The pre-global financial crisis notion of strategic asset allocation (SAA) and asset liability management studies contained a more static element and trustees considered portfolio suitability based on triennial reviews. Since the crisis, superannuation funds have been moving towards shorter term time frames and absolute return-based investing, traditionally the domain of hedge funds and multi-asset funds.

However, they are burdened by regulatory and peer constraints in terms of liquidity, fees, transparency and asset allocation. This requires increasingly advanced modeling tools that go beyond mean-variance analysis and seamlessly integrate these tools:

Risk factor analysis Examining portfolios on a risk factor, rather than asset class basis (for example, the equity and credit factor risk embedded in hedge funds).
Regime switching models  Examining stressed regimes, incorporating correlation compression and flight-to-quality effects into government bonds.
Market-aware assumptions  Allowing investors to benefit from mean reversion in market cycles over medium-term time frames.
Liquidity tests  Allowing investors to understand how a fund manages liquidity in a steady state (through the cycle) and stressed environment, a mandatory requirement for super funds under the new regulations.


A forward-looking approach

Every major crisis in modern times has been exacerbated by instruments that only gained widespread acceptance a few years prior to the event. For example, portfolio insurance in 1987, statistical arbitrage in the case of Long Term Capital Management in 1998, internet stocks in 2000 and collateralised debt obligations in 2008. The one constant is the boom-bust cycle inherent in human behaviour.

This behaviour can be modeled by adopting a forward-looking, mean-reverting simulation rather than relying purely on historical outcomes. This enables the examination of many possible economic conditions and their impact on asset and portfolio returns over multiple periods. In addition, alternative asset classes may not possess a rich data history, but paths can be simulated using factor models.

Increasingly, investors may prefer to incorporate the medium-term outlook into their assumption set, depending on a specific time horizon. Such assumptions can start from current valuations and mean revert over the market cycle, thereby introducing a certain amount of volatility into return forecasts.

These assumptions suit investors such as defined contribution funds that require a flexible approach to asset allocation, a so-called “floating SAA”, or defined benefit funds with a finite life span. Post-GFC, there is increased appreciation that markets can, from time to time, significantly deviate from fair value in the medium term as they are affected by sentiment and momentum.

As such, including an element of current market indicators can assist in positioning different portfolios in these ways.

For equities Normalised earnings and reversion to long-term valuation bands are critical. Without normalisation Australian equities would look cheap at the top of a mining boom.
For bonds Mean reversion from current to long-run steady-state yield curves can be assumed. Such models suggest lower returns for bonds in the earlier years of the simulation.
For alternative assets Factor models seamlessly integrate the mean reverting risk factors into alternative asset returns.
For currencies A combination of purchasing power parity and interest rate parity can be used to forecast returns of unhedged assets.


It has to be bespoke

Good stochastic modeling tools are essential to engage markets in a forward, rather than backward, looking approach as the system constantly evolves and adapts to both the investment environment and the specific needs of individual investors. However, stochastic analysis is not a cure-all and does contain an element of model risk.

Therefore, it needs to be coupled with scenario and path-specific analysis to answer such questions as what the impacts of a slowdown in China or an Australian housing-market collapse would be. There is room for both stochastic and deterministic models. Whatever the case, the best models will provide unique insights into how capital markets operate and allow for tailoring to investors’ specific needs. Bespoke advice should be offered for both defined contribution and defined benefit funds in this respect.

Dr Harry Liem is a principal at Mercer.

Join the discussion