Media bias and electoral winners: evidence from Italy

In light of today's Italian parliamentary elections, here is a good recent paper by three Italian economists on the effects of media bias on electoral outcomes, concerning in particular the electoral results of former PM Berlusconi.

Here is from their VoxEU column:

"...all Italian voters know that Berlusconi owns the major commercial TV network since the early eighties. We exploit exogenous variation in viewers' exposure to Berlusconi bias using idiosyncratic deadlines to switch to digital TV from 2008 to 2012. At the deadlines, old analogue signals were switched off and only digital signals kept on airing. ... Most digital channels are aired by new media companies, which have no ties to Berlusconi. After switching to digital TV, many Italian households changed their viewing habits. From June 2008 to June 2011, the share of viewers of Berlusconi-controlled channels dropped from 84% to 71%. ... We explore the causal effect of the shock to bias exposure on voting behaviour at regional elections in March 2010, the first elections held during the switch-off process. ... This setup allows us to compare the electoral outcome of Berlusconi’s party between municipalities that were exposed to the new digital channels and those where the digital switch-off had not taken place yet. In order to make the comparison more credible, we compare municipalities close to the switch-off geographical boundary. These municipalities are only a few kilometres apart, and they differ only in terms of access to the new television technology.

Although the control of most pre-digital outlets by Berlusconi was widely known, the switch caused a drop in his coalition vote share by 5.5 to 7.5 percentage points. This effect is economically and statistically significant. Scaling the effect by 2005 Berlusconi supporters and non-voters who watched new channels in 2010, we estimate that at least 30% of them changed their voting behaviour after the switch-off of analogue TV. The effect was stronger in towns with older and less educated voters. At least 30% of digital users had not filtered out the bias from 1994 to 2010."

They apply a robust strategy to compare the municipalities with and without exposure to digital channels, allowing them to control for a range of possible unobservable characteristics, and tend to refer to it as a quasi randomized experiment (*in econometrics,  the closer you can get to observing a randomized experiment, the better). Even though their results are very strong and suggest that the switch of the media bias did have a significant effect on Berlusconi votes, there still could be a possibility of an omitted variable bias that could have led to this change in voting patterns. 

However, one needs to keep in mind that Italy is a specific example in which media bias was playing a particularly important role, simply due to the fact that during Berlusconi's primeministership he managed to control most of it. This is why Italy is a fruitful filed for a range of political economy research that looked at how this media monopoly and its necessary bias in favour of Berlusconi affected his party's electoral chances. Here is a working paper from Larcinese (2005) that showed how voters who watched Mediaset (Berlusconi's private media company, largest in Italy) were much more prone to vote for Berlusconi in the 2001 elections (the second time he won office). Even though this could be explained by initial ideological divergence of those who choose to watch it in the first place, the results do imply significant effects of media control on electoral outcomes. This becomes even more important when taking into consideration that after seizing power more strongly (Berlusconi's first primeministership lasted a bit more than 6 months, while the second one lasted for 5 years), he got more control over the public broadcasting company RAI, thereby completely monopolizing the media market. Similar to Larcinese's paper is this one by Durante and Knight (2009) which draws similar implications for ideological exposure of Italian voters to news programs.

Italy is definitely not a stand alone case. A very good paper addressing this media bias effect in the US is by Della Vigna and Kaplan  (2007) "The Fox News Effect: Media Bias and Voting", QJE (the link is for their 2006 working paper), which has found that the introduction of Fox News has gained the Republicans from 0.4 to 0.7 percentage points in the 2000 elections in towns which broadcast Fox News. They have also found that Fox News convinced from 3% to 28% of its viewers to vote Republican. 

Enikolopov, Petrova and Zhuravskaya (2011) find similar implications in Russia, where independent TV decreased the aggregate vote for the government party by 8.9 percentage points, for the 1999 elections. In a slightly different paper observing media capture by politicians (so a case close to Berlusconi), McMillan and Zoido (2004) looked at how Peru's president Fujimori was able to bribe the media, the judges and the legislators and thus keep himself in power for a long time (until this was discovered by one media owner that didn't sell out). In the paper they list the amount of bribes Fujimori was paying to some judges and newspapers. 

There is a whole range of other papers observing this relatively new and under-researched field of political economy - the role of mass media on policy outcomes. For anyone interested more deeply in the subject, I recommend Prat and Stromberg's (2011) overview of the current literature.

Going back to our first paper, it will be interesting to write a follow-up taking into consideration the results of today's (and tomorrow's) elections in Italy. Simply to see whether the effect of the switch to digital TV has made a lasting influence. However, this will be hard to do since in the meantime Berlusconi had lost a lot of support due to other things (like almost bankrupting the country in November 2011, for example). But without doubt Italy will still provide us a lot of potential research on the media bias story in years to come.