Date of Award

2015

Document Type

Thesis

Publisher

Edith Cowan University

Degree Name

Doctor of Philosophy

School

School of Business

Faculty

Faculty of Business and Law

First Supervisor

Associate Professor Robert Powell

Second Supervisor

Dr Lee Lim

Abstract

This thesis models the dependence risk profile, investment risk and portfolio allocation features of seven 20-stock portfolios from the mining, energy, retail and manufacturing sectors of the Australian market in the context of the 2008-2009 global financial crisis (2008-2009 GFC) and pre-GFC, GFC, post-GFC and full sample period scenarios revolving around it. The mining and energy portfolios are the base of the study, while the retail and manufacturing are considered for benchmarking purposes. Pair vine copula models including canonical vines (c-vines), drawable vines (d-vines) and regular vines (r-vines) are fitted for the analysis of the portfolios’ multivariate dependence and their underlying sectors’ dependence risk dynamics. Besides, linear and nonlinear optimization methods threaded with the variance, mean absolute deviation (MAD), minimizing regret (Minimax), conditional Value-at-Risk (CVaR) and conditional Drawdown-at-Risk (CDaR) risk measures are implemented to examine the portfolios’ investment risk and optimal portfolio allocation features.

The vine copula modelling of dependence aims at examining the dependence risk profile of the portfolios in specific market conditions; studying the changes of the portfolios’ dependence structure between pairs of period scenarios; and recognizing the vine copula models that best account for the portfolios’ multivariate dependence. The multiple risk measure-based portfolio optimization seeks to identify the least and most investment risky portfolios, single out the portfolio that offers the best risk-return trade-off and recognize the stocks in the portfolios that are good candidates for investment.

This thesis’ main contributions stem from the “copula counting technique” and “average model convergence” perspectives proposed to handle, analyse and interpret the portfolios’ dependence structure and portfolio allocation features. The copula counting technique aside from simplifying the analysis and interpretation of the assets’ dependence structure, it enables an in-depth and comprehensive analysis of their underlying dependence risk dynamics in specific market conditions. The average model convergence addresses the optimal stock selection and investment confidence problems underlying any type of portfolio optimization, and faced by investors when having to select stocks from a wide array of optimal investment scenarios, in a more objective manner, through model convergence and model consensus. Both, the copula counting technique and average model convergence are new concepts that introduce new theory to the pair vine copula and multiple risk measure-based portfolio optimization literatures.

The research findings stemming from the vine copula modelling of dependence indicate that the each of the portfolios modelled has dependence risk features consistent with specific market conditions. Out of the seven portfolios modelled the gold mining and retail benchmark portfolios are found to have the lowest dependence risk in times of financial turbulence. The iron ore-nickel mining and oil-gas energy portfolios have the highest dependence risk in similar market conditions. Out of the energy portfolios the coal-uranium is significantly less dependence risky, relative to the oil-gas. Out of the mining portfolios the iron ore-nickel is the most dependence risky, while the gold portfolio has the lowest dependence risk. The retail benchmark portfolio is significantly less dependence risky than the manufacturing benchmark portfolio in both, tranquil periods and non-tranquil periods. In terms of investment risk, the oil-gas energy portfolio is the most risky.

The “copula counting technique” is acknowledged for simplifying the analysis and interpretation of the portfolios’ dependence structure and their sectors’ dependence risk dynamics. The average model convergence provides an alternative avenue to identify stocks with large weight allocations and high return relative to risk. The research findings and empirical results are interesting in terms of theory and practical financial applications. Portfolio managers, risk managers, hedging practitioners, financial market analysts, systemic risk and capital requirement agents, who follow the trends of the Australian mining, energy, retail and manufacturing sectors, may find the obtained results useful to design investment risk and dependence risk-adjusted optimization algorithms, risk management frameworks and dynamic hedging strategies that best account for the downside risk the mining and energy sectors face during crisis periods to the pair vine copula and multiple risk measure-based portfolio optimization literatures.

The research findings stemming from the vine copula modelling of dependence indicate that the each of the portfolios modelled has dependence risk features consistent with specific market conditions. Out of the seven portfolios modelled the gold mining and retail benchmark portfolios are found to have the lowest dependence risk in times of financial turbulence. The iron ore-nickel mining and oil-gas energy portfolios have the highest dependence risk in similar market conditions. Out of the energy portfolios the coal-uranium is significantly less dependence risky, relative to the oil-gas. Out of the mining portfolios the iron ore-nickel is the most dependence risky, while the gold portfolio has the lowest dependence risk. The retail benchmark portfolio is significantly less dependence risky than the manufacturing benchmark portfolio in both, tranquil periods and non-tranquil periods. In terms of investment risk, the oil-gas energy portfolio is the most risky.

The “copula counting technique” is acknowledged for simplifying the analysis and interpretation of the portfolios’ dependence structure and their sectors’ dependence risk dynamics. The average model convergence provides an alternative avenue to identify stocks with large weight allocations and high return relative to risk. The research findings and empirical results are interesting in terms of theory and practical financial applications. Portfolio managers, risk managers, hedging practitioners, financial market analysts, systemic risk and capital requirement agents, who follow the trends of the Australian mining, energy, retail and manufacturing sectors, may find the obtained results useful to design investment risk and dependence risk-adjusted optimization algorithms, risk management frameworks and dynamic hedging strategies that best account for the downside risk the mining and energy sectors face during crisis periods.

Share

Paper Location

 
COinS