Use this URL to cite or link to this record in EThOS:
Title: Testing a new approach to construct international financial market indices : an application to Asian-Pacific economies
Author: Wang, Qing-Chao
ISNI:       0000 0004 6422 1793
Awarding Body: SOAS University of London
Current Institution: SOAS, University of London
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
The import price is conventionally predicted mainly by world trade price and other trade-related variables in macro-econometric models. This thesis seeks to enhance the predictive power of those models through explicit inclusion of the financial market information. The task entails construction of external Financial Condition Indices (FCIs). Import price indices of six Asian-Pacific economies - Singapore (SG), Korea (KOR), Taiwan (TW) , Thailand (TH), Indonesia (ID), and Malaysia (MA) are modelled in this research experiment. The external FCIs are aggregated from high-dimensional financial indicators, which are selected to represent markets of developed economies - US, Europe, UK, and Japan. Monthly data for the period of 1991M1-2013M9 are used. In addition to predictive power, two other goals are also targeted during the experiment: stability of indicator weights during regular data updates, and economic interpretability of FCIs. The thesis starts construction of FCIs by following the widely adopted Principal Component Analysis (PCA), based on the Dynamic Factor Model approach (PCA-DFM). Two weaknesses of PCA-DFM are noticed: (1) indicators weights are unstable and given a single set of external financial indicators, and (2) the resulting FCIs are identical to all six economies. Some improvement in the predictive power is observable, but marginal and inconclusive with respect to all six economies. Subsequently, the thesis explores the Partial Least Squares (PLS) approach in three parts: (1) PLS regression method to construct economic-specific FCIs; (b) Simple Dynamic Sparse PLS method (SDS-PLS) to relax the dynamic, synchronised restriction among all indicators in both PCA-DFM and PLS regression; and (3) Revised Dynamic Sparse PLS method (RDS-PLS), which replaces the reflective mode in SDS-PLS by the formative mode. In order to examine weight stability of indicators, FCIs are also concatenated. Once the predictive power of concatenated FCIs is established, economic interpretability of individual indicators, especially those with stable weights, are investigated.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral