Stock market volatility and monetary policy.

摘要

This thesis comprises three essays. The first essay examines the effect of federal funds rate surprises on implied stock market volatility using U.S. data. While volatility is measured using two popular implied volatility indices (VIX and VXO indexes), different techniques are employed to measure federal funds rate surprises from federal funds futures data at the daily and monthly frequencies. We find that the surprises significantly increase volatility, even when timing uncertainty is accounted for. Consistent with the efficient markets hypothesis, we find that the expected component of a target rate change; as well as the target rate change itself, do not significantly affect volatility. Nonlinearities and asymmetries are explored in the response of volatility to the direction of the rate change and the sign of the surprise. The evidence of asymmetries and nonlinearities is found to be weak.;The third essay uses Canadian data to examine risk premiums and predictability in futures contracts (BAX futures) on short-term Canadian interest rates (Bankers' Acceptances). While evidence for a constant risk premium is found, the predictive regressions employed only uncover weak signs of predictability (and time-varying risk premiums) in returns on BAX futures. This result is confirmed by forecast efficiency regressions. Lastly, out-of-sample forecasting of Bankers' Acceptances returns is undertaken. Forecasting results reveal the superior predictive ability of the model exploiting the restrictions of economic theory in comparison to random walk, autoregressive and error correction models.;The second essay investigates the dynamic response of U.S. stock market variables to monetary policy shocks and the transmission of monetary policy shocks to the stock market using vector autoregressive models. We find that volatility is increased and excess returns are decreased contemporaneously due to a monetary policy shock but that the persistence of the effect depends on the model used. A daily analysis using conditional heteroskedasticity models confirms the results found with vector autoregressive models.