Fernando Castro de Campos Roriz.
03/09/2014
Orientador: Marcelo Medeiros.
Banca: Carlos Viana de Carvalho. Marcelo Fernandes. Ruy Monteiro Ribeiro. João Felipe Mergulhão. Ricardo Dias de Oliveira Brito .
http://www.dbd.puc-rio.br/pergamum/tesesabertas/1022001_2014_completo.pdf
Nível: Doutorado
This dissertation is composed of three articles in financial economics.The first article tests for the effects of political uncertainty into the stock market. Using an index of political uncertainty constructed from the winning probability of U.S. presidential candidates calculated in the months prior to the 2004 Presidential Elections, we find that an increase in this index tend to depress considerably more stocks with higher exposure to the market factor, even after controlling for partisanship effect and overall market volatility. Thus, it seems to be the case that when investors face an increase in political uncertainty they prefer stocks with smaller exposure to the market risk factor, once these assets increase (or decrease less) their hedging ability in periods of high uncertainty. Given the lack of a state contingent market where individuals can fully diversify the electoral risk, these findings show that consumers could respond to the wealth uncertainty generated by the electoral process holding stocks less exposed to the market factor. In the second article, I develop a model in which elections and the electoral process move asset prices. The model allows the party in power to have an impact over the profitability of firms, interpreted as the partisanship effect. This feature makes stock prices respond to the uncertainty caused by the electoral race and the upcoming election. Under this source of uncertainty, investors start requiring a higher risk premium to hold assets subject to the election result. Also, it is shown that in general, regardless of which party is best for firms protability, investors demand a risk premium to hold these assets during the election. Finally, we discuss some results of the literature using the predictions of the model. The third article explores another strand of the financial economics literature and compares the performance of standard, naive and skeptical financial models when plugged into a Markowitz optimization framework. Using an economic value approach as in Fleming et al (2001,0), we find that naive portfolios perform better than traditional models like the CAPM or the Fama-French 3 factor model. Besides, we consider another approach to select a single model containing several factors and estimated using the LASSO technique (Tibshirani (1996)). We show that depending on the penalty parameter used to select the relevant factors the LASSO technique performs considerable better than the usual factor models used in the financial literature.
