Tuesday 19 August 2014

Wren-Lewis and Syll on Prediction.


Prof. Simon Wren-Lewis (economics professor at Oxford University, and a fellow of Merton College) has long been posting on the topic of prediction. In a recent post he writes:
"Macroeconomic forecasts produced with macroeconomic models tend to be little better than intelligent guesswork. That is not an opinion - it is a fact. (…) In other words, model based forecasts are predictably bad.
"The sad news is that this situation has not changed since I was involved in forecasting around 30 years ago."
His conclusion is that, even tough unreliable, macroeconomic forecast/prediction is "probably no worse than intelligent guesses". Although he doesn't specify, he is presumably speaking of predictions based on New Keynesian/Neoclassical models.

In reply, Prof. Lars Pålsson Syll (professor of civics at Malmö University) wrote about the impossibility inherent in forecasting/prediction:
"The future is inherently unknowable - and using statistics, econometrics, decision theory or game theory, does not in the least overcome this ontological fact. The economic future is not something that we normally can predict in advance. Better then to accept that as a rule 'we simply do not know.'
"So, to say that this counterproductive forecasting activity is harmless, simply isn't true."
Syll goes further than Wren-Lewis: he seems to make a case for the impossibility of prediction/forecast in economics in general, and, unlike Wren-Lewis, accuses forecasting/prediction of being harmful.

Regardless of their philosophical differences, or whether New Keynesian/Neoclassical models are to be blamed, both men seem to agree on two things: (1) prediction is essentially out of the question. What's more, if I am not mistaken, both (2) tend to favour government intervention in the economy.

----------

While I disagree with them on what refers to prediction/forecasting, I am not taking sides in their debate, neither am I in their league and I won't pretend otherwise.

I can't, however, avoid this question: if prediction is at least very unreliable (if not outright impossible and harmful, as Syll seems to claim) -- i.e. point (1) above -- how can an active government  -- i.e. point (2) -- predict the effects of its policies?

This is not a matter of simply making a general statement like "plan A is the way to go", and leave things at that, for if the government faces mutually exclusive courses of action (say, plan A, B, and C), it must decide which is the most beneficial; for that, it seems, it would need some kind of quantitative estimate: say, how many jobs it predicts each alternative would create.

If econometric methods are not reliable, what kind of methods would be used instead?

In the absence of good answers to those questions (and it would be incumbent upon both men to provide them), it seems to me the ultimate implication of their reasoning is that government intervention could be as unreliable and potentially as harmful as the predictions on which it is based on. To me, this seems a pretty good argument for a hands-off government: laissez faire.

What's more, although they focus on macroeconomic policy, I see no reason why their argument should be limited to that. Urban planning, for instance, uses pretty much the same kind of data and forecasting methods: should it be abandoned? How do public transport authorities decide how many lanes a new road should have? How many beds should a new hospital have?

2 comments:

  1. "In the absence of good answers to those questions (and it would be incumbent upon both men to provide them), it seems to me the ultimate implication of their reasoning is that government intervention could be as unreliable and potentially as harmful as the predictions on which it is based on. To me, this seems a pretty good argument for a hands-off government: laissez faire."

    Let's be careful here. There was no such thing as "macroeconomics" before Keynes invented it in 1936 with his General Theory. And econometric models came still later. Therefore, the question is not whether governments should act, but whether they should look to econometric models to determine whether/how to act, given that governments managed to do so for millenia without the "aid" of such "prediction." It seems to me that the social contract that exists at least implicitly between most Western nations and their citizens requires governments to act in a manner that promotes the general welfare, Doing nothing often is not consistent with that mandate, and doing nothing is exactly what the people do not believe in "the general welfare" are actively promoting.

    Pilkington had his own discussion on econometrics at NC today:

    http://www.nakedcapitalism.com/2014/08/philip-pilkington-econometricians-financial-markets-and-uncertainty-an-anthropological-view.html

    On the limitations of the point probability theory that undergirds the utilitarian econometrics of modern economics, you may want to check out some papers by Michael Emmet Brady: http://papers.ssrn.com/sol3/cf_dev/AbsByAuth.cfm?per_id=1033456

    ReplyDelete
  2. I am afraid I don't understand. What does Keynes have to do with what Wren-Lewis and Syll said?

    ReplyDelete