Friday, May 2, 2014

Paul Krugman Really Blows It

I still do not have a copy of Piketty's book, but watching the ongoing debates it has triggered has been quite fascinating, with the recent subtext of Paul Krugman (and a few others, notably Simon Wren-Lewis) arguing with various heterodox economists over it, particularly Jamie Galbraith and most recently Tom Palley, standing out quite noticeably.  I was frankly not all that impressed by Palley's latest bit on flim-flam, but in attempting to rebut it, Paul Krugman has just up and blown it in such a massive and embarrassing way I simply cannot resist commenting on it.

So, he comments on the "hangups of the heterodox," which I just tried to link to, but it did not work (sorry about that, but Mark Thoma will probably link to it).  Anyway, Krugman says a number of not unreasonable things, such as noting that he has noted that when an economy is in a liquidity trap, flexible prices may be destabilizing.  But then at the end, when he attempts to deliver what he obviously considers to be the ultimate coup de grace, he does it, not only blows it, but falls flat on his face in a massive error.  I quote his final paragraph:

"And what's going on here, I think, is a fairly desperate attempt to claim that the Great Recession and its unfortunate aftermath somehow prove that Joan Robinson and Nicholas Kaldor were right in the Cambridge controversies of the 1960s.  It's a huge non sequitur, even if you think they were indeed (which you shouldn't.)  But that's what seems to be happening."

OK, I do not think that is what Galbraith or Palley are arguing, although Jamie in particular is making a case that Piketty wrongly ignores the capital theory debates, and Krugman has been whonking on about how the marginal productivity theory of income distribution is a "good first pass" on at least the factor income part of it.  But no, that is not where Krugman falls down.

It is in his parenthetical aside, that "you shouldn't" think that Robinson and Kaldor (who was not one of the main participants in that debate from the Cambridge, UK side, and I say that as one who has recently been labeled a "Kaldorian") were right in the debate.  The problem is that Paul Samuelson agreed that in fact Robinson and Piero Sraffa (and Garegnani and Pasinetti) were in fact right.  The possibility of reswitching does undermine profoundly the marginal productivity theory of factor income distribution, especially for capital.  He did so in his "Summing Up" paper after the symposium on reswitching in the QJE in 1966.  His final sentence of that paper, after going through the arguments in several papers was "The foundations of economic theory are built on sand."

Krugman makes a lot of good points, but he really needs to get it together about what went down during the Cambridge capital theory controversies.  Robinson and Sraffa were right, and Paul Samuelson said so.  Period.

Barkley Rosser


Peter Dorman said...

Barkley, long ago I read a piece by Frank Hahn that I thought really nailed the matter. Basically, both sides were wrong. Cambridge MA was wrong to claim that heterogeneous physical capital could be collapsed into a single magnitude that doesn't depend on interest rates in order to generate an equilibrium interest rate. But Cambridge UK was wrong to think that Sraffa had produced a model that could test the properties of an aggregate capital measure. There is no dynamic aspect to Sraffa, and there is no meaningful conception of capital (as opposed to specific capital goods) without time. If I remember correctly, Hahn showed that Sraffa's capital was constrained to reproduce exactly its same composition from one period to the next (which is why they aren't really different periods), and this renders his model a special case.

Those who are more on top of this literature can correct me. said...


You are talking about Sraffa's basic commodity. I did not say that Sraffa necessarily provided a full alternative. The biggest problem indeed was that his analysis was always done in terms of comparing long run steady states.

However, he was the key player in the critique, along with Joan Robinson, with such figures as Garegnani and Pasinetti the junior supporting actors, along with Harcourt as sort of a summarizer and publicist. Kaldor was not involved with the debate at all.

As it was, the first time I met Paul Samuelson, I gave him a hard time about all this (about 40 years ago in Madison, Wisconsin when he showed up to give a talk at the Memorial Union theater). He freely admitted that Cambridge, UK was right and that the way to go was to simply model using heterogeneous capital.

It was always noted by the more careful Sraffians such as Steedman that the conditions under which one could use aggregate capital were essentially the same for both Marx and the neoclassicals. This was the source of the banging back and forth between the Sraffians and the Marxists back then.

Peter Dorman said...

I don't want to get into a thing about it, but no, Hahn was not discussing the basic commodity. He was writing about the aggregate value of capital.

If you reflect on it, the notion of a "long run steady state" implies a severe restriction on the outcomes one can model. The value of capital depends on its future returns, which are a function, among other things, of future prices. Sraffa imposes the condition that these have to be the same as today's prices. What holds for Sraffa is interesting only if it also holds in a world in which the price structure is allowed to change. That, as I recall, was Hahn's point.

ProGrowthLiberal said...

I had noted over at the comment section of Thoma's blog that Piketty's book would likely revive the Cambridge capital controversy - and so it has.

Peter Dorman said...

Incidentally, I had to revisit this issue when writing the micro volume of my textbook. I decided to keep it simple and obvious: book value (capital as stuff) doesn't coincide with market value (present value of profit stream if you believe that fundamentals provide the central tendency), and we end up with Tobin's q. But this is still enough to invalidate the interpretation of the supply and demand for capital as being determined by time preference and marginal productivity of capital, since it mixes up real and financial capital. said...


Here to correct myself, blame late night blogging. The issue is the standard commodity, not the basic commodity. It is the standard commodity that allows for both a possible aggregation (with some saying it overcomes index number problems) and also allows for a clear measure of value, unsurprisingly not unrelated.

That said, even when one has the standard commodity, an odd creature not clearly observable in the real world, this does not save one from the paradoxes of capital theory in the form of non-monotonicities between aggregate capital and r, or, for that matter, between the savings rate and r in a neoclassically formulated growth model. This is where the problems arise.

Apologies to all for my misstatement last night, for which I should be duly whipped, especially if I am going to pick so hard on Paul Krugman for what was ultimately a parenthetical aside, although it links to a core delusion that he maintains regarding the supposed virtues of the marginal productivity theory of distribution.

BTW, anyone wanting to get my real bottom line on the capital theory debates should read Chap. 8 of my 1991 From Catastrophe to Chaos: A General Theory of Discontinuities, which reappears in the second edition in 2000.

Owen Paine said...

The standard commodity is a fairly straight forward construct

But it is fragile eh?

Like the equilibrium construction

It is in essence timeless
A meta static concept as bill viceroy called such

Each is So fragile one needs
a chain of such constructs
To approximate dynamic systems

And indeed what do you have after all this calculation

Samuelson simply retreated into physical specification
Of the production system

I submit
the universalization of any rate
---In particular i include the rate of surplus value--
Is to avoid the real historical process
A dynamic system like a market system
may have regulating And persistent forcings
..lawful motions...
These are the objects of analysis

A series of differences
That drive
a market mediated
For profit firm embodied
production system
Induced endogenous changes
can use the term surplus or residue for these firm level differences

The sum of these differences has it's own significance of course

Sandwichman said...

I've only read accounts of the CCC from the British side and those long ago. I may be wrong but the impression I get is that you CAN'T "model" indeterminacy determinately. You can build a model of a special case to demonstrate that a general model is wrong but you can't simply generalize from that special case model. I can prove that your specimen of a unicorn is not a unicorn without having to produce a "real" unicorn.

Sandwichman said...

And, yes, I agree that Krugman blew it. Unless, of course, his intention was to distract attention from the issues and attract attention to himself.

Anonymous said...

I don't have enough of a background in Economics to comment on the validity of the underlying theories - but I'll note that claiming something is true because a respected figure said so is more of a religious argument than a scientific one.

I don't believe in evolution because Darwin said so - I believe in it because of the evidence (finch beaks and tortoise shell shapes in the Galapagos, relative populations of light/dark moth color variations depending on how much soot is on the trees, etc.

If economists are going to call their field a science as opposed to a philosophy, it would be nice to see more arguments based on observed facts rather than the musings of respected thinkers. said...


I did see your comment there. What I do not remember is whether or not it preceded or followed Jamie Galbraith's bringing up the issue, which certainly brought it forward quite publicly, if not to universal approval or agreement.


I agree the construct is straightforward. I did not say that it is fragile, but that it may be hard to observe. In any case, even if one can pin it down, it does not resolve the deeper problems raised at the British Cambridge regarding capital theory, which may or may not be lurking in Piketty's analysis.


I don't think this has too much to do with Keynesian uncertainty, although it certainly does reduce the certainty of using standard capital theory in a serious way.

Regarding PK, I should not pick on him further. He certainly does like to draw attention to himself, though, and he has had a bad track record about giving proper credit to others for ideas that he has espoused and even received famous prizes for.

That said, I think he sort of views himself as a modern day Samuelson to some extent, and certinaly knew Samuelson at MIT. Not only has he done theoretical work within the neoclassical framework that has won him a Nobel Prize, but he has written a Principles text and writes a widely read column, more widely read than the one Samuelson wrote (and PAS never had a blog), although the influence of his Principles book nowhere approaches that of Samuelson's. One thing they share is a desire to assert the supremacy of standared neoclassical economics, if occasionally noting its limits, and that is where I think Krugman has gotten into trouble, trying to imitate the Master, and falling on his face while doing so, not actually understanding what the Master had to say on this matter.


I appreciate your distrust of merely quoting respected thinkers, but what is involved here is Krugman in effect channeling Samuelson. So, what Samuelson said and meant is indeed crucial. Krugman posed that Cambridge USA had "won" the debate with Cambridge UK, with Samuelson being the acknowledged leader of the USA side. But that leader himself completely disagreed with Krugman's assessment, so that is indeed relevant, indeed decisive.

Harking back to paine's comment, it is only partially true that Samuelson "retreated to physical specification," although that is in effect what simply saying "model heterogeneous capital" amounts to. In the 1966 debate Samuelson had already retreated from the physical production function, hence all the discussion of the "surrogate production function," and indeed for anybody really knowing the history of this, which Samuelson certainly did, the formulation of the Cobb-Douglas production function was never based on any physical or engineering concept, but was back-derived from aggregate factor income distribution data with an eye to make it look like that was consistent with some sort of economy-wide optimizing aggregate agent, a very peculiar idea. said...

Let me add a quick note on "whatever happened to the Cambridge critique," which might also partly explain Krugman's good on this matter of "who won?"

So, in 1966 Samuelson clearly accepted that the Cambridge UK critique was valid. It is not clear if Solow ever did, although he clearly had to at least theoretically, but not too long after this he basically went silent on the issue, which seriously undermined his Nobel Prize-winning growth model based on an aggregate production function, the sort of thing lying at the heart of nearly all RBD DSGE models. In effect what happened was that the critique was ignored, a "let's move on to estimating those aggregate production functions, folks, and pay no more attention to this unfortunate roadside accident." As time passed, only a few heterodox grad programs taught this stuff so that now very few really know what it was all about, and the absence of discussion of it and the widespread use of Solovian functions can easily lead one to conclude that somehow Cambridge USA "won" the debate. It may well be that this at least partly explains how it came to pass that Krugman fell into deep doo doo in his discussion of this matter.

Anonymous said...

Peter, what was the title of the piece by Frank Hahn?

Tom Hickey said...

Solow in "The last 50 years in growth theory and the next 10." p. 4-5:

Here is where the anecdote that I promised comes in. I spent the year 1963 – 4 in Cambridge, England, engaged in one interminable and pointless hassle with Joan Robinson about some of these issues. Interminable is bad enough, pointless is bad enough, and putting them together is pretty awful. The details are too lurid to be told to young people. At one point, however, I realized that the discussion had become metaphysical and repetitive, and I decided to try a new tack. So I buttonholed Joan in her office one day and said: ‘Imagine that Mao Tse-Tung calls you in’—she was in her Chinese period then—‘and asks a meaningful question. The People’s Republic has been investing 20 per cent of its national income for a very long time. There is now a proposal to increase that to 23 per cent. To make a correct decision, we need to know the consequences of such a change. Professor Robinson, how should we calculate what will happen if we increase our investment quota and sustain it?’

‘So what will you tell Chairman Mao?’ I asked Joan. She baulked and bridled and dodged and changed the subject, but for once I was relentless. ‘Come on, Joan, this is Chairman Mao asking a legitimate economic question; the future of the People’s Republic and possibly of mankind may depend on the answer. What do you tell him?’ Finally, she grumbled: ‘Well, I guess a constant capital–output ratio will do.’ It made my day; I knew I could do better than that, and I knew she had been forced by practicality, even imaginary practicality, to give up the metaphysical ghost. I was smiling all the way home to tell my wife that Joan had buckled, and violated her own metaphysics.

One of her major contentions had been that it was illegitimate to think of ‘capital’ as a factor of production with a marginal product. Yes, a single capital good (or its services) was a productive input. But aggregating those goods, whose services are yielded over their remaining lifetimes, introduces all sorts of complications. It is always a problem in economics to navigate between pure and abstract conceptions (how would a concept like ‘capital’ fit into a complete and formal description of an economy) and the needs of practical calculation (Mao’s hypothetical question). It can (almost) never be done perfectly. I thought that Joan Robinson had been unfairly playing on that difficulty in order to undermine the ‘neoclassical’ attempt to construct a usable model of investment and growth. Faced with the need to be pragmatic, she had no recourse but the kind of statement that she had criticized in others.

That is what I mean by making it plausible: a simple, clear model should tell you how to get from empirical beliefs to practical conclusions. said...

Interesting, Tom, and consistent with my gut sense that Solow has taken a harder line than Samuelson did. It may be Solow who spread the idea that Cambridge USA won based on something like this anecdote, although offhand she says nothing about marginal products of capital as Solow claims she did, only that she allowed for the existence of a measurable aggregate capital. It should be kept in mind that Joan Robinson did write a book on growth theory that dealt with investment issues, so Solow's proclaimed triumph here looks a bit overblown.

But, it could be that Krugman got his ideas from some filter down from Solow, weak as Solow's claim appears to be.

Peter Dorman said...


I really have to apologize, it has been at least a couple of decades since I read the Hahn piece, and the title escapes me. I vaguely remember it was in the CJE, but I could be wrong. To be honest, my interest in this bit of academic disputation has really waned, and I don't want to spend the time it would take to track down the precise article. I'll leave this to you.

To be very clear though, the problem of unchanging price structures applied directly to the Sraffian conception of capital; this is not about the standard commodity. My "first dissertation" (a most unpleasant story) came to grief on exactly this point. said...

I would agree with Peter that it is not per se a solution to the "price index problem," and I would note my careful wording that some said it was. It was not.

Sandwichman said...

Tom Hickey,

Fascinating quote! I find it fascinating that Solow didn't realize the implications of his question. Maybe Robinson's "baulking, bridling, dodging and subject changing" wasn't so pointless, after all. "Capital" would have an entirely different meaning in a (cough) socialist economy than in a capitalist one. Or, in more technical terms, valuation proceeds differently in a command economy than it would in a market economy. Solow's question was irrelevant and suggests he didn't really understand the difference between physical product and value, output and revenue.

Anonymous said...

I cannot access the full text, but this one opens with Hahn's assessment of Sraffa's _Production of Commodities by Means of Commodities_

Anonymous said...

I found a useful set of papers on the Cambridges Capital Controversies here:

In particular the 2004 one "Whatever Happened to the Cambridge Capital Theory Controversies?".

BTW most people misunderstand the "commodities by means of commodities" booklet, as it never had an aim of creating a realistic model of the political economy. its aims were:

#1 To show fairly rigorously that even trivial comparative-static neoclassical multi-commodity models as the ones described in the booklet suffered from Wicksell effects invalidating JB Clark's "3 parables".

#2 To show that in trivial comparative-static neoclassical multi-commodity models even if they suffered from Wicksell effects it was possible to build a "numeraire" that was independent of the distribution of income, even if it was quite awkward.

I guess that #1 was just an entertaining "putting down" for the author, and J Robinson even argued that it was a bit of an irrelevance considering the much more serious objections that could be made; but #2 had been one of the great open question of political economy studies, and the author not only gave a proof that such a "numeraire" existed, but even gave a *constructive* proof.