By Solomon Kavuma
2009
Recent growth in central bank reserve holdings have ignited the need to manage reserves in an optimal and efficient way. Reserves are now no longer seen as a store of wealth but as a source of income for the central bank. Appropriate choice of the currency composition of the reserves is one way in which efficient allocation in this regard can be achieved.
The paper identifies an applied framework that maximizes the currency allocation problem. It identifies the issue of currency allocation among central bank reserves as a multi-objective function that involves a lot of matrix computations and simulating a theme against specified risk parameters. This kind of problem makes the mean variance framework the ideal candidate for working out the inherent parallelism in the computations.
The framework is based on the computation of the expected return, which is simulated with the variance and covariance matrices to build the efficient frontier. It discusses some of the advantages and limitations to the approach and finally sets the ground for the choice of the currency composition benchmark based on following risk metric considerations: highest risk adjusted return, capital preservation probability, VAR and the conditional VAR under specified confidence levels.
Keywords: Optimization, variance covariance matrices.