This dissertation consists of two essays. The first essay tests the mixed jump diffusion, a mixture of normals, and an ARCH-type process for characterizing exchange rate changes. The parameters of these three models are estimated using maximum likelihood techniques and specification tests are applied to evaluate these models. The processes are evaluated over varying differencing intervals to determine the influence of time frames on the distributional form of exchange rate changes. Results indicate instability for fitted models of spot exchange rate price changes. The best fitting stochastic process changes as the sampled time period changes and as the length of the differencing interval is changed. In contrast to previous research, the mixed jump diffusion model is not supported by these results. Examination of the mixed jump diffusion probability density function reveals that the poor fit of this model is an artifact of the maximum likelihood process. / The second essay examines the relationships between distributions of exchange rates and fundamental variables. This essay develops two models that relates fundamental economic variables to explain either the ARCH effects, the jumps, or the mixture of normals. Data sets are developed that isolate the effect that the candidate process explains. Regression models are used to try to explain the jumps and the ARCH effects. Multiple discriminant analysis is used to examine the shift between normals. The regression results indicate that explanatory power increases by isolating the jumps. No improvement was found in the ARCH regressions. The application of discriminant analysis to explain the shift between normals is not very successful. Prior probabilities play an influential role in the discriminant analysis results. In most cases there are large differences in the prior probabilities. Given such predominance the error rate is minimized by assigning all the observations to the normal with the largest prior probability. Implications of results and improvements in methodology are discussed. / Source: Dissertation Abstracts International, Volume: 55-11, Section: A, page: 3593. / Major Professor: Elton Scott. / Thesis (Ph.D.)--The Florida State University, 1994.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_77287 |
| Contributors | Johnston, Ken Peter., Florida State University |
| Source Sets | Florida State University |
| Language | English |
| Detected Language | English |
| Type | Text |
| Format | 114 p. |
| Rights | On campus use only. |
| Relation | Dissertation Abstracts International |
Page generated in 0.0111 seconds