A Comprehensive Framework for Global and Domestic Asset Allocation


There are two main streams to deal with traditional asset allocation strategies i.e. theoretical approach and implementation approach. These approaches are the prime focus of this study. Portfolio optimization is based upon two fundamental ingredients i.e. estimation of return vector and covariance matrix. This study compares the 12 covariance matrix under four categories i.e conventional methods, factor models, portfolio of estimators and shrinkage approach. This study also compares the performance of 7 alternative ways for estimation of return vector. Study also develops portfolios based on mean-variance optimization, minimum variance portfolios, constraints portfolios and naïve diversification. This study first time introduces the ‘country risk’ as unprice risk factor in the Black-Litterman model and uses this augmented Black-Litterman formula (BL-CR) for the estimation of expected return vector. The comparison of asset allocation strategies are base upon the financial efficiency and diversification dimensions using 10 asset classes from 5 emerging Asian countries i.e. India, Indonesia, Pakistan, Philippines & Thailand, 4 asset classes from global environment and 22 asset classes from Pakistan. Study reveals that factor models as a group outperform the competing covariance estimators in all the emerging countries. From the number of positive and negative weights to asset classes, maximum and minimum value of weights, other diversification measures of the mean-variance framework, it is reveal that mean variance portfolios are concentrated, mostly counterintuitive, results more short positions and highly sensitive to the choice of input. Similarly the financial efficiency of these portfolios is also highly sensitive to the input estimates. Results of asset allocation strategies suggest that, on an average, equally weighted portfolios result a competitive strategy in Pakistan and in global environment. Therefore study also recommends that investment managers and academia should at least consider the naïve diversification as a first obvious benchmark in comparison with other asset allocation strategies. The BL-CR model outperform the original model as it has relatively less short positions, more number of positive weight, less variance, low value of Herfindahl index and high value of excess sharp ratio. Therefore BL-CR model is more appropriate on mathematical and empirical ground in asset allocation than original model to disperse country risk. This study also recommends that investment managers and academia should consider the Black-Litterman model under country risk for tactical asset allocation decisions in emerging Asian countries.

Download full paper