This poetic idea hides two very important assumptions: that market participants act in rational self-interest; and that you can predict correlations by looking at recent data. Almost all portfolios are built on these ideas. And both are wrong, according to Joachim Klement, CIO at Wellershoff & Partners.
“If you look at markets, you look at human beings,” says Klement. “To understand markets, you have to understand humans. Typically people assume that professional investors are more sophisticated and, as a result, more rational than private or unsophisticated investors. I do not make that observation, to be perfectly honest.
“As expert investors, we are subject to exactly the same behavioural biases as any unsophisticated investor. The only advantage we have is that we can deal with them in a more professional way by having processes in place that try to avoid the worst mistakes.”
Joachim Klement is partner and CIO of Wellershoff & Partners.
In this role he works with private banks, family offices and institutional investors to develop asset allocation strategies and manage portfolios.
Before joining Wellershoff, he was head of the UBS Wealth Management Strategic Research team and, until the end of 2007, head of equity strategy. In these roles Klement was responsible for the matic research within UBS globally and the equity market and sector strategies. He also worked as an investment consultant.
He publishes frequently in academic journals and is a regular contributor to different media and newspapers.
Klement believes a lot of people in the investment industry are trying to act as if nothing changed in 2008-09. They try to go back to business, trying to run the same portfolios and strategies as before, preferring to bury their heads in the sand.
“Too many people in this industry are still stuck with theories like capital asset pricing model, like rational expectation hypothesis, efficient market hypothesis, and they still apply them despite the fact they have been completely discredited by empirical data over the past ten years.”
For the industry to move forwards, it is important people embrace new ways of analysing investors – such as behavioural finance; and new ways of managing portfolios, including ideas such as minimum variance and risk parity portfolios.
So how does he actually implement ideas such as behavioural finance in a real-life situation? Interestingly, the most important impact has not been in how he creates portfolios, but in how he talks to his clients. A typical situation for Klement would be managing a pension on behalf of a board of trustees.
“But trustees, even though they are not managing their own money, are just as sensitive to losses as anybody else,” he says. “What most people do is present a pie chart to the trustees and show the performance of the overall portfolio in the last quarter. Then what happens is a year like 2008 comes and the pension fund loses 10% to 20% of its assets and everybody freaks out.”
Instead, what Klement does is split the pension into three buckets and present them separately: one in cash or cash equivalent, one in short term investments, and one in more risky, longer-term investments (see box).
Performance measurement is another problem that has behavioural finance issues. The typical trustee will say they have a three to five-year time horizon, but if it starts to under perform over six months, then questions will be asked.
“Here, behavioural finance tells us that the more frequently you compare yourself to any form of artificial benchmark, the more loss aversion kicks in and the less likely you are to favour long-term strategies,” explains Klement. “In my opinion the best thing would be to look at the portfolio once a year, especially if you have long-term investment horizons,” he says.
On to the second big issue – portfolio construction. Most investment models are extremely intolerant of estimation errors, according to Klement.
We do panic from time to time. Usually that is the case when we see something happening in the market that is not included in our models
CIO, Wellershoff & Partners
“This super-sensitivity to assumptions about future returns is combined with the problem that I don’t believe anybody in this world can predict equity returns in the future with the precision of 1% or less. I haven’t met a single person who has this ability and, to be perfectly honest, it’s way beyond anyone’s capabilities at this point in time.”
So portfolio construction is very sensitive to predicted returns but nobody can predict returns. An insoluble problem? No – but new ways of investing have to be used. Klement does not believe classic Markowitzian mean variance optimisation works. Instead, he creates portfolios that try to take estimation errors and prediction errors systematically into account.
“There are various methodologies to do that; we use ‘resampled efficient frontier’, developed by Robert and Richard Michaud in the early ’90s.” What you do is you create an artificial future for your different asset classes – like a Monte Carlo simulation. You create an efficient frontier for that simulation and then you do that 10,000 times, so you get 10,000 efficient frontiers, one for every outcome the future might hold. You then average over these 10,000 efficient frontiers.
“We analyse historical data,” he explains. “Specifically, we look at every years since 1970. Then we gear our portfolios towards an environment of weak economic growth and interest rates that are constant or rising. Those are our two big macro calls.” With this data, Klement’s team create a potential five-year future from randomly picking out historic data. So in one there could be an equity boom, a property bubble, a commodities crash. All possible outcomes are there.
“That’s why we wanted to go back to the ’70s when you had hyperinflation, gold bubbles and market crashes.” This methodology creates a return forecast for every asset class and tends to create portfolios that are very well-diversified.
Made to suit
This has a specific and positive impact on the behavioural biases of Klement’s investors. One thing he has noticed is that professional investors have a hard time working with volatility, value at risk and other classical measures. What they really care about is drawdowns (losses) and time under water (how long it takes to get back to zero after a loss) and the diversified portfolio that his system creates tends to reduce those two metrics. This all sounds too calm to be true – is he never surprised?
“We do panic from time to time,” he confesses. “Usually when we see something happening in the market that is not included in our models. Last year, the one thing we did was put a stop-loss on an awful lot of our portfolios in August when the US debt ceiling discussion broke down and we had that sell-off.
“Political events, wars, debt crises, euro problems: that’s when we temporarily get out of our portfolios. But do we change our strategy because of that? No, we only would change our strategy if our long-term growth and interest rate strategy changed.”
Looking in the past
Klement does a lot of what he calls financial archaeology. For his models he goes back in history as far as possible. For his international portfolios, he decided to stop at 1970 mostly because of Bretton Woods. Because if you have international assets, things change dramatically once Bretton Woods is implemented.
“But if I’m in just one currency or asset class then I try to go much further back, which means going way beyond what you get in Bloomberg or Reuters. We use a lot of data that is available in the archives of central banks and we now use Global Financial Data to some extent.”
Most investors are hamstrung by their data providers. For instance, most bond indices start in the mid- ’80s. But since then all we’ve had is a time of continuously falling interest rates – a deceptively inaccurate sample. For more perspective, you have to go at least back to the ’70s, according to Klement. He gives other examples .
“If you invested in US Treasuries in 1972, you had to wait until 1982 to get your money back without inflation compensation,” he says. “Another story – which I haven’t been able to confirm, unfortunately – was that in 1954 the Canadian Government offered to exchange all war debt for one 30-year bond with a 4% coupon. People took it.“That 30-year bond was quoted at par value only on two days in 30 years: the day it was issued and the day it was paid back. Everyday in between you had massive losses because of rising interest rates.
“If you look at measures such as drawdown, VAR, time under water in the ’70s, you start to wonder why we ever considered bonds to be a safe investment for people with a short investment horizon.” The solution to falling into the same trap? Do what Klement does: doubt everything, demand proof of everything and don’t be a dreamer.