Is the discipline of economics still fit for purpose? Do economists have a meaningful role to play in companies devoted to helping investors build their wealth? Economic forecasts are a staple in investment commentaries, yet economists have been increasingly getting things wrong – so wrong that, as Richard Whittle highlights, they have become something of a laughing stock. However, the flaws have always been there. Until now, though, we have been collectively mesmerised by the intelligent-sounding lingo and the seemingly inhuman ability of economists to read what is happening in complex financial markets. Economists tend to have a way of words that helps us make sense of global political currents, too. Non-expert investors have put economists on a pedestal, eager to believe them because we are looking for confirmation of our biases or short-cuts to decision-making. However, as the rise of Donald Trump and the surprise of the financial crisis of 2008 and other shocks have reminded us: economists are fallible just like the rest of us. Let’s not be too hard on them for occasionally getting the wrong end of the stick. Instead, we should remind ourselves that economics is more art than science – and that economists are not fortune-tellers. No-one can predict with great accuracy what tomorrow will look like. – Jackie Cameron
By Richard Whittle*
At its heart, the so-called “crisis” in economics is simply a result of the flaws in our species. Simply put, the variables used by economists are inherently problematic as they are attempts to model human decision making. What this should tell us is that the value in economics is not in some magical ability to divine the future.
The trouble is, forecasting models are very attractive. They help investors assess risk, help central banks decide policy and allow politicians to justify ideological flights of fancy. And it’s in this last group that the caveats, warnings and limitations of these models are so often ignored.
Let’s look at those inherent problems in human decision making a little more closely. Let us say that an economy has grown by 3% every year for the past 20 years. A forecasting model, based on historical growth will rightly forecast future growth with a high probability of about 3%. Does this mean this is guaranteed? Of course not.
The model does not take into account that GDP is a product of human decision making. Just because we have constantly performed one action over and over again, are we destined to in perpetuity? GDP is an observation of our confidence, our tastes, our decisions to spend or save and so much more. These factors are in turn affected by countless others, the isolation of which is impossible due to their constantly changing nature.
I have something at stake in the debate. Right now, I am in the process of devising a model of the UK economy which incorporates relevant psychological drivers of decision making. In my model, I use proxies for people’s confidence and try to build in measures for this confidence to disappear very quickly – for instance how a major news story about people falling into negative equity might dent house buying confidence.
I also try to build in herd-like behaviour. And there are a host of other psychological effects to consider too. Did you know that people normally browse a holiday on a mobile phone or tablet but prefer to book it on a home computer? This is the same for most large purchases and the cumulative effects on the economy can be considerable.
It’s those pesky humans again, making everything complicated. You are less likely to impulse buy online if you have to enter your card details. Again this has a noticeable effect on the economy as new financial technology – such as contactless payments – makes unnecessary purchases more likely.
Doctoring the models
Now, this model performs very well in tests designed to assess the robustness of forecasting. It identifies a general slowdown in confidence throughout 2017 culminating in a large downturn in general economic indicators at the end of the year. It seems plausible. However, I show this model to my students as an example of overconfidence in forecasting.
You see, it might produce statistically robust forecasts, but it cannot present the full picture and so its use as a means of prediction is limited. The unknown conditions of Brexit may play a part, the weather may do as well. Without knowing these outcomes, and people’s response to them, my model is incomplete.
So, do I throw up my hands and curse economics as a futile endeavour? Just what is the value of the academic discipline to which I have (so far) devoted my career? Well, the analogy of a medical doctor is useful here.
Designing this model gives a better understanding of the economy even if it can’t guide it down a path of unblemished progress. In the same way, a doctor cannot definitely prevent illness, but can offer advice on prevention and hopefully offer a cure if you do get ill. This is the same for the work economists do.
Economists can offer advice on preventing crises or slowdowns but cannot definitively prevent them from happening. Economists can also offer robust advice on restoring growth, although when the advice is that the economy has grown too fast and should slow, it is often not welcomed by policy makers. This advice is built on a strong evidence base, however just like a doctor prescribing a cure, it is foolhardy to say exactly when the cure will definitely work, or if it will adapt to changing conditions.
Perhaps the hard part is getting people to acknowledge these realities. There remains a prevailing view that an economics model makes a definitive forecast for the future – some economists themselves are guilty of maintaining an ideological belief in a method regardless of empirical observation. In fact, economics models simply suggest a version of the future, and incorporate the likelihood of that future occurring.
My model meets a strenuous robustness check, but I still query the use of it as a forecasting tool. I can see, however, why these forecasts are extremely attractive to policy makers.
There is comfort in statistics, and our processing of probabilities is flawed. Someone making a bet with an 80% likelihood of success will be disproportionately disappointed if they lose. This is because we tend to overweight such high probabilities as a certainty and already expect the winnings before the outcome, whereas in reality there is a one in five chance of a loss. The opposite is true too. The very small chance of all the factors responsible for an economic crisis converging together at a particular time reassures people that it will never happen.
Policy makers are as susceptible to this as anyone, and should appreciate that the true value of the economist lies not in mystical fortune telling, but in achieving a better understanding of the nature of the economies in which we live and work.