Murphy Responds to Krugman on the Austrian Theory of the Business Cycle

|Peter Boettke|

Bob Murphy responds, stressing again the vital importance of an Austrian understanding of capital theory and providing some basic data that is consistent with the Austrian theory of the Business Cycle.  I find Murphy persuasive in this format, and would love to see this debate taken to the next level of scientific discourse.  I am very glad that Bob is using his considerable talents to engage the critics in the way he is. It would simply be awesome to see Bob take his argument to outlets like REStat, let alone the AER, QJE, or JPE.

Larry White now has a group of young and eager graduate students working on monetary theory, policy and history that should chime in, and Jesus Huerta de Soto as well has a group of students and former students working on monetary theory, policy and history that have something to say.  It seems as if a renaissance in Austrian macroeconomics is about to take place with all these young and eager minds entering into the profession and attempting to overturn received opinion by taking the ideas of Mises, Hayek, and Rothbard to a new generation of thinkers and addressing new policy challenges.  We know that these ideas have shown up in papers in the NBER working papers and written by faculty at UCLA and Chicago.  So it makes sense that the younger economists of the Austrian tradition should be emboldened to submit their work to the top outlets and engage the debate at the highest levels of scientific discourse in economics.

Comments

[Greg Ransom]

Don't forget, William White has taken Selgin & Hayek macro/money points right to the commanding heights of the banking and macro community.

And a broad range of top macro guys have been making Hayekian points of all sorts, usually with footnotes, e.g. John Taylor, Frydman, Phelps and others.

So this is a multifront revolution.

[Jasper]

Zero Hedge just reprinted Murphy's post.

[Jasper]

Hyper link didn't come through: http://www.zerohedge.com/article/robert-murphys-retort-paul-krugman-austrian-business-cycle-theory

[Mario Rizzo]

People who want to do this kind of work with influence in the profession need to learn contemporary macreconometrics.

[Current]

What do folks think about Chris Cotter's proposed evidence?

[Angel Martin]

Mario said: "People who want to do this kind of work with influence in the profession need to learn contemporary macreconometrics."

Could you please elaborate a bit more on that? I understand the need for Austrian economists to learn the tools mainstream economists use. But I don't get what your point is in your comment.
Thank you.

[Dee Williams]

I don't understand Murphy's Response to Tyler Cowen.

Specifically, he says that capital need not go under maintained during the boom period. Therefore, the fact spending on capital maintenance doesn't fall during he boom doesn't contradict his predictions. But this does not make sense to me.

#1. Roger Garrison (who Murphy is obviously relying on for theoretical input) says explicitly that the increase in consumption and new investment during the boom will be funded in part by "by the undermaintenance of existing capital".
http://www.auburn.edu/~garriro/strigl.htm

#2. If capital is not under maintained during the boom, I don't see why output would fall when the investment bubble collapses. For example, in responding to Cowen's critique, Murphy uses the example of corn growers and silo owners. He says that during the boom the corn grower expands production, but silo owners (and other entrepreneurs in the middle stages of production) do not expand production to accommodate. That may explain why output doesn't increase during a recession, but it doesn't explain why production *declines*.

I am glad that Murphy is attracting Krugman's attn, but I am afraid he is misrepresenting he Austrian position (or at least not defending it as strongly). Maybe I misunderstanding something?

[anon]

Current - who's Chris Cotter? And which paper are you referring to?

[George Selgin]

I hope, Pete, that you're right about those young Austrian monetary-macro-economists: may they may make up for the "lost generation"! But they'd better not have any illusions about how easy it is to publish unorthodox money-macro in those top journals!

[Bob Murphy]

[Billy?] Dee Williams, I am working on a new model--in between my sushi story and _The Pure Theory of Capital_--that will answer your wildest dreams.

There is no contradiction between Garrison and my work. His thesis and my antithesis will emerge as a grand synthesis in the fullness of time.*

* A month or two.

[MichaelB]

Bob Murphy is getting lots of exposure. Here The Economist takes note: http://www.economist.com/blogs/freeexchange/2011/01/americas_jobless_recovery_3

[Current]

Chris Cotter wrote this:
http://adamgmartin.com/Homepage/Essay_Contest_files/CotterABCEmpirical.pdf

It's quite interesting.

[Observer]

"People who want to do this kind of work with influence in the profession need to learn contemporary macreconometrics."

That is what people typically do in grad school.

[Observer]

@Bob Murphy

A model? Like a formalized, in a mainstream outlet publishable model?

I'm really looking forward to this kind of stuff.

[Roger McKinney]

Dee Williams, I agree. I think that's why Hayek's Ricardo Effect is so important to the ABCT.

[Daniel Kuehn]

For what it's worth, my concern about Cotter and lots of other empirical work like it is that this finding about the sensitivity of capital goods industries is not new and its not exclusively Austrian. That's a relationship that Keynesians rely on! I imagine its an implication of RBC models too.

What Austrian empirical work needs to do is demonstrate the link between interest rates and the capital structure itself, and then the relationship between readjustments of that capital structure and economic downturns. That's very different from showing that - surprise! surprise! - capital investment responds to the interest rate.

[Observer]

I agree. The paper like that of Cotter does not deliver any new input to the debate or provide an empirical support of the debate.

[Peter Boettke]

First, I think encouraging people to do the necessary research work is a start, but it doesn't mean it has been done.

Second, the coverage that Bob Murphy is getting is fantastic because (a) it shows off his very good skills at writing and thinking clearly about these issues, and (b) it will encourage others to pick up the topic and try to dig deeper (including Murphy himself). Thus we move from the discourse in The Economists (which is awesome) to the discourse in the journals. BTW, Bob was extensively trained in formal theory and econometric analysis at NYU, and he obviously has a deep understanding of Austrian economics from his education at Hillsdale and his studies at NYU and career since with LvMI and whatnot. Along with Andy Young (now at WVU), but is well situated to engage in the sort of model and measure exercise with respect to ABCT that potentially could move that discourse from The Economist to the journals.

Third, to George --- I believe that timing as well as ambition might work in the favor of these young economists to keep at it with respect to the journals. If you take a look at recent GMU graduates, the commitment to publishing is evident in their work, and this is constantly encourage by our faculty core of Chris, Pete, Virgil, Larry, Dick and myself. So I am hoping that those kids working with Larry will follow this publishing ethos in the same way their peers outside of monetary and macroeconomics have been doing.

Fourth, I wonder whether the move toward agent based modeling that was also picked up by The Economist this summer, is also a possible avenue for younger Austrian macroeconomists to pursue. I am sure Roger Koppl would think so, as does my colleague Dick Wagner. I also think there are experimental possibilities following on the heels of Vernon Smith's experiments with regard to bubbles.

Fifth, I also think the analytic narrative approach to economic history is another approach to empirical work on the Austrian theory of the business cycle.

As Lin Ostrom has championed, a multiple methods methodology has tremendous opportunities to explore a set of ideas that circumstances has seemed to put back into play among economists. As editor of the RAE, I hope we see lots of ABCT papers. But as an economist I also hope to see lots of ABCT papers get in the AER as well.

[Current]

Roger, Bob, Dee,

The normal thought experiment behind ABCT is to suppose a non-progressing economy as a simplification. Often ERE/general equilibrium is supposed. In that situation capital maintenance is important. There are no search costs in such a theoretical situation.

But in progressing economy ABCT can occur too. There is no theoretical reason why growth can prevent it. But, what growth and technological change means is that maintenance isn't the issue. In a growing economy lack of maintenance is a perfectly normal situation. As new technology replaces old and as old industries die some capital equipment is not maintained.

What happens in an ABCT boom is that low interest rates make some things relatively cheaper than they would be otherwise. Roundabout and capital investments are made cheaper as are risky investments. Later it may be that those investments can't be sustained at the same level they were. Also, the simple need for reallocation of resources causes costs itself such as retraining which Murphy mentions.

It isn't that capital is necessarily undermaintained in a physical sense. It's that as old capital is replaced for new that new capital is built on the assumption of lower long-run real interest rates than actually occur. When the bust comes that capital can't be used economically.

In the boom I was involved in a typical project of this sort. The PC company I worked with were planning a line of very stylish and slim computers to compete with Apple. Most of that work was subsequently thrown away.

This issue is related to the debate between Mises and Schumpeter on forced saving. Schumpeter proposed that it could only be beneficial because during the boom more capital will be invested. As far as I can remember Mises pointed out pretty much what I've said above. More investment may occur even in real terms, but it won't necessarily be sustainable.

It's interesting to speculate if an ABCT bust could occur in an economy where all capital goods last forever. Oddly this could occur because of the effects Gene Callahan and I pointed out in the discussion of Say's law a while ago. If the utility of working or extracting natural resources falls then output could fall too.

[Greg Ransom]

Is there any hope that the example of biological science will someday be imitated?

Darwinian biologists do field research -- in the field -- in order to understand the real world.

Field research on rent control was Hayek's first real hint at the coordinating role of relative prices.

It would help understanding immensely if real field work research had been done on the rolling effects of the housing boom and bust in Orange County, CA.

It's pathetic that economists read Michael Lewis and other journalists to get an understanding of what happened in the real world of the financial economy over the last ten years.

[Current]

Regarding Cotter's paper....

I'm not sure everyone talking about it has read it. Anyway...

> For what it's worth, my concern about Cotter
> and lots of other empirical work like it is
> that this finding about the sensitivity of
> capital goods industries is not new and its
> not exclusively Austrian. That's a
> relationship that Keynesians rely on! I
> imagine its an implication of RBC models too.

I agree that disambiguating Austrian and RBC evidence is difficult.

> What Austrian empirical work needs to do is
> demonstrate the link between interest rates
> and the capital structure itself, and then the
> relationship between readjustments of that
> capital structure and economic downturns.
> That's very different from showing that -
> surprise! surprise! - capital investment
> responds to the interest rate.

What Cotter's paper does is to:
* 1. Estimate the natural interest rate.
* 2. Compare that to the market interest rate and derive a gap.
* 3. Chart the output of various classes of goods in the periods after that gap opens up.

Cotter found that initially consumer goods and long-run capital goods are stimulated (though not mining which runs counter ABCT). Then after a period of growth that spreads to all classes of good there is a period of recession afterwards.

This is a very different thing from saying simply that capital investment responds to the interest rate. He's not simply using the market interest rate, but rather the gap between that and an estimate of the natural rate. And he's not saying simply that there is a rise in output, he's saying that there is a rise followed by a fall.

[McKinney]

Hayek's Ricardo Effect demonstrates the connection between interest rates and the capital structure. It's not that businessmen see interest rates fall 2% and borrow all they can. Hayek points out that businessmen rarely care about interest rates. They borrow because the lower interest rate changes the relative prices of capital and consumer goods, and therefore the profit opportunities. Businessmen are highly tuned to prices and profits.

[McKinney]

Pete, I think agent based modeling is exciting and hope to see Austrians using it. And I'll put in another plug for my favorite model comparison technique - Structural Equation Modeling because it was designed to compare different theoretical models of the same data.

[Daniel Kuehn]

McKinney - I've heard this from you before about SEM and I'm curious how that works - could you briefly explain?

When I learned SEM it was framed as a way of doing path analysis - multiple causal relationships with potential feedback loops - with variables that you didn't necessarily have observable data for. So you had latent variables instead of explicitly measured variables.

My understanding is the only way that can be used to "test theories against each other" is if you have a different theories on the causal pathways you're looking at. How easily adaptable is this to the task at hand? Is that essentially what you have in mind - a common framework with some shifting of the causal pathways that one would test? How does this all play out?

[McKinney]

Daniel, yes that's the SEM I had in mind. I haven't actually done it on macro data, but I have done a lot of SEM on business data. As you know, the latent variables just correct for measurement error. So for example you would create latent variables of historical data that represent Keynesian aggregates and model their interactions. Then create latent variables that represent Austrian variables and model them. Then use the model fitting stats to see which model fits the data best.

You may need to disaggregate some data for the Austrian model, then in the Keynesian model you would aggregate it again. You can work with longitudinal data as well as cross section data.

[Daniel Kuehn]

McKinney -
OK. First, I'd just say there's nothing that would require aggregation in the Keynesian models. If you think of the "Keynesian models" that have actually been run - Brookings, Klein's models, whatever the Fed works with, etc. - they're going to be far, far more disaggregated than anything any Austrian empiricist has ever worked with. Think of it this way: every piece of data any empirically minded Austrian has ever put to use was most likely collected for Keynesian modeling purposes!!! So the modeling is at least as disaggregated - if not more. This disaggregation obsession, I think, is a highly, highly misleading way of differentiating the perspectives. Theoretical Keynesian models can have "i" industries and "i" prices - set "i" as high as you want. The important thing is the dynamics, not the level of aggregation.

That having been said - I think what people normally think of in terms of "Keynesian modeling" falls into two primary camps: (1.) parameter estimation for old-school systems of equation, and (2.) multiplier estimates.

Both of these involve substantial identification problems. The identification problem associated with the parameter estimation approach revolves around simultaneous determination: how do you know these slopes and elasticities if you have a bunch of data at the equilibrium? - you need an identifying disturbance. The identification problem associated with the multiplier approach is an endogeneity/counterfactual problem: how do you the impact of government spending if you spend precisely when the economy starts turning down? - again you need some sort of identifying disturbance.

In any sort of causal modeling like SEM or path analysis, I think you largely just finesse these simultaneous determination or endogeneity problems away. You can't run the model if you don't make enough causal model specification decisions to fully identify the model. I don't think any Keynesian econometrician is just willing to finesse the problem away like this (I don't think any econometrician is). Government spending IS endogenous. Prices and quantities ARE simultaneously determined.

So let's assume away all these identification problems and assume we're talking about the same set of variables so we can compare. What is the real difference between them? Do Keynesians and Austrians expect a different relationship between investment and the interest rate? Do they expect a different relationship between money supply and the price level? Do they expect a different relationship between investment and consumption (I'm told the co-movement "embarassment" is not an embarassment at all)? No, no, and no. And we wouldn't expect them to come up with different answers to these relationships because any theory that survives a couple decades or more HAS TO be able to produce these relationships.

I don't pick at this to pester. I actually pick at this because I am genuinely interested in an empirical Austrian program. Maybe it's just because I'm a Keynesian that buys the logic of ABCT, but I'm not particularly interested in "comparing" the two. I expect both theories "happen" in the real world. I personally place more value on modeling the actual dynamics of the adjustment and re-adjustment of the capital structure and getting a sense of how that's related to employment fluctuations. I started working out a modeling approach last fall but had to drop it when a few other things came up... maybe I'll start thinking about that again.

[Jim Rose]

Perhaps two better empirical challenges to ABCT are by Friedman, and by Ohanian, albeit indirectly.

Friedman's plucking model of business fluctuations suggests a bust-boom business cycle below an output ceiling while the ABCT theory suggests a boom-bust cycle around a trend.

The literature on Friedman's plucking model has burgeoned since the 1993 Economic Inquiry symposium in honour of Friedman that had his and other papers on the plucking model. Garrison replied to Friedman's plucking model in 1996 in the Economic Inquiry stressing which level of aggregation is a good test.

On Lee Ohanian, he explains the depth of the 1930s great depression in the USA as follows:
• The industrial decline began in 1929 before the large monetary contraction or the banking panics - the conventional culprits.

• Friedman and Schwartz date the first banking panic occurring from November 1930 until January 1931, but this first episode is after industrial hours worked have fallen 30%.

• Ohanian attributes the massive drop in manufacturing industry hours in 1929 and early 1930 to President Hoover's labour policies, which kept nominal and real wages high.

• Economists cite monetary contraction (Friedman and Schwartz, 1963) and banking panics (Bernanke, 1983) as important determinants of the Depression, but US manufacturing industry and employment was significantly depressed before these factors was quantitatively important.

• A factor other than monetary contraction or bank runs was central in initiating the Depression. Importantly, says Ohanian, this factor impacted very differently across sectors. The great depression did not start as a garden variety recession for manufacturing industry. The drop in hours was immediate and deep.

• Hours worked in agriculture were roughly unchanged during the early 1930s, with large falls in agricultural real and nominal wages, which indicates that the initiating factor behind the great depression was sector-specific. Ohanian's labour market data indicate that this initiating factor prevented the industrial labour market from clearing.

The ABCT is a candidate reply to Ohanian (2009) because ABCT posits sector specific effects with the large fluctuations concentrated in the capital good industries.

My doubts about the Ohanian (2009) hypothesis that large manufacturers kept wages high from 1929 to 1931 in return from Hoover's protection from unions is collusive agreements are vulnerable to cheating. Unions played along because of wages did not fall. Roosevelt needed a massive new regulatory framework to enforce his cartelisation and wage fixing policies as is explained in Cole and Ohanian (2004). Regulation by stealth is overrated and the real thing is far more effective.

The notions by Ohanian (and by Rothbard) that Hoover could secure agreement a high wage policy by a wide enough group and be able to detect and punish cheating by incumbents, new entrants, and smaller lower cost manufacturers for any length of time are problematic.

ABCT has a better chance of explaining a sectoral collapse in employment and hours worked after a long boom. Ohanian said a monetary explanation of the Great Depression requires a theory of a very large and very protracted monetary non-neutrality.

ABCT identifies a large and protracted monetary non-neutrality: a long boom under the mask of stable prices followed in by a monetary contraction in 1928. From 1933, the new deal is a massive supply-side shock that depressed the economy further.

The hurdle for the ABCT explanation for the great depression is Friedman's view that long booms are not followed by deep recessions. The 1991 and 2001 recessions were mild follow-ups to long booms.

[McKinney]

Daniel, you are more focused on how Keynesian and Austrian econ complement each other, and that's fine. But you have to admit that there are some areas in which they disagree. And as Jim Rose points out, monetarism disagrees with both.

Isn't the gist of Keynesian econ the AD/AS model, which is built up from IS/LM? I'm suggesting to use a similar model for Keynesian econ and develop one for Austrian econ that emphasizes relative prices between capital and consumer goods.

It would involve more than just changing paths between latent variables. It would require very different latent variables.

Jim Rose's Friedman, Schwartz, Ohanian model is a good example of too much aggregation of data. Using just gdp as the dependent variable will hide a lot of important stuff going on beneath the surface that the gdp figures distort but which determine gdp.

[McKinney]

Daniel: "I expect both theories "happen" in the real world."

As I wrote above I think that is true, just at different times. However, the problem happens when you take an explanation for one aspect of the business cycle and make it the explanation for the whole thing. That's where most theories go wrong and cause the conclusions and policy to be wrong.

"I personally place more value on modeling the actual dynamics of the adjustment and re-adjustment of the capital structure and getting a sense of how that's related to employment fluctuations."

I guess you have read Hayek's Ricardo Effect? I don't think ABCT is complete without it.

[Current]

How does Keynesianism ever predict a bust of the sort that ABCT predicts? Doesn't it just suppose the the Animal Spirits may become less spirited from time-to-time?

[Troy Camplin]

"The hurdle for the ABCT explanation for the great depression is Friedman's view that long booms are not followed by deep recessions. The 1991 and 2001 recessions were mild follow-ups to long booms."

Might it not matter where the misallocation of capital occurs? The boom leading to the 2001 recession was caused in no small part by investment in internet startups. Anyone remember Pets.com? But that was a situation where mostly rich people were making risky investments, and when they lost their money, it only affected some slightly, and those who were affected a lot were small in number. The people who lost jobs were high tech people who quickly got jobs in other tech sectors -- albeit as employees making much less income. This recession came out of a housing boom, affecting a lot of people, since lots of people of all incomes buy houses, the poor were being encouraged to buy houses, and blue collar workers were building the houses, using skills not easily transferable.

[Bob Murphy]

Current, yes that's the synthesis I have in mind--I want to tell something simple like my sushi story, but in a progressing economy. In that setting, the critical "intermediate" capital goods are those that will be needed to maintain the higher level of output, and since they aren't being made properly, that rising level of output is going to crash, though not to pre-boom levels.

[Bill Woolsey]

Friedman's view is only a problem for the really, really bad versions of Austrian business cycle theories.

If you identify booms as periods of growing nominal quanties of money and matching nominal quantities of credit, then rising prices are constantly requiring that malinvestments get liquidated. Suppose the purchasing power of money wasn't falling. And all of the increase in the quantity of money was real, and the matching increase in credit supply was real. Well, then there would be growing malinvestments.

But in reality, in a sustained boom, the rising prices of goods means that a given increase in the nominal quantity of money and matching increase in the nominal supply of credit isn't buying as much. It can't distort as much.

The progressively higher level of nominal money and credit isn't causing a progressively higher level of malinvestment. It is the growth of nominal money and credit perhaps is causing a level of malinvestment.

Think of the malinvestments as being some segment of the economy on an unsustainable growth path. It has got to get back down to a sustainable one sooner or later. It isn't that there is some kind of growing gap that must get bigger the longer the boom.

I am sure everyone knows about the claim that it takes accelerating money growth to sustain the boom. Well, suppose that doesn't happen? It is possible for the malinvestments to be liquidated with nominal money and credit growing at a constant rate. Relative prices readjust, real output shrinks, the unemployment rate rises, and then and the nominal quantity of money and credit and prices all continue to rise.

If you miss this, say, you thought it all had to do with increased japanese competition with us care companies, then the boom ended. There aer no more malinvestments.

The quantity of money is rising. But in an inflationary equilibirum all prices are rising all the time. It is necessary to borrow more dollars to buy the same stuff, even if it isn't a malinvestment. This is the other side of the coin of the need to accelerate money growth to maintain a boom. If you don't the boom ends, the malinvestments get liquidated, by there is still growth in the quantity of money and credit.

Money growth just continues, and as prices rise, the real impact of that money growth becomes less and less. Only if the quantity of money stays ahead of the price inceases is their room for an increase in the real quantity of money and the real supply of credit that could impact the real allocation of resources.

[McKinney]

Bill: "But in an inflationary equilibirum all prices are rising all the time."

In the long run all prices will have risen. But in the medium term they never rise at the same time. Malinvestment happens because lower interest rates makes investment in capital goods more profitable. But the hiring of more workers causes them to spend on consumer goods, so the prices of consumer goods rise.

Prices for consumer goods rise faster and stay ahead of price increases for capital goods. That causes the Ricardo effect to kick in.

[Current]

I agree with what Bill Woolsey has said to a certain degree. If there has been a steady, established rate of price increase then that will be expected. The money creation that produces it won't lower the real rate of interest because the price increase will be factored into expectations (the Fisher Effect).

That doesn't mean that steady inflation is just as good as steady deflation. Accounting around x% is much harder than accounting around 0. Account falsification will still creep in even if it is much weakened by expectations.

But, Bill was going much further in other posts in the past few days. I don't agree with him about that. I don't think that people can "expect" their way around nearly anything.