It’s that time of year when every self-respecting student of markets blows the dust off their crystal ball, and sets about attempting to forecast the direction of markets for the next year. Some pore over charts, others examine the history of presidential election years, some analyse market data and others, surprisingly large numbers of others, consult star charts.
Financial forecasters suffer from the same problems as earthquake forecasters – they don’t understand the underlying principles of what they’re trying to analyse. While the latter generally recognise this and have the sense to put in place the appropriate caveats the former mainly seem to be blind to the possibility of error. Of course, the fact that over-confident earthquake analysts may lose their careers and reputations while their equivalent in finance will merely have to wait another year to assay an attempt at guru status may have something to do with this.
There’s a long history of earthquake prediction, which has mainly served to reveal that we’re totally useless at predicting earthquakes. The basic theory behind earthquakes – plate tectonics – wasn’t accepted until the 1960’s, although it had been proposed as far back as 1912. As with so many scientific theories it took an overwhelming amount of evidence – fossils, paleomagnetism, ocean floor bathymetry and so on – before it became generally accepted.
Earthquakes are generally concentrated in Wadati-Benioff zones, areas around the edges of tectonic plates, where one plate is moving underneath another one. Friction causes the plates to stick and earthquakes occur when the friction is overcome and the plates slip. So we understand why earthquakes occur, but that doesn’t mean we can predict them. In fact we have never predicted a major earthquake because the variables are just too darn complex to model, at least given our current levels of knowledge and computing power. As D.A Freedman and P.B. Stark put it in What is the Chance of an Earthquake?:
“Forecasts are based on a complicated mixture of geological maps, rules of thumb, expert opinion, physical models, stochastic models, numerical simulations, as well as geodetic, seismic, and paleoseismic data”.
Of course, the fact you can’t actually predict an earthquake and that we know why we can’t do so doesn’t stop people from attempting to. After all, the kudos that you’d get from successfully guessing where the next Big One is going to hit would be phenomenal, while the downside for most people would be minimal. Scientists, though, have good reason for not waving around their crystal balls because they risk their careers by doing so. This is the upside of scientific conservatism – it may mean that it takes 50 years to accept plate tectonics, but when it does so we can be pretty sure that it’s correct.
Financial forecasters, on the other hand, lose absolutely nothing by predicting anything. The evidence, such as it is, suggests that most forecasts are complete garbage and biased to the upside. Given that markets tend to rise about twice as much as they fall, likely an outcome of basic human overconfidence, then if you don’t have a clue what you’re doing it’s a pretty decent bet to guess markets will rise.
Even better, as no one has a clue then nearly everyone will always guess that markets are going up so if you’re wrong then you’re wrong in a bunch, and your failed forecasts are most likely going to be forgotten about. Certainly you’re not going to find yourself on the end of a bout of peer-reviewed hyper-criticism by a bunch of your fellow forecasters pointing out your methodological ineptitude and ruining your career once and for all. Indeed, if you go against the general trend and get it wrong you may well get noticed, and that would be extremely bad, because the last thing anyone trading on luck wants to do is to draw attention to the fact.
Financial forecasting is simply testimony to the relentless stupidity of human beings, who seem to be endlessly gullible. Without this gullibility the majority of the securities industry would collapse, because there’s virtually no evidence of any skill in forecasting from anyone. If earthquake predictors went around demonstrating the same cavalier approach to peoples’ lives and property they’d quite rapidly get their asses sued off. In fact, a manslaughter lawsuit is being prosecuted in Italy at the moment
, based on a failure to predict an earthquake, which may have profound implications for research and prediction studies around the world.
Of course, economists rely on more than the gut feel and finger in the air hopelessness of many financial forecasters, and have a whole range of models available to them. Unfortunately these have turned out to be less like the robust descriptions of physics, where it’s difficult to distinguish the model from the reality, and more like earthquake prediction models.
This corresponds with the analysis of Goyal and Welch
which suggests that no standard economic forecasting model had any useful predictive capacity – which included (deep breath): dividends, earnings, stock variance, cross-sectional premia, book value, net issuing activity, t-bills, long-term yield, corporate bond returns, corporate bond yields, inflation, investment to capital ratio, a combination of all of the aforementioned variables and consumption, wealth, income ratio:
“We believe that the evidence suggests none of the academic models we reexamined warrants a strong investment endorsement. Most models not only cannot beat the unconditional benchmark, but also outright underperform it.”
Unfortunately economists, unlike scientists, don’t get demonstrably disowned if they come up with models that turn out to have no genuine underpinning, although this may be because they don’t actually realise the problem since they spend their whole lives working with theories that don’t work. A study by Paul Söderlind
has shown that a panel of economists demonstrated no ability to predict the movements of the stockmarkets but nevertheless remained convinced that they had some skill in what is essentially an exercise in coin-tossing. In fact these findings were predictable, which goes to show that economists are just as fallible and subject to behavioral bias as the rest of us.
As the Freedman and Stark paper above points out, spending our time looking at earthquake predictions is a dubious activity:
“Another large earthquake in the San Francisco Bay Area is inevitable, and imminent in geologic time. Probabilities are a distraction. Instead of making forecasts, the USGS could help to improve building codes and to plan the government’s response to the next large earthquake. Bay Area residents should take reasonable precautions, including bracing and bolting their homes as well as securing water heaters, bookcases, and other heavy objects. They should keep first aid supplies, water, and food on hand. They should largely ignore the USGS probability forecast.”
What goes for earthquake forecasting does for its economic equivalent, with the added caveat that anyone making scientifically unsound statements around earthquakes is likely to find themselves looking for a new career. Investors should spend their time taking reasonable precautions rather than reading investment forecasts.