Balancing the risk budget

By now, many Japanese pension funds have concluded their annual review for the fiscal year 2006. With the average return for the year at around 4.5%, many boards will be moderately satisfied about what will be the fourth consecutive year of positive investment returns after three consecutive negative years from fiscal 2000-02, which brought turmoil and despair to the industry. Many pension funds have now reached a solvency level that allows them to take a step back and re-assess the risk profiles of their portfolios to avoid having to face under-funding of the severity seen after the IT bubble burst in 2000.

Nearly 10 years have passed since the abolition of the 5-3-3-2 rule (a portfolio allocation with at least 50% in ‘safe' assets, 30% in equities, 30% in non-yen assets and 20% in real estate) for funds. The abolition of these restrictions came at an awkward time. Having obtained more freedom, pension funds added to their equity exposure as the IT bubble developed and burst.

Regarding liabilities, a similar observation can be made: Japan introduced new financial accounting rules for pension liabilities in 2000; in fact, it was one of the first countries to bring its rules closer in line with IAS. With 10-year Japanese government bond yields already below 2% at that point, the vastness of the market value of liabilities (previously not directly apparent as they remained off balance sheet for financial accounting purposes and were discounted at 5.5% for pension accounting purposes) became all too clear.

With Japanese equities reaching a 20-year low in April 2003 and the 10-year yield on JGBs hitting 0.43% three months later, funds were harshly confronted with what are still the main elements of a pension fund's risk-budget: interest rate risk and equity beta risk.

In previous columns I elaborated on the structural changes sponsors and pension boards have implemented to lower the risk level of the schemes, with the introduction of cash-balance plans and DC plans as the main result. With the solid investment returns of the past four years and long-term yields at stable albeit still low levels, pension funds are starting to reassess their risk budget as an additional tool to manage solvency risks.

Looking at portfolio allocations, the annual survey by the Pension Fund Association has shown a shift from the traditional assets (insurance GICs, domestic bonds, international bonds, domestic equities and international equities) to non-traditional ones. The latest survey shows the average pension fund now has a 4.2% allocation to hedge funds and 3.2% to other assets such as real estate. These are averages and disguise the wide divergence in the level of innovation among schemes: individual funds could have as much as 50% in non-traditional assets.

What can be concluded from the average is that the risk budget in an asset-only context of the typical scheme remains heavily tilted to equity beta (31% domestic and 18% international), since allocations to risk assets other than equity are minimal. Highly unconstrained mandates remain rare and risk allocation to hedge funds (which for simplicity we assume to represent pure alpha) remains low in relation to the total risk budget. If we also consider liabilities, the risk budget most probably is dominated by interest rate risk and equity beta risk. ‘Most probably,' because although the introduction of cash balance benefit formulas may have shortened the duration profile of liabilities somewhat, on average the maturity of the typical final salary DB liability will be much longer than the 5.8 years duration of the Nomura BPI benchmark used for all but a few domestic fixed income portfolios.

If the risk budget of a Japanese scheme is mainly comprised of interest risk and equity beta risk, is this an efficient use of the risk budget?

One may have the view that under the current Japanese yields and equity-valuation environment, having a risk budget tilted to these elements makes sense.

One could also argue that a more diversified composition is required. Over the past 25 years the Japanese Topix returned 5.4% on an annualised basis with a standard deviation of 24%. Calculating an optimistic (risk-free rate assumed 0%) Sharpe ratio would result in 0.23 as risk-adjusted return for the domestic equity beta risk portion of the portfolio. For the domestic bond asset class, the return and risk figures for the past 25 years are 5.2% and 4.4%, leading to a Sharpe ratio of 1.2 in an asset-only context.

The above numbers are probably unrealistic going forward, especially where it concerns returns. With 10-year yields at 1.6% one would not expect a medium to longer term return on domestic bonds of 5.2%; on the other hand, given that the past 25 years includes Japan's lost decade, a 5.4% return assumption on Japanese equities might be too conservative. The risk profiles of the asset classes suggest a Sharpe ratio for these two beta categories of 0.3 to 0.5, which coincides with empirical research.

If we talk about adding alpha to a risk budget, the 0.3 to 0.5 range of risk-adjusted return appears to be a hurdle most active managers seem confident they can beat on a three- to five-year horizon. Meanwhile asset classes such as commodities may provide comparable expected Sharpe ratios, but their correlation with other parts of the risk budget makes inclusion highly efficiency-enhancing.

Having a balance of beta and alpha risks in your portfolio, taking into account expected Sharpe and information ratios and correlations between the various strands of risk, will be key to avoiding a recurrence of the dire straits the industry was in only four years ago. Further beta diversification via allocations to sources such as real estate/infrastructure, credit and commodities in combination with larger tracking error budgets to alpha sources with high and stable information ratios and diversifying properties such as hedge funds, currency alpha and GTAA alpha are now being embraced by more pension funds as they strive towards stability of solvency levels painstakingly regained over the past four years.