A Debate on Mainstream Economics: A Gadget is a Dangerous Thing

Anthony Bonenby Worldly Philosopher Anthony Bonen

Recently, Paul Krugman (here, here and here) has taken a keen interest in a heterodox economist’s critique of mainstream economics (and not for the first time). In reponse, Tom Palley has issued solid rejoinders to Krugman and Simon Wren-Lewis. However, I fear the brilliant and (rightfully) esteemed Nobel laureate has missed the key point of heterodox’s frustration with marginal productivity theory. So, I follow his call to “continue to debate how we do economics.”

A key issue in distribution debates is the fair remuneration of capital and labor. Marginal productivity theory says, very basically, that the real wage and the (real) rate of profit are equal to the marginal increase in production added by the last unit of labor and capital, respectively.

Many economists, Krugman points out, do not believe this to be a cardinal truth of capitalist economies. Rather, he says [my emphasis]: 

"there are plenty of economists who are willing to use marginal-product models (as gadgets, not as fundamental truth) who don’t at all accept the sanctity of the market distribution of income."

That word – gadget – really stunned me. It sounds benign, innocuous even. Yet marginal product models are no mere “gadget”. They are the entrée for modern macroeconomics.


Top economic journals will simply not publish a paper in which profits are determined by, for example, Kaleckian-style mark-ups (and don’t even ask about the extraction of surplus from labor). Choosing to publish only in more ‘pluralistic’ journals will limit academic employment opportunity. The reality for many young economists is that we have no choice but to adhere to this “gadget” through years of grad, doc and post-doc work. Not long after, you can soon reach a point at which you don’t even want to hear a criticism of this gadget that you’ve worked with for so long. Accepting it would undercut years (decades, perhaps) of your work!

So why not just suck it up? Let profit equal the marginal product of capital in my models, but then acknowledge this in no way reflects reality. Two reasons:
1) it is not – as Krugman implies – an innocuous assumption, and;
2) marginal product valuation of factors of production is essentially a tautology.

First, the problem is not that I or anyone else think New Keynesians believe capitalists and workers get what they deserve (i.e., what is fair). Rather, ‘getting what they deserve’ is necessarily and precisely what marginal-product models say. There is no ambiguity about this. If a model (not the modeler!) says that capital receives its marginal product, then there is no moral basis within the analytical framework to argue capital should get more or less.

The Salt Water folks may eschew the optimal distribution implied by marginal-product models, but using them in the “Civil War” (Krugman’s term) means New Keynesians have forfeited all battles over mathematical tools to the Fresh Water side – group who does buy into the sanctity of market distribution. So, why should progressive economists use a mathematical framework that is inherently divorced from their view of the drivers of distribution? The standard answer is that it’s easier to do so (“tractable” in the jargon). It was disheartening to hear a renowned economist whom I greatly admire suggest precisely this path.

More importantly, marginal product theory is an irrefutable tautology. We will never delineate the partial derivative additions of capital and labor to output. Therefore, we will never be able to test the hypothesis that profit equals the marginal product of capital. We go on assuming (only in our models, of course) that capital gets what it adds and adds what it gets. This is the beautiful result of free and equal agents maximizing their payoffs. Problem with marginal product theory is not just that it implies market forces could bring about a fair distribution of income, but that this outcome will necessarily be defined as “fair” because labor got what it added, and capital added what it got. Point finale.

Marginal product theory, at its core, recasts distributional outcomes as a technological issue – decidedly outside the scope of economic departments. If economists are serious about discussing, modeling and testing distribution theory then we must move beyond the limited framework provided by marginal productivity. The rigorous and realistic modeling of income distribution is and will remain a great challenge for years to come, but it must be done. Perhaps I’ll try after I get tenure.

I have learned and have used marginal-product models, but not because I believe them to be useful gadgets, it is because I must. It is an ill-founded theory with strong implications for distributional ethics masquerading as an agnostic gadget. Yet, I cannot avoid it – and that’s not the fault of the Fresh Water brood.

Submit to FacebookSubmit to Twitter


#2 Anthony Bonen 2014-05-08 10:30
@Daniele - thanks for the comments!

Your first point is well taken. I definitely glossed over the common (pervasive even) use of monopolistic competition and attendant mark-ups (over marginal products, of course). I avoided this 'rabbits hole' because it would have led to a discussion of competitive equilibrium framework that is also part and parcel of the standard and necessary mainstream modelling approach. But for wont of brevity.

I disagree with your second point, however. I've not read the Foley and Michl book, but the Basu paper (which is based on it) is interesting, but it does not directly test the notion that w = MPL, r = MPK. He proxies the marginal products by a switch point relationship (built from classical theory) which is operationalized by the growth rates of labour and capital productivity precisely because MPL and MPK are not observable (which is my point here). Further, he says quite plainly his results are ONLY applicable to the distributional implications of Solow-neutral technical change (p. 601). But there's a HUGE mainstream literature focusing on skill-biased technical change, and many neoclassical models incorporate Harrod- and Hicks-neutral technical change -- all of which are not subject to Basu's test. (also, his rejected null hypothesis is a little too easy, but that's another story).

There are other approaches of course. Yakovenko (2012) of econophysics fame shows very convincingly that 97% of the distribution (wage-earners) follow a Gibbs distribution, but the top 3% (profit-earners ) follow a Pareto/Power distribution, which is very suggestive of two distinct classes whose returns are governed by very different processes. But, again, r = MPK is not being directly tested because MPK is not measureable! Despite the the ubiquitous use of normal shocks in models and errors in testing, the normal distribution is, as far as I know, not at all essential to neoclassical/Ne w Keynesian theory -- it's only used for "tractability", ya know... its just a gadget :)
#1 Daniele 2014-05-05 21:24
A friendly but critical NSSR Econ alumn response. While I would agree with the general tone of the article, I would be a little more careful in making some very strong statements. For instance, saying that markup equations don't feature in top economic journals. New Keynesian models, as well as mainstream endogenous growth models of the long run, are full of monopolisticall y competitive firms with constant price elasticity of demand. These equations give a parametric markup, just like Kalecki's. The main difference, in this respect, is that distribution is exogenous in Kalecki, while it comes from marginal products in mainstream theories (even if one removes perfect competition, as it is the case here). An alternative would be Kaldor, in which distribution is endogenous in that it adjusts to accommodate the "natural" rate of growth, while at the same time it has nothing to do with marginal products. In both frameworks, growth can be cyclical in endogenous fashion, that is even in the absence of "shocks". A second statement that appears uncalled for is the one about empirical tests. In their 1999 book, Foley and Michl actually provide a "back of the envelope" test of the Neoclassical view of distribution against the "Classical" view. An econometric extension of the test (which refutes the Neoclassical view, thus strengthening your point) appears in a 2010 Metroeconomica article by Deepankar Basu.

Add comment

Security code