Dissertation

Essays on time series and financial econometrics

Ho, Kin Yip

2008

Overview

This dissertation is on the topic of multivariate conditional heteroskedasticity modeling in time series and financial econometrics. It is of interest to consider multivariate volatility models because it is widely accepted that financial volatilities move together over time across assets and markets. There are several challenges associated with this class of models, such as ensuring positive-definiteness of the conditional variance-covariance matrix in practice, interpreting the model parameters, and incorporating the various features such as long-range persistence and asymmetries in volatility. Given these challenges, it is not surprising that theoretical analysis and empirical applications of multivariate models have lagged behind the univariate version and knowledge of the asymptotic (large-sample) and finite-sample properties of these models remains fairly limited. This dissertation fills the gap in the literature on multivariate conditional heteroskedasticity models by considering their finite-sample properties and empirical applications in different contexts. Chapter 1 analyzes the finite-sample properties of various models that not only incorporate features such as asymmetries, persistence, and time-varying correlations but also ensure the positive-definiteness of the conditional variance-covariance matrix at every point in time. The main conclusion is that their finite-sample properties do not perform too poorly in terms of bias and mean square error. Chapters 2–4 present three different empirical applications of this class of models. In particular, Chapter 2 considers the volatility dynamics of the greater China equity markets and notes that the stylized facts established for financial markets in developed economies are not necessarily applicable to those in developing economies. The next chapter analyzes cross rates that do not involve the US dollar as the numeraire currency. It is concluded that empirical regularities for the bilateral US dollar exchange rates could be different from the volatility and correlation properties of cross currency rates. The final chapter examines macroeconomic fluctuations of the Australian economy using a group of multivariate volatility models. The correlations among the fluctuations of the national income components are observed to be time-invariant based on the Lagrange Multiplier and Information Matrix tests. These components also do not display significant asymmetries in their response to macroeconomic shocks.

• Publisher: ProQuest Dissertations & Theses
• Subject: Economics / Finance
• ISBN: 0549970495, 9780549970491