A Historical Event Analysis of the Variability in the Empirical Uncovered Interest Parity (Uip) Coefficient

This dissertation, "A Historical Event Analysis of the Variability in the Empirical Uncovered Interest Parity (UIP) Coefficient" by Wai-kee, Yuen, 袁偉基, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Abstract of thesis entitled: A Historical Event Analysis of the Variability in the Empirical Uncovered Interest Parity (UIP) Coefficient Submitted by: YUEN Wai Kee For the degree of: PhD At the University of Hong Kong in (April 2006) This paper documents and investigates the "abrupt and sharp variability of uncovered interest parity (UIP) coefficient estimates," on the basis of "rolling" sub- samples using daily data for Germany and the United States in the recent floating regime. The empirical failure of UIP is well-known. It is also well-known that the departure of the empirical UIP coefficient from the theoretical one varies across sampling periods. Documenting and explaining the variation of the empirical UIP coefficient could help us understand the various reasons for the departure of the empirical UIP coefficient from the theoretical one across sampling periods, and shed light on the reasons for the empirical failure of UIP. We pay particular attention to abrupt and sharp changes in empirical UIP coefficients across rolling sub-sample, because such changes could indicate change in empirical UIP coefficients that may be linked to important historical events, as distinct from small changes that may simply be due to sampling errors. We caution that an observed episode may result from either a deleted observation in the last rolling sub-sample or an added observation in the next rolling sub-sample. To cope with this issue, Wong (2000) suggests a "Spikes causality test" to identify the ivepisodes that are responsible for the abrupt and sharp changes by experimenting with slightly different rolling window size. We find substantial variability in the empirical rolling UIP coefficients. The abrupt and sharp variability of sub-sample UIP coefficient estimates appears to be associated with four identified causes: (1) fluctuations in the business cycle, (2) shifts in monetary policy, (3) shifts in exchange rate policy and (4) expectations of realignments. Since the rolling sampling technique does not suggest a specific form for either the type of shocks or the mechanisms by which these shocks might influence the relationship. So we experiment with three slightly different augmented models by allowing the identified causes to: (1) shift the constant term, (2) shift the slope, and (3) shift both the constant and the slope of the UIP model. We find that the identified causes are statistically significant in the augmented UIP model. Thus the identified causes are partially responsible for the poor performance of the conventional UIP model empirically. Words: 362 By: YUEN Wai Kee _________________ DOI: 10.5353/th_b3642420 Subjects: Interest rates - Econometric models Foreign exchange rates - Econometric models