Marc Lavoie’s New Book

Article of the Fortnight from Fixing the Economists

by Philip Pilkington, Fixing the Economists

Marc Lavoie’s new book on the foundations of Post-Keynesian economics is out entitled Post-Keynesian Economics: New Foundations. I learned a lot from the last version of this book and Marc has told me that he’s been working hard to update the new one in light of more recent developments in both Post-Keynesian and mainstream economics.

The first chapter is available on the publisher’s website. It looks pretty good too. It contains discussions of the SMD (Sonnenschein–Mantel–Debreu) theorem. He also discusses the fact that tests for the aggregate production function have recently been exposed as being based on regressing national accounting identities on the national accounts data. If that sounds like gobble-dee-gook to you it basically means that much of the empirical results of mainstream economics since the 1960s is hokum — i.e. it’s a big f-ing deal! Lavoie writes:

Neoclassical economists are claiming to measure something, but are really measuring something entirely different. Their theories, such as the necessary negative relationship between real wages and employment, seem to be supported by the data, whereas the negative relationship arises straight from the identities of the national accounts, with no behavioural implication for the effect of higher real wages on employment. I have discussed these issues with a few of my neoclassical colleagues. The most genuine answers have been that without these elasticity estimates they could no longer say anything. But they would rather continue making policy proposals based on false information than make no propositions at all. In other words, they would rather be precisely wrong than approximately right. (p. 36)

Lavoie also discusses the empirical claims of the mainstream. Through a survey of a variety of literature, especially the literature of so-called ‘meta-regressions’ which seeks to find data manipulation, conscious or unconscious. The findings are damning and support the claims often made on this blog: econometric regressions run on models is largely meaningless the vast majority of the time.

Most post-Keynesians demonstrate scepticism when it comes to empirical and econometric research. Still, one cannot but be impressed by the huge quantity of empirical work that seems to provide support for orthodox theory. This section has shown that this cynicism with regard to orthodox econometric research is largely justified, as many of the studies that appear to verify or confirm orthodox theory are just artefacts. What is an ‘artefact’? The most common definition, relevant to science, says that an ‘artefact’, or ‘artifact’, is a spurious finding caused by faulty procedures. Meta-regression analysis has certainly demonstrated that many of the empirical proofs of orthodox theory were phoney and arising from defective procedures. The word ‘artefact’ is also used in the fantasy and sorcery literature. There, an ‘artefact’ is a magical tool with great power, like a magic wand. This definition seems to be particularly relevant to neoclassical production functions since all the predictions that can be drawn from a model of perfect competition cannot be refuted, even when we know that the required conditions do not hold. (p. 70)

Lavoie stops short of saying that the techniques should be given up altogether. But it seems to me difficult to draw any other conclusion. In High Finance these techniques, applied to macroeconomic models, are known to be GIGO (Garbage-In, Garbage-Out) which is why they are not taken seriously when there is money on the table. Why is it that policymakers and academics continue to insist upon using them? Perhaps because there is a certain amount of magical thinking going on in the highest echelons of the profession and the practitioners are not being honest about what they can and cannot say with relative surety?

This book will probably turn out to be a key reference text for the growing student movement against current economic teaching.