Essays on information and prediction

The dissertation is entitled: "Essays on information and prediction" and consists of a total of four chapters. The central issue is the study of the role that certain mechanisms of aggregation of information, such as predictive markets or elections, can play in improving the accuracy of predictive algorithms. The first chapter of the thesis is entitled: "Economic Determinants of Political Alternation: A Panel Data Analysis of OECD Countries" and studies the relationship between economic performance and political alternation. The initial hypothesis is that the stochastic process that determines the political alternation is not independent of the economy, so that the recent evolution of the macroeconomic variables would be very relevant when explaining the electoral results. We are implicitly considering the "responsibility hypothesis", by which voters are aware of the economic information since they consider that the government is responsible, through its actions, for the good or bad economic situation of the country. In this way, economic variables can predict the probability of political alternation. Throughout the chapter, a critical review of the literature is presented, with special attention to the article Brender & Drazen (2008) and later the estimation of a re-election probability model is proposed, using macroeconomic indicators. The contributions of this chapter to literature are diverse. On the one hand, the results obtained contradict those found in the "economic vote" literature. On the other hand, there are also methodological contributions: the use of a discount rate to allow voters to give more weight to the most recent data, propose an alternative way of measuring political alternation and the use of structural deficit data to solve the problem of multicollinearity between the independent variables..