Work needs to be done to develop optimal asset allocation and liability-management strategies for sovereign wealth funds, argues Lionel Martellini. The first step is to define each fund's purpose and source of revenue

By some estimates, the total size of sovereign wealth funds (SWFs) currently stands at more than $3trn, more than twice the estimated size of the world's hedge fund industry, but only one-seventh of the global investment fund industry. The growth of SWFs is likely to continue; research published by Deutsche bank in September 2007 estimated that they could reach around $5trn in the next five years and $10trn within the next decade.

This rapid growth poses challenges for the international financial markets and sovereign states. In particular, an outstanding challenge remains with respect to improving our understanding of optimal investment policy and risk management practices for SWFs. Recent academic research conducted by EDHEC-Risk Institute in co-operation with Deutsche Bank, suggests it is desirable to analyse the optimal investment policy of a sovereign wealth fund in an asset-liability management framework that allows one to formalise the impact on the optimal allocation policy in the presence of risk factors affecting the state surplus dynamics, and the (implicit or explicit) liabilities the fund is facing.

Broadly speaking there are three main kinds of SWFs. The first group contains the natural resources funds, with an estimated 70% of total SWF asset holdings in the hands of resource-rich countries, such as the United Arab Emirates and Norway. The focus of these funds is to maintain economic stability against commodity price fluctuations and ensure that future generations are not disadvantaged by the exploitation of natural resources by the current generation.

The second group relates to the foreign reserve funds, which notably includes a number of Asian countries such as China, Korea and Singapore. The focus of these funds should be to hedge away the impact of risk factors behind these commercial surpluses, and also to generate higher returns than local sterilisation bond costs related to the issuance of sovereign debt aimed at reducing the monetary base expansion linked to capital inflows.

The last group of funds, accounting for a marginal fraction of total sovereign wealth, is the pension reserve funds for countries such as New Zealand, France or Ireland, which have set aside a portion of their pension funds and manage them separately to prepare for an ageing society.

Intuition suggests that allocation policies should differ for different kinds of SWFs. Recent advances in dynamic asset-pricing theory have, in fact, paved the way for a better understanding of optimal dynamic asset allocation decisions for such long-term investors by precisely taking into account the stochastic features of the SWF endowment process (where the money is coming from), the stochastic features of the SWF's expected liability value (what the money is going to be used for), and the stochastic features of the assets held in its portfolio.

In fact, it can be shown that optimal asset allocation strategy involves a state-dependent allocation to three building blocks: a performance-seeking portfolio (also known as the PSP and typically heavily invested in equities); an endowment-hedging portfolio (also known as EHP and customised to meet the risk exposure in the sovereign wealth fund endowment streams); and a liability hedging portfolio (LHP) which is heavily invested in bonds for interest rate hedging motives, and in assets exhibiting attractive inflation-hedging properties when the implicit or explicit liabilities of the sovereign wealth funds exhibit inflation indexation. While the first building block is the standard highest risk-reward component in any investor's portfolio, the other two need to be customised for each specific SWF.

For example, in the case of the Norwegian fund, a natural resource fund that has been set up to help meet future pension payments, the optimal allocation strategy should involve a short position in oil/gas commodity futures, or a long position in stocks of companies such as airlines that benefit from decreases in oil prices, so as to diversify away some of the risk exposure in the country's revenues. It should also include a long position in inflation-linked securities, which will help the sovereign state to hedge away some of the inflation uncertainty in future pension payments.

This PSP/EHP/LHP approach is the equivalent for SWFs of the pensions industry's liability-driven investment strategies (LDI), and from an academic perspective it is yet another example of the fund-separation theorem.

Investable proxies
In general, uncertainty in the endowment stream is not entirely spanned by existing securities. For example, in the case of an SWF managing commercial surpluses, the endowment stream is related to worldwide economic growth, the fluctuations of which are not replicable by a traded asset. This induces a specific form of market incompleteness, which makes the dynamic asset allocation problem more complex, and raises the challenge of designing investable proxies for the hedging of unexpected changes in risk factors that would be likely to affect fund revenues. For example, in the case of a foreign reserve SWF, where revenues are related to trade balance surplus from the sovereign country (for example, China or Singapore), the risk factors that have an impact on the contributions to the sovereign wealth funds would be related to world economic growth, inflation differentials and changes in currency rates, among others.

Overall, it appears that the development of an asset-liability management analysis of sovereign wealth funds has potentially important implications for investment banks and asset managers, which could implement financial engineering techniques for the design of customised building blocks aimed at facilitating the implementation of genuinely dedicated ALM and risk management solutions for these long-term investors.

Top-down and bottom-up
The aforementioned asset allocation decisions relate to the design of the long-term strategic allocation SWFs, with an associated optimal exposure to rewarded risk factors. Additionally, it is legitimate for SWFs to seek alpha opportunities, and/or consider reaching strategic stakes in selected target companies. In fact, long-term equity holdings can be a natural source of alpha generation, given that SWFs are better placed to benefit from any temporary mispricing opportunity thanks to a longer-term investment horizon, and, unlike pension funds, a higher margin for error and the absence of regulatory constraints. Eventually, however, such security selection decisions can lead to a strong bias, such as a recent overweighted exposure to financials.

These unintended bets on market, sector and style returns can have a significant impact on the portfolio return and need to be quantitatively measured, and optimised. Alternatively, these biases can be adjusted for in a multi-factor setting through a completeness portfolio designed to fill in the difference between the portfolio allocation and the long-term strategic benchmark allocation. In this context, index futures can be used as cost efficient vehicles for dynamic adjustment of portfolio exposure to market risk.

Long-term investors, short-term constraints
The recent crisis has had a profound impact on the investment strategy of many SWFs that were built on the idea that they could hold on to investments over long time horizons and cash in on a liquidity premium. The current economic crisis is illustrating that this strategy might not always work out. With increasing economic problems in the home country, the calls for investing in distressed assets at home become louder. At the same time, contributions to many funds are likely to decrease, given a drop in demand for natural resources and shrinking fiscal sources. As a consequence, some sovereign funds have to draw back from investments abroad, even if they are marked at a loss at the time, in order to finance investments in their home country. Furthermore, states are increasingly calling on their funds to bridge gaps in their housekeeping, or to finance economic stimulus packages. This, again, might make ill-timed divestments necessary.
These elements suggest that, beyond the elements described above, SWFs would also benefit from the latest insights regarding dynamic risk-controlled allocation strategies that are designed to help long-term investors meet a number of short-term goals and constraints. These insights are becoming increasingly used by pension funds, where a shift from a static LDI strategy to dynamic LDI strategies is taking place, and they are likely to have an impact on SWF allocation policies in the years ahead.

Lionel Martellini is professor of finance at EDHEC Business School and scientific director of EDHEC-Risk Institute