Empirical essays on inferring information from options and other financial derivatives

Abstract

This thesis consists of three essays on inferring information from option contracts and other financial derivatives in the U.S. market as well as in the international markets. The first essay examines corporate bankruptcy probabilities inferred from option prices and credit default swaps (CDS) spreads around the 2008 financial crisis in the U.S. market. Option pricing framework is used where the risk-neutral density of the underlying asset is assumed to be a mixture of two lognormals augmented with a probability of default, to calibrate to the market option prices. The CDS model assumes a constant default probability which is solved from the non-linear equation that equates the present value of expected premium payments with the present value of expected payoffs. The essay documents that both sources provide ex-ante bankruptcy probabilities, but there is no significant evidence suggesting one predicts the other. The second essay constructs volatility indices for 15 markets around the world and examines implied volatility spillover between these markets. Volatility indices are constructed using option prices based on the new VIX methodology with modification to address its limitations. Spillover effects are then examined using vector autoregressive analysis, impulse response functions and forecast error variance decomposition. Empirical results show that the U.S. is unambiguously the dominant source of uncertainty in the world. Correlation between markets largely depends on geographical proximity. The findings support the notion of informationally efficient international stock markets, in that information transmitted from one market to another is processed within one or two days. The third essay further investigates spillover effects in variance risk premiums, which has been interpreted as the difference between the realised variance under the physical measure and the risk-neutral measure. Realized variance under the physical measure is constructed for each market using the HAR-RV model, which is able to capture long-memory characteristic of volatility. Risk-neutral expectation of future variance is approximated by a portfolio of option contracts, as calculated in the second essay. Steps are taken to address serial correlation and dependence, and variance risk premium spillovers are examined using vector autoregressive analysis, impulse response functions, and Granger Causality tests. The findings are consistent with those found in implied volatility spillovers. The U.S. market is the distributor of uncertainty in the global market. Information transmitted from one market to another is quickly digested, but it may take longer in crisis period due to greater uncertainty.