It’s fashionable these days for commentators to observe that active managers as a whole do not add value to portfolios over and above market returns. In part, this is because of the received wisdom from modern portfolio theory that all the traders of stocks in a market must be involved in a zero-sum game.
This reminds me of the joke about two economists walking along in the City, when one said to the other “Look! There’s a £50 note lying on the street”. The other replied: “Don’t be stupid. If it were a real £50 note, someone would have picked it up.” The point is, that even if the market were perfectly efficient, it would not mean that no-one picked up the £50 note.
In fact, for a number of reasons, equity markets are not in practice fully efficient.
Some investors clearly do make long-term returns which are better than the market. Warren Buffett, for instance, has, over several decades, produced long-term returns of 22% per year. That he has done so implies that he possesses skill, not just luck. Clearly, not every investor can boast Buffett’s record. So how do trustees identify the sort of manager that will add value in the long run and avoid those who will not? To answer this, it’s important first to identify the main imperfections, which might lead to market inefficiency.

One major distorter of markets is short-termism. Many investors act as if asset management was all about stock tips, or investing in the latest ‘hot’ sector. They are especially keen to emphasise their performance over periods of as short as a quarter or a month. This type of behaviour is what the modern portfolio theorists were really exposing as a waste of time and money. Short-term changes in stock prices are random and not consistently predictable. Yet a whole industry of traders goes on furiously buying and selling stock, not because of a rational evaluation, but because of fashion, or bogus ‘signals’ in the ‘price action’. To be fair, day trading is not utterly useless: the continual buying and selling helps determine the equilibrium price for stocks and provides liquidity. But it isn’t a reliable source of value for long-term investors.
The interesting thing about Buffett is that he doesn’t trade his portfolio. It embodies a selective ‘buy and hold’ strategy. The skill he applies delivers value over the long run and doesn’t require high-frequency trading. So perhaps one key characteristic of managers who deliver long-term value is that they focus on judgements for the long term, rather than on meeting short-term performance or volatility targets.
What I want to focus on now is a surprising source of market inefficiency. It arises from a perfectly reasonable concern with minimising risk. But it has led many investment managers away from genuinely active management and into quasi or closet-passive management. This is the use of ex ante (expected) tracking error as a control mechanism for equity portfolios.
There are two essential problems with reliance on ex ante tracking error to control the relative risk in the portfolio.
The first problem is technical – fat tails. It would be very nice if reality conformed to the symmetrical expectations of the bell curve, but life and markets just aren’t like that. I remember working in an investment bank in London in 1987, when the stock market crash began. A young quantitative analyst came to the morning meeting the next day and said: “What happened yesterday will not happen again today. That was a four standard deviation event.”
What followed, of course, was several days of similar-sized moves. The supposedly stable ranges for historical volatility had exploded. To make matters worse, historical correlations also broke down. Market makers who thought they had hedged their
long positions found that instead of dampening volatility, the hedge actually amplified it.
So the assumptions of normal distribution and stable correlation, which underlie the use of ex ante tracking error as a control tool, are very suspect. Events are surprisingly often distributed in the tails of the bell curve. What this means is that use of ex ante tracking error gives the plan sponsor a false sense of security that the portfolio is ‘risk controlled’. When risk control is needed most – during a crisis – ex ante tracking error will frequently let you down. Indeed, in general, ex ante tracking error is a poor predictor of ex post (realised) tracking error.
Now there are some very sophisticated people working in financial markets. They can be expected to respond to this weakness by using non-normal distributions for analysis, or by stress-testing their tracking error under different scenarios. However, even if they were to improve on their quantitative risk controls, there’s a more fundamental issue at stake – does managing risk relative to a benchmark control the risks that really matter?
Before considering this, it’s important first to distinguish between absolute and relative risk. Absolute risk is the chance that the investment made will go down in value. Relative risk is the chance that its performance will vary from that of the benchmark used to measure the manager’s performance. It’s inevitable and right that the plan sponsor will want to measure the manager against some objective benchmark. However, when this is a high frequency observation and when short-term variations can result in a manager being fired, the manager has a strong incentive to pay too much attention to short-term risks relative to the benchmark. Excessive attention to relative risk will in turn have adverse consequences for investment performance, not least in the form of bunching around the peer group or the benchmark.
Let’s go back to the point of holding assets in a pension fund. The objective is to more than to cover the liabilities in the long run, or to minimise the cost of meeting them. So, like any long-term saver, a pension fund has an interest in exploiting the equity risk premium. It also wants to limit the downside risk of holding volatile assets during periods of weakness. Hence its asset managers are typically asked to hit a performance target over and above a benchmark, with the constraint that the ex ante tracking error against the benchmark must remain within a defined range.
The problem is that using ex ante tracking error as a control mechanism means that the portfolio cannot be sufficiently different from the benchmark to protect the value of assets during bear markets. More important, it also means that the pension fund will probably buy into overvalued assets when the market is in bubble mode.
To illustrate this, here are two examples. The first is Japan in the early 1990s. The country then represented more than 40% of the world market cap index. If a pension fund’s global equity manager was attempting to keep within a tracking error range – say 3-5% – then a substantial part of the portfolio would have been exposed to Japan. This would have had very adverse consequences for the portfolio’s value in subsequent years.
Now think of the late 1990s and the bubble in US technology stocks. Cisco Systems rose rapidly in price and market weight. At its peak it represented 4% of the S&P500 and was briefly the world’s largest company by market cap.
At that point, the safe thing to do, from a tracking error point of view, was to hold Cisco Systems. However, this was not the safe thing to do if the objective was to preserve the portfolio’s value. By the start of 2000, Cisco was massively overvalued at more than 100 times earnings. Over the following three years it lost nearly 80% of its value. Those who held it in 2000 because of their concern to minimise tracking error did not cover the portfolio from the real risk – that the value of its holdings might go down.

So, beyond the technical issue of fat tails, there is a more profound case against the use of ex ante tracking error as a risk control. Tracking error begins to make investment decisions for the manager. Because of the tendency for markets to overshoot ‘fair’ value, these often turn out to be bad investment decisions. Addressing risk in an absolute sense, rather than relative to the benchmark, will serve the portfolio and the client better in the long run.
Now let’s return to the sources of reward to active management. Some managers beat the market over the long run because of superior information. The quality of their research gives them a competitive edge. This is perhaps most pronounced with small cap or emerging market equities, but there are examples of success among managers of large-cap stocks in the developed world as well.
Other managers have exceptionally good systems for processing the available information. In a complex world, they are adding value because of the accuracy of their models in identifying the sources of stock performance.
But it is possible also to gain from active management without superior information or systems. Making consistent judgements about stock value independently of the composition of the benchmark, and without the straitjacket of ex ante tracking error can, we believe, produce a return which rewards this form of active management. In effect, benchmark-independent judgement benefits from the market-distorting habits of others.
David Boal is regional head, UK & continental Europe, Bank of Ireland Asset Management in London