6
Distributing News and Political Influence
David Strömberg
The mass media affect both who receives political information and what informa-tion they receive. This influences public policy, because politicians tend to favor informed voters and well-covered issues. Politicians do so for good reason: not only are well-informed voters more likely to vote than uninformed voters, but they are also more likely to vote for those candidates who further their interests. For example, Delli Carpini and Keeter (1996) found that in the 1988 U.S. presidential election nearly 9 out of 10 of the most knowledgeable 10 percent of survey respon-dents voted, while only 2 out of 10 of those in the least informed decile did so (see also Larcinese 2001). In addition, Stein and Bickers (1994) found that people who are well informed in general are also more likely to be aware of new projects in their districts. Furthermore, voters who are aware of new projects are more likely to vote for their incumbent representative, controlling for the actual increase in awards to the district. This chapter discusses my recent work on how the mass media may affect policy formation.
profits and on how their news coverage depends on politicians’ actions. To analyze media influence I (Strömberg 1999, 2001a, forthcoming) combine a model of mass media competition with a model of political competition (for other media models see Anderson and Coate 2000; Masson, Mudambi, and Reynolds 1990; Spence and Owen 1977; Steiner 1952). In this hybrid model the distribution of informed and uninformed voters arises endogenously through the deliberate and purposeful actions of the mass media, voters, and politicians. It indicates that some common features of the mass media have important political consequences.
One such feature is that the mass media operate under increasing returns to scale. For example, once a television program has been produced, the extra cost of an additional viewer is small. For a newspaper the cost of gathering, editing, and writing news to produce the first newspaper is high, but once this fixed cost has been borne, the variable cost of selling additional newspapers is just the cost of printing and delivering them (for the cost structure of newspapers see Litman 1988;
Rosse 1970). This cost structure induces media motivated by profits to cover issues that concern large groups while frequently neglecting minority groups and special interests.
The resulting news bias has political consequences. For example, in a world with-out mass media trade policies are likely to ignore dispersed consumer interests and favor special interests with highly concentrated benefits from trade barriers (see Lohmann 1998; Olson 1965). In a country without mass media, for a politician to advocate reducing trade barriers may be difficult. Few consumers have individual incentives to keep themselves informed about the effects of trade barriers and of politicians’ positions on this issue. Special interests will, however, surely keep them-selves informed, thereby inducing pressure on the politicians. The mass media may counter this bias, because they provide politicians with a megaphone that reaches the large, dispersed consumer groups. In a similar vein, without the mass media we might expect policies to ignore dispersed taxpayers’ interests and favor those that receive concentrated benefits from some small government program. The mass me-dia might counter this bias, because their cost structures make covering politicians’
positions on taxes more profitable than covering their positions on small govern-ment programs.
If this was the only aspect of news reporting, newspapers would never report on, say, operas, whose audiences consist of a tiny share of the population, yet clearly they do. One reason for this may be that a main source of revenue for many newspa-pers and television stations is advertising, and for advertisers not only the size, but also the characteristics of the audience are important. In the newspaper industry examples abound of newspapers that have increased their sales only to see profits fall as a consequence of falling advertising revenue. One of the most cited examples involves the English newspaper the Times. Michael Mander, deputy chief executive of the Times in the late 1960s put it this way (Mander 1978, p. 75):
Distributing News and Political Influence 97
From 1967 to 1969 the Times . . . sales shot up from 270,000 to 450,000—a remarkable achievement. But its higher sales made it no more attractive as an advertisement medium . . . adding to the readership just watered down the essential target group and increased the cost of reaching it. A reversal of policy changed the situation with a consequent dramatic improvement of profitabil-ity. The circulation is back down to 300,000.
In a frequently cited case from American television the show “Gunsmoke” was canceled even though it had high ratings. The show’s audience was evidently too old and too rural to be worth much to advertisers (Barnouw 1978, p. 73). Apparently the mass media try to cover issues that groups valuable to advertisers find interesting.
This media bias translates into a political bias favoring these groups.
A third feature arises because surprising events are more newsworthy than ex-pected events. In the model, news about government programs is valuable to read-ers because these programs affect their everyday lives and they need to take actions to adjust to them. For example, news about rural public works may help famine victims find employment. Similarly, early news about changes in agricultural subsi-dies help farmers produce the right crops to realize the full value of these subsisubsi-dies.
In this setting unexpected developments will therefore be valuable and thoroughly covered by the media. By contrast, news about a program that develops as expected will not be very valuable and will be scantily covered by the media. This behavior of the media induces politicians to boost unexpected increases in programs and moder-ate unexpected cutbacks. The reason is that unexpected spending hikes attract the attention of the media, making such hikes more politically profitable. Similarly, un-expected program cutbacks receive extra news coverage, making them more politi-cally costly. These information management concerns induce politicians to launch a few large spending initiatives in the hope of garnering media coverage, balanced by small cuts in a large number of programs in the hope of avoiding media coverage.
Because of this convex response, over time the media will induce politicians to focus more on programs whose high variability in demand attracts frequent media cover-age, such as famines, and less on programs with a constant demand and scarce me-dia coverage, such as endemic hunger. This particular case has been argued by Drèze and Sen (1990), who found that India, which has free media, has avoided famines, but not endemic hunger, more successfully than China, which lacks free media.
A final feature is that the amount of news coverage depends on the cost of deliv-ering news. Although trivial, this has important and testable implications, because the major mass media—radio, newspapers, and television—have very different de-livery costs. As distributing radio waves is less expensive than distributing newspa-pers to remote areas, the model predicts that radio will increase the share of well-informed rural voters, and that this should cause an expansion in programs that benefit these voters. This hypothesis is tested in the following section.