Diversification – too much is just right

“Diversification is the only free lunch in finance” is a cliché which is easy to dismiss but one which captures an important truth: in a world where it is difficult to predict returns, lowering volatility without reducing return is extremely valuable.

The distinction between diversification (where all ideas are expected to make a positive return) and hedging (where ideas are deliberately offsetting both in risk and return) is extremely important. Here we are referring to genuine diversification: good investment ideas that have low correlation with each other.

Investors can sometimes be dismissive of this – they worry that diversification lowers volatility levels and imply through a naïve information ratio type analysis that this must lead to lower return potential. At its heart this is based on the false premise that market returns are truly independent through time (I.I.D. as it is referred to in statistics).

We know that in the real world returns trend and so investment ideas can have a negative correlation, but both return positively. An obvious real-world example of this can be seen in US Treasuries and the S&P500 over the past 20 years; they have a negative correlation over most time periods, but both have made substantial positive returns.

Despite the apparent obvious benefits, it is common to hear people refer to “over diversification”. There may be three primary reasons that investors are sceptical of the benefits of true diversification:

• While an idea may look diversifying in a particular environment, changing correlations through time may make this transitory. An example of this might be structured products which pay a large coupon but embed a tail risk. Most of the time the coupon payments looks uncorrelated to markets but it is only when the extreme tail event strikes that they draw down alongside other risk assets.
• If they are used to considering investment universes which are highly constrained then the degree to which genuine diversification is available may colour their opinion. For example, within a long only equity portfolio the amount of genuine diversification will be highly constrained compared to, say, a long/short multi asset portfolio.
• Hindsight bias. While diversification is good, just having bought the asset which performed best is better! Clearly this is a weak argument however it is implicit in a lot of the criticism of diversified portfolios. By their very nature a proportion of investment ideas will have lost money over the particular timescale you are considering, but this does not necessarily mean it was not a good idea to hold them given imperfect foresight.

So, while true diversification is undoubtably beneficial it is also difficult to find; locating genuinely independent sources of return takes effort. Of course, in addition to providing diversification, it is important that ideas actually provide return.

We need to start with good ideas and then see what provides diversification rather than simply look for diversifying ideas regardless of merit. As the incremental benefit of adding new independent returns diminishes it would be natural to think there was a sweet spot; get enough diversification to get most of the benefits without the additional cost/effort of adding more.

However, this is where another extremely important market property comes into play: diversification always reduces in a crisis. Without fail, it will turn out that some of the diversification that you thought existed in a portfolio did not persist as some of the returns turn out to not be so genuinely independent.

People often characterise this the “correlations go to one in a crisis” adage, however the truth is rather more nuanced: diversification does always diminish but it does not disappear. Figure 3 shows how the effective number of independent positions within a particular multi asset universe has evolved through time.

So, if we know that diversification will likely reduce when we need it most and that there is a non-linear impact of a reduction in diversification, this leads us to a very important conclusion: in order to construct a truly robust portfolio it appears prudent to maintain “too much” diversification during the good times in order that you still have enough when you need it. To illustrate this, consider two starting points:

• “Enough” diversification – get most of the benefits of volatility reduction without “wasting” effort going further, illustrated as 4 factors which provide a 50% diversification benefit.
• “Too much” diversification – extra diversification which doesn’t seem to bring a lot of benefit, illustrated as 10 factors which only provide an extra 28% diversification benefit for (at least) 2.5x as much effort in sourcing ideas.

But when a crisis emerges and diversification decreases, the benefit becomes clear; starting at a higher level of diversification provides a buffer which avoids hitting the “tipping point” that occurs at a small number of independent factors and hence volatility remains at reasonable levels.

In this illustration, two of the factors turned out not to be independent, so the number of independent factors reduces by two; this increases the volatility of the green portfolio to 41% while the pink portfolio volatility only increases by 12%.

While this is clearly a stylised example it illustrates a powerful point: by maintaining a buffer of true diversification a portfolio can be constructed in a much more robust manor and hence provide a much more stable journey for investors.

This is a philosophy we embed within our portfolio construction through a combination of both quantitative techniques (extensive use of hypothetical scenario analysis as well as traditional risk models) and qualitative judgement (on the fundamental drivers of the return of an idea, of the behaviour of other market participants, etc.), which form an essential part of being able to deliver an absolute return across different market environments and regimes.

Learn more about the strategy