The failure of DSGE macroeconomics

24 Jan, 2014 at 10:56 | Posted in Economics | 5 Comments

As 2014 begins, it’s clear enough that any theory in which mass unemployment or (in the US case) withdrawal from the labour force can only occur in the short run is inconsistent with the evidence. Given that unions are weaker than they have been for a century or so, and that severe cuts to social welfare benefits have been imposed in most countries, the traditional rightwing explanation that labour market inflexibility [arising from minimum wage laws or unions], is the cause of unemployment, appeals only to ideologues (who are, unfortunately, plentiful) …

wrong-tool-by-jerome-awAfter the Global Financial Crisis, it became clear that the concessions made by the New Keynesians were ill-advised in both theoretical and political terms. In theoretical terms, the DSGE models developed during the spurious “Great Moderation” were entirely inconsistent with the experience of the New Depression. The problem was not just a failure of prediction: the models simply did not allow for depressions that permanently shift the economy from its previous long term growth path. In political terms, it turned out that the seeming convergence with the New Classical school was an illusion. Faced with the need to respond to the New Depression, most of the New Classical school retreated to pre-Keynesian positions based on versions of Say’s Law (supply creates its own demand) that Say himself would have rejected, and advocated austerity policies in the face of overwhelming evidence that they were not working …

Relative to DSGE, the key point is that there is no unique long-run equilibrium growth path, determined by technology and preferences, to which the economy is bound to return. In particular, the loss of productive capacity, skills and so on in the current depression is, for all practical purposes, permanent. But if there is no exogenously determined (though maybe still stochastic) growth path for the economy, economic agents (workers and firms) can’t make the kind of long-term plans required of them in standard life-cycle models. They have to rely on heuristics and rules of thumb … This is, in my view, the most important point made by post-Keynesians and ignored by Old Old Keynesians.

John Quiggin/Crooked Timber

5 Comments

  1. Professor — You seem to be on a massive campaign to delegitimize DSGE models. I wholeheartedly applaud your efforts. Like you, I find that the inability of these models to account for involuntary unemployment makes them worse than useless. And their widespread acceptance among economists makes those economists worse than useless; they are dangerous.

    • The reason they do not account for involuntary unemployment is that they implicitly assume based on ideology that the probability of an involuntarily unemployed to withdraw from life, i.e. die from not having access to medicine, energy resources, food, etc., is 1. The lucky ones will migrate or volunteer for going on a one-way trip to Mars.

      • 🙂

  2. It is unfair to characterize macro models strictly in terms of political viewpoints. Of course there is a correlation between models used and beliefs, but dismissing “right-wing” dismisses a lot of models beyond what John Quiggin addresses.

    For example, the quintessential “right-wing” explanation, the Austrian, does not require unions or minimum wage. It requires malinvestment in the lead up to the crash and inhibitions to the recovery through continued distortions of the capital market.

    This is not to say the Austrian explanation is right. We should just not limit our search of possible models so easily to Keynesian models.

  3. I would wish to distinguish between reasonable and unreasonable heuristics. Mathematical models have supported heuristics masquerading as ‘great truths’. But could there be a role for logic and mathematics in helping to select or improve heuristics? Isn’t this what Keynes did, and shouldn’t we?


Sorry, the comment form is closed at this time.

Blog at WordPress.com.
Entries and comments feeds.