Thesis (M.B.A.)--University of Hong Kong, 1998. / Includes bibliographical references (leaf 64-66).

Thesis (M.B.A.)--University of Hong Kong, 1995. / Includes bibliographical references (leaves 124-125).

Thesis (Ph. D.)--University of California, San Diego, 2000. / Vita. Includes bibliographical references.

Thesis (Ph. D.)--University of Chicago, 1987. / Includes bibliographical references (leaves 109-112).

Thesis (Ph. D.)--Stanford University, 1991. / Includes bibliographical references (leaves 174-185).

Thesis (M.Soc.Sc.)--University of Hong Kong, 1986. / Also available in print.

No description available.

M.Com. (Financial Economics) / Recently, there has been a growth in the bond market. This growth has brought with it an ever-increasing volume and range of interest rate depended derivative products known as interest rate derivatives. Amongst the variables used in pricing these derivative products is the short-term interest rate. A numbers of short-term interest rate models that are used to fit the short-term interest rate exist. Therefore, understanding the features characterised by various short-term interest rate models, and determining the best fitting models is crucial as this variable is fundamental in pricing interest rate derivatives, which further determine the decision making of economic agents. This dissertation examines various short-term interest rate models in continuous time in order to determine which model best fits the South African short-term interest rates. Both the linear and nonlinear short-term interest rate models were estimated. The methodology adopted in estimating the models was parametric approach using Quasi Maximum Likelihood Estimation (QMLE). The findings indicate that nonlinear models seem to fit the South African short-term interest rate data better than the linear models

Much depends on the future course of interest rates. The decisions of families to make major purchases, the willingness of businesses to expand and invest, the rise and fall of the economy and stock market, the ability of lesser developed countries to repay their debts, the tenure of presidents and prime ministers--all of these may turn on whether interest rates increase or decrease in the months ahead. Several decision functions developed in the dissertation permit the direction of change of interest rates on long-term U. S. government bonds to be forecast correctly about 60% of the time. When the different models are combined, effectiveness is increased, and when the forecasts are dollar-weighted, performance in excess of 70% is possible. The results are evaluated in comparison with a Bayesian forecasting model and a 10,000-event Monte Carlo simulation of a random decision rule. The forecasting ability of the models is statistically significant at the 99% level of confidence. The dissertation reports on one of the first application of powerful techniques recently developed in cybernetics and engineering to forecasting the direction of change in interest rates. Two forecasting algorithms, called linear decision functions or linear classifiers, are derived using the principles of pattern recognition. Because they are recursively updated, both algorithms operate dynamically and adapt their performance to changes in the economic environment. One classifier, a modification of the widely used least-mean-squared-error algorithm, is adapted to permit monthly revision and to allow larger movements of interest rates to have greater weight in future decisions. The second algorithm permits refinement of the parameter estimates generated by the first. These formal, mathematical constructs are then supplied with financial variables--leading indicators of inflation and investment activity--to permit unconditional, ex-ante forecasts of the direction of change of interest rates on long-term government bonds over a one-month time horizon throughout the period 1969-82. The results should be of interest to investment managers, speculators, corporate treasurers, policymakers, economists and forecasters.

M.Sc. (Mathematical Statistics) / In this dissertation we investigate the South-African interest rate market by analyzing the Johannesburg Interbank Agreed rate (JIBAR) and the South-African three-month treasurybill rate. In particular, we assess the goodness-of-fit of some well known parametric singlefactor short rate models. In all the data sets investigated, we firstly found the interest rate increments to exhibit severe nonnormalities, which is also found to be the case in numerous other empirical studies. Secondly, we reject the model fit for the various parametric short rate models tested. Thirdly, we found strong evidence to support the presence of jumps in all the data sets and that interest rate increments mostly exhibit fat-tailed distributions. Consequently, we tested the ability of diffusion models, driven by Brownian motions, to generate jump induced nonnormalities via a nonparametric test of diffusion model fit. The nonparametric short rate model was rejected, i.e diffusion models are misspecified. Lastly, since diffusion models are misspecified we investigated whether a jump-diffusion model can be fitted to the data with higher accuracy. In conclusion we find that a jump-diffusion model, namely the Vasiˇcek jump-diffusion model, can adequately generate the jump induced nonnormalities present in each of the data sets.