Doctor of Philosophy
Dr. Lars Stentoft
Dr. Timothy Conley
My thesis consists of three chapters describing volatility forecasting during periods of financial booms and busts, the economic and statistical benefits of flexible data generating process of index returns, and multivariate model of exchange rate returns and their options. In the first chapter, I propose a non-linear threshold model for realized volatility of S\&P 500 index, allowing us to obtain a more accurate volatility forecast, especially during periods of financial crisis. The changes in volatility regimes are driven by negative past returns, where the threshold equals approximately $-$1\%. This finding remains robust to different functional forms of volatility and different set of indices from both developing and developed countries. The additional flexibility of the model allows me to produce a more accurate one and multiple-days-ahead forecasts compared to the linear specification and GARCH family models. Finally, I derive an approximated closed form solution for multiple-step-ahead forecast, which is based on the normal-inverse Gaussian conditional distribution of returns. In the second chapter, I develop a novel discrete-time model for the asset return based on the high-frequency data and mixture of normal (MN) distributions of the latent volatility. This model accurately replicates distributions of both returns and realized volatility under the objective measure. To compute option prices, I specify a Radon-Nikodym derivative, which includes both Gaussian and non-normal innovations, correspondingly. Crucially, my approach avoids calibration of all model's parameters. I price European Put options using Monte Carlo simulations and assess pricing performance of MN and nested Gaussian models during turbulent financial markets in 2008-2011 years. MN model does not only substantially reduce option pricing errors compared with Gaussian model, but provides an appealing econometric framework to assess evolution of investors' risk. Next, I show a novel approach for predicting returns distribution by exploiting informational content of option prices and MN model. Finally, I build a simple quantitative strategy, which substantially outperforms returns of S\&P 500 index (76\% compared with 2\%) during turbulent 2008-2011 years, while remaining market-neutral and had the same volatility as a benchmark returns. In the third chapter, which is a joint work with Chang, Feunou and Fontaine, we propose a new multivariate factor model of exchange rate returns and their option-implied variances. This model documents a tight factor structure in the variance of exchange rate returns and then relate it to the economic factors. In particular, we show that the common factors driving variances of exchange rate returns include the variances of global factors and the common factors driving variances of country-specific shocks. We build a tractable multivariate asset pricing model based on these stylized facts for the underlying exchange rate returns. Our multivariate model provides a reasonable fit compared with performance of univariate models estimated for each series separately. Crucially, our model has a number of appealing benefits which are not attainable in the univariate framework. For example, this model can be used to devise a better portfolio construction or hedging strategy for a portfolio containing both currencies and currency options.
Pypko, Sergii, "Volatility Modelling with Applications to Equity and Foreign Exchange Markets" (2016). Electronic Thesis and Dissertation Repository. 4314.