« Keynes's General Theory, chapter three | Main | The multiplier is "one" »
Keynes's General Theory, chapter four, *The Choice of Units*
This chapter may seem cryptic but the key is the tiny footnote to Hayek; this chapter is Keynes obsessing over capital theory and the Austrians.
Hayek argued that an economic downturn should be understood as a discombobulation of the capital structure and here is Keynes arguing against that approach. When you cut through the terminology, Keynes says that capital heterogeneity isn't needed to generate aggregate demand analysis and that his core mechanisms will operate in any case.
Keynes admits that with economic development labor gets very specialized, or very closely connected to particular capital goods, so yes there are capital complementarities of the Austrian kind. But Keynes thinks such fragilities will only help his argument, while rendering the analytics too messy. He declares his intention to proceed with homogeneous magnitudes of capital and labor.
This chapter often fails to receive its proper due; it is very important for understanding the location of Keynes in the history of economic thought.
With this one chapter, Austrian capital theory falls off the map.
Posted by Tyler Cowen on December 11, 2008 at 07:46 AM in Economics | Permalink
Comments
Isn't Keynes simply avoiding the capital measurement problem by defining
capital as labor value? This does not preclude high interest rates
discounting future cash flows.
Posted by: gnat at Dec 11, 2008 8:49:58 AM
If only Austrian Capital Theory could be struck off the map.
In this chapter Keynes fixes the capital stock. That means, obviously, that it will be maximally utilized at full employment. It also means that any event that causes anything less than full employment will cause a drop in economic output from it's maximum.
Underemployment is clearly the problem in an economic model where it is defined as the only problem.
It is here that economics becomes about the short run and it is here that it becomes hydraulic.
Posted by: Current at Dec 11, 2008 9:11:43 AM
With this one chapter, Austrian capital theory falls off the map.
And depressions were vanquished once and for all!
Posted by: Bob Murphy at Dec 11, 2008 9:28:07 AM
I clicked on the comments just because I wanted to see what Bob Murphy would write.
Posted by: scott clark at Dec 11, 2008 10:06:51 AM
From what I see, this chapter actually illustrates Keynes' muddled thinking. Although he is correct to point out that it is not possible to get a meaningful aggregate by adding non-homogeneous capital or commodities he makes an assumption that his argument is not valid for labour and clams that economists can come up with meaningful 'labour-units' by adding up non-homogeneous labour.
The chapter and the entire book are full of such muddled ideas that cannot hold up to careful scrutiny. Keynes and his followers believe that they can understand complex, non-linear systems by making simple assumptions that would allow them to use mathematical tools that are totally inappropriate. That is why Keynesian economics has failed to make meaningful positive contributions to understanding how the real world works and has led to misery and underperformance whenever it was taken seriously.
Posted by: Vangel at Dec 11, 2008 10:09:57 AM
In other words.
Punt.
Posted by: Greg Ransom at Dec 11, 2008 10:23:02 AM
The biggest problem for Keynes is that he would have had to LEARN capital theory.
Too much work.
Better strategy -- punt.
ALL evidence on this matter points to the fact that Keynes never learned the work of Menger, Bohm-Bawerk & Hayek on capital structure -- and never knowing it, never understood it.
Never.
Posted by: Greg Ransom at Dec 11, 2008 10:25:33 AM
Hayek repeatedly said that as far as he could tell, Keynes never knew any 19th century economics outside of Marshall -- Keynes's essential ignorance of the work of Menger, Bohm-Bawerk, and any other German economist who wrote in the 19th century was one big reason Hayek had this deeply negative view of Keynes as an economist.
Posted by: Greg Ransom at Dec 11, 2008 10:28:15 AM
When are we doing the next chapter(s)?
Posted by: Tim at Dec 11, 2008 10:46:07 AM
Bob Murphy: "And depressions were vanquished once and for all!"
Vangel: "can come up with meaningful 'labour-units' by adding up non-homogeneous labour."
These points are inter-related. Many Keynesian economists have claimed that all labour can be rebased towards unskilled labour. It is not at all clear how this can be done. So, labour, like capital is inhomogeneous.
Bob says that once Austrian capital theory has gone "depressions were vanquished". He is right.
Let's suppose that Austrian capital theory is wrong and that capital inhomogeneity does not matter. In that case only unemployment or underemployment can cause a fall in output.
Look though at what happens when firms cut back on staff. They do not cut all equally. They cut back on those that produce the least, at the margin. They don't necessarily cut back the lowest paid. Rather they attempt to remove those who, in management's opinion, give the lowest "value add".
Given this how is a depression ever likely to occur? Let's say unemployment rises by 1%. Does that mean that output will fall by 1%, no, even Keynesians don't think so. Given that the 1% in question are the least productive workers and that capital equipment remains the same it is implausible that this would lead to anything close to a 1% drop in output.
Add to this is the bald fact that output has fallen without major falls in employment. This points to the aggregation of capital being a bad analytical assumption to start from.
Posted by: Current at Dec 11, 2008 10:52:58 AM
I clicked on the comments just because I wanted to see what Bob Murphy would write.
Yeah, me too.
Posted by: Blackadder at Dec 11, 2008 11:00:12 AM
Tyler, you touch on a recurrent issue here when you suggest that Chapter 4 is a response to Austrian capital theory. Lots of people do it (yes, I am talking to you, little Johnny Rawls, and especially you, Georgie Hegel) but Keynes is particularly culpable--carrying on an argument with a predecessor without letting the audience in on the argument or the identity of the predecessor (Chapter 2 is a quarrel with Pigou, not so?). You'd do us a kindness if you would try to flush out those quarrels when you think you see them.
Posted by: Buce at Dec 11, 2008 11:49:23 AM
Here's my suggestion. Let's do conduct economics as a quantitative, experimental "science". Here's the experiment. Let's fire every single Keynesian macrooeconomist in the country -- force them to get another line of work -- and then get out our measuring tape and look to see if "labor" is indeed a homogeneous input that can be quantified in "labor units".
Vangel wrote:
"and clams that economists can come up with meaningful 'labour-units' by adding up non-homogeneous labour."
Posted by: Greg Ransom at Dec 11, 2008 12:16:43 PM
"carrying on an argument with a predecessor without letting the audience in on the argument or the identity of the predecessor"
An important question is why Keynes routinely does this. These are some of the reasons. 1. Keynes in many important instances had never mastered true character and detail of the predecessor argument; 2. engaging the content and force of the predecessor argument would betray the weakness and sometimes even vacuity of Keynes'response; 3. Keynes was a professional political "spinner", excelling at "persuasion" in the popular press -- and he used the same techniques in his books and economic papers.; and 4. Keynes stood in a position of great power and status within the economics profession in England, and he knew he could get away with all sorts of professional behaviors which lesser folks wouldn't even attempt.
No doubt more could be added.
Posted by: Greg Ransom at Dec 11, 2008 12:30:32 PM
Cutting through the terminology FAIL "Keynes says that capital heterogeneity isn't needed to generate aggregate demand analysis and that his core mechanisms will operate in any case."
Posted by: ckstevenson at Dec 11, 2008 12:46:11 PM
OK, so "Keynes says that capital heterogeneity isn't needed to generate aggregate demand analysis and that his core mechanisms will operate in any case."
The first part of the sentence is correct "capital heterogeneity isn't needed to generate aggregate demand analysis," but this just indicates the problem of such analysis. The second part of the sentence is Keynes la-la land. There are no such things as "core mechanisms" working at a macro level based on aggregate equations.
Its pure fantasy.
Posted by: liberty at Dec 11, 2008 1:20:59 PM
From the point of view of empirical method Chapter 4 is a brilliant application of classical economic theory. Keynes lays the foundations for what we now know as macroeconomics by directly confronting the challenges of aggregating economic data.
His solutions: Focus on monetary values which can be aggregated. Avoid relying on the measurement of capital since the problem of measuring depreciation is a total boondoggle. To deal with differences in labor quality, he proposes that workers be weighted by their wage (or marginal product -- a brilliant application of classical theory). Having weighted workers by their marginal product, he can now claim that an increase in employment will imply an increase in output.
This latter conclusion may hold even for small changes in the capital stock (I'd have to think about it more to be sure). Overall Chapter 4 establishes Keynes as a brilliant economist.
Posted by: ccm at Dec 11, 2008 2:28:26 PM
"To deal with differences in labor quality, he proposes that workers be weighted by their wage (or marginal product -- a brilliant application of classical theory)."
I propose that workers be weighted by their height, which similarly doesn't get around the problem that most people are only willing or able to undertake a narrow range of jobs.
Marginal product or wage may be the best way of aggregating labor, but it isn't clear how it directly confronts the problems of aggregating data. After all, this is the same Keynes who -- one page before -- claims that the aggregate price of goods sold (C+I+G+Xn, say) is a bad way to measure national income, because the output of society is a "non-homogeneous complex which cannot be measured, strictly speaking".
So how is a similar aggregate of the price of one factor of production a "homogeneous complex" which can be measured?
Posted by: Eric at Dec 11, 2008 3:22:28 PM
ccm: "by directly confronting the challenges of aggregating economic data"
Alas he doesn't.
ccm: "Avoid relying on the measurement of capital since the problem of measuring depreciation is a total boondoggle."
Depreciation is not the only problem of capital though. Capital may have different outputs. It is like a giant junkyard that history bequeaths us. We have to sort out what of it is useful. It may serve our present needs or it may not.
If we ignore it then whatever theory is produced is only short term. We can say that the capital today is roughly the same as that last month, but not the same a five years ago.
ccm:"To deal with differences in labor quality, he proposes that workers be weighted by their wage (or marginal product -- a brilliant application of classical theory)"
Well, if that works for Labour why not use it for Capital too?
ccm:"Having weighted workers by their marginal product, he can now claim that an increase in employment will imply an increase in output."
You don't need to weight them by marginal product to do that. Even if you don't the answer is the same.
It's easy. Consumption goods are made by capital goods and labour. If you hold capital goods constant then clearly any difference is down to labour.
Posted by: Current at Dec 11, 2008 3:34:48 PM
Eric: "So how is a similar aggregate of the price of one factor of production a "homogeneous complex" which can be measured?" Precisely because Keynes has embraced the assumption that wage = marginal product of labor. You can argue that this is a bad assumption -- but then you are also arguing with the whole of classical and neo-classical economic theory. In fact you may be taking on the Austrians too.
Current: "Alas he doesn't." Needless to say he does not resolve all the problems, but it was a good enough start to build on.
Current: "If we ignore it then whatever theory is produced is only short term." Perhaps. I do wonder to what degree weighting by marginal product mitigates this problem.
Current: "Well, if that works for Labour why not use it for Capital too?" A very good question. Clearly his decisions are motivated by his agenda -- but I'd like you to find a economist who doesn't share this fault.
Current: "Consumption goods are made by capital goods and labour. If you hold capital goods constant then clearly any difference is down to labour." This is only true because we have developed a way of measuring non-comparable forms of labor. If the number of doctors falls by 10 and the number of assembly line workers rises by 10, are you really sure that GDP stays at exactly the same number?
Posted by: ccm at Dec 11, 2008 4:03:42 PM
Let me try this one again.
Current: "Well, if that works for Labour why not use it for Capital too?"
Labor has the advantage that the unit one worker hour is well defined and the wage of that unit is easily measured. While capital's rent is easily measured, capital does not have comparable units: machinery is much less homogeneous than humans; how does one address working capital and liquid capital where rents may be harder to calculate. Furthermore we rarely have market values for "the junkyard that history bequeaths us", so there is no reliable dollar value of capital, which can be weighted by the rent.
In short, the units in which labor is measured are inherently more homogeneous than units of capital.
Posted by: ccm at Dec 11, 2008 4:17:01 PM
Let me try this one again.
Current: "Well, if that works for Labour why not use it for Capital too?"
Labor has the advantage that the unit one worker hour is well defined and the wage of that unit is easily measured. While capital's rent is easily measured, capital does not have comparable units: machinery is much less homogeneous than humans; how does one address working capital and liquid capital where rents may be harder to calculate. Furthermore we rarely have market values for "the junkyard that history bequeaths us", so there is no reliable dollar value of capital, which can be weighted by the rent.
In short, the units in which labor is measured are inherently more homogeneous than units of capital.
Posted by: ccm at Dec 11, 2008 4:17:56 PM
//I clicked on the comments just because I wanted to see what Bob Murphy would write.//
I wanted to see what Greg Ransom would post. And I was not disappointed! But I do have a question: I get your point that nobody except for you (and Bob Murphy???) understands Hayek's capital theory, blah blah...but answer this: Exactly how doesn't Keynes understand it? Non-chalantly throwing out accusations like "he doesn't understand," without explaining, makes it seem like you don't understand it either, so do expand please. Or put it in your blog instead of flooding a comments section with something that doesn't contribute to the body of economic knowledge at all.
Posted by: Tim at Dec 11, 2008 5:23:17 PM
"Or put it in your blog instead of flooding a comments section with something that doesn't contribute to the body of economic knowledge at all."
This is coming from soneone who is praising a chapter in General Theory.
*blushes profusely*
Posted by: Alex at Dec 11, 2008 7:32:14 PM
ccm: "His solutions: Focus on monetary values which can be aggregated."
Sorry? How do you know that things that cannot be aggregated can be safely ignored? Keynes said you so?
Posted by: Vasile at Dec 11, 2008 7:46:43 PM
