Following a large body of literature on cross-market crisis linkages, this thesis investigates the potential effects of ''flight-to-safety'' phenomena in two strands of financial modelling: volatility forecasting and bond pricing. In particular, we consider the switching from equity to safe haven assets for modelling the dynamics of financial market volatility and US Treasury prices.
Chapter 2 examines the relationship between the occurrence of flight-to-safety (FTS) episodes and subsequent equity return volatility. We assign to each day in the sample the status of flight-to-safety based on either stock-bond or stock-gold return comovements. The link between flight-to-safety and future equity return volatility is assessed by including the lag of the FTS conditioning variable in a model for the daily realised variance of the S&P 500 index returns. The results are consistent with our expectations: there is a positive, statistically significant interaction between FTS and the next-day volatility of the stock market. Out-of-sample, we find that the one-day ahead volatility forecasts based on FTS information outperform those of some of the most common models proposed in the literature.
Chapter 3 extends the research scope by examining how flight-to-safety can improve the modelling of both equity and safe haven return volatility in a multi-period setting. To this end, we propose a new proxy for FTS based on the realised semi-covariance computed from negative intraday equity returns and positive intraday safe haven returns. We use the FTS proxy to model directly the conditional variance of equity returns, but we also allow for an effect on the safe haven by means of a volatility spillover from the stock market. The incremental value of employing flight-to-safety for volatility forecasting is not only statistically but also economically significant. When comparing with a benchmark, we find that the predictions of our approach are more accurate, and their application to asset allocation leads to better portfolio performance for an investor implementing volatility-timing strategies over horizons as long as one month.
Chapter 4 turns the attention to the role of equity jump tail risk in pricing government bonds. We estimate a term structure model for US interest rates where bond prices are determined also by an equity left tail factor that reflects investors' fear of abrupt negative return shocks to the international stock market. The model shows that equity tail risk is priced in the US term structure.
Consistent with the theory of flight-to-safety, we find that the response of Treasury bond yields and future excess returns to a shock to the equity tail factor is negative and opposite to what happens in the stock market. The evidence of flight-to-safety is stronger at the short end of the US yield curve where equity tail risk has significant explanatory power for Treasury risk premia and is responsible for large term premium drops since the recent financial crisis.
Overall, the findings of this thesis contribute to the literature on flight-to-safety by showing that the switching from equity to safe haven assets in times of stress has important implications for volatility forecasting. Specifically, accounting for flight-to-safety allows for improved predictions of return volatility and better performance of a volatility-timing strategy. Furthermore, this thesis provides guidance on how the safety attribute of Treasuries varies with the maturity of the bonds. Lastly, this study sheds light on the existence of a common predictor between equity and bond markets, whose existence can be justified by the safe haven potential of US Treasuries.
|Date of Award||25 Jun 2019|
- The University of Bristol
|Supervisor||Evarist Stoja (Supervisor) & Nick J Taylor (Supervisor)|