Foreign reserves strategic asset allocation : a Bayesian approach

  • Timbul Budi Santoso

Student thesis: Doctoral Thesis


Despite massive global foreign reserves accumulation since the late 1990s, a study on strategic asset allocation from a central bank perspective remains relatively limited. Many countries have built up foreign reserves greater than their trading and financing sufficiency; consequently, there is increased public awareness as to how central banks manage their reserves. This research addresses the issue of asset structure of foreign reserves portfolios; currency composition and strategic asset allocation. This thesis provides an alternative strategic asset allocation framework that focuses more on liquidity and safety objectives, without sacrificing return objective of the existing central banks' portfolios. This research contributes to the theoretical literature that combines mean-variance Bayesian framework with the spirit of the transaction theory of central bank foreign reserves. First, using the unconditional Bayesian approach with external transaction constraints, I investigate risk reduction benefits to both minimum variance and optimal portfolios for a given existing return for central banks'currency allocation. Our results show that before imposing currency weight constraints, all the diversification strategies considered provide significant benefits, regardless of the current benchmark used. The results when the trade constraints and debt constraints are imposed reveal that there is the potential for diversification benefits to be obtained from the currency portfolio optimisation. However, the choice of the current benchmark, and both trade and debt constraints, play an important role in the decision as to with which diversification strategy central bank reserves managers should proceed. Second, using the similar objective functions and diversification benefits measures, the framework is then applied to define asset allocation policy for government bond portfolio. The results from the optimal portfolio for a given existing benchmark return show that there is significant risk minimization benefit by adding developed market longer maturity, quasi-government, emerging and inflation-linked government bonds altogether to the current benchmark portfolio.The benefits could not be found if each asset class is added directly to the current portfolio policy. From the same spanning strategies as the optimum portfolio, the global minimum variance portfolio analysis shows that the diversification benefits are mainly driven by United States Treasury Bills. Third, I investigate risks minimization of government bonds portfolio for a given benchmark returns in various investment policies and risk preferences. This research examines the important role of budget constraints, liquidity buffer allocation, global financial crisis investment opportunities, investing in ultra-long government bond and non-bond investment. Analysis on the impact of the budget provides policy implications that central bank should not to tranche their reserves, but express liquidity requirements in the form of constraints on the portfolio optimisation framework. Investigation on the role of liquidity allocation confirms earlier findings that the risk reduction benefits are mainly driven by US Treasury Bills. The impact of global financial crisis since 2008 emphasises the need for central banks to diversify their foreign reserves benchmark beyond its current setting. We could not find benefits for the central bank to invest in ultra-long maturity government bonds. Lastly, we found that non-government bond investments provide significant benefits when it compares to the existing benchmark, and incremental benefits from the well-diversified government-related bond portfolio.
Date of Award4 Jan 2017
Original languageEnglish
Awarding Institution
  • University Of Strathclyde

Cite this