Author: Philip Edwards

Defensive diversification

“Diversification is the only free lunch” is one of the most over-used – and perhaps dangerous – phrases in finance. The idea derives from Modern Portfolio Theory (MPT), developed by Harry Markowitz in the 1950s, for which he shared the Nobel prize in Economics in 1990. MPT provided mathematical tools, in the form of mean-variance optimisation, to help investors maximise their expected return for a given level of risk. In essence, such tools embedded the idea that by diversifying across a range of assets, investors could obtain a more attractive risk-return trade-off.

The legacy of MPT is that diversification has become one of the easiest and most commonly used justifications for any investment decision. However, rather than being seen as a universal good, diversification should more accurately be viewed as a trade-off: we can reduce the downside of being wrong, but only at the cost of reducing the upside from being right. As Buffett once remarked: “diversification is protection against ignorance; it makes little sense if you know what you’re doing.”

Putting this a little less bluntly, we might think of diversification as a useful defence against uncertainty; but one that dilutes the impact of our highest conviction insights and often raises the governance burden. The dilution effect is clear in relation to actively managed portfolios: the more diversified a portfolio, the less impact any high conviction positions can have on overall performance. The governance cost relates to the additional work – in the form of due diligence and ongoing monitoring – that results from adding new assets to a portfolio.

That diversification, conviction and governance considerations are in tension is perfectly natural. However, in the presence of principal-agent dynamics, the temptation is for asset managers to over-diversify portfolios – by moving in the direction of a flawed benchmark – resulting in a less attractive risk-return profile for the end investor.

Defensive decision-making

In his 2014 book, Risk Savvy, the decision-making expert Gerd Gigerenzer defined defensive decision-making as follows: “a person or group ranks option A as the best for the situation, but chooses an inferior option B to protect itself in case something goes wrong.” Much of the literature on defensive decision-making focuses on the medical setting, where examples of “defensive medicine” include ordering unnecessary tests and over-prescribing medications in order to reduce the risk of being sued for malpractice. However, Gigerenzer points out that defensive decision-making is commonplace in the modern world, often arising in relation to political or corporate decisions where blame can easily be assigned.

Defensive decision-making flourishes in situations where decisions are taken by one party (the “agent”) on behalf of, or that impact, another party (the “principal”). The danger is that instead of assessing the risks and opportunities in a balanced way, the agent chooses to focus primarily on managing downside risk, fearing that poor outcomes will be felt more intensely by the principal than good outcomes. Such principal-agent problems abound in the investment industry, where trustees often take responsibility for decisions on behalf of beneficiaries, and trustees invariably delegate stock selection to external asset managers – sometimes further mediated by a fiduciary manager. The aggregate effect of these layers of intermediation is a potentially serious misalignment of interests which creates the conditions for defensive decision-making.

An example of “defensive diversification” would be where an asset manager holds a security primarily in order to reduce the risk of underperforming the benchmark, but where there is little or no expectation of positive long-term returns. Diversifying a portfolio to manage benchmark-relative risk in this way may reduce the portfolio manager’s career risk, but does nothing to improve the end-saver’s long-term return prospects. Indeed, there is a growing body of academic work suggesting that over-diversification imposes a cost on the end saver in the form of lower returns.

In Best Ideas, Cohen, Polk and Silli show that active equity managers’ highest conviction positions outperform the market and other stocks in their portfolio by a large margin (c.4-10% p.a. depending on the methodology and benchmark employed). The results are statistically and economically significant and robust across a range of specifications. The authors suggest “that while the typical manager has a small number of good investment ideas that provide positive alpha in expectation, the remaining ideas in the typical managed portfolio add little or no alpha.”

The authors put forward a number of reasons why managers might proactively choose to dilute their best ideas with what turn out to be value-destructive diversifying positions. These include a desire to avoid extreme benchmark-relative performance; attempts to maximise risk-adjusted returns (as opposed to just returns); or a commercial preference for running a strategy with a large volume of assets (which precludes holding a small number of relatively concentrated positions).

Related research by Cremers and Petajisto (2009) showed that the most active stock pickers – as measured by Active Share – delivered the greatest long-term outperformance. Managers seeking to remain within a certain distance of the benchmark ultimately delivered disappointing long-term returns. This work supports the idea that an overreliance on benchmarks as a solution to the principal-agent problem creates a situation in which diversification undermines managers’ ability to deliver alpha.

This is not to lay all the blame for defensive diversification at the feet of the asset management community. In many ways, managers are simply responding to the instructions and incentives provided by asset owners. In particular, any shared understanding between the asset owner and manager on what would constitute a tolerable limit for short-term underperformance (whether formalised in a tracking error constraint or not) would tend to mitigate against holding a concentrated portfolio dominated by high conviction positions. Procyclical hire and fire decisions on the part of trustees – terminating managers after a bad 3 years, hiring after a good 3 years – also provides a strong motivation for managers to avoid concentrated bets that might give rise to extended periods of benchmark underperformance.

Furthermore, the widely followed and reported Morningstar Ratings disincentivise concentrated portfolios by penalising volatility in their rating methodology. In order to maintain a high Morningstar Rating, managers need to avoid benchmark underperformance over 3- and 5-year periods and to minimise the overall volatility of their portfolio. So while most institutional investors should be able to diversify manager-level volatility themselves, asset managers are presented with a host of reasons to reduce portfolio volatility – both in absolute and benchmark-relative terms – in the process diluting their highest conviction positions.

Mitigating defensive diversification

While MPT has created the impression that diversification is always and everywhere a good thing, asset owners would benefit from a more sceptical attitude. Indeed, the argument outlined above suggests that over-diversification favours managers – giving them the ability to raise larger funds while managing career and business risk – at the expense of returns to investors.

Long-horizon investors, whose beliefs and governance structure enable them to tolerate substantial benchmark-relative risk, may wish to consider whether their managers are running portfolios that adequately reflect their high conviction views. Clearly there is no “right” level of diversification, but asset owners conscious of the incentives leading managers to over-diversify, should be willing to challenge their managers over the extent to which their best ideas can have a meaningful impact on performance.

Concentrated portfolios offer no guarantee of superior long-term performance. But asset owners pursuing an active investment strategy need to ensure that principal-agent conflicts do not undermine their ability to deliver long-term returns in the interest of their beneficiaries. This requires careful scrutiny and engagement with managers at the outset of any relationship, a monitoring approach that identifies performance-chasing and defensive diversification, and a decision-making process that can cope with volatile benchmark-relative performance at the individual manager level.

Asset mispricing: the ignored externality

As part of the growing movement to recognise environmental, social and corporate governance (ESG) risks, the concept of externalities has become increasingly relevant. The English economist Arthur Pigou was one of the first to develop the idea of externalities in the early 20th century, pointing out that sparks from railway engines sometimes ignited nearby woodlands or farmland, destroying timber or crops. Because the owners of the land received no compensation for the damage caused, neither the railway companies nor their customers recognised the social costs associated with their transaction.

The externalities of greatest interest in contemporary discussion typically fall under the E, S and G headings. However, this leaves a critically important externality – that of asset mispricing – largely ignored. As well as polluted air, we have polluted asset prices. In this case, the offending parties are asset management strategies that are driven by prices rather than cashflows, contributing to destabilising positive feedback effects and chronic mispricing.

Adherents of efficient markets thinking will argue that the extent of asset mispricing is so small as to make any negative spillover effects trivial. But the existence of asset price bubbles, large and persistent profits to simple momentum strategies, and the inversion of risk and return across many asset classes (often described as the low volatility anomaly) completely undermines the picture of markets as more-or-less efficient.

If we instead accept that financial assets can be mispriced, and at times severely so, then it is worth considering what economic damage might result from this source. We can think about this at two levels: at the macro level of asset price bubbles and at the micro level of corporate decision-making.

Asset price bubbles are the ultimate manifestation of mispricing and are widely regarded as economically disruptive. However, the conventional wisdom in central banking – epitomised by the Greenspan doctrine – had been to use monetary policy to limit the damage after a collapse in asset prices, rather than attempting to head off the damage as a bubble inflates. This view remained intact in the aftermath of the Tech bubble of the late 90s, but was fundamentally challenged by the experience of the financial crisis. In a speech in 20091, Janet Yellen said that “not dealing with certain kinds of bubbles before they get big can have grave consequences. This lends more weight to arguments in favour of attempting to mitigate bubbles, especially when a credit boom is the driving factor.” It remains to be seen to what extent central bank policy will embrace this more interventionist mindset and how effective the approach will be.

While less obvious than asset price bubbles, the damage done by the impact of mispricing on corporate decision-making may be just as great. If prices do not reflect fundamental value, corporate executives are faced with a dilemma: do they maximise the share price in the short run or cashflows in the long run. In the presence of lucrative share option schemes, many choose the former. There has been much hand-wringing over the causes and effects of corporate short-termism, with the focus often on egregious compensation packages or the problems of quarterly earnings guidance. But the effects of asset mispricing on the private sector have received relatively little attention.

Asset owners have an opportunity to reduce both the macro and micro level social costs described above through their investment strategy choices. As we have argued elsewhere, investment decisions based on past price movements, as opposed to a thoughtful assessment of future cashflows, can cause asset prices to depart substantially from fair value. Common examples of price-only strategies include momentum and trend-following, tracking error constraints, and procyclical performance-chasing by both asset managers and asset owners. Performance-chasing is more a by-product of career risk considerations than an explicit strategy choice, but nonetheless deserves more attention than it typically receives. By identifying and removing such strategies from their portfolios, asset owners could help reduce the feedback effects that contribute to asset mispricing and thereby reduce the externalities that accompany it.

Amidst the broader shift in attitudes towards business – with a greater expectation that companies should behave in a socially responsible and sustainable way – investors need to be cognisant of the externalities that arise from their investment approach (and not just from the underlying companies they hold). Attention has so far been directed towards environmental and social issues, but over time, questions will increasingly be asked of the social utility of various aspects of finance – especially if financial market excesses contribute to an economic downturn in the years ahead. Asset owners and asset managers therefore have good reason, on both private and social grounds, to review the wider impact that their investment approach has on the world around them.