Regular readers of FT Alphaville know our fascination with — and deep skepticism of — macroeconomic forecasting.
To keep the meme going, we pass along three recent entries on the topic from around the econoblogosphere.
First up is Ragu Rajan, who tries to explain why economists missed the crisis:
I would argue that three factors largely explain our collective failure: specialization, the difficulty of forecasting, and the disengagement of much of the profession from the real world.
Like medicine, economics has become highly compartmentalized – macroeconomists typically do not pay attention to what financial economists or real-estate economists study, and vice versa. Yet, in order to see the crisis coming, you had to know something about each of these areas, just like it takes a good general practitioner to recognize an exotic disease. Because the profession rewards only careful, well-supported, but necessarily narrow analysis, few economists try to span sub-fields. …
What is hardest to forecast, though, are turning points – when the old relationships break down. While there may be some factors that signal turning points – a run-up in short-term leverage and asset prices, for example, often presages a bust – they are not infallible predictors of trouble to come. …
And it may well be that academic economists have little to say about short-term economic movements, so that forecasting, with all its errors, is best left to professional forecasters.
For the reasons we laid out here, we doubt that forecasting should be “left to” anyone. See Noahpinion for further critiques of Rajan’s view.
Moving on, Mark Thoma has a more general point, and this one we do agree with:
There’s a good reason why I try to avoid forecasts. In the past, whenever I’ve tried to predict the path the economy would take, I’ve found myself reading subsequent data releases in a way that supports the forecast. I think that once you make a forecast, it affects your objectivity, and I think that applies generally, not just to me.
We mentioned previously that it would be great if people could take the useful parts of forecasts, assessing their underlying data and logical coherence, while paying little attention to the actual prediction. But the mere existence of the forecast has psychological effects that distort our thinking.
Finally, we liked Stephen Gordon’s admonition not to confuse policymaking models that include counterfactuals with forecasts:
In order to do a proper evaluation of the policy, you need the proper counterfactual: what would have happened without the policy?
If Star Trek were still on the air, this would be a simple enough problem to solve: just wait for an episode in which the crew stumbles across a parallel universe in which the policy of interest hadn’t occurred, and compare it with the one we happen to be in. But since it’s not, the policy analyst is obliged to construct the relevant parallel universe on her own. …
This distinction between policy evaluation and forecasting is crucial: a model that makes bad forecasts can still be useful for policy analysis.
Gordon goes on to explain that a model’s conclusions aren’t automatically consigned to irrelevance simply because of errors in predictions.
Consider, for instance, the famous Romer-Bernstein graph of what unemployment would look like with and without the stimulus.
The fundamental point of the graph — that unemployment will be higher without the stimulus — isn’t necessarily disproved by the fact that unemployment ended up going higher than originally forecast even with the stimulus.
Had the graph properly accounted for the severity of the economy’s problems at the time, unemployment for both scenarios simply would have been higher. (And no, we’re not revisiting the whole did-the-stimulus-work-or-not debate — just making a point here.)
Of course, the graph looks silly in hindsight, and not everybody is aware of the point Gordon is making.
So perhaps another lesson is that if policymakers are going to rely on such models, they should be prepared to explain later why there is a big divergence from what the model predicted.
Related links:
Economic forecasting delusions – FT Alphaville
Nouriel Roubini and the 40% rule – FT Alphaville
Too bullish, too bearish, whatever… – FT Alphaville
Can we predict a financial crisis? – FT Alphaville