Essays on structural credit risk modelling and financial econometrics

The first essay studies whether credit rating downgrades feed back on the asset value of the downgraded companies and thus cause real losses. To investigate this issue, I construct a structural credit risk model incorporating rating changes and their associated feedback losses. A maximum likelihood estimation method based on time series of equity prices and credit ratings is then developed for the credit rating feedback model. Evidence from a sample of US public firms downgraded from investment grade to junk shows strong support for the existence of feedback losses. The estimated feedback losses are significant for a third of our sample, and the cross-sectional mean of the feedback loss is 7%.; In the second essay, the transformed-data maximum likelihood estimation (MLE) method for structural credit risk models developed by Duan (1994) is extended to account for the fact that observed equity prices are likely contaminated by trading noises. With the presence of trading noises, the likelihood function based on the observed equity prices can only be evaluated via some nonlinear filtering scheme. A localized particle filtering algorithm is devised for the structural credit risk model of Merton (1974) to execute this task. Applying the estimation method to the Dow Jones 30 firms and 100 randomly selected US public firms, the findings suggest that ignoring trading noises can lead to significant over-estimation of the firm's asset volatility.; The third essay empirically studies a jump-diffusion model for stock price movements using high-frequency data. The stock price is assumed to follow a jump-diffusion process which may exhibit time-varying volatilities. An econometric technique is then developed for this model and applied to high-frequency time series of stock prices that are subject to microstructure noises. The estimation method is based on first devising a localized particle filter and then employing fixed-lag smoothing technique in the Monte Carlo EM algorithm to perform the maximum likelihood estimation and inference. Evidence based on the intra-day IBM stock prices in 2004 suggests that high-frequency data is crucial to disentangling frequent small jumps from infrequent large jumps. Furthermore, accounting for microstructure noises becomes important as the sampling frequency increases.