Forget the forecasts

Richard Exell thinks a recession is likely. Personally, I couldn’t care less.  Economic forecasts are both futile and irrelevant.
They’re futile because we simply don’t have the intellectual resources to predict (pdf) GDP to within the (narrow) margin of error that makes a difference between a “good” and bad “forecast”. This is not because of the shortcomings of neoclassical economics, many as they are. It’s because human affairs generally aren’t predictable. Political scientists can’t forecast (pdf) revolutions, criminologists can’t forecast riots, marketers can’t forecast which products (pdf) will sell and which not. And so on.  As Ambrose Bierce put it 100 years ago in his Devil’s Dictionary:

Predict, v.t. To relate an event that has not occurred, is not occurring, and will not occur.

A big reason for this is that social behaviour is an emergent process; quite simple individual behaviour generates complex structures. In economics there are (at least) three types of problem here:
- What’ll happen to technical progress? If there are popular new products or processes, we can get a boom in consumer spending or investment as people spend on these. But we can’t predict the pace or success of innovation.
- (How) irrational will people be? One issue now is whether bosses’ overconfidence will raise capital spending by more than spare capacity, the apparent dearth of investment opportunities and uncertainty depresses it. This is not precisely quantifiable.
- Will individuals’ behaviour be correlated or not? If lots of people buy at the same time, we get booms and bubbles. If they try and sell at the same time, we get crashes. But predicting such correlations is impossible.
You might object here that some people did foresee the crisis. I’m not sure. For one thing, if enough people throw darts, some will hit a treble 20 without having any skill. Those who called the crisis were, at best, folk who reasonably pointed out risks and, at worst, charlatans who got lucky without great insight: as far as I can tell (and please feel free to direct me to evidence otherwise), a lot of these did little more than go “Debt. Woooo.”
Maybe they’ll come a time when we get sufficient knowledge (pdf) of complexity (pdf) to make reliable forecasts. Maybe not. Until, then, though, forecasting is just a way of risking making fools of ourselves.
And, what’s more, for no very good purpose.
For almost all policy purposes, we don’t need to know next year’s GDP. What’s the right tax-benefit system, or planning regime, or bank regulation? A GDP forecast is irrelevant to these.
Even orthodox macroeconomic policy doesn’t really need forecasts. The conventional Taylor rule doesn’t use them. And the debate about whether to loosen fiscal policy or not is about the risks of a debt crisis versus the desire to reduce unemployment, rather than hinging on any forecast.
I suspect that the desire for macro forecasts is a historic legacy. When governments thought they could tweak growth by monetary and fiscal policy means, it needed them. But we are at least 30 years past that stage now. And when firms were big, inflexible behemoths, they needed to forecast next year’s demand. Fordist firms needed forecasts. But Zaraist ones don’t. It’s time we moved into the Zaraist age.

Comments

[Luis Enrique]

up to a point, surely Chris? If we thought European economies were about to boom, or continue to flounder, that makes a difference to important things like how we regard the state of public finances and bank balance sheets, which is a forward-looking matter. Another way of putting the same point is to say that expectations matter, in all manner of ways, and expectations are a forecast. The fact expectations are frequently wrong does not negate their importance. Haven't you yourself said that the right way to think about forecasts is as a probability distribution - we cannot avoid having subjective probability distributions. You don't think UK GDP is going to halve or double next year, for example.

[dilberto]

Economics is a science based on observed human behaviour like philosophy, what is unique about economics and philosophy is that they have the potential through their insights to change that which they observe (non-linear feedback). Knowledge of how people behave under particular economic conditions can be advantageous to those who possess such knowledge, so when that knowledge becomes widely known the behaviour of people is likely to change in order to take advantage of that knowledge making the insights on which that knowledge is based no longer valid.

That is why economic forecasting can never be an exact and useful science.

[chris]

I think we assess debt sustainability on the basis of trend long-run growth, not next year's - and we can more sure of that.
Yes, I do think in terms of probability distributions - but these are very different from the "recession next year" type forecasts.

[Andrew]

Are you then saying that forecasts should be a little more nuanced in terms of detailing probability distributions rather than being simplifications of same?

Is there any evidence that for example the BOE's "fan graph" forecasts are any more accurate or useful than a single figure summary? (the "central" case).

Aren't decisions often more or less binary, and a forecast merely a way of summarising an analysis of perceived probability distributions into a recommendation for action?

[Andrew]

You seem to be objecting to the simplification inherent in analysis, and therefore analysis itself.

Even operating on the basis of a non-forecasting rule relies on an implicit forecast that conditions will maintain such that the rule is useful.

[gastro george]

Interesting, because Osborne's economic policy is based on the "fact" that the people can forecast the "fact" that his economic policies will lead to lower taxes in the future so can confidently go out and spend more money.

[ortega]

Seems very hard to convince people, many economists included, that the task of economists is not to predict the future. As soon as one of them tells you what will happen, better stop reading.

[Orin T.]

Are you saying that Macroeconomics is like creationism; i.e. that it explains everything and predicts nothing?