Does diversification fail us?

Diversification is often described as the only free lunch in finance. Many market participants and academics advocate diversifying by asset class, geography and/or risk factor to aim to achieve higher returns with a lower level of risk.

However, as recent events have shown, when there is a significant downturn, the price of growth assets can sometimes fall across the board. This can lead to larger declines in a portfolio’s value than expected, with diversification seemingly offering little protection. This has led many investors to ask the question: does the free lunch disappear when you are hungriest? If so, how beneficial really is diversification for investors?

Correlations are not static

One method used to assess the potential diversification benefits of an asset class is to look at its correlation with other assets: the lower the correlation, the stronger the implied diversification benefits. Observers of recent events and financial crises in general will note that there appear to be few ‘hiding places’ when the market is in turmoil, and this is supported by data. As shown below, correlations of asset classes with equities tend to increase in the worst 5% of days for equity performance, but reduce for the strongest 5% of days.

Why is it still important to diversify?

Here are some of the key reasons why we believe diversification remains important:

Relevance of time horizon

A classic quote from Benjamin Graham is that ‘In the short run, the market is a voting machine but in the long run, it is a weighing machine.’ Returns are driven by a combination of fundamentals and sentiment, and it is the former that should matter most to long-term investors.

Looking at daily or monthly returns can be misleading due to sentiment dominating short-term price moves, as there can be a tendency in times of market stress for investors to sell everything at once, either due to panic or because they are forced to (for example by their investment mandate), which increases correlations in the short term.

Despite criticisms of using volatility as a measure of risk, long-term investors should also avoid focusing too much on the shape of outcomes. The short-term shape of outcomes washes out in the long run, as explained in this blog. Investors should also keep the effect of large short-term moves on ongoing solvency in mind, particularly if drawing down on assets.

But in summary, if you can look through short-term volatility, we believe the jump in short-term correlations is less relevant over the long term.

The recovery is just as important as the fall

Diversification mitigates the risk of being invested in the slowest-recovering asset. The consequences of a lower but still positive return can be just as powerful in the long run: we calculate that a 1% lower return for 40 years has about the same impact as an instantaneous fall of one third. As an example, an investor holding solely Austrian equities in 1913 would have taken 97 years to get their money back in real terms.

Don’t overstate the impact of a spike in correlations

While the jump in correlations during a market downturn is undesirable, any correlation less than 1 still offers a diversification benefit, even if that benefit is reduced. Using the correlations in the chart above, we compared the volatility of an equally weighted portfolio (across the six asset classes) using downside correlations versus correlations based on the whole dataset. Holding all else equal, the volatility was only about 0.5% higher. The larger driver of risk in a downturn is that overall volatility spikes.

How investors can prepare

An increase in correlations during downturns makes diversification less attractive, but this doesn’t make it useless. It is rather like wearing a seatbelt: it won’t help you much if you drive off a cliff, but it will protect you to some degree in many other scenarios!

But investors shouldn’t ignore short-term tail risk, especially “knock-out” risk (the level of poor performance they could not withstand). Unfortunately these risks are inherently hard to model as there is scant data.

One approach for dealing with this is to conduct stress tests using deterministic scenarios. At LGIM, a panel of investment experts across multiple teams construct a broad range of plausible tail event scenarios to stress test portfolios. This is then used to identify potential tail events on the horizon, ascertain weaknesses or risk concentrations in a portfolio, and then highlight potential tail hedges.

This won’t stop portfolios experiencing poor performance in a downturn, but it might just help them reach the other side in better shape.