Nonparametric approach to portfolio diversification: the case of Australian equity market.

This study investigates the portfolio diversification possibilities among Australian sectoral, size and style indexes and between Australian aggregate equity index and selected international indexes. Two analytical methods are used – nonparametric cointegration that appears to be the most appropriate for the financial data analysis, and principal component analysis (PCA) that is suitable for detecting relations among a large number of variables and for clustering co-moving variables. Having identified linear and non-linear unit roots in the time series data we show that based on Bierens’ nonparametric cointegration the number of cointegrating relations between respective indexes increases (and the portfolio diversification opportunities diminish) in the post-GFC period (2007-2012) relative to the historic average (1992-2012). Regarding sectoral diversification, the PCA results suggest that sectoral relations underwent minor changes in the post-GFC period with few additional diversification opportunities appearing.