Describing the relevance of factor investing to the realm of institutional investment and defined benefit pensions is the aim of a new paper from academics at the EDHEC-Risk Institute.
The paper, entitled Factor Investing in Liability-Driven and Goal-Based Investment Solutions from Lionel Martellini (pictured), director at the institute and Vincent Milhau, research director at EDHEC, starts by pointing out how it has already been established that factor investing determines asset allocation by risk factor as opposed to standard asset class separations.
This, the authors say, is a widely accepted idea among institutional investors but questions remain over the exact role of risk factors in an asset-liability management investment process.
Here we turn to modern portfolio theory. It would suggest that “adopting a factor lens” can offer new insights on both the performance-seeking and liability hedging portfolios.
In the world of institutional investing and defined benefit pension, a portfolio is intended to “strike a balance” between performance and risk relative to the benchmark. That benchmark is the value of liabilities for investors facing commitments.
Factor investment drivers
The interest in factor investing has, according to Martinelli and Milhau, been driven by, first, an increasing recognition that traditional cap-weighted indices have a “rather bad” risk-return profile which can be improved upon by investing in stocks that have shared characteristics (i.e. the basis for today’s factor zoo).
Second, with concern over active management fees on the rise, and active’s related failure to provide the performance to justify it, investors have sought out vehicles which can provide performance with “solid economic arguments” and which can be provided at a cheaper cost.
Third, the Great Financial Crisis in 2008 led to renewed interest in sound risk-management practices, leading to more questions being asked about the risks faced for the returns sought. (Beyond the scope of this report, the recent Covid-19-related crash might reinforce this trend even further.)
“These changes have attracted attention to systematic factor investing, now regarded as an approach that blurs the line between passive investing, which involves replicating a cap-weighted index, and active investing, which involves proprietary selection and/or timing skills,” the duo write.
So we move on to the needs of the institutions. Martinelli and Milhau point out the objective for any performance-seeking portfolio (PSP) is to “harvest risk premia across and within asset classes in the most efficient way possible”. In equities, at least, a number of factors have been shown to do just that. Martinelli and Milhau list size, value, momentum, volatility, investment and profitability.
In theory, the PSP should be one that maximises the Sharpe ratio, regardless of the existence or nature of the investors’ liabilities. But, and there is always a but, this is difficult to implement because the maximum Sharpe ratio portfolio depends on the expected returns of constituents, which are “very hard to estimate” and its out-of-sample performance is “severely plagued by estimation errors”.
Alternative theories to help with portfolio composition have been devised, including the Capital Asset Pricing Model or, to avoid the blunt tool of cap weighting, ‘smart weighted’ equity indices have been devised.
Here, factors come in. “By selecting securities with the appropriate characteristics, one can construct portfolios with expected returns above the market without using active management,” Martinelli and Milhau explain.
They add, moreover, that factors are not only useful for designing models to estimate risk and expected return parameters, “but they now can also be regarded as building blocks for the construction of a well-rewarded performance-seeking portfolio”.
The second element of the research moves on to the second building block, that of using a liability-hedging portfolio (LHP) to replicate the performance of a liability. This necessarily needs to be customised to each individual and from a factor standpoint, the relevant factors are also not necessarily the same as for the PSP.
Martinelli and Milhau suggest that at the allocation stage in liability-driven investing, the degree of overlap between assets and liabilities can be measured with a “suitable multi-class factor model”.
“The model can also be used to construct equity portfolios with better liability-hedging properties than a standard broad cap-weighted index,” the team continue. “With these more ‘liability-friendly’ portfolios, investors can in principle allocate more to the performance-seeking portfolio for the same risk budget and therefore enjoy higher returns.”
The last stage, then, is to choose an allocation to the PSP and LHP and this depends on risk budgets. The risk of the LDI strategy, Martinelli and Milhau contend, depends on the investment policy and the risk of each building block and their correlation.
“For instance, its tracking error with respect to the liabilities for a given allocation increases with the tracking error of the LHP, but also with that of the PSP, which is not controlled at the portfolio construction stage since fund separation principles recommend that the PSP be designed with no hedging concern in mind.”
This has the consequence of implying that by decreasing the tracking error of the PSP pillar with respect to its liabilities, an investor can allocate more to this element while also staying within the limits of a given risk budget.
“This leads to an increase in upside potential, unless the performance of the more ‘liability-friendly’ PSP that replaces the original PSP is too inferior.”
After modelling a set of factors – including size, value and momentum as well as interest rates, term spread and credit spread – Martinelli and Milhau say a variety of new PSPs can be constructed using different weighting systems.
They conclude by suggesting that while factor investing and liability-driven investing relate to two very separate traditions in the academic literature, there is a case for combining the approaches.
“Each of the three steps of a liability-driven investing process, namely the construction of a well-rewarded performance-seeking portfolio, the construction of a safe liability-hedging portfolio and an efficient allocation to these building blocks, can be better addressed by taking a factor perspective,” they say.
This, they conclude, is the “first step towards the introduction of a comprehensive investment framework blending liability-driven investing and factor investing.”