Unqualified truth is very rare in the world of investment, which is why investment beliefs are critically important for investors, in particular those who view themselves as long-horizon investors.
But let me propose one here: all genuine long-horizon investors experience underperformance (if they measure investment performance frequently enough).
Let me start with a colourful hypothetical example borrowed from Nassim Nicholas Taleb’s brilliant book Fooled by Randomness. Consider a dentist setting up a trading room in his attic – perfectly rational behaviour, as he is a truly outstanding investor. He is able to outperform short-term bonds by 15% pa, albeit with a volatility of 10% pa. He therefore has a probability of making money in any one year of 93%, which would keep most of us happy. However if we shorten the time frame for measurement, the story starts to sound very different. Measured over a minute, his probability of being ahead shrinks dramatically to 50.17%. Over a second? The very same statistic goes down to 50.02%, basically a coin flip. With this monitoring frequency, all investors will experience underperformance; literally in a matter of seconds.
Of course no investors monitor performance that frequently so let me show you some real-world data. A study conducted by Brandes Institute examined a sample of 145 international equity funds and their long and short-term performance. It discovered that the top 15 funds with the highest 15-year returns all underperformed the index and their peers significantly during shorter periods. All of them showed up in the worst decile for at least one quarter. When measuring rolling three-year returns eight out of the 15 fell into the worst decile at least once. Their conclusion is that short-term underperformance is “as normal as death and taxes” and simply an inherent by-product of the long-term investment process.
With that I think it is reasonable to argue that for long-horizon investors, short-term underperformance is not something they might encounter; it is something they will encounter.
Unfortunately it is well established that human brains don’t treat losses and gains the same. There is a technical term here introduced by Amos Tversky and Daniel Kahneman: loss aversion. It refers to people's tendency to prefer avoiding losses to acquiring equivalent gains – the emotional wear and tear caused by the losses outweighs the boost from the gains.
If we marry loss aversion with frequent performance measurement we then get another technical term that starts to reveal one of the fundamental difficulties with regards to long-horizon investing: myopic loss aversion.
Remember in our example the dentist has a 50.17% probability of being ahead (ie outperforming short-term bonds) over a minute. Assuming he spends eight hours a day in front of his screen, he will have (on average) 241 pleasurable minutes against 239 unpleasant ones. Not only will our dentist be emotionally drained by the end of each day from the sheer volatility of the ups and downs, but he will feel the losses far more keenly that any boost he gets from gains. Our dentist will simply not survive this emotional onslaught, and heaven forbid may even be tempted to change the portfolio (which if left alone has a 93% chance of finishing the year ahead).
To summarise, myopic loss aversion leads to “selling low” – terminating prematurely a sound long-term investment position – and that is exactly the behavioural trap long-horizon investors should guard themselves against.
There is a simple solution, at least in theory: recognise the value of inactivity and evaluate investment performance less often. In practice fiduciary duty can make it hard to argue that you are acting responsibly in respect of someone else's investments if you don't even know what the performance looks like. A remedy to that would be shifting the focus of reporting/measurement from short-term metrics to long-term outcomes – eg extending the term over which performance is measured. Instead of reading too much into the performance for the last quarter, try to put it in the context of the long term by focusing on for example the average return for the past seven or ten years.
Better statistical tests can be designed so as not to draw erroneous conclusions from data with abnormally high noise. These tests should be pre-specified with an agreed confidence interval, and be sensitive to the changing degrees of freedom as we collect more data.
The tension for long-horizon measurement is to stay focused on achieving long-term goals while still providing short-term checks and balances / ongoing review. To overcome the short-term noise issue it is important to incorporate subjective qualitative assessment alongside more objective performance data points. In essence, it requires looking at non-performance elements and seeking to answer whether there is anything about the investment proposition now that leads us to believe is will make a positive (or negative) performance contribution in the future. Has the investment strategy executed been consistent with stated investment beliefs and thesis? Did anything happen to affect the qualitative, forward-looking skill rating of the (both internal and external) asset managers? Has the investment team been stable and has team culture remained positive and strong?
Long-horizon investors should study the past but it is the past experience that is informative and valuable; not the past performance.