Who trusts Berlusconi? An econometric analysis of the role of television in the political arena

The objective of this paper is to improve our understanding on how the former Italian prime minister formed and kept his exceptional political consensus despite the never-ending series of scandals in which he has been involved. Specifically, we carry out an empirical assessment of the relationship between trust in television and trust towards the prime minister. Many Italian commentators argue that television does not exert any particular bias on public opinion. If the assumption of television's irrelevance were true, one could reasonably expect to find no significant relationship between trust in TV and trust in Mr Berlusconi as prime minister. Contrary to this hypothesis, our empirical analysis shows that trust in television was the strongest predictor of trust in the Italian prime minister in Spring 2011, thus suggesting that Mr Berlusconi's media empire played a key role in the building of his political consensus. This result is based on probit and ordered logit estimates and is robust to different specifications. However, there are three reasons for which this finding could be interpreted as the fruit of a spurious correlation. First, it is difficult to distinguish the effect of trust in television from that of other phenomena that potentially influence trust in the prime minister. To deal with this problem, we include in the main equation a wide set of individual and household control variables. Second, individual effects, such as individuals' exogenous shocks, may be correlated with trust both towards television and the prime minister, thus creating a common bias. Third, as we explain in the paper, it is reasonable to assume the existence of reverse causality. To deal with the last two problems, we turn to instrumental variables (IV) estimates. When we address endogeneity in IV estimates, trust in television remains the best (most significant and strongest) predictor of trust in the prime minister. By contrast, we find that the dependent variable is significantly and negatively correlated with educational qualification, trust in the judicial system and with a measure of tolerance towards immigrants. Even if these results pass robustness checks and hold in IV estimates, it must be remarked that the cross-sectional design of the research dictates extreme caution in the interpretation of correlations as causal relationships. Nonetheless, the paper contributes to the literature by carrying out the first econometric investigation into the role of television in individuals' political opinions, with a focus on trust towards the Italian prime minister. The relationship between television and political attitudes and beliefs is an important topic for economics, in view of the unquestionable role that political institutions, particularly the government and the prime minister, play in a country's economic performance. © 2012 Blackwell Publishing Ltd