Media activity and public spending

Christian Bruns Oliver Himmler JEL Classification Politicians seeking reelection need voters to know what they have done for them. Thus, incentives may arise to spend more money where media coverage is higher. We present a simple model to explain the allocation of public spending across jurisdictions contingent on media activity. A politician seeking to maximize the probability of reelection will shift more money to jurisdictions where an extra dollar raises more votes because a larger share of the electorate is informed about his policy. The main prediction of the model is that media activity is higher in the core areas of media markets. This implies higher spending levels there and lower spending levels in remote jurisdictions. Empirical support for this prediction is found using United States data on county-level federal grant allocation, Designated Market Areas and the location of licensed television stations. - Media activity and public spending 1 Introduction Do politicians channel more funds to places where media activity is higher? Political actors need voters to know what they have done for them. Since voters nowadays heavily rely on commercial mass media such as television, newspapers or radio to obtain political information (Horrigan et al. 2004), politicians strive for media attention. However, the mass media typically bias coverage towards certain groups (Hamilton 2004; Strmberg 2004a). Thus, political benefits to those groups are more likely to attain media attention and make the affected voters consider the benefits when casting the ballot. Consequently, the payoff from funds in terms of votes is higher in these informed groups. This is why politicians face incentives to provide favorable policies to voters targeted by mass media. In this paper, we examine whether local television news, which are the most popular source of political information in the United States (Napoli and Yan 2007), affect the allocation of federal grants. More precisely, we are interested in whether a geographic pattern of grant allocation exists which reflects a well-documented bias in local TV news programming: local news content focuses on the core areas of TV markets where the majority of TV stations is located (Adams 1980; Alesina and Trautman 2008; Heider 2000; Kaniss 1997). Thus, political actions in the core area are more likely to receive coverage which suggests that voters in that area are better informed about local politics than voters in remote regions. Due to this publicity effect, the amount of funds a location is awarded should systematically decline in distance to the core area of a TV market. We analyze the relation between TV markets and public spending both theoretically and empirically. First, we present a simple model based on Strmberg (2004a,b) which serves to establish the link between local TV news and grant allocation. We show that television stations bias their news programs towards the markets core areas because news production is less expensive there. As a consequence, in these regions there are more voters who are informed about their local politics. Then, the model explains how this information structure in the electorate affects a politicians incentives for strategic grant allocation. The main prediction of the model is that politicians direct more funds to areas close to media centers. We test this prediction empirically using US county-level data. The dependent variable is the amount of federal grants per capita allocated to a county. Using a countys distance to the nearest media city in the market as our main explanatory variable, we find a substantial effect of distance on grant spending. A county which is located twice as far away as the mean distance receives 3.4% less in grants per capita which makes for an average total loss of about $2.8 million per county per annum. Our analysis builds on the pioneering work by Strmberg (2004a,b) on the influence of mass media on fiscal policy. In a formal model, Strmberg (2004a) shows that certain groups of voters receive favorable policies when the mass media report more heavily on campaign promises concerning these groups. Analyzing data from the United States, Strmberg (2004b) tests empirically whether better informed groups in the electorate receive favorable policies. He finds that counties where many households owned a radio received more funds from a major New Deal program in the 1930s. In contrast to Strmbergs empirical paper, our paper examines how policy is affected by the geographic distribution of information which results from biased news programming in a market where everyone has access to the medium. Other research that deals with the medias role for accountability includes Snyder and Strmberg (2010), who find empirical evidence that newspaper coverage affects voters information about their US House representatives. A better fit of newspaper markets with districts is found to increase information levels, resulting in more accountable Congressmen. Besley and Burgess (2002) also deal with the medias role for political accountability. They find that state governments in India provide more public food and calamity relief in hard times when newspaper circulation in the respective state is higher. We proceed as follows. While the next section presents the theoretical model, Sect. 3 gives a description of the data and estimation approach. Empirical results are presented in Sect. 4. Section 5 concludes. 2 Model The model builds on the framework by Strmberg (2004a,b). A politician distributes a given budget within his constituency. From now on, we refer to the politician as a US governor who receives grants from the federal government and allocates the money across counties in his state. The governor distributes grants strategically in order to win the next election. He knows that funds generate more votes in counties where many voters attribute the benefits derived from grant spending to him. He further knows that the share of these informed voters is larger in counties which are extensively covered by local television news. The model shows that economic forces make local television stations broadcast more news from counties located in the core area of a TV market. This geographical bias in local news programming yields lower shares of informed voters in remote counties and larger shares of informed voters in core areas. Thus, incentives arise for the governor to spend more in these core areas.1 The timing of the game which is described in the next section goes as follows: the governor distributes grants across counties. Then, the TV stations decide on their reporting strategy and voters choose a station to watch. Finally, an election takes place. The governors allocation problem. The governor aims at winning votes by allocating total grants G across the counties c = 1, 2, . . . , C in his state such that pc gc = G, where gc denotes grant spending per capita and pc the population in (...truncated)