Standard Life Investments

Evaluating Multi-Asset Strategies


An increasing number of institutional investors recognize the benefits of a multi-asset approach, including the potential to improve diversity, increase liquidity and reduce volatility. The strategy fits readily alongside a variety of investment approaches and asset class categories. But adopting a multi-asset strategy comes with challenges: in particular, how to evaluate outcomes. Relying on only one or two measures for evaluation can lead to misinterpretation of the historical results. Instead, it is advisable to use a variety of evaluation techniques. One of these – correlation – is widely used, but also widely misused in a multi-asset context. It fails to capture the many dimensions of multi-asset investing. There are a number of more useful performance and risk analytics, both historical and predictive, that can help in understanding what drives multi-asset investment outcomes.

The need for a better approach

Multi-asset strategies are successfully gaining ground within institutional portfolios, and their role is likely to grow. A 2014 report by McKinsey estimated that assets allocated to multi-asset strategies will increase by 10% annually over the next several years, making it one of the most rapidly growing investment approaches in the US. A study of US institutional investors commissioned by Standard Life Investments underlined the many benefits of multi-asset strategies (Greenwich Associates [2015]). These include the potential for improved diversity, greater liquidity and reduced volatility. Also advantageous is their ability to fit readily alongside a variety of investment approaches and asset class categories.

That said, multi-asset strategies come with challenge and even defining the term ‘multi-asset’ is not straightforward. Accurately evaluating the strategy outcome is vital and should begin with the recognition that there are no short cuts. Relying on only one or two measures for evaluation can lead to misinterpretation of the historical investment results achieved. Instead, it is advisable to use a variety of evaluation techniques. One of these – correlation – we analyze in depth, as we believe it is misunderstood in the many dimensions of multi-asset investing. We go on to examine some of the more useful performance and risk analytics that can help us understand what drives multi-asset investment outcomes.

Multiple measures of multi-asset

Our investigation covers the following stages.

  • We review methodologies that can help us evaluate strategies beyond basic return and risk analysis. We focus this analysis on strategies that aim to provide downside protection, meaningful diversification to the broader portfolio, and an absolute return.
  • We set out the limitations of correlations in evaluating strategies and markets. These are a result of complexities, tail dependence, regime shifts and sensitivity to outlier events.
  • We review different ways to evaluate multi-strategy outcomes, using a range of historic and predictive measures. Historical evaluation analyzes the impact of past events on current holdings. Predictive techniques extrapolate from the past to suggest how the current portfolio might perform under different circumstances in the future.
  • We address why the best evaluation measures will vary.

Conclusions

Institutional investors will benefit from a more nuanced analysis of multi-asset portfolios. In particular, correlation can be misleading when viewed in isolation. This means investors analyzing historical correlations – and variations in correlation between risk assets and the multi-asset portfolio – should not use this information to drive decision-making in isolation. Rather, correlation should be complemented by other measures.

Investors can assess tail behavior to understand performance during market stress. Analysis of market capture in both rising and falling markets provides insight into long-term returns. Asset class attribution and market participation analysis corroborates – or rejects – findings on correlation. Predictive measures, such as risk modelling and tail-behavior modelling, complement historical methods. Investors should also consider the placement and role of the strategy within their overall portfolio to determine which evaluation techniques are most important.

Employing a combination of different approaches provides investors with a better assessment of the true risks in their portfolios. They are tools that allow us to challenge and enhance our understanding of portfolio behavior in extreme scenarios.

Download the full paper from the Journal of Portfolio Management, December 2017