Learning from your mistakes
Cliff Asness is the managing and founding principal of AQR Capital Management, based in Greenwich, Connecticut. They are one of the managers recently appointed by BT in Australia and New Zealand to replace Putnam as a sub-manager for core global investments, including global equities. Their style is a value and momentum blend, with the emphasis on value.
The AQR process is described as ‘a differentiated, best in class approach', providing a lot of active return in a very efficient way. They aim to exploit risk in the widest range of options. AQR has developed a quantitative approach designed to exploit stock/industry, country and currency risk more effectively. This is achieved by managing these three key components in independent strategies combined within one overall portfolio. In this way, the thinking goes, investors are not exposed to elements of unwanted or unmanaged currency or country risk. BT's CIO Stewart Brentnall says of AQR: "We have rarely seen a research methodology that is so pervasive through their entire range of products."
Cliff Asness is a contrarian at heart, a self-confessed heretic, who has paid a heavy price in the past for going against the grain. As a value investor at the turn of the decade, he stuck by his belief that stock markets were grossly over-valued. He launched the company's first fund during the bull run of the late 1990s and notes that, "our ‘only' bad period was the first 18 months".
He even wrote a technical paper in 1999 for the Financial Analysts' Journal attempting to burst the tech bubble. He called it ‘bubble logic', but it was not published. "To do something contrarian and then try to write articles about; it came across as sarcastic and quite bitter," he admits.
"Bubble logic was my disparaging term for the tortured stories and sometimes outright lies necessary to justify prices in 1999-2000." Asness describes the market bubble as a painful time for him professionally, having to defend his ideas against the frenzy of the market and then to try to recover from a 41% fall in his NAV over the first 19 months of a new product. "So now that the bubble has burst, the most important question is: what have we learned from this devastating experience?"
His attempt to answer this question has finally been published in the Financial Analysts' Journal, this time called ‘rubble logic': "It's still sarcastic, but happier."
His thesis centres on what investors ought to expect from stock investments over the long term; as opposed to what the sell-side tells them to expect. He questions the methods used to illustrate likely future returns. "On the face of it, using historical stock returns as a forecast for the future seems unimpeachable. It is certainly common, and the method is simple and clear," he says.
But although seemingly reasonable, this method can produce some strange results. "For instance, they would have found an average annualised compound real return on the S&P 500 Index from 1946 through to the end of 1999 of 8.4% over inflation. That's a nice return to expect in the future! But over the next few years - from the end of 1999 to the end of 2004 - the average annualised real return on the S&P 500 turned out to be minus 5%.
"Although intentionally extreme, it makes the important point that forecasts derived from past averages, even if long-term averages, are often actually backwards.
"Consider the old adage about equity returns - that stocks never lose if held for the long term. Well, if a decade is your idea of the long term, then this adage is true, only when prices start out in the lower valuations. When prices start out more expensive, there are decades when stocks not only lose to inflation but lose big."
Investors expecting stocks to return 7-10% over the long term are deluded, he believes. "The problem with expectations of 7-10%, or more, is that they are based on assumptions that include extreme market conditions. Long-term averages are fine for forecasting, except near the top and bottom of bubbles."
To find a true market average, you need to strip out of the extreme highs and lows."Equities have, in fact, never lost to inflation over any 20-year historical period. But as Asness says: "That fact is more a statement about their past average returns than about their risk. If the equity risk premium is much smaller in the future, the probability of equities losing over a 20-year period is much larger than if the premium were higher.
"Secondly, investors need to use the lower equity premium when making plans. So, pension funds with their assumed market returns and individuals with their ‘when can I retire' spreadsheets have to assume equities will rise at about a nominal 6-7% a year (and a whole portfolio, including bonds, with costs, taxes, and inflation, will rise considerably less). It is inconsistent to believe that stock prices do not have to fall now (because people have learned to accept a lower equity risk premium) and still to use a 10+% a year return assumption."
It is not all bad news, though, and Asness says that over time, a value approach wins out. But he warns against complacency at this time. "If we look at current valuations and ignore the bubble (see chart above) the long trend line is near the high. An immediate logical conclusion might be to expect a zero real return on stocks over the next decade."
If stock prices are already high, the question is whether this means we face permanently low returns or near-term very bad returns followed by more historically normal results. That is one of the most crucial puzzles currently facing capital markets, says Asness. "Little can change the high probability that high prices lead to lower expected returns in the future, but the timing is very much up in the air."
Asness also suggests investors should disbelieve the suggestion that earnings can grow at a steady 10% a year. Once again, forecasting earning per share (EPS) growth using long-term averages is misleading at best. According to Asness: "During the bubble, the aggregate five-year earnings growth forecast from Wall Street analysts hit 15% a year for the S&P 500 and 30% a year for the NASDAQ 100 Index. Those figures exemplify the forecasting of growth above any possible reality. And it occurs often, albeit usually in a milder form."
History tells another story. The past 75 years, a period marked by a great performance in the US economy and stock markets, demonstrate that 10% is way too high. Realised EPS growth for S&P 500 stocks has been less than 2% above inflation for this period.
"We need to translate real EPS growth into nominal growth, which is what most people use in practice. If we take a forecast of 2% long-term real EPS growth and add it to an assumed steady 2-3% inflation from now on, we find that history favours a 4-5% long-term nominal EPS growth for the future. I bet Wall Street does not agree."
So what sort of return does Asness suggest investors should expect going forward 10 years? He says estimates based on a Gordon-type dividend discount model (which effectively assumes steady-state valuation) lead to forecasts of about 4% real returns (6-7% nominal returns if inflation stays in the 2-3% range).
In the short term, "stock returns are basically unforecastable," says Asness. "And in the long run, only valuation works, which is just bad TV. Imagine a Wall Street guru saying, ‘Stocks will return an expected 6-7% over the long haul, but frankly, we, and everyone else, have no idea what will happen in the next few weeks or months or even years'. Now, think of it repeated every day. Not only is it not bullish; what is worse, it is boring!"
Think about that next time your are watching CNBC.