THE INTER-TEMPORAL STABILITY OF ASSET CORRELATION IN SINGAPORE

Abstract

Under the Modern Portfolio Theory framework, the portfolio selection model is based on the optimisation of the portfolio mean-variance relationship The true application of the model would require ex-ante data, meaning that data employed needs to be projected or forecasted. However, the inherent difficulties of obtaining ex-ante data means that often, the many studies which exhort the benefits of portfolio diversification utilise ex-post or historical data. In particular, past studies supporting the benefits of including property in a portfolio presented historical low positive or negative correlations between property equities and other assets as evidence. In the usage of such historical data, caution is needed. This is because unless the observed structural relationships are stable over time, the use of such ex-post data co-movement patterns as proxies for the ex-ante relationships is not feasible. In view of this, this dissertation sets out to examine the inter-temporal stability of correlations among assets in Singapore, especially between property and equities. The stability of the asset correlation matrix is important as the presence of such stability over time facilitates the selection of an ex-ante optimal investment strategy. In particular, if considerable instability is found in asset correlations over time, the benefits of diversifying over different assets, especially property, may be less significant than previously found After reviewing the relevant literature and carrying out inter-temporal stability tests of correlations among major assets in Singapore, it was found that the correlations were generally stable over time, even as the investment horizon was lengthened from short term through the mid term to the long term This implies that the ex-post patterns of co-movement are useful proxies for the ex-ante co-movements of the different assets.