For years, researchers in behavioural finance have pointed out that our investment decisions are clouded by countless cognitive biases which cause us to fall short of fully rational behaviour. However, some new research suggests that such biases are not always very expensive.
Olga Bourachnikova and Marie Pfiffelmann at Strasbourg Business School compared two sets of portfolios. One set are those obtained from the portfolio optimisation first proposed by Harry Markowitz 60 years ago, in which investors use assets' volatilities and covariances to minimise risk for an expected level of return. These are the benchmarks of rational portfolios. The other set are those derived from the behavioural portfolio theory proposed by Hersh Shefrin and Meir Statman at Santa Clara University. These portfolios disregard covariances and are intended to protect investors from worst-case scenarios while offering a chance of a big gain; they tend to have the same risk-return properties as combinations of bonds and lottery tickets (albeit ones with better odds than state lotteries).
In theory, the latter portfolios should be inefficient, in the sense of being more risky than Markowitz's portfolios for a given level of expected return. But Bourachnikova and Pfiffelmann found that this was not always so. Yes, the behavioural portfolios tended to have high risk, but they were reasonably efficient.
Perhaps, then, irrationality works better in practice than in theory. Rough and ready portfolios don't necessarily lose us money. The findings of researchers in behavioural finance might therefore be like lots of things in economics - true, but not very much so.
Their evidence is not unique. Lorenzo Garlappi at the University of Texas at Austin and colleagues have found that simple portfolios that split money evenly across assets work better than more complicated asset allocation strategies.
There's a reason for this. Rational optimising strategies require us to know assets' volatilities and covariances. But we can't know them for sure because the distribution of past returns isn't necessarily a good guide to future risks. Entirely rational investors - who probably exist only in computer simulations - therefore don't have the tools needed to make optimal decisions. And if you're travelling in the dark, a bridge with a missing span is worse than just useless. In the second-best world of unavoidable ignorance, strategies that are sub-optimal in theory need not be so in practice.
What's going on here is what the Bank of England's Andy Haldane recently likened to a dog catching a frisbee. It doesn't need to solve equations to do so, but just keeps its eye on the target. Simplicity, then, works better than complicated 'rational' thinking. As Gerd Gigerenzer of the Max Planck Institute has shown, simple rules of thumb based on limited information - what he calls "fast and frugal heuristics" often do as well or better than sophisticated thought. For example, he says, the question: "which is the larger city, Hamburg or Cologne?" is often answered more correctly by non-Germans than by Germans. This is because non-Germans ask themselves "which has a team in the Bundesliga?" and this triggers the correct answer - Hamburg - whereas Germans try to use more information some of which misleads them.
All this, though, poses the question. If non-rational investing based on ignoring information can be as good as rational investing, why is it that Brad Barber and Terrance Odean, two US economists, have found that most retail investors underperform the market?
It's because some of the cognitive biases identified by behavioural finance economists are not efficient heuristics but do instead cost us money.
One of these is the overconfidence that we can identify good stocks. This can cause us to trade too much and so incur unnecessary costs.
Another is that investing equal amounts across assets works only if you are careful about which assets you select from. Shlomo Benartzi and Richard Thaler, two US-based economists, have found that pension fund investors who were offered a choice of lots of bond funds and few equity funds invested lots in bonds, whereas those offered lots of equity funds and few bond funds held lots of equities. Naïve diversification, they concluded, "does not assure sensible or coherent decision-making".
There's also a danger with behavioural portfolio theory. 'Lottery stocks' - those with big upside potential - tend on average to underperform other shares, as do stocks on the verge of bankruptcy which would do well if they could be refinanced. This means that investors who chase very big gains can lose a lot. Aim stocks, which tend to be more speculative and lottery-like, have under-performed the main market for years. Trying to get rich quick in stocks markets is therefore dangerous.
Todd Feldman at San Francisco State University points to another expensive bias - the recency effect, the belief that recent market conditions will persist. If you're a stock-picker, this bias can actually make you money, because it encourages you to buy momentum stocks, those that have recently done well, and momentum has often paid well in the past. But if you're an asset allocator, it can be dangerous because it can lead you to buy at the top of markets and sell at the bottom.
The message here is simple. Some sorts of mistakes are expensive, and some aren't. The trick in investing is not to avoid mistakes - that's impossible - but rather to minimise the expensive ones.
MORE FROM CHRIS DILLOW...
Chris blogs at http://stumblingandmumbling.typepad.com