Resumen de Essays on the economics of the media: media bias, competition and tax anticipations
This thesis is a collection of empirical studies on the Economics of the Media. Chapter 1 and Chapter 2 are studies on the determinants of media bias. Chapter 1, entitled "Multidimensional Media Slant: Complementarities in News Reporting by US Newspapers", is a study of the determinants of the daily choice of front page news made by editors of major US printed newspapers. Chapter 2, entitled "Media Content Response to Entry: Evidence from Spanish Newspapers" and joint work with Diego García, is an empirical study of the effects of entry of a new competitor in the content decisions of printed newspapers in Spain. The second part of this thesis is a study on the application of media data to the empirical study of fiscal multipliers. Chapter 3, entitled "The Effects of Taxes on Economic Activity: A Narrative Approach to Frequent Anticipations", collects the work of this second line of research.
Front page news not only have a special role in determining the public awareness of events, but they are also a clear-cut observable outcome on a daily basis, and therefore amenable to systematic scrutiny. Little is known about the determinants of the specific choice of lead news, while they receive major public attention every day. Chapter 1 is devoted to the study of front page choices made by editors of major newspapers in the US. I document that newspapers are biased to certain combinations of news on top of biased to certain news. To identify my measures of bias, I exploit the variation in the media relevance of news across different topics and days. To measure the news relevance I use lead news choices of other US mass media. As a consequence, my measures of bias are relative to the overall media bias. In the current implementation I use a multinomial logit model. I also provide a reader-maximization model for front page decisions that I use to interpret the empirical biases as preferences of the population of target readers of each newspaper. From my estimation, I recover maps of complementarities among pairs of topics for each of the major US newspapers. I find that complementarities between news contribute largely to the probability that news on a topic appears in the front page.
Regulators are concerned with the lack of competition hindering the actual informational role the media is meant to play. Economic theory does not offer a unique answer to the expected product changes after the entry of a new competitor in the industry, in contrast, answers depend on the relative importance of assumptions about the market.
In Chapter 2, joint work with Diego García, we study the effects of entry of a new competitor in the content decisions of a established printed newspapers duopoly in Spain in the 1980s. We exploit a unique context which offers a quasi-natural experiment: El Mundo became a major player in the daily newspaper market very quickly. To analize the newspapers’ reaction in Chapter 2 we collect newspapers original text articles and pages from the newspapers’ libraries. Rather than using standard metrics of bias, we use automatic text analysis on the opinion sections and estimate a topic model that we borrowed from the machine learning literature (Blei, Ng and Jordan, 2003). This empirical strategy quantifies the relative frequency of topics for each leading newspaper and day for the period of interest. Using a difference-in-difference (DD) strategy on the editorial content measures of particular topics, we exploit the before-and-after ElMundo and between newspapers variation in decisions. We find that the two incumbants, ABC and El País, move farther apart after the entry of El Mundo; ABC intensified its coverage of entertainment and domestic content while El País that of economy. We also document that former incumbants both start to cover more international and less political issues in the editorial pages. Our results are consistent with a superiority of demand over supply forces in the market of major printed newspapers in Spain in the 1980s. This is a new piece of evidence that is in line with the results of Mullainathan and Shleifer (2008) model of the market for news.
The media plays a relevant role in the transmission of public information to society.
As reporters and analysts of fiscal bills during their process of elaboration and until approval, they provide information that may be worth to the public’s forecast of the success of potential new fiscal policy. This is an anticipation channel that has not been considered before in the empirical literature of fiscal multipliers. The second part of this thesis, materialized Chapter 3, is a study of the macroeconomic effects that mass media news about future fiscal tax bill approvals have on US economic activity during the period 1968-2007. In this paper I combine text data and the space devoted to fiscal policy news to construct a new measure of beliefs on tax approvals. I exploit the textual information to predict tax approvals in the US congress at a particular month and do so by estimating a classification algorithm for high-dimensional data that I borrow from the machine learning literature. Since this information typically flows faster than standard measures of GDP, I propose a mixed frequency dynamic factor model to estimate both the economic activity latent factor and the effects of anticipated tax shocks on it. In my empirical analysis, I have control of all potential effects of tax changes from the moment they are born as tax bills until they are effectively implemented. I find that one-month-ahead media anticipations of tax rise approvals significantly stimulate current economic activity in the short-run, while longer-horizon forecasts have opposite effects on economic activity. The effects are the opposite for tax cuts.
