top

Asiasn review of Financial research

Past Issues

HOMEPast Issues Past Issues

Asian Review of Financial Research Vol.31 No.3 pp.415-448
Factor-based Strategic Asset Allocation for Pension Funds
Joon-Haeng Lee Professor, Department of Economics, Seoul Women's University
Kinam Park* Researcher, Korea Asset Pricing
Key Words : Asset Allocation,Alternative Investment,Risk Factor,Factor Exposure,Risk Management

Abstract

The essence of strategic asset allocation (SAA) is to derive the optimal asset allocation taking into account the diversification effect of each asset class under the same risk and return profile structure. The mean variance optimization method is useful when the structure of the risk and return of assets is homogeneous. However, due to the emergence of various alternative investments, methods for determining optimal asset allocation are the subject of controversy. We review the problems of the traditional SAA method that is used by pension funds and propose a new SAA method that considers multiple factors. We examine how to apply the new model to domestic pension funds. We provide a new asset allocation paradigm that provides a reasonable weighting for alternative investments and a basis for determining foreign currency hedge ratios for foreign investments. A review of domestic pension funds' portfolios, assuming a similar target rate of returns and risk tolerance limits, identifies big differences in the weights given to alternative investments. The variation occurs because most pension funds predetermine the portion of the portfolio that will be alternative investments and then perform an SAA only for the remaining asset classes. As decisions on the weights of alternative investment consider the entire portfolio, it is necessary to improve the current asset-based SAA method that performs the allocation after excluding the alternative investment. Factor-based SAA can be a useful method for incorporating alternative investments into the asset allocation framework and can identify methodologies for integrating risk with existing assets such as stocks and bonds. There are four steps in factor-based SAA: selecting an asset mapping the asset class to each corresponding factor determining the target factor weight and re-mapping it to the asset class. In the first step, an appropriate factor must be selected to explain the payoff of each asset class. We use a principal component analysis to extract the common elements in all of the asset classes and then conduct a correlation analysis with the factor candidates to determine growth, inflation, real interest rates, credit, and FX. It is interesting to note that the inflation factor has a higher negative correlation with the other factors during an economic crisis, confirming that investing in the inflation factor during a crisis can effectively hedge a portfolio's returns. Second, to estimate the factor coefficients for each asset using a multifactor model it is necessary to map the asset class to each corresponding factor. The estimated coefficients are then used as a link in determining the target factor exposure and to determine the optimal asset allocation. The current six asset classes are re-defined and divided into 17 asset classes based on the cash flow and risk characteristics of each asset class. In particular, we classify the alternative investments into equity and debt types. In this way, we can make the risk and profit structure of each asset class as homogeneous as possible. Third, the target factor weight is determined by selecting an appropriate factor exposure for the risk profile of the pension fund. The target factor weight should be consistent with the pension fund's investment objectives and risk tolerance limits and is also a key element in determining the optimal allocation of investment assets. To determine the target factor exposure, we first derive a factor-based efficient frontier. Once the factor-based efficient frontier has been acquired, the optimal factor exposure with the highest Sharpe ratio is selected from the alternatives that meet the target returns and risk tolerances of the pension fund. In this case, the target factor exposure is based on the implied factor exposure obtained from the current portfolio's starting point. Fourth, to remap to the asset class, we determine the asset class that meets the target factor weight. This process usually uses an optimization method. We set up a hypothetical portfolio of pension funds and analyze these portfolios. The implications of the case analysis are summarized as follows. First, to increase the Sharpe ratio, it is desirable to reduce the domestic equity investments and increase the portion of overseas investments. Second, the foreign exchange hedge policy of the pension fund is too conservative and should be expanded. Third, the impact of alternative investments on the portfolio's risk-return profile varies by its composition, such that the weights should be distributed based on the correlation of factors and factor exposures. In addition, factor-based SAA allows monitoring of the exposure of investment assets to each of the corresponding risk factors at all of the times when effective risk monitoring for alternative investments is expected.
LIST
Export citation