| dc.description.abstract | My thesis is divided in three chapters. In the first I extend the application of Bandi and Tamoni (2014)‘s time series decomposition to other asset classes, such as fixed income, credit and credit derivatives, and other models, such as the Fama and French three factor model. I document a significant increase in R squared from using the decomposition across all the asset classes and models. I also exploit the time-domain properties of the decomposition to compute time-varying betas and analyse the determinants of risk across time. In the second chapter I present some stylized facts about political risk, as proxied by Baker, Bloom and Davis’ (2016) Economic Policy Uncertainty index. The log differential is stationary and exhibits a leptokurtic distribution, with heavy tails on both sides. The index also exhibits a time varying conditional variance, as highlighted by an ARMA(1,1)-GARCH(1,1) model. I also investigate the persistent nature of the index with various Heterogeneous Autoregression (HAR) specifications: the basic version, augmented with factors, and taking into consideration asymmetric effects. I finally investigate the out of sample conditional predictive ability of these models. In the third chapter, I introduce a model that allows separating political risk from other risks, by extracting two uncorrelated factors in a two factor Heston setting. I employ the Economic Policy Uncertainty index by Baker, Bloom and Davis (2016) as a noisy proxy of political risk. The model allows to price options and recover the impact of political risk on the whole P-distribution of asset returns. I find that political risk has a sizeable effect on the distribution of the S&P 500 index returns and that it impacts all moments. I also find that political risk has a positive and statistically significant impact on the S&P 500 index Equity Risk Premium. | en_US |
