| In my dissertation, I contribute new models and methodologies for the estimation and forecast of market risk, the factors that drive risk; and the real-sector investors' response to risk generated from commodity price fluctuations. My objective is to bring together themes and measurement issues for risk modeling in both the financial and the real sector. In a new model that combines return predictability with stochastic volatility, I estimate and forecast risk with a new Monte Carlo Markov chain algorithm. The model achieves fast convergence of the parameters and the volatility to the true values by relying on the multi-move sampling of the volatility. This model allows an investor to form optimal portfolios and generate out-of-sample portfolio returns. When applied to a two-asset portfolio with S&P500 as the risky asset, it improves returns by 42%. To identify sources of risk and return essential to portfolio balancing, I develop a factor model that estimates a forward-looking measure of portfolio risk. The factors comprise market, investment style and industry classification. I employ a Newey-West estimator to obtain the factors correlation matrix. Among the factors that aggregate Japanese equity portfolios risk and returns, short-term debt sensitivity and being a multinational or a firm with micro-capitalization are significant. Based on the predicted factor returns, investment in a portfolio of undervalued securities would have provided an average 80% cumulative return. In the last part of my dissertation, I estimate real-sector investors' response to commodity price fluctuations. I develop a model for the commodities demand that allows for correlation of the demand shocks and time-variability of the demand parameters. The complexities of the estimation of a large number of parameters are overcome with the Markov chain Monte Carlo methodology. The partial Bayes factor finds the new model as an improvement over the basic model with constant parameters. This new model turns out new information about the energy demand elasticity: During the 2001' heavy energy price fluctuations, the energy demand elasticity increased reaching close to unity, in contrasts with the highly inelastic energy demand established in the last decades. |
