger and more powerful vehicles and the additional injuries and fatalities that would result from downsizing vehicles to meet sharply higher fuel economy standards. The status quo was advantaged in public politics, and the legislative efforts failed in 2002. After failing to achieve higher CAFE standards in Congress, the Sierra Club turned to private politics. It sought to mobilize the public against the automakers’ refusal to improve the fuel economy of their vehicles. The Sierra Club began a three-year campaign to put public pressure directly on the big-three automakers. As Carl Pope, executive director of the Sierra Club, explained “we’re going now to the customers.” The Sierra Club hired Haddow Communications to conduct an advertising campaign challenging the automobile companies. 2 The radio ads specifically called on CEO Bill Ford, “Now more than ever, America needs cars that get better gas mileage. That’s why we’re asking Bill Ford, head of the Ford Motor Company, to do his part and to produce more fuel efficient, SUV’s and pickup trucks.” Separately, another group aired television commercials arguing that buying an SUV supported terrorism and threatened national security by increasing the dependence on imported oil. News coverage of the activist campaigns increased the attention given to the issues. As The New York Times observed, “the message is attracting attention through 2 Television spots featured former Nebraska senator Bob Kerrey and retired Navy vice admiral Jack Shanahan. In one ad Kerrey said, “Its time for us to tell the auto industry that we want to break the grip of oil-producing countries and reduce our oil use.” In another Kerrey stated, “We ask our young men and women to sacrifice their safety and perhaps their lives to fight the war against terrorism. We all know that our dependency on imported oil is part of the problem and we know that increasing the fuel economy of the cars we driveispartofthesolution.” 3 news coverage.” 3 TheSierraClubobtainedleveragethroughmediacoverageof thecampaignandthrough a variety of actions intended to emphasize recent decreases in fleet fuel economy and the seriousness of the global climate change issue. The news media regularly reported on fuel economy data and regulation, and the activists and the industry provided information to influence those reports. For example, environmental activists railed against the introduc- tion of the Hummer H2, which was sufficiently heavy that it was not subject to any fuel economy standards. Earlier, the Sierra Club had campaigned against the Ford Excursion, which it dubbed the Ford Valdez. Ford decided to drop the Excursion. In 2003 the Bush administration increased the light truck standard to 22.2 mpg by model year 2007. The administration argued that the increase would save 2.5 billion gallons of gasoline, but critics claimed that the automobile industry was already planning increases greater than in the new standard. Daniel Becker of the Sierra Club called the action by the Bush administration “irresponsible.” The big three automakers criticized the new standard as costing more than it was worth and costing the nation badly needed jobs. III. Overview A. Literature on the News Media The impact of the news media on public politics is broadly observable (Graber (2000)). For example, following the oil crises in the 1970s Erfle and McMillan (1990) found that media coverage affected the price of home heating oil relative to the price of bunker oil sold to electric utilities. They concluded that oil companies restrained their price increases to forestall a public reaction and possible new regulation. Stromberg (2001) has shown that as the U.S. radio audience increased in the 1930s more relief funds were allocated to counties with more listeners. Besley and Burgess (2002), Besley and Prat (2001), and Besley, Burgess, and Pratt (2002) studied models of political agency focusing on the relation between an incumbent officeholder and an electorate that obtained information from the news media. Besley and Burgess considered a model in which the media provides information about government performance to the electorate, and this induces the government to be responsive to the 3 The New York Times, January 8, 2003. 4 interests of the public. They test the model using data from India and find that state governments are more responsive to the public’s needs the broader is newspaper readership. Besley and Pratt focused on the opportunity for the government to capture the media and found that inefficiency and malfeasance can result. In their model the government is a source of information for the media and hence may manipulate that information. In the model considered here the sources of information are interest groups. Stromberg (2002) presented a model in which increasing returns to scale and adver- tising incentives drive the media to provide more information to large and higher income groups than to smaller and lower income groups. These incentives for media bias can affect public policy. Media bias can also arise from within the news organization. 4 The American Association of Newspaper Editors (1999) (www.asne.org/index.cfm?d=2632) conducted a survey and concluded, “The public suspects that the points of view and biases of journalists influence what stories are covered and how they are covered.” Although the public broadly views the news media as biased, individuals differ considerably in their perception of the nature and direction of that bias. Bovitz, Druckman, and Lupia (2002) used a hierarchical model of a news organization to explain the sources of bias and the forces that act to control it. The sources of bias are the ideological orientations of media elites, but for those orientations to influence public opinion, editors must either accept an ideologically-biased report by a journalist or bias a report to serve the ideological orientation of the owners of the media organization. Bias can be mitigated by the career concerns of editors and by the audience. The audience must be willing to read the report, and for it to have influence the media must bias information in a direction contrary to the initial beliefs of the audience. Mullainathan and Shleifer (2002) provide a model in which bias can arise both from an ideological orientation of the news organization and from incentives to tell an interest- ing story. The ideological bias arises from the preferences of journalists, and the bias in their stories can be neutralized if there is an ideological balance among competing news organizations. The incentives to tell an interesting story come from a preference to be well 4 Attention has also been given to the effect of media competition on programming (Spence and Owen (1977), Owen and Wildman (1992), Noam (1985)(1987)) and to the wel- fare analysis of media competition (Anderson and Coate (2001), Hansen and Kyhl (2001), Dukes and Gal-Or (2001)). 5 thought of by readers, and in a psychology-based model of the audience these incentives can produce bias. In the model considered here bias results from the assumed responsibilities of the media to serve the public, in this case by mitigating a market failure and a government failure. A number of cross-country studies have been conducted on the influence of the media on politics and the economic performance of countries. In the realm of private politics Dyck and Zingales (2002) consider the effect of the news media on the behavior of corpo- rate executives and directors in choosing “socially responsible” corporate activities. They conclude that the media has substantial impact on the environmental policies of firms and on governance. In the model presented here the news media influences the public in both their consumption decisions and in their collective choice on regulation. Hamilton (1996) argued that television programming exhibits market failures in edu- cational programming for children, public affairs coverage, and indecent and violent pro- gramming. The market failures result in an oversupply of programming with negative externalities and an undersupply of programming with positive externalities. He argued for regulation to correct the market failures. In the model presented here the media acts in response to a market failure arising not from within the media market but among the public. B. Overview of the Model The model focuses on the competition between activists and firms and the role of the news media in that competition. The model describes how the media informs the public based on information provided by an activist, firms, and possibly its own investigative journalism. The model provides an explanation of media bias as a function of media preferences and the opportunity to influence the public. It also provides an explanation for the communication strategies of the activist and the firms. Ultimately, a theory should identify the role of the news media in interest group competition as a private institution with “officeholders” who self-select into the profession of journalism. The current model provides a first step in that direction. At the center of the competition between the activist and the firms are a private consumption decision of citizen consumers and a collective choice to regulate the product they consume. The private decision pertains to a product that has associated with it an 6 unregulated externality. The collective choice is whether to regulate the externality, where the form of regulation is a product standard. The private decisions of citizen consumers may be thought of as whether to purchase a low fuel economy vehicle, and the public decision may be thought of as whether to impose a CAFE standard. Citizenconsumershaveincompleteinformationabouttheseriousness of theexternality, and the news media provides a soft report on the seriousness based on information from its sources and its own investigative journalism. Citizen consumers are rational and take into account the possible bias in the report from either the news media or its sources. The sources are an activist, whose interests are in dampening the demand for the product and regulating the externality, and firms that produce the product and oppose regulation. The activist and the firms are better informed than is the media, and they act strate- gically to influence the media’s news report. Their influence takes the form of advocacy, where they communicate favorable (hard) information to the media and may conceal un- favorable information. The media may investigate the seriousness of the externality if it does not receive hard information from the activist or the firms. Investigative journalism both informs the media and provides an opportunity to expose possible concealment by its sources. If concealment is exposed, the activist or the firms suffer reputation damage. The firms are assumed to compete in the market but to act collectively as an industry in dealing with the activist, the media, and the public. 5 Citizen consumers are sophisticated and anticipate the strategies of the activist, firms, and media. They subscribe to the media’s service only if they expect to benefit from it, and those who do not subscribe benefit from an informational spillover. The preferences of the activist are to mitigate the externality, and the firms maximize their profits. The objectives of the news media are a combination of profits, journalistic responsibilities, and journalistic performance. The responsibilities pertain to its role in society to serve the public by providing information to citizen consumers to improve their private and collective choices. Journalistic performance pertains to exposing concealment by sources, which can enhance the media’s reputation. The model focuses on the case in which the media can influence the collective decision. 5 Collective action by firms in public politics is allowed under the Noerr Pennington doctrine based on the right to free speech. 7 That is, in the absence of the media the collective choice is not to regulate, but if the media’s report indicates that the externality is very serious, the collective choice is to regu- late. Because of its responsibility to serve citizen consumers in their private and collective choices, the news media may have an incentive to bias its report in the direction of greater seriousness of the issue when there is a collective choice failure and thus lead citizens to regulate the externality. The news media has an opportunity to bias its report only when it does not have hard information, either from its sources or its own investigation. Figure 1 illustrates the model and identifies the influences on private and public politics. The news media also might be able to affect the private decisions of citizen consumers by causing them to take the externality into account in their consumption decisions. This moral suasion can be a substitute for regulation. C. Results Demand for the firms’ product and industry profit are decreasing in citizen consumers’ beliefs about the seriousness of the externality and the extent to which they internalize the externality associated with their purchases. The support for regulation is decreasing in consumers’ valuation for the product, since regulation reduces the value of the product. The number of citizens who prefer regulation is increasing in the seriousness of the externality but is decreasing in the extent to which consumers internalize the externality. The news report thus can directly affect the private decisions of consumers and the industry as well as the collective choice on regulating the externality. The activist and the industry may provide hard information to the media, and their communication strategies take the form of advocacy, where the sources present favorable information and may conceal unfavorable information. The media’s beliefs that the exter- nality is serious are increasing in the probability that the industry conceals unfavorable information and decreasing in the probability that the activist conceals unfavorable infor- mation. These properties carry over to the public’s beliefs given a news report that the externality is serious. The media reports hard information when it has it, but when it has no hard informa- tion, it can bias its report. The incentive for bias is due to preferences that depend on the aggregate welfare of the public, whereas the collective choice is determined by the pivotal voter. This incentive is only present when the media can influence the collective choice, so 8 bias is both instrumental and opportunistic. In that case, the media biases its report in favor of regulation to mitigate both a market failure represented by the externality and a government failure due to a (super)majority requirement. Media bias thus can be welfare enhancing. The media’s bias provides incentives for concealment by its sources, but in equilibrium it is only the activist that conceals information. Concealment by the activist increases the opportunity for the media to bias its report in favor of regulation. Bias thus induces an alignment of interests between the news media and the activist; i.e., with the party whose interests are favored by the bias. Concealment and bias are thus synergistic. When it biases its report, the media need not fear being exposed because the interest group that could expose it has preferences for regulation that are aligned with those of the media. Because the bias is only in one direction, the public believes a report that the issue is not serious and is skeptical of a report that it is serious. The public becomes more skeptical of the media’s report the more likely is the activist to conceal unfavorable information. Since the news media biases its report against the interests of the industry, the industry has no incentive to conceal unfavorable information. The activist, however, conceals with positive probability, and that probability is limited only by the risk of being exposed for having concealed information. The activist may conceal unfavorable information with prob- ability one if the damage to its reputation from being exposed is small. The informational competition between the activist and the industry is thus one-sided. The media is assumed to conduct investigative journalism only if it receives no hard information from its sources. Investigative journalism is costly, and the incentive to bear the cost comes from the possibility of both becoming better informed and enhancing its reputation by exposing concealment by one of its sources. If it could conceal the hard information from its sources or its investigative journalism, the media will do so only in the same event in which the activist has an incentive to conceal its hard information. The equilibrium for the case in which the media can conceal information is thus qualitatively the same as when it cannot conceal hard information. Moral suasion by the news media reduces the harm from the externality and decreases the public support for regulation. Moral suasion is thus a substitute for regulation and for biased media reports. Moral suasion may eliminate the media’s informational influence 9 andhenceitsincentivestobiasitsnewsreport. In the model the bias of the news media is due not to the personal preferences of those in the news organization but is instead due to the objective of serving the interest of the public. 6 That is, the media provides information about a market failure and a collective choice that benefits the public as a whole but is not preferred by a sufficient majority to overcome the status quo advantage in pivotal politics. This comes at the expense of truthful reporting. IV. Time Line and Information Structure Figure 2 identifies the sequence of actions in the game. In the information stage the ac- tivist and the firms search for information about the issue, and in the communication stage they send messages to the news media. The media then sets a price for a subscription, and citizen consumers who expect to benefit subscribe to the media reports. The media decides whether to conduct investigative journalism, and if it does, it searches for information. The media then provides a report. Citizen consumers next collectively choose whether to regulate the product, and given that decision, the firms set prices for their products, and citizen consumers make their consumption decisions. A state of nature θ cantakeontwovaluesθ ∈{θ L ,θ H }, θ L <θ H ,andthecommon knowledge prior probability is p = Pr(θ = θ H ). The interpretation is, for example, that θ H corresponds to global climate change being very serious, whereas θ L means it is not serious. The state θ H is favorable to the activist’s cause of dampening demand and obtaining regulation of the externality, and the state θ L is favorable to the interests of the firms. In the information stage the activist and the firms search simultaneously for informa- tion about the state, and each search is successful with probability q ∈ (0,1) and unsuccess- ful with probability 1 ? q. For the news media, investigative journalism is successful with probability q M . A successful search provides hard information that is conclusive evidence that the state is θ H or θ L , and an unsuccessful search yields no information, which will be denoted by φ. The cost to the public of verifying hard information is assumed to be high, 6 Journalists may also have individual preferences that influence news stories. Pat- terson and Donsbach (1996), for example, surveyed journalists in five countries and con- cluded “there is a significant correlation between journalists’ personal beliefs and their news decisions.” 10 so the activist and the firms cannot communicate their information directly to the public. The public relies on the soft report by the news media. The activist and the firms thus communicate to the public through the news media. The news media is assumed to be able to evaluate the information provided by the activist and the firms, so the sources cannot fake information. The rest of the model is introduced in the following sections beginning at the end of the game. Because of the complexity of the strategic situation the model is formulated in a relatively simple form. V. The Public A. The Public as Citizen Consumers Each member of the public makes a private decision whether to purchase the product, e.g., a low fuel economy vehicle, and collectively the public chooses whether to regulate the product, e.g., impose a fuel economy standard. Citizens differ only in their valuation v for the product, and the public is assumed to consist of N citizens with valuations distributed uniformly on the interval [0,?v]. That is, the distribution is Nf(v)wheref(v)= 1 ?v ,v∈ [0,?v]. Purchasing a low fuel economy vehicle generates one unit of emissions, which is assumed to be a pure public bad that costs each citizen θ. Regulation in the form of a standard s ∈ (0,1) can be imposed to reduce the emissions proportionately to (1 ? s)θ.Regulation also affects the value of the product. The valuation with regulation is assumed to be given by v(1?αs), where α ∈ [0, 1 s )isaparameter. 7 The automakers argue that α>1, reflecting downsizing, reductions in power, and reduced safety. An indication that α>1wasgivenincongressionaltestimonybyEdwardB.Cohen,vice president of Honda North America. 8 He stated, “If the current car fleet were still at 1981 performance, weight and transmission levels, the passenger car CAFE would be almost 36 mpg instead of the current level of 28.1 mpg. The trend is particularly pronounced since 1987. Based on EPA’s data, technology has gone into the fleet from 1987 to 2000 at a rate that could have increased fuel economy by about 1.5% per year, if it had not 7 The specifications (1?s)θ fortheemissionsand(1?αs)v for the valuation are to be understood as local rather than global properties of regulation. 8 Honda is the only automobile company operating in the United States that is not a member of the Alliance of Automobile Manufacturers and the only one not opposing a substantial increase in fuel economy standards. Honda has the highest fleet fuel economy. 11 instead focused on other vehicle attributes demanded by the market. There is no reason why this technology trend of improved efficiency (as opposed to fuel economy) should not continue.” 9 He indicated that average vehicle weight had increased by 12% from 1987 to 2000 and average horsepower increased by 70%. This testimony implies that the auto industry has responded to consumer demand by using improved fuel efficiency to provide increases in vehicle size and power, resulting in no improvement in fuel economy. This suggests that higher fuel economy standards would have substantially reduced the value of the vehicles to consumers. The analysis thus focuses on the case of α≥1. B. Private Choice Citizen v’s expected utility u v from consumption is specified as u v = z v (v(1? αs + )?y ?(1? s + )θ + ηN), where y is the price of the product, θ + denotes the mean of v’sposteriorbeliefsaboutθ given the media’s news report, s + ∈{0,s} is the standard corresponding to θ + ,andz v ∈{0,1} is an indicator variable with z v = 1 denoting a purchase and z v = 0 denoting no purchase. 10 The cost to citizen v of the externality from her purchase is (1 ? s + )θ + ,andη ∈ [ 1 N ,1] denotes the extent to which the citizen takes into account the effect of her purchase on others. Thus, η = 1 N corresponds to ignoring those effects, and η =1correspondstotaking them fully into account. The parameter η is thus a measure of other-regardedness. A free-rider problem is present when η<1. The parameter η is a characteristic of preferences and is independent of information about the seriousness of the issue. Initially, η is treated as fixed and small, and in Section X it is viewed as responsive to moral suasion. A citizen with a valuation v satisfying v ≥ v ? (s + ,θ + ,y) ≡ y +(1?s + )θ + ηN 1?αs + purchases the product, and the others do not. The demand X(s + ,θ + ,y) for the product is then X(s + ,θ + ,y)=N parenleftBig 1? v ? (s + ,θ + ,y) ?v parenrightBig . (1) 9 Statement of Edward B. Cohen, vice president, Honda North America, before the Senate Committee on Commerce, Science and Transportation, December 6, 2001. 10 To simplify the model, s + is assumed to be either 0 or s.Thatis,s is the standard when the media reports that the externality is very serious, and otherwise the standard is 0. 12 The price is established by Cournot competition conditional on θ + and s + .Eachof the n firms in the industry is assumed to be identical, and to simplify the model each is assumed to have zero cost. The profit pi i of firm i then is pi i = P(X)x i , where x i is the output of firm i, X =Σ n i=1 x i , P(X) is the inverse demand function obtained from X(s + ,θ + ,y)andv ? (s + ,θ + ,y). The equilibrium price y ? (s + ,θ + )is y ? (s + ,θ + )= 1 n +1 parenleftBig ?v(1?αs + )?(1?s + )θ + ηN parenrightBig , (2) so the price reveals the information the firms have. From (1) information that the ex- ternality is more serious (higher θ + ) reduces demand, and the firms respond with lower prices. Despite the lower price, demand is lower the higher is θ + . The marginal valuation v ? (s + ,θ + ) ≡ v ? (s + ,θ + ,y ? (s + ,θ + )) is v ? (s + ,θ + )= ?v n +1 + n(1?s + )θ + ηN (n +1)(1?αs + ) , (3) which is strictly increasing in θ + and in s + for α>1, so regulation decreases the demand for the product. Demand X(s + ,θ + ) ≡ X(s + ,θ + ,y ? (s + ,θ + )) is X(s + ,θ + )= n n +1 N (1?αs + )?v parenleftBig ?v(1?αs + )?(1?s + )θ + ηN parenrightBig , (4) which is positive if ?v> (1?s + )θ + ηN (1?αs + ) , which will be assumed to be the case when η is small. 11 That is, there is some consumer who purchases. The externality θ + X(s + ,θ + ), however, could exceed ?v; i.e., the externality could be quite serious. The profit pi ? i (s + ,θ + )offirmi is pi ? i (s + ,θ + )= Ny ? (θ + ) 2 (1?αs + )?v = N(?v(1?αs + )?(1?s + )θ + ηN) 2 (n +1) 2 (1?αs + )?v . (5) The price in (2) is decreasing in θ + , η, α, s + ,andn and increasing in ?v. Similarly, the profitisdecreasinginθ + , s + for α>1, the competitiveness (n) of the industry, and the extent (η) to which consumers internalize the externality. 12 11 The proportion of consumers who purchase is increasing in ?v and lim ?v→∞ v ? ?v = 1 n+1 . 12 Profit is decreasing in the standard only if α>1. The increases in CAFE standards during the 1970s and 1980s shifted considerable market share and profits from the big three automakers to the producers of smaller, higher fuel economy vehicles. 13 All citizens with valuations at least v ? (s + ,θ + ) purchase the product, so the externality is (1 ? s + )θ + X(s + ,θ + )=(1? s + )θ + N(1 ? v ? (s + ,θ + ) ?v ). Regulation directly reduces the emissions and also reduces demand which further reduces emissions. The externality is stronger the more competitive is the industry and weaker the more consumers internalize it. Since the externality is a pure public bad, every citizen incurs the harm. Consequently, the utility U 1 (s + ,θ + ;v) of a citizen with v≥v ? (s + ,θ + )is U 1 (s + ,θ + ;v)=v(1?αs + )?y ? (s + ,θ + )?(1?s + )θ + X(s + ,θ + )?(1?s + )θ + (ηN ?1) =(1?αs + ) parenleftBig v? ?v n +1 parenrightBig + 1 n +1 (1?s + )θ + ηN ?(1?s + )θ + (ηN ?1) ? parenleftBig 1? v ? (s + ,θ + ) ?v parenrightBig (1?s + )θ + N. (6) The utility U 0 (s + ,θ + ;v) of a citizen who does not purchase is U 0 (s + ,θ + ;v)=?(1?s + )θ + X(s + ,θ + ). (7) The news media has an opportunity through the report it presents about θ to affect both private and public politics. Private politics pertains to the consumption decisions of citizens, and public politics corresponds to the collective choice of the regulation. One influence of the news media on private politics is to affect demand for the product. As considered in Section X, the media might also be able to affect demand through moral suasion. The influence on public politics is through the portion of the public that supports regulation. C. Collective Choice Public politics governs the choice of the regulation, where the status quo is no regula- tion (s + = 0). Pivotal politics theory (Krehbiel (1998)(1999)) predicts that a change in the status quo can require a supermajority to overcome a filibuster in the Senate or override a presidential veto. A supermajority may also be required to overcome vote buying (Grose- close and Snyder (1996)). The relation between citizen preferences and the supermajority required to change the status quo is not known, however, so instead of modeling the in- stitutions, the supermajority γ of the public required for regulation to be enacted will be taken as a parameter where γ≥ 1 2 . Each citizen is assumed to act as if she is pivotal. 14 A citizen who does not purchase the product prefers regulation because she only bears the harm from the emissions, as indicated in (7). The valuation v o (s + ,θ + ) of the citizen who purchases and is indifferent between s + = s and no regulation is, using (6), defined by U 1 (0,θ + ;v o (s + ,θ + ))?U 1 (s + ,θ + ;v o (s + ,θ + ) ≡ 0. Solving for v o (s + ,θ + ) yields v o (s + ,θ + )= ?v n +1 + θ + N (n +1)α?v(1 ? αs + ) parenleftBigparenleftBig n? η +(n +1) parenleftBig η ? 1 N parenrightBigparenrightBig ?v(1 ? αs + ) ?(2?α? s + )nθ + ηN parenrightBig . (8) Appendix A presents conditions such that v o (s + ,θ + )<?v and v o (s + ,θ + )>v ? (s + ,θ + ), and these conditions will be assumed to be satisfied. Since U 1 (·)isincreasinginv, all citizens with v>v o (s + ,θ + ) oppose regulation. The support for regulation is decreasing in the com- petitiveness of the industry and is increasing in the number N of citizens affected by the externality. The support for regulation is decreasing in ?v,andlim ?v→∞ v o ?v = 1 n+1 . The valuation v o (s + ,θ + ) of the citizen who is indifferent between regulation and no regulation is increasing in s for α>1, so a higher standard receives less support from among the purchasers of the product. The valuation v o (s + ,θ + ) of the indifferent citizen is increas- ing in θ + when ?v > 2(2?α?s + )nθ + ηN (n?η +(n +1)(η? 1 N ))(1?αs + ) , (9) in which case support for regulation increases as the externality becomes more severe. This condition is satisfied when η is small. If the failure to regulate the externality is viewed as a government failure, a news report that the externality is serious may mitigate a government failure when (9) is satisfied. The support for regulation also depends on the extent to which citizens internalize the externality. The derivative is dv 0 (θ + ,s + ) dη = θ + Nn α(n +1)?v(1?αs + ) (?v(1?αs + )?(2?α?s + )θ + N). (10) If the externality is costly (θ + N>?v), the derivative is negative, so greater other-regardedness reduces the support for regulation. Support for regulation decreases because an increase in η shifts demand downward, which lessens the effect of the externality. To the extent that 15 moral suasion can affect η, the news media as well as the activist and the industry have incentives to offer moral arguments to persuade consumers to act or not to act privately in the market. Moral suasion is not treated as a strategy here but instead is considered qualitatively in Section X. D. Media Influence Basedontheirpriorinformation ˉ θ = pθ H +(1?p)θ L , citizens with v<v ? (s + , ˉ θ)donot purchase and prefer regulation s + = s,andthosewithv≥v ? (s + , ˉ θ) purchase and a subset of them, those with v ∈ [v ? (s + , ˉ θ),v o (s + , ˉ θ)), prefer regulation. Those with higher valuations prefer no regulation. The status quo of no regulation is assumed to be the collective choice with prior information, which requires that v o (s, ˉ θ)<γ?v or ?v parenleftBig γ? 1 n +1 parenrightBig > ˉ θN (n +1)α?v(1?αs) parenleftBigparenleftBig n?η+(n+1) parenleftBig η? 1 N parenrightBigparenrightBig ?v(1?αs)?(2?α?s)n ˉ θηN parenrightBig . (11) Consequently, s + = 0 corresponds to θ + = ˉ θ. For the media to influence the collective choice, a proportion γ of the public must prefer regulation when the media reports that the issue is serious. Letting the mean of the public’s posterior beliefs be denoted by ˉ θ H when the media reports that the issue is serious, regulation s + = s is preferred if and only if ˉ θ H N (n +1)α?v(1?αs) parenleftBigparenleftBig n?η+(n+1) parenleftBig η? 1 N parenrightBigparenrightBig ?v(1?αs)?(2?α?s)n ˉ θ H ηN parenrightBig ≥ parenleftBig γ? 1 n +1 parenrightBig ?v, (12) where s + = s corresponds to θ + = ˉ θ H . If (11) and (12) are satisfied, the news media influences the collective choice when it reports that the issue is serious. 13 This situation is illustrated in Figure 3. VI. The News Media: Subscribers and Sources 13 A necessary and sufficient condition for both (11) and (12) to be satisfied is ( ˉ θ H ? ˉ θ) bracketleftBigparenleftBig n?η +(n +1) parenleftBig η? 1 N parenrightBigparenrightBig ?v(1?αs)?(2?α?s)( ˉ θ H + ˉ θ)nηN bracketrightBig > s parenleftBig n?η +(n +1) parenleftBig η? 1 N parenrightBig α?v?nηN ˉ θ H parenrightBig . This is satisfied when ˉ θ H > ˉ θ, which will be the case in equilibrium, and (10) is satisfied for θ + = ˉ θ H + ˉ θ 2 . 16 A. Subscribers The media influences private politics because purchase decisions depend on the in- formation provided; i.e., v ? (s + ,θ + )dependsonthenewsreportthroughθ + . The media influences public politics if its report affects the collective choice between regulation s and no regulation as considered in Section V.D. The pivotal voter, however, must receive the report of the news media. Public and private politics depend on who and how many citizens see the news report. The members of the public must have a reason to spend the time to comprehend the news report. They may do so if subscribing to the media’s service is expected to benefit them by allowing them to make better decisions or because the news report has entertainment value. 14 The news media’s report is also a public good, since in principle a subscriber could share the information with others. The case considered here is that in which the news report spills over to all citizens. Because of this spillover the revenue of the media is problematic. Every citizen prefers to free-ride, yet if no one subscribes to the media service, no news report will be forthcoming. One perspective is that those who subscribe are those who expect to benefit in their private decisions. The media then prices its service given its potential subscribers, and those who expect to benefit by more that the price less the entertainment value subscribe. Appendix B presents a model of this perspective. Citizens must subscribe to the media service prior to receiving a news report, and hence the media’s revenue is independent of the report it eventually provides. 15 The subscribers are those citizens who expect to benefit in their private decisions by at least the price charged by the media less the entertainment value. The expectations of citizens are assumed to be rational and consistent with the equilibrium strategies of the interest groups and the news media. As indicated in Appendix B, the set of subscribers is increasing in the entertainment value and in the extent to which consumers internalize the externality. The profit of the news media is increasing in the expectations of extreme reports the media might provide. B. The Media’s Sources 14 This perspective is also adopted in Bovitz, Druckman, and Lupia. 15 This would not be the case in a repeated game. 17 When the media receives hard information, it is assumed to report that information to the public. If it concealed hard information provided by a source, it could be exposed for having distorted that information. Such exposure is assumed to be prohibitively costly to the media. As indicated in Section IX, if the media faced no risk of exposure if it concealed hard information, the equilibrium would be qualitatively similar to that in which the media reports hard information faithfully. That is, the event in which hard information is concealed by a source is the same event in which the media would conceal. The same conclusion results if the media investigates and finds hard information. Consequently, the media is assumed to report hard information. The only opportunity for discretion by the media thus occurs when neither the industry nor the activist provides hard information and either investigative journalism is not conducted or an investigation is conducted and it is unsuccessful. That is, the news media exercises discretion in its report only when it has no hard information. Theactivistandthefirmscouldcommunicatedirectlywithcitizensratherthanthrough the news media. However, because of their incentives to advocate their side of the issue, the credibility of their communications would be in doubt. Furthermore, citizens may have to incur prohibitively high costs to verify the content of the communications. In contrast, themediamaybeabletoevaluatemoreefficientlytheinformationpresentedbytheactivist and the industry. In addition, the media can investigate the issue if the activist and the industry provide no hard information. The media thus both reduces costs for the public and adds credibility through its new reports. Moreover, the news media can be a low-cost means of disseminating information to the public. The activist and the industry thus are assumed to communicate through the news media. Given the results of their searches, in the communication stage the activist and the firms acting as an industry independently send to the news media a message m i ∈M,i= A,F,whereA denotesactivistandF denotes firms and the message space is M = {θ L ,θ H ,φ} with φ denoting no information. If a search by either the activist or the in- dustry is unsuccessful, the only message it may send is m i = φ. If a search is successful, the activist and the industry are not required to report their information; i.e., they can conceal it by sending the message m i = φ. The activist and the industry thus can advocate their position by communicating favorable information to the media, but neither is forced 18 to provide unfavorable information. 16 A communication strategy of the industry is a probability distribution over the message set M.Thestateθ L is favorable to the industry, so if its search reveals that state, the industry has a dominant strategy of sending m F = θ L , in which case there will be no regulation. If the industry’s search reveals the unfavorable state θ H , it may send the message m F ∈{θ H ,φ}. That is, the industry cannot fabricate information, e.g., that θ = θ L , but it can remain silent about θ H by sending m F = φ. Let the probability that the industry sends m F = φ when θ H is observed be denoted by ρ F ∈ [0,1]. Similarly, the state θ H is favorable to the activist, and if its search reveals that state, it has a dominant strategy of sending the message m A = θ H .Ifθ = θ L , the activist conceals (m A = φ) the unfavorable information with probability ρ A ∈ [0,1] and sends m A = θ H with probability 1?ρ A . The activist and the industry may choose not to conceal unfavorable information if the media can independently search for information through investigative journalism. The information the media obtains from an investigation is denoted m M ∈{θ L ,θ H ,φ}.Ifits investigation is successful and one of the sources might have concealed information, the media can so state in its news report. This harms the reputation of the source that may have concealed the information, which mitigates the incentive to conceal information. C. Media Beliefs The media has no hard information when it receives the messages m A = φ and m F = φ and either does not investigate (m =(m A = φ,m F = φ)) or investigates unsuccessfully and has the information m prime =(m A = φ,m F = φ,m M = φ). The media’s beliefs p φ when it does not investigate are p φ = Pr(θ = θ H | m A = m F = φ)= p(1?q + qρ F ) p(1?q + qρ F )+(1?p)(1?q + qρ A ) . (13) It is straightforward to show that when it unsuccessfully investigates Pr(θ = θ H | m A = m F = m M = φ)=p φ , since the probability q M of a successful investigation is independent of the state. The media’s beliefs are thus the same when its sources provide no hard information and when it investigates and receives no information or does not investigate. 16 See Dewatripoint and Tirole (1999) for a theory of advocacy. 19 The media’s beliefs have posterior mean ˉ θ o given by ˉ θ o = p(1?q + qρ F )θ H +(1?p)(1?q + qρ A )θ L p(1?q + qρ F )+(1?p)(1?q + qρ A ) , (14) which is strictly increasing in ρ F and strictly decreasing in ρ A . That is, if the industry conceals with higher probability, it is more likely that m F = φ resulted from state θ H .If the activist concealed with higher probability, it is more likely that m A = φ resulted from state θ L . Also, the higher is the quality q of the search the greater is the opportunity for the activist and the industry to conceal unfavorable information. The posterior mean is strictly increasing in q if and only if the industry conceals with a higher probability than the activist. As demonstrated below, in equilibrium only the activist conceals, so ˉ θ o is decreasing in the quality of the search. The difference between the media’s posterior mean and the prior mean is ˉ θ o ? ˉ θ = p(1?p)q(ρ F ?ρ A )(θ H ?θ L ) p(1?q + qρ F )+(1?p)(1?q + qρ A ) . (15) The direction of influence of the activist and the industry is as expected. The greater the probability the activist (industry) conceals the smaller (larger) is ˉ θ o ? ˉ θ. D. News Reports and the Public’s Beliefs To identify the public’s beliefs, the report set of the media must first be identified; i.e., the set of reports the news media can effectively send. If the media sends a report r/∈{θ L ,θ H }, the public, which is rational, infers that the media has no hard information, so if r ∈ (θ L ,θ H ), the public makes the correct inference ˉ θ o about θ. Consequently, if the media has no hard information, its effective set of reports is M = {φ L ,φ H , ˉ θ o }.Anybias by the news media thus must take the form of reporting r = θ H or r = θ L when it has no hard information. The public’s beliefs depend on whether the media conducts investigative journalism, which is not observable to the public, and what it reports when it has no hard information. Let ξ ∈ [0,1] denote the probability the media investigates when (m A = m F = φ), and let ω denote the probability the media reports r = θ H and 1 ? ω denote the probability the media reports r = ˉ θ o when m A = m F = φ and the media does not investigate. Similarly, let λ denote the probability the media reports r = θ H and 1?λ denote the probability the 20 media reports r = ˉ θ o when the media investigates unsuccessfully. The public’s posterior beliefs then are Pr(θ = θ H | r = θ H )= pPr(r = θ H | θ = θ H ) pPr(r = θ H | θ = θ H )+(1?p)Pr(r = θ H | θ = θ L ) , (16) where Pr(r = θ H | θ = θ L )=((1?q)qρ A +(1?q) 2 )(ξ(1?q M )λ +(1?ξ)ω), (17) and Pr(r = θ H | θ = θ H )=q(2?q?ρ F (1?q))+(1?q)(1?q(1?ρ F ))(ξq M +ξ(1?q M )λ+(1?ξ)ω). (18) The probability Pr(θ = θ H | r = θ H ) in (16) is strictly decreasing in ρ A if ω>0or λ>0and,ifω<1orλ<1, is strictly decreasing in ρ F .Givenr = θ H , a higher probability of concealment by the activist reduces the confidence of the public that θ = θ H ,sincethe activist conceals when θ = θ L . A higher probability ρ F of concealment by the industry reduces the public’s posterior probability that the issue is serious because there is a lower probability of being presented with hard information that θ = θ H . If, however, the media reports r = θ H with probability ω = λ = 1 whenever it has no hard information, the probability in (16) is constant in ρ F . If ω = λ = 1, that is, the media biases its report whenever possible, the probability Pr(θ = θ H | r = θ H )isincreasinginξ. That is, investigative journalism increases the confidence of the public in the report because the opportunity for bias is smaller. The higher the probability (ω or λ) that the media biased its report, however, the lower is the probability in (16). The public thus exhibits skepticism. The public’s beliefs ˉ θ H about the posterior mean when it receives a report r = θ H are ˉ θ H = Pr(θ = θ H | r = θ H )θ H +(1?Pr(θ = θ H | r = θ H ))θ L = θ L +(θ H ?θ L )Pr(θ = θ H | r = θ H ). (19) From (16)-(18) the public’s posterior mean inherits the properties of Pr(θ = θ H | r = θ H ). The posterior mean ˉ θ H is strictly decreasing in ω and λ, so the more likely is the news media to bias its report, the more skeptical is the public. Conversely, the higher is the quality q M 21 of the media’s investigation the higher is ˉ θ H , since the public has more confidence in the media’s report. If the media reports r = θ H with probability ω = λ =1whenithasno hard information, ˉ θ H is constant in ρ F and is strictly increasing in ξ. E. Exposing the News Media for Bias If the news media biases its report when it has no hard information, the activist or the industry could expose the bias if it had hard information. Neither, however, has the incentive to do so. The opportunity for exposure is when the activist or the industry has hard information and concealed it, so the party that concealed the information would be exposing itself as well. More importantly, the party that concealed the information has no incentive to expose the media, since that would negate the reason for the concealment. Suppose that the activist concealed the unfavorable information θ L and the news media reported r = θ H . This report serves the interests of the activist, and hence it has no incentive to expose the bias. If the media were to report r = θ L , the activist could not expose the media because the report was accurate. Similarly, if the industry concealed the unfavorable information θ H , it has no incentive to expose a report r = θ L and cannot expose an accurate report r = θ H . Consequently, the media faces no risk of exposure. VII. News Reports and Investigative Journalism A. Media Preferences As indicated in Section III, the news media is widely viewed as biased, although poll respondents disagree about the nature and direction of that bias. Bias could result from an ideological preference for regulation among the journalists in or the owners of the news organization or because of their beliefs about the seriousness of the issue. Bias in such a case comes directly from individual preferences or beliefs and begs the question of why journalists and the news organization have those preferences or beliefs. The approach taken here is to represent the preferences of the news media in terms of fundamental measures of profits and journalistic considerations. The media as a private institution provides information to the public to help people make better decisions both in their roles as consumers and their roles as citizens. The consequences of their decisions will be represented by aggregate public welfare W,whichis identified below. The media also has a role in discovering misrepresentation in information 22 provided by its sources. Investigative journalism is the instrument for uncovering misrep- resentation, but investigative journalism is costly. These considerations are represented by the utility U M of the media specified as U M = pi ? M + ξ(?c + χW ξ=1 + δJ)+(1?ξ)χW ξ=0 , (20) where pi ? M is profit as identified in Appendix B, c is the cost of investigating, χ is the weight given by the media to aggregate public welfare, J is the utility from journalistic performance, δ is the weight attached to it, and W ξ=1 and W ξ=0 are, respectively, the aggregate welfare if the media investigates or not. As indicated in Appendix B, profits result from subscriptions based on rational expectations of subsequent media reports and hence are unaffected ex post by the media’s investigative journalism or news report. The media’s journalistic preferences are specified below. B. News Reports The media’s opportunity for influence is when the activist and the industry provide no hard information and the media either does not investigate or investigates without success. Consider the strategy of biasing its report by reporting r = θ H when it has no hard information. This strategy has two implications. First, the news media may have a weaker incentive to investigate because investigating reduces the probability it will have the opportunity to exercise its discretion. Second, the activist may have a greater incentive to conceal θ = θ L so as to increase the opportunity of the news media to exercise its discretion in the direction the activist prefers. The industry has no counteractive incentive because it can decrease the media’s opportunity for discretion only by reducing the probability it conceals unfavorable information θ H . Consequently, there is a synergy between the media’s incentive to bias its report and the group whose interests are advantaged by the bias. That group has an incentive to conceal unfavorable information, so as to increase the opportunity for the media to bias its report. Journalistic standards and the media’s responsibility as a institution serving the pub- lic, however, call for unbiased reports. 17 If the news media is concerned about violating standards by biasing its report, it may incur a cost in the form of a reduction in employee 17 Codes of ethics and responsibilities for journalists can be found at www.asne.org/ideas/codes. Also see the Poynter Institute Web site www.poynter.org. 23 morale or a loss of pride in its professionalism. Goldberg (2002), however, argued that the news media biases its reports but does not understand that it is doing so. The media could, for example, just believe that a report r = θ H is warranted when it has no hard information because that helps mitigate the externality. Because the cost of providing a biased report is unclear, only a bound on that cost will be identified. The media’s news report comes after its decision to investigate, so the report depends only on aggregate public welfare. The media evaluates aggregate public welfare in terms of its own information, which when it has no hard information is the posterior mean ˉ θ o in (14). The media’s evaluation W(s + ,θ + , ? θ) of aggregate public welfare when it has no hard information, has beliefs ? θ ∈{θ L ,θ H , ˉ θ o }, and induces beliefs θ + ∈{ ˉ θ L , ˉ θ H , ˉ θ o } of the public is 18 W(s + ,θ + , ? θ)= integraldisplay ?v v ? (s + ,θ + ) ((1 ? αs + ) ? (1 ? s + )(ηN ? 1)θ + )v N ?v dv ? integraldisplay ?v 0 (1 ?s + ) ? θN parenleftBig 1 ? v ? (s + ,θ + ) ?v parenrightBig N ?v dv = X(s + ,θ + ) bracketleftBig (1 ?αs + ) parenleftBig ?v + v ? (s + ,θ + ) 2 parenrightBig ? (1 ? s + )(ηN ? 1)θ + ? (1 ?s + ) ? θN bracketrightBig . This is linear and strictly decreasing in ? θ. To determine the direction of possible media bias, consider first the alternative of reporting r = θ L when the media has no hard information. This would result in no regulation, since the posterior mean ˉ θ L is less than ˉ θ o . In addition, more consumers would purchase, worsening the externality. Consequently, the media will not report r = θ L when it has no hard information. The media thus reports r = θ L only when it has received hard information, so Pr(θ = θ L | r = θ L )=1. This implies that the conditional mean is ˉ θ L = θ L . If the media reports r = ˉ θ o , the public believes that the media has no hard information. As will be demonstrated below, in equilibrium the industry has no incentive to conceal, so ρ F =0.Then,from(15) ˉ θ o < ˉ θ, so citizens choose not to regulate. If the media reports r = θ H , the public understands that the report could be biased and has posterior beliefs ˉ θ H given in (19). Although ˉ θ H <θ H , the posterior beliefs are still 18 Aggregate welfare includes the profit of the firms, which are assumed to be owned by the public. The media’s profits are not included, since they are determined ex ante and are independent of its subsequent actions. 24 greater than the prior mean; i.e., ˉ θ H > ˉ θ. The public thus can be influenced by the report even though it knows that the information might be biased. The media then prefers to report r = θ H when it has no hard information if W(s, ˉ θ H , ˉ θ o )? W(0, ˉ θ o , ˉ θ o ) is positive. This condition is W(s, ˉ θ H , ˉ θ o )?W(0, ˉ θ o , ˉ θ o )=X(s, ˉ θ H ) bracketleftBig (1?αs) parenleftBig ?v + v ? (s, ˉ θ H ) 2 parenrightBig ?(1?s)(ηN ?1) ˉ θ H ?(1?s) ˉ θ o N bracketrightBig ?X(0, ˉ θ o ) bracketleftBig ?v + v ? (0, ˉ θ o ) 2 ?(ηN ?1) ˉ θ o ? ˉ θ o N bracketrightBig . (21) To illustrate this condition, consider the case in which α = 1 and consumers do not take into account (η = 1 N ) the effect of the externality on others. Then, (21) is W(s, ˉ θ H , ˉ θ o )?W(0, ˉ θ o , ˉ θ o )= nN (n +1) 2 ?v parenleftBig (n +2)?v? ˉ θ o (2(n +1)N ?n) 2 parenrightBig ((1?s)(?v? ˉ θ H )?(?v? ˉ θ o )), which is positive when N is large. The media then is influential, which is the case of interest here. When the media reports r = θ H when it has no hard information, the maximum cost ?c J of violating journalistic standards the media is willing to incur is defined by ?c J ≡ χ(W(s, ˉ θ H , ˉ θ o )?W(0, ˉ θ o , ˉ θ o )). If the cost is greater than ?c J , the media reports truthfully. When the cost is below ?c J ,the media reports r = θ H with probabilities ω ? =1andλ ? = 1. The media prefers to bias its report because of its effects on private and public politics. Through private politics the report reduces demand, and through public politics it leads to regulation which increases public welfare. The remainder of this section considers the case in which the cost is less than the bound, so the media biases (λ ? = ω ? = 1) its report when it has no hard information. C. Journalistic Performance and Investigative Journalism The media may gain prestige if through investigative journalism it uncovers conceal- ment by the activist or the industry. The media cannot uncover concealment without error, however, because it does not observe the information received by its sources. The media knows the equilibrium strategies of the activist and the industry, however, and knows that 25 the activist will never conceal information θ H and the industry will never conceal θ L .Con- sequently, if the media investigates and observes θ H and the industry has sent the message m F = φ, the industry may have concealed. If the media receives the messages (m A = φ,m F = φ,m M = θ H ), it believes the industry concealed with probability ˉρ F given by ˉρ F = qρ F qρ F +1?q . (22) Similarly, the posterior probability ˉρ A the activist concealed when (m A = φ,m F = φ,m M = θ L )is ˉρ A = qρ A qρ A +1?q . (23) Journalistic performance J in (20) will be represented as the gain in prestige for the news media from exposing concealment by its sources, weighted by the posterior probability that the source concealed information. Letting b i ,i= A,F, represent the prestige, the gains from exposing the activist and the industry, respectively, are ˉρ A b A and ˉρ F b F . The media is assumed to investigate only when it has no hard information. The expected utility EU M (ξ) of the media as a function of the probability ξ of investigating conditional on (m A = φ,m F = φ)isthen 19 EU M (ξ)=pi ? M + ξ bracketleftBig q M parenleftBig p φ (δb F ˉρ F ?c + χW(s, ˉ θ H ,θ H )) + (1?p φ )(δb A ˉρ A ?c+ χW(0,θ L ,θ L )) parenrightBig +(1?q M )(?c + χW(s, ˉ θ H , ˉ θ o )) bracketrightBig +(1?ξ)χW(s, ˉ θ H , ˉ θ o ), (24) where p φ is given in (13), W(s, ˉ θ H ,θ H ) is the media’s evaluation of aggregate welfare when its search reveals θ = θ H and it reports r = θ H and W(0,θ L ,θ L )iswelfarewhenitssearch reveals θ = θ L . 20 19 Note that the media receives b i ,i = A,F, only if it investigates and not when one interest group provides hard information and the other provides no information. If one of its sources provided hard information, investigative journalism would be meaningless. Moreover, professionalism would likely prevent the media from falsely claiming that it had investigated. 20 In (20), W ξ=0 = W(s, ˉ θ H , ˉ θ o ), and W ξ=1 = q M (p φ W(s, ˉ θ H ,θ H )+(1?p φ )W(0,θ L ,θ L ))+ (1?q M )W(s, ˉ θ H , ˉ θ o ). 26 The media’s optimal strategy ξ ? for investigative journalism satisfies ξ ? ? ? ? ? ? ? ∈ (0,1) if dEU M (ξ) dξ =0 =0 if dEU M (ξ) dξ ≤ 0 ? ξ ∈ (0,1] =1 if dEU M (ξ) dξ > 0 ? ξ ∈ [0,1), where from (24) dEU M (ξ) dξ = q M bracketleftbig p φ δb F ˉρ F +(1?p φ )δb A ˉρ A + χ?W bracketrightbig ?c (25) and ?W reflects the informational consequences of an investigation or ?W ≡ p φ W(s, ˉ θ H ,θ H )+(1?p φ )W(0,θ L ,θ L )?W(s, ˉ θ H , ˉ θ o ). The sign of ?W cannot be determined unambiguously, but the expected demand with a successful investigation can be shown to be greater than with an unsuccessful investigation. Investigative journalism is thus motivated by uncovering concealment by the activist or the industry and possibly by greater expected public welfare due to the better information it provides. If ?W<0, the only incentive to investigate is due to journalistic performance. VIII. Communication: Competition Between the Activist and the Industry The activist and the industry compete to influence the public by sending messages to the news media. Each has an opportunity to conceal unfavorable information and thus present a more favorable case to the media. The opportunity to conceal unfavorable infor- mation, however, occurs only if the other player has not observed that information. For example, suppose the state is θ H . If the industry observed θ H , it could send the message m F = φ. If the activist observed θ H , however, it has a dominant strategy of sending m A = θ H . Consequently, the industry can conceal the state only if the activist does not observe it. In addition, if the activist’s search reveals no information, the media may in- vestigate and observe m M = θ H , which would then cast suspicion on the veracity of the industry’s message m F = φ. The communication strategies of the activist and the industry also depend on the media’s reporting strategy. In the case considered, the media reports r = θ H whenever it has no hard information, so the public’s beliefs ˉ θ H are constant in ρ F . The industry then cannot influence the report received by the public. The public’s beliefs ˉ θ H are decreasing 27 in the activist’s concealment strategy ρ A , so the public becomes more skeptical of a news report r = θ H when the activist conceals θ = θ L with higher probability. This results because concealment by the activist gives the media a greater opportunity to provide a biased report. The news media’s incentive to bias its report also influences its investigation strategy,whichinturnaffectstheconcealmentstrategies. Concealment poses a risk. For example, the industry may incur reputation damage H F ≥0 if it is suspected of having concealed; i.e., the media reports r = θ H and the industry sent m F = φ. Thatdamagewillbeassumedtooccurthroughthenewsreportandin proportion to the posterior probability ˉρ F that the industry concealed given either m A = θ H or (m A = φ,m M = θ H ), which is given in (22). The expected utility U F (ρ F ;m i = θ H ,i= AorM) of the industry conditional on having observed θ H and concealed with probability ρ F is U F (ρ F ;m i = θ H ,i= AorM)=npi ? i (s, ˉ θ H )? ˉρ F H F , (26) since the media reports r = θ H with probability 1 when it receives m A = θ H or m M = θ H and regulation s results from the public politics. To determine the industry’s optimal concealment strategy, suppose the industry ob- served θ H , concealed it, and the media investigated with probability ξ. With probability q M the media observes θ H , in which case the utility of the industry is that in (26). With probability 1?q M the media observes nothing and reports r = θ H , so the industry’s profit is given by the first term in (26). This is also the outcome if the media does not investigate. Consequently, regardless of the concealment strategy of the industry the report received by thepublicisr = θ H , and the industry’s profits are npi ? i (s, ˉ θ H ). Since the industry suffers damagetoitsreputationifitconcealsandtheactivistorthemediareceiveshardinfor- mation that the state is θ H , the industry cannot gain through concealment. Consequently, ρ ? F = 0 is optimal. The activist is in a different position. If it conceals unfavorable information θ L ,it increases the opportunity for the media to report r = θ H . This opportunity is present only in the events in which the industry observed nothing and the media did not investigate or investigated and observed nothing. Concealment also poses a risk to the activist if the industry sends m F = θ L or the media investigates and observes m M = θ L . The damage to the reputation of the activist is denoted H A ≥0, and the expected damage is assumed to be 28 proportional to the probability ˉρ A in (23) that the activist concealed. The activist is assumed to have preferences over both regulation and the state, but the activist conceals only when it observes θ L , so its preferences are conditional on θ = θ L . The activist has an incentive to conceal only if it prefers regulation s with r = θ H to no regulation with r = θ L . In this case, its preferences satisfy U A ( ˉ θ H )>U A (θ L ), where the activist’s utility function is denoted by U A (·). 21 The expected utility EU A (θ L ) of the activist conditional on having concealed θ L is EU A (θ L )=(q +(1?q)ξ ? q M )(U A (θ L ) ? ˉρ A H A )+(1? q)(1 ?ξ ? q M )U A ( ˉ θ H ), where the probability q corresponds to m F = θ L and the probability (1?q)ξq M corresponds to the media investigating successfully given m F = φ. The expected utility EU A (ρ A ;θ L ) of the activist as a function of its concealment strategy is EU A (ρ A ;θ L )=ρ A EU A (θ L )+(1?ρ A )U A (θ L ). Thederivativeis dEU A (θ L ) dρ A =(1?q)(1 ? ξ ? q M )(U A ( ˉ θ H ) ?U A (θ L )) ? (q +(1?q)ξ ? q M )ˉρ A H A . (27) The first term in (27) is positive, and the second term is negative and represents a check on the incentive to conceal resulting from the possible damage to the activist’s reputation. Evaluating (27) at ρ A =0yields dEU A (θ L ) dρ A |ρ A =0 =(1?q)(1?ξ ? q M )(U A ( ˉ θ H )?U A (θ L )) > 0, so the equilibrium concealment probability ρ ? A is positive. The activist thus conceals unfa- vorable information with positive probability. This results regardless of whether the media investigates. The optimal concealment probability equals 1 if the damage H A is small, and a greater H A can result in ρ ? A ∈ (0,1). An interior equilibrium then can exist if the derivative in (25) equals 0 for (ξ ? ,ρ ? A ) ∈ (0,1)×(0,1). The equilibrium concealment and investigative journalism strategies identified by (25) and (27) could take a variety of forms. The activist conceals with probability 1 if the 21 For example, aggregate welfare could be greater with regulation when θ = θ L . 29 damage to its reputation is no greater than the bound ˉ H A given by ˉ H A ≡ (1?q)(1?ξ ? q M )(U A ( ˉ θ H )?U A (θ L )) q(q +(1?q)ξ ? q M ) , (28) where ξ ? is the corresponding investigation probability. 22 Then, from (25) if the cost of investigative journalism is low, the media investigates with probability 1; i.e., c ≤ q M ((1?p φ )δb A q + χ?W | ξ=1 ). Consequently, for some parameter values the equilibrium is (ρ ? A =1,ξ ? =1).Asthequality q M of the investigation increases, however, the bound ˉ H A in(28)goesto0whenξ ? =1, in which case ρ ? A <1. Similarly, for higher c, ξ ? <1. In the equilibrium the public is skeptical of the report r = θ H because the activist conceals θ = θ L with positive probability. That is, using (18) and (19) with ρ ? F =0, λ ? =1,andω ? = 1, the posterior mean is ˉ θ H <θ H . If off the equilibrium path the media were to report r ∈ (θ L ,θ H ), the public would believe that the media’s posterior mean was ˉ θ o ,andfrom(15) ˉ θ o < ˉ θ. The public then would not choose regulation. Regulation thus results with ex ante probability Pr(s)=p +(1?p)(1?q)(1?q(1?q(1?ρ ? A + ρ ? A ξ ? q M )). (20) IX. Concealment by the Media The media may have the opportunity and ability to conceal hard information provided by one of its sources, although professionalism may prevent it from doing so. When pro- fessionalism does not prevent concealment, the media may face the risk of exposure by one of its sources. This risk could take the form of a lawsuit for defamation or the source taking its information to the public or to its allies in government. Even if that were not the case, the media may have no incentive to provide a report inconsistent with the hard information it receives. This results from the media’s preferences for regulation when its beliefs are θ = θ H . 22 In this case, if the media observes an off-the-equilibrium-path message m A = θ L ,its beliefs are assumed to be that θ = θ L . Since it then “knows” the state, its report would be r = θ L . 30 The hard information messages the media may receive from its sources are presented in Table 1, and associated with each message pair is whether the media has an incentive to conceal. The first column pertains to the case in which the public’s welfare is greater (W(0,θ L ,θ L )≥W(s,θ + ,θ L )) without regulation when θ = θ L , and the second column is the case in which welfare is greater with regulation. When regulation is not preferred in θ = θ L , the media has no incentive to conceal because a truthful report results in the maximal public welfare. Also, when θ = θ H , the media also has no incentive to conceal. Table 1 Concealment by the Media Incentive to Conceal Messages (s + =0) (s + = s) (m A = θ H ,m F = θ H )no no (m A = θ H ,m F = φ)no no (m A = θ L ,m F = θ L )no yes (m A = θ L ,m F = φ)no yes (m A = φ,m F = θ H )no no (m A = φ,m F = θ L )no yes When regulation is preferred in θ = θ L , the media has an incentive to conceal a message m i = θ L ,i = A,F, and report r = θ H , resulting in regulation. If the media conceals the message m F = θ L , however, the industry would expose it, which is assumed to be prohibitively costly to the media. 23 Only in the case (m A = θ L ,m F = φ)will concealment not be exposed. This case, however, is the one in which concealment takes place in the equilibrium characterized above. The qualitative effect on private and public politics would thus be the same whether the activist or the media concealed θ L . The same conclusion results if the media can conceal the results of its own investigative journalism. It has no incentive to do so when m M = θ H ,andwhenm M = θ L ,ithasan 23 For example, NBC’s Dateline fabricated pickup truck fires due to side-impact collisions and was exposed by General Motors, leading to the firing of the head of NBC News and others. See the case “General Motors Like a Rock (A)” by Keith Krehbiel in Baron (2003a), pp. 85-88. 31 incentive to conceal only when it prefers regulation when θ = θ L . If it prefers regulation, the media faces a risk of exposure if the industry had sent m F = θ L . Consequently, the media has an incentive to conceal only when the industry report is m F = φ. But, in this case the media has no incentive to investigate. That is, if m A = θ L and m F = φ, the media knows the state and has no incentive to investigate. If m A = θ H and m F = φ, the media also has no incentive to investigate, since a truthful report results in regulation. Consequently, the only incentive for the media to investigate is due to journalistic performance to expose concealment from its sources when neither provides information. This is the case considered above in Section VII.C. X. Moral Suasion The strategies of the activist, the firms, and the news media could affect the private and collective decisions of the public not only through beliefs but also through preferences. For example, greater recognition of the externality could lead some citizen consumers to assume a duty to internalize a portion of the externality. For example, voluntary recycling has become widespread over the past two decades as a result of the recognition of the externalities associated with a variety of products. In the model, this is represented by the parameter η, which could depend on the treatment given to the issue in the news report. This treatment could take the form of normative arguments that affect the preferences of the public beyond the information about the seriousness θ of the externality contained in the news report. When the media treats the issue in a normative manner, it is participating in private politics through moral suasion. In the model, moral suasion that affects η also affects citizens’ preferences for regulation, so moral suasion affects public politics. The responsiveness of citizen consumers to the media’s treatment of the issue also could provide an incentive for informational bias, and this could provide an incentive for the activist to conceal unfavorable information. The activist and the industry could also offer normative arguments, but because of credibility concerns their arguments might not have the same influence as those of the news media. Moral suasion still is attempted by activists and firms, however. The cost of contacting the public through advertisements is moderate, particularly when the news media views the advertisements themselves as newsworthy. For example, in January 2003 the Detroit Project led by Arianna Huffington spent nearly $200,000 for television adver- 32 tisements linking SUVs to the support of terrorism. The ads asked “What is your SUV doing to our national security?” Showing armed terrorists in a desert, the ads stated “And these are the terrorists who get money from those countries every time George fills up his SUV.” Other campaigns were also underway. The Evangelical Environmental Network (EEN) declared that “transportation is a moral issue” and ran TV advertisements asking, “What would Jesus drive?” The answer is not an SUV. 24 Also, “Earth on Empty, a group of Boston artists, plasters fake parking tickets on S.U.V. windshields that instruct drivers to ‘try to get honest with yourself.’ ... The posters at a recent antiwar rally in San Fran- cisco said, ‘Draft S.U.V. drivers first.’” 25 Whether these arguments have influenced either private or public politics is unclear, but activists as well as their targets frequently make such arguments to the public. In the context of the model, η could be represented as an increasing function of the forcefulness of moral suasion about the importance of internalizing some portion of the ex- ternality; i.e., about other-regardedness. Then, based on the above analysis, moral suasion has the following effects: 1. Moral suasion reduces the demand for the product and hence the harm from the exter- nality. Moral suasion is thus a substitute for regulation. 2. Moral suasion decreases the public support for regulation because it reduces the extent of the externality. Moral suasion thus affects public politics. 3. Moral suasion can eliminate the informational influence of the news media, and hence the incentive to bias its report, if the pivotal voter no longer prefers regulation when r = θ H . 4. Moral suasion shifts downward the demand function for the product which results in a lower price and lower profits in the industry. 5. Moral suasion increases the profit of the news media by shifting the demand function; i.e., increasing the number of subscribers and allowing a higher price to be charged. 26 Moral suasion through the news media could be either a complement to or a substitute for bias in media reports. It could give the activist a stronger incentive to conceal informa- 24 BoththeEENandDetroitProjectcampaignswereconductedby FentonCommunications. 25 The New York Times, February 8, 2003. 26 See Appendix B. 33 tion to give the news media a greater opportunity to have moral as well as informational influence. Moral suasion itself could be viewed as another form of media bias. XI. Conclusions The news media obtains information from sources as well as from its own investigative journalism. Its news reports influence private and public politics, and this provides incen- tives for its sources to provide information strategically. The media has discretion in its choice and presentation of the news and thus can bias its reports. Bias could result from the preferences of journalists or the owners of the media organization, but it could also arise in response to the role in society the media has assumed—as a protector of democracy and a provider of information to the public. In response, the media could report in an unbiased manner, but it could also act to mitigate collective action problems. The collective action problem considered here is a free-rider problem associated with an externality. By biasing its reports the news media could mitigate both a market failure and a government failure by reducing demand and by affecting preferences for regulation, respectively. The media’s incentives for bias are fully taken into account by citizen consumers in both their consump- tion and their collective choices, yet biased reports have influence because citizen consumers can neither verify the reports nor the information provided by the media’s sources. Media coverage then can provide incentives for sources to advocate their interests by presenting favorable information and concealing unfavorable information. In the model the group whose interests are aligned with the fulfillment of the media’s role in society has an incentive to conceal unfavorable information, whereas the group whose interests would be harmed by the public’s response to the bias has no incentive to conceal information. An alignment between the media and certain interest groups thus results. This alignment could, but need not, be seen as an ideological orientation of the media through behavior that favors the causes of certain interest groups. An ideological orientation could also result from attracting journalists with particular personal preferences that are consistent with the alignment arising between the media’s objectives and those of interest groups. This has implications for interest group strategies. The interest group aligned with the media’s objectives may focus on private politics, whereas the opposing group may be forced to focus on public politics and rely on the skepticism of the public. 34 The news media and the interest groups may also have an opportunity to influence the preferencesof citizenconsumers throughmoral suasion. Inthe model moral suasionpertains to the extent to which consumers take into account the effect on others of the externality associated with their own consumption. Moral suasion in the model is a substitute for regulation in two senses. First, it reduces the magnitude of the externality by reducing demand. Second, it diminishes the public support for regulation. The strategic competition among interests and the role of the news media in providing moral arguments to the public remains a subject for future research. The opportunity to bias news reports leaves a burden on the news media to be certain that its bias is indeed in the interests of the public rather than in the interests of the journalists and media owners who comprise the news organization. The institutions of government and the collective action of citizens are intended to serve the interests of the public, and the fundamental issue is whether the role of the news media should extend to market and nonmarket failures that governments fail to address. Thomas Jefferson wrote, “Were it left to me to decide whether we should have a government without newspapers or newspapers without a government, I should not hesitate to prefer the latter.” 27 In the presence of a government, however, should the role of the news media in society be more limited? 27 Letter to E. Carrington, 1787. 35 Appendix A Characterization of the Indifferent Citizen in Collective Choice To show that some purchasers prefer regulation and others do not, note that the difference v o (s + ,θ + )?v ? (s + ,θ + )ispositiveif ?v parenleftBig (1?αs + ) parenleftBig 1? n +1 nN parenrightBig ?(α?1)η parenrightBig ?θηN(2?α?s + ) > 0. (A1) The condition in (A1) is satisfied for ?v ≥ θ + ηN(2?α?s + ) (1?αs + )(1? n+1 nN )?(α?1)η, whichissatisfiedforη small. The condition for ?v>v o (s + ,θ + ) is identified by the difference ?v?v o (s + ,θ + )= n (n +1)α?v(1?αs + ) bracketleftBig ?v(1?αs + ) parenleftBig α?v?θ + N parenleftBig 1+η? n +1 nN parenrightBigparenrightBig +(θ + ) 2 N 2 η(2?α?s + ) bracketrightBig . (A2) The term in brackets is assumed to be positive. Appendix B Subscribers and Media Pricing A. The Subscribers The subscription decision must be made prior to any report by the media, so the subscribers are those who expect to benefit from the reports by enough to bear the cost. That cost is the difference between the price charged by the media and the entertainment value of its reports. To identify the subscribers, consider those consumers who would gain from the reports provided by the media. The news media can make three generic reports r ∈{θ L ,θ H , ? θ}, where ? θ ∈ (θ L ,θ H ) is the report of the media in the event that neither the industry nor the activist has presented hard information nor has an investigation by the media been successful. As demonstrated in Section VI.C, the report ? θ leads to beliefs ˉ θ o in (14). The citizens who purchase corresponding to the three reports are identified based on the cutoffs v ? (0, ˉ θ L ),v ? (s, ˉ θ H )andv ? (0, ˉ θ o ), respectively, where, as identified in Section VII, ˉ θ j is the 36 posterior mean given a report r = θ j ,j = L,H,ands + = 0 corresponds to the posterior mean ˉ θ o . The cutoffs can be shown to be ordered as v ? (0, ˉ θ L )<v ? (0, ˉ θ)<v ? (s, ˉ θ H ), where v ? (0, ˉ θ) is the cutoff based on the prior mean ˉ θ in the absence of any media report. Also, v ? (0, ˉ θ L )<v ? (0, ˉ θ o )<v ? (s, ˉ θ H ), but v ? (0, ˉ θ o )couldbegreaterorlessthanv ? (0, ˉ θ). To reduce the number of cases considered, only that in which v ? (0, ˉ θ o )<v ? (0, ˉ θ)isanalyzedhere.This condition is satisfied in the equilibrium. A citizen’s subscription decision depends on her expectations about the reports she might receive from the media. Those expectations are necessarily formed prior to sub- scribing and are assumed to be rational in the sense that they are consistent with the equilibrium strategies of the activist, firms, and media. The expectations are denoted by ψ =(ψ L ,ψ H ,1 ? ψ L ? ψ H ), where ψ i is the probability of r = θ i ,i = L,H.Those expectations are identified in Section C of this Appendix. Those consumers with v ∈ [v ? (0, ˉ θ L ),v ? (0, ˉ θ o )] benefit from the news report r = θ L , since they will purchase, whereas based on prior information or reports r = θ H or r = ˉ θ o they would not. Each consumer is assumed to understand the subscription strategies of other citizens and anticipates the collective decision. A consumer’s gain thus is only due to the change in her consumption strategy. That gain is v? 1 n+1 ?v? n n+1 η ˉ θ L N. The expected gain G L is G L = ψ L parenleftBig v ? 1 n +1 ?v ? n n +1 η ˉ θ L N parenrightBig = ψ L (v ?v ? (0,θ L )), if v ∈ [v ? (0, ˉ θ L ),v ? (0, ˉ θ o )]. (B1) Similarly, a consumer with v ∈ (v ? (0, ˉ θ),v ? (s, ˉ θ H )] gains when the report is r = θ H , since in that event she will not purchase, whereas she would have purchased based on prior information as well as when r = θ L or r = ˉ θ o . The expected gain G H is G H = ψ H parenleftBig n n +1 (1?s) ˉ θ H ηN ?(1?αs) parenleftBig v? 1 n +1 ?v parenrightBigparenrightBig = ψ H (1?αs)(v ?v ? (s, ˉ θ H )), if v ∈ (v ? (0, ˉ θ),v ? (s, ˉ θ H )]. (B2) Aconsumerwithv ∈ [v ? (0, ˉ θ o ),v ? (0, ˉ θ)] gains both if r = θ L or r = ˉ θ o ,inwhichcases she will purchase whereas she would not have purchased based on prior information. The 37 corresponding expected gain G( ˉ θ o )is G( ˉ θ o )=ψ L parenleftBig v ? 1 n +1 ?v ? n n +1 η ˉ θ L N parenrightBig +(1?ψ L ?ψ H ) parenleftBig v ? 1 n +1 ?v ? n n +1 ˉ θ o ηN parenrightBig , (B3) where s + = 0 is assumed to correspond to ˉ θ o . The difference ?G in the gains G L and G( ˉ θ o )in(B2) and (B3) for valuations v L ∈ (v ? (0, ˉ θ L ),v ? (0, ˉ θ o )] and v ∈ [v ? (0, ˉ θ o ),v ? (0, ˉ θ)], respectively, are ?G = G( ˉ θ o )?G L =(1?ψ H ?ψ L ) parenleftBig v ? 1 n +1 ?v ? n n +1 ˉ θ o ηN parenrightBig + ψ L (v ?v L ) > 0, since v L <v. Consequently, the gains to consumers with valuations closer to the prior mean are greater than the gains to consumers with lower valuations. This means that as the media raises the price for a subscription the consumers who decide not to purchase are those with the more extreme valuations, as identified next. B. The Media’s Pricing The media is assumed to charge a price k for a subscription. A subscription could also have an entertainment value e≥0, in which case the net price is k ?e.Ifthenewsreports are provided at a price of zero as in the case of television and radio broadcasts, k could be interpreted as advertising revenue per subscriber. From the consumer’s perspective k then represents the cost of enduring the advertising that is interspersed with the news reports. If k>e, the number of subscribers S(k ?e) includes those citizen consumers with v ∈ [v ? (0, ˉ θ L ),v ? (s, ˉ θ H )]whohavegainsofatleastk ?e, as illustrated in Figure 4. Using (B1) and (B2) the number of subscribers are S(k ?e)= N ?v parenleftBig v ? (s, ˉ θ H )? k ?e ψ H (1?αs) ? parenleftBig v ? (0, ˉ θ L )+ k ?e ψ L parenrightBigparenrightBig = nηN 2 (n +1)?v parenleftBig ˉ θ H (1?s) 1?αs ? ˉ θ L ? (n +1)(k?e) ηnN parenleftBig ψ L + ψ H (1?αs) ψ H ψ L (1?αs) parenrightBigparenrightBig . (B4) As expected, the number of subscribers is increasing in e. The profit pi M of the news media is pi M = kS(k?e), 38 where the production cost is assumed to be zero to simplify the model. The profit depends on the public’s beliefs about θ and not on the media’s beliefs. The profit-maximizing price k ? is 28 k ? = ηnNψ H ψ L ( ˉ θ H (1?s)? ˉ θ L (1?αs)) 2(n +1)(ψ L + ψ H (1?αs)) + e 2 , (B5) which is strictly increasing in ψ H and ψ L . The profit pi ? M is pi ? M = N ?v parenleftBig ψ H (1?αs)+ψ L ψ L ψ H (1?αs) parenrightBig k ?2 . (B6) The media’s profit is strictly increasing in ψ H and ψ L , so profit is greater the more likely are the extreme reports r = θ L and r = θ H . This results because the set of subscribers is greater the higher is the probability that a consumer will alter her private decision as a result of the news report. The news media’s report is made ex post taking the set of subscribers as fixed, and the more extreme expectations those reports support, the greater are ex ante profits. The media cannot commit ex ante to its reporting strategy, but instead it benefits because citizens expect that their consumption decisions are likely to be changed by the news report. The price and profit of the media are increasing in the stringency s of regulation if α≤ ˉ θ H ˉ θ L . The media’s price and the profit are also increasing in the competitiveness of the industry producing the product, since the lower price in a more competitive industry increases the number of potential subscribers (v ? (s, ˉ θ H )?v ? (0, ˉ θ L )) N ?v who could gain from the news report. The profit in (B6) is also increasing in e, so from a profit perspective the media has an incentive to make its stories entertaining. This both increases the number of subscribers and allows a higher price to be charged. The entertainment dimension of its stories is not incorporated into the model but instead is taken as a fixed characteristic of any news report. 28 This is optimal provided that e < ηnNψ L ψ H ( ˉ θ H (1?s)? ˉ θ L (1?αs)) 2(n +1)(ψ L + ψ H (1?αs)) , which will be assumed to be satisfied. If this condition is not satisfied, it is optimal to price at k = e in which case all citizen consumers subscribe. The profit is then pi e = e N ?v .This profit is assumed to be less than that in (B6). 39 The profit is also increasing in η, so the more citizen consumers take into account the externality associated with their consumption the greater is the media’s profit. This results because the media increases its price as η increases, since an increase in η shifts upward the media’s demand function. 29 The net effect on the number of subscribers cannot be determined analytically, but for high η more citizens subscribe the greater is η. C. Rational Expectations of the Public The public has been assumed to form rational expectations (ψ L ,ψ H ,1?ψ L ?ψ H )about the possible reports r ∈{θ L ,θ H , ˉ θ o }. In equilibrium the report r = ˉ θ o is never observed, so ψ L + ψ H =1. The probability ψ H is given in (29). The probability ψ H is increasing in the probability that the activist conceals unfavorable information and decreasing in the probability that the media investigates. 29 The demand function shifts upwards because although v ? (s, ˉ θ H )andv ? (0, ˉ θ L )both are increasing in η, the increase is greater the greater is the posterior mean. 40 References American Association of Newspaper Editors. 1999. Examining our Credibility: Examining Credibility, Explaining Ourselves. Reston, Va. Anderson, Simon P. and Stephen Coate. 2001. “Market Provision of Public Goods: The Case of Broadcasting.” Working paper, University of Virginia. Baron, David P. 2002. “Private Politics and Private Policies: A Theory of Boycotts.” Working paper, Stanford University. Baron, David P. 2003a. Business and Its Environment, 4th ed. Upper Saddle River, NJ: Prentice Hall. Baron, DavidP. 2003b. “PrivatePolitics.” Journal of Economics and Management Strategy. 12 (Spring): 31-66. Besley, Timothy and Robin Burgess. 2002. “The Political Economy of Government Re- sponsiveness: Theory and Evidence from India.” Quarterly Journal of Economics. 117: 1415-1451. Besley, Timothy, Robin Burgess, and Andrea Prat. 2002. “Mass Media and Political Accountability.” in The Right to Know: Institutions and the Media, Roumeen Islam, ed. New York: World Bank, forthcoming. Besley, Timothy and Andrea Pratt. 2001. “Handcuffs for the Grabbing Hand? Media Cap- ture and Government Accountability.” Working paper, London School of Economics. Bovitz, Gregory L., Druckman, James N., and Arthur Lupia. 2002. “When Can a News Organization Lead Public Opinion? Ideology versus Market Forces in Decisions to Make News.” Public Choice,forthcoming. Dewatripoint, Mathias and Jean Tirole. 1999. “Advocates.” Journal of Political Economy. 107: 1-39. Dukes, Anthony and Esther Gal-Or. 2001. “Minimum Differentiation in Commercial Media Markets.” Working paper, Carnegie-Mellon University. Dyck, Alexanderand Luigi Zingales. 2002. “The Corporate Governance Role of the Media.” Working paper, Harvard Business School. Erfle, Stephen and Henry McMillan. 1990. “Media, Political Pressure, and the Firm: The Case of Petroleum Pricing in the Late 1970s.” Quarterly Journal of Economics. 41 115-134. Friedman, Monroe. 1999. Consumer Boycotts.NewYork:Routledge. Goldberg, Bernard. 2002. Bias: A CBS Insider Exposes How the News Media Distort the News. Washington, DC; Regnery Publishing. Graber, Doris A., ed. 2000. Media Power in Politics. Washington, DC; CQ Press. Groseclose, Tim and James M. Snyder. 1996. “Buying Supermajorities,” American Politi- cal Science Review, 90 (June): 303-315. Hamilton, James T. 1996. “Private Interests in “Public Interest” Programming: An Eco- nomic Assessment of Broadcaster Incentives.” Duke Law Journal. 45:1177-1192. Hansen, Claus and Soren Kyhl. 2001. “Pay-per-view Broadcasting of Outstanding Events: Consequences of a Ban.” International Journal of Industrial Organization. 19: 589- 609. Krehbiel, Keith. 1998. Pivotal Politics: A Theory of Pivotal Politics.Chicago:University of Chicago Press. Krehbiel, Keith. 1999. “Pivotal Politics: A Refinement of Nonmarket Analysis for Voting Institutions.” Business and Politics,1:63-81. Mullainathan, Sendhil and Andrei Shleifer. 2002. “Media Bias.” Working paper, NBER. Noam, Eli M., ed. 1985. Video Media Competition: Regulation, Economics, and Technol- ogy. New York; Columbia University Press. Noam, Eli M. 1987. “A public and private-choice model of broadcasting.” Public Choice. 55: 163-187. Owen, Bruce and Steven Wildman. 1992. Video Economics. Cambridge, MA; Harvard University Press. Patterson, Thomas E. and Wolfgang Donsbach. 1996. “News Decisions: Journalists as Partisan Actors.” Political Communication. 13:453-68. Spence, Michael and Bruce Owen. 1977. “Television Programming, Monopolistic Compe- tition and Welfare.” Quarterly Journal of Economics. 91: 103-126. Stromberg, David. 2001. “Radio’s Impact on Public Spending.” Working paper, Institute for International Economic Studies, Stockholm University. 42 Stromberg, David. 2002. “Mass Media Competition, Political Competition and Public Policy.” Working paper, Institute for International Economic Studies, Stockholm Uni- versity. 43 Figure 1 The Players and the Game consumption regulation activistindustry The Public (citizen consumers) news media r news report m A government market m M search search private politics public politics price m F investigative journalism Figure 2 Time Line Citizen con- sumers decide on regulation Firms price and consumers purchase Media prices and citizens subscribe Activist and the industry search Activist and in- dustry communi- cate to media Media investigates and reports Nature selects state Figure 3 Media Influence γ v ^ vv*(0,θ L ) v*(0,θ o ) v*(0,θ) v*(s,θ H ) Figure 4 Subscribers and Media Pricing and Profits vv*(0,θ L ) v*(0,θ o ) v*(0,θ) v*(s,θ H ) subscribers pay k* π M = k*A(k*-e) media profit