Once again, I find myself slightly behind the curve, with Scott Sumner (and again, and again, and again, and again), Nick Rowe and Bill Woolsey out there trying to face down an onslaught of Austrians rallying under the dreaded banner (I won’t say what color) of Cantillon Effects. At this point, the best I can do is some mopping up by making a few general observations about the traditional role of Cantillon Effects in Austrian business cycle theory and how that role squares with the recent clamor about Cantillon Effects.
Scott got things started, as he usually does, with a post challenging an Austrian claim that the Federal Reserve favors the rich because its injections of newly printed money enter the economy at “specific points,” thereby conferring unearned advantages on those lucky or well-connected few into whose hands those crisp new dollar bills hot off the printing press first arrive. The fortunate ones who get to spend the newly created money before the fresh new greenbacks have started on their inflationary journey through the economy are able to buy stuff at pre-inflation prices, while the poor suckers further down the chain of transactions triggered by the cash infusion must pay higher prices before receiving any of the increased spending. Scott’s challenge provoked a fierce Austrian counterattack from commenters on his blog and from not-so-fierce bloggers like Bob Murphy. As is often the case, the discussion (or the shouting) produced no clear outcome, each side confidently claiming vindication. Scott and Nick argued that any benefits conferred on first recipients of cash would be attributable to the fiscal impact of the Fed’s actions (e.g., purchasing treasury bonds with new money rather than helicopter distribution), with Murphy et al. arguing that distinctions between the fiscal and monetary effects of Fed operations are a dodge. No one will be surprised when I say that Scott and Nick got the better of the argument.
But there are a couple of further points that I would like to bring up about Cantillon effects. It seems to me that the reason Cantillon effects were thought to be of import by the early Austrian theorists like Hayek was that they had a systematic theory of the distribution or the incidence of those effects. Merely to point out that such effects exist and redound to the benefits of some lucky individuals would have been considered a rather trivial and pointless exercise by Hayek. Hayek went to great lengths in the 1930s to spell out a theory of how the creation of new money resulting in an increase in total expenditure would be associated with a systematic and (to the theorist) predictable change in relative prices between consumption goods and capital goods, a cheapening of consumption goods relative to capital goods causing a shift in the composition of output in favor of capital goods. Hayek then argued that such a shift in the composition of output would be induced by the increase in capital-goods prices relative to consumption-goods prices, the latter shift, having been induced by a monetary expansion that could not (for reasons I have discussed in previous posts, e.g., here) be continued indefinitely, eventually having to be reversed. This reversal was identified by Hayek with the upper-turning point of the business cycle, because it would trigger a collapse of the capital-goods industries and a disruption of all the production processes dependent on a continued supply of those capital goods.
Hayek’s was an interesting theory, because it identified a particular consequence of monetary expansion for an important sector of the economy, providing an explanation of the economic mechanism and a prediction about the direction of change along with an explanation of why the initial change would eventually turn out to be unsustainable. The theory could be right or wrong, but it involved a pretty clear-cut set of empirical implications. But the point to bear in mind is that this went well beyond merely saying that in principle there would be some gainers and some losers as the process of monetary expansion unfolds.
What accounts for the difference between the empirically rich theory of systematic Cantillon Effects articulated by Hayek over 80 years ago and the empirically trivial version on which so much energy was expended over the past few days on the blogosphere? I think that the key difference is that in Hayek’s cycle theory, it is the banks that are assumed somehow or other to set an interest rate at which they are willing to lend, and this interest rate may or may not be consistent with the constant volume of expenditure that Hayek thought (albeit with many qualifications) was ideal criterion of the neutral monetary policy which he favored. A central bank might or might not be involved in the process of setting the bank rate, but the instrument of monetary policy was (depending on circumstances) the lending rate of the banks, or, alternatively, the rate at which the central bank was willing lending to banks by rediscounting the assets acquired by banks in lending to their borrowers.
The way Hayek’s theory works is through an unobservable natural interest rate that would, if it were chosen by the banks, generate a constant rate of total spending. There is, however, no market mechanism guaranteeing that the lending rate selected by the banks (with or without the involvement of a central bank) coincides with the ideal but unobservable natural rate. Deviations of the banks’ lending rate from the natural rate cause Cantillon Effects involving relative-price distortions, thereby misdirecting resources from capital-goods industries to consumption-goods industries, or vice versa. But the specific Cantillon effect associated with Hayek’s theory presumes that the banking system has the power to determine the interest rates at which borrowing and lending take place for the entire economy. This presumption is nowhere ot my knowledge justified, and it does not seem to me that the presumption is even remotely justifiable unless one accepts the very narrow theory of interest known as the loanable-funds theory. According to the loanable-funds theory, the rate of interest is that rate which equates the demand for funds to be borrowed with the supply of funds available to be lent. However, if one views the rate of interest (in the sense of the entire term structure of interest rates) as being determined in the process by which the entire existing stock of capital assets is valued (i.e., the price for each asset at which it would be willingly held by just one economic agent) those valuations being mutually consistent only when the expected net cash flows attached to each asset are discounted at the equilibrium term structure and equilibrium risk premia. Given that comprehensive view of asset valuations and interest-rate determination, the notion that banks (with or without a central bank) have any substantial discretion in choosing interest rates is hard to take seriously. And to the extent that banks have any discretion over lending rates, it is concentrated at the very short end of the term structure. I really can’t tell what she meant, but it is at least possible that Joan Robinson was alluding to this idea when, in her own uniquely charming way, she criticized Hayek’s argument in Prices and Production.
I very well remember Hayek’s visit to Cambridge on his way to the London School. He expounded his theory and covered a black board with his triangles. The whole argument, as we could see later, consisted in confusing the current rate of investment with the total stock of capital goods, but we could not make it out at the time. The general tendency seemed to be to show that the slump was caused by [excessive] consumption. R. F. Kahn, who was at that time involved in explaining that the multiplier guaranteed that saving equals investment, asked in a puzzled tone, “Is it your view that if I went out tomorrow and bought a new overcoat, that would increase unemploy- ment?”‘ “Yes,” said Hayek, “but,” pointing to his triangles on the board, “it would take a very long mathematical argument to explain why.”
At any rate, if interest rates are determined comprehensively in all the related markets for existing stocks of physical assets, not in flow markets for current borrowing and lending, Hayek’s notion that the banking system can cause significant Cantillon effects via its control over interest rates is hard to credit. There is perhaps some room to alter very short-term rates, but longer-term rates seem impervious to manipulation by the banking system except insofar as inflation expectations respond to the actions of the banking system. But how does one derive a Cantillon Effect from a change in expected inflation? Cantillon Effects may or may not exist, but unless they are systematic, predictable, and unsustainable, they have little relevance to the study of business cycles.
Uneasy Money 
David
A good closing of the “discussion”.
LikeLike
David:
Did you see Horwtiz’s comment?
http://www.coordinationproblem.org/2012/12/sumner-murphy-richman-and-cantillon-effects.html
When I read it, I was thinking along the same lines as the thrust of your post. At Horwitz’s level of abstraction, it would seem like monetary policy could have some effects, though it seems like entrepreneurs are being a bit myopic about investment. Buy new specific capital goods based upon last week’s sales?
I certainly believe that banks (particularly if led by a central bank) can set short term interest rates. Somehow, in your discussion of the comprehensive determination of interest rates and the entire yield curve, expectations of future short term interest rates didn’t stand out. To me, the key error in the Austrian approach is the assumption that everyone expects the current bank rate to persist forever. The banks set the short rate and everyone expects it to last forever. Pretty much myopic.
Perhaps I mistunderstood your argument here, but some of the notion that monetary policy (or the banks) cannot impact interest rates might be due to thinking of monetary policy in terms of a specific change in the quantity of money. That is the way of most of the recent blog debate. But suppose the policy is instead to create as much money as needed to target a specific short term interest rate? It is no longer how can this small increase in the quantity of money impact the vast and huge markets for both financial and real assets. If heroic changes in the quantity of money are “needed,” then that is what happens.
As for my notion of myopia, suppose the policy is to keep this short term interest rate pegged forever and ever.
But with those caveats, applying the theory to historical events, such as the housing boom and Great Recession, would be pretty useless. While the quantity of money wasn’t constrained, the “too low for too long” very short term interest rates were most decidedly not set at that rate forever and ever.
LikeLike
Monetarists are not convincing in answering the Austrians because they insist in not understanding how the monetary and fiscal policy work.
You write: newly printed money enter the economy at “specific points,” – yeah, at the same specific points assets of equal value exit the economy as all the Fed ever does is *buying assets* from eager sellers. Money in, assets out. Net worth unchanged. In every transaction.
This is different from fiscal policy that is authorized by Congress and in which the Fed acts as an agent of Congress: yes, fiscal policy is newly printed money entering in arbitrary points (spending) and leaving in arbitrary points (taxing). But this has nothing to do with the monetary policy.
LikeLike
Is r determined by stocks or flows?
Draw a PPF, with the flow of new capital goods produced on one axis, and the flow of new consumption goods produced on the other axis. Both are measured in physical units, not value units. Is that PPF straight or curved?
If that PPF is a straight line, it pins down the relative price of capital goods vs consumption goods. There is only one rate of interest at which the stock of capital goods will be willingly held, at which r=MPK/Pk. The investment demand curve in the flow loanable funds market is perfectly elastic. Changes in saving do not affect r. The IS curve is horizontal. Shifts in the LM curve cannot change r. The central bank cannot set r, even with sticky prices/wages. We have a stock model of r.
If that PPF is curved (bowed out) movements along the PPF, increasing I relative to C, increase the marginal cost of production of capital goods, and so raise the price of capital goods Pk. The rate of interest is no longer pinned down by MPK/Pk, because Pk is not pinned down by technology. The investment demand curve now slopes down. Changes in saving affect r, because an increase in S and hence I will cause Pk to rise and hence MPK/Pk to fall. The IS slopes down. Shifts in the LM curve will change r. The central bank can set r, temporarily, provided prices/wages are sticky. We have a flow model of r.
I think the PPF is curved, rather than straight. I think Hayek thought so too. (And I think Joan Robinson never got her head straight on this.) But Hayek never got his head straight on the sticky price/wage question, and how monetary policy affected employment and output. Because if monetary policy does affect employment and output, you won’t just be moving along a given PPF. And if prices/wages are perfectly flexible, a change in monetary policy can’t even move you along a PPF, unless you think in terms of forced savings, which implies a systemic relationship between the savings rate and seigniorage (seigniorage revenues are saved, rather than consumed). Which is where Cantillon effects come in. But since seigniorage revenues are peanuts, any forced saving coming from Cantillon effects must be peanuts too.
LikeLike
In fact, *all* Austrians think that PPF is curved. Because if that PPF were a straight line, the whole of Austrian capital theory would be nonsense, because r would be determined by technology (and expectations) independently of time preference.
LikeLike
“Cantillon Effects may or may not exist, but unless they are systematic, predictable, and unsustainable, they have little relevance to the study of business cycles”
I agree with this David.
And here’s my systematic, predictable, and unsustainable Cantillon effect – commodities. I’m not an Austrian by any means, But this is one relative inflation I worry about.
Ryan Avent tweeted yesterday about how the price increases in services and commodities completely overshadows the price increases of ‘stuff’ i.e consumer durables.
Services are healthcare, finance, education (and govt itself). There is some Baumol disease type story there. But a more plausible candidate seems to be the information asymmetry + govt-corporate complex creep. That’s the great microeconomic public policy challenge of our times.
As for commodities, while there is indeed a China demand story to the supercycle, it seems to me that the standard structure of monetary easing – flushing liquidity through the universal banking system – is a recipe for commodity volatility + a generic shaky floor under commodity prices in these financialized times.
But leaving aside Austrian type Cantillon effects, Scott’s argument also depended on arguing that ‘holding fiscal policy constant’ CB asset purchases don’t increase the price of those assets. I thought it was a rather funny definition of holding fiscal policy constant. I wonder if you agree with Scott’s view. If you don’t, and you’re on-board with the more mainstream New Keynesian view of monetary easing being net wealth, surely which asset is purchased matters.
You could call it a ‘fiscal effect’. In which case let’s just say that all monetary action, as normally conceived, has fiscal effects. The only *pure*, fiscaly neutral injection of money is a helicopter drop. Which, paradoxically, is viewed as a quasi-fiscal action.
LikeLike
This post is yet another reminder that I appreciate the perspective you bring to these discussions. You help to turn “shouting sessions” with no clear “winners or losers” into a learning experience.
LikeLike
David: “Cantillon Effects may or may not exist, but unless they are systematic, predictable, and unsustainable, they have little relevance to the study of business cycles.”
I would add: and they have got to be big too. Big enough to matter.
LikeLike
Nick writes,
“I would add: and they have got to be big too. Big enough to matter.”
This is why it matters whether our conceptual imagination has the concepts of yo-yo-ing shadow money, endogenous purchasing power creation, and the choice between longer production processes promising superior output and shorter processes promising inferior output.
It also helps to the the concept of non-economic goods, and to understand who some inputs to production would lose their economic good status when the production structure is shortened.
The “HOW BIG” question requires a conceptual universe that does not exist in the imagination of most grad school trained economists who grew up answering test questions with “K” all the other magically related homogeneous aggregates — and who still fill their whiteboards and peer reviewed publications without any room for the causal dimensions of the real world I’ve mentioned.
LikeLike
David, you leave out the direct connection Hayek draws between changes in the structure of investment and the expansion and contraction of shadow money and the endogenous creation of money and purchasing power:
See, for example, this discussion and these numbers:
http://hayekcenter.org/?p=2954
It isn’t all about interest rates — interest rates are stand out because of their importance as formal elements in simple models — and the role of the financial sector goes beyond that of a central bank setting an interest rate.
LikeLike
Having read Hayek’s original Cambridge lecture, and most all discussions of this episode, I can’t don’t see any plausibility in your suggestion about what Joan Robinson was alluding to in the ‘raincoat’ comment. Hayek in the lecture was talking about demand for short term consumption goods and longer term consumption goods, and the competition for resources to complete those production processes. It’s all very self evident if you know that lecture and the explanatory context.
David writes,
“I really can’t tell what she meant, but it is at least possible that Joan Robinson was alluding to this idea when, in her own uniquely charming way, she criticized Hayek’s argument in Prices and Production.”
LikeLike
Here as direct link to “Long Shadows: Collateral Money, Asset Bubbles and Inflation” by Jonathan Wilmot, James Sweeney, Matthias Klein, and Carl Lantz.
Click to access Sweeney%20-%20Money%20supply%20and%20inflation.pdf
LikeLike
Thanks for writing this, David. A very clarifying discussion of what should have been the subject of discussion from the beginning.
LikeLike
I clicked on the link to my post to see just how “fierce” my counterattack was. I confess I don’t see it. (I’m being serious.) If I disagree with Scott, is that being “fierce”?
LikeLike
I think it might be useful to examine the opposite, when the Fed tightens, and weigh the differences between them from the point of view of average Joe. In that regard, I think monetary stability certainly has its merit; and we need to consider that any movement in a more positive direction would only be in respect to reversing the effects of severe tightening that should never have occurred.
LikeLike
I have the same question Bob Murphy has. David, why did you characterize his response as “fierce?”
LikeLike
David,
Am I correct in saying that Hayek is arguing that in a multi-good economy (capital goods and consumer goods), who gets the money first will affect the relative prices of both types of goods. Suppose we have three agents that will spend their money and receive their income accordingly:
Agent #1 – Spending – Capital Goods: 60% Consumer Goods: 40%
Income – Capital Goods: 100% Consumer Goods: 0%
Agent #2: Spending – Capital Goods: 80% Consumer Goods: 20%
Income – Capital Goods: 50% Consumer Goods: 50%
Agent #3: Spending – Capital Goods: 10% Consumer Goods: 90%
Income – Capital Goods: 0% Consumer Goods: 100%
If we give each agent $1.00 at the beginning of period #1 then:
Agent #1 – Spending – Capital Goods: $0.60 Consumer Goods: $0.40
Agent #2 – Spending – Capital Goods: $0.80 Consumer Goods: $0.20
Agent #3 – Spending – Capital Goods: $0.10 Consumer Goods: $0.90
Total Capital goods spending = $1.50
Total Consumer goods spending = $1.50
Agent #1 – Income – Capital Goods: $1.00 Consumer Goods: $0.00
Agent #2 – Income – Capital Goods: $0.50 Consumer Goods: $0.50
Agent #3 – Income – Capital Goods: $0.00 Consumer Goods: $1.00
All agents end with what they started with and the relative prices of all goods are unchanged.
If instead we give Agent #1 $3.00 at the beginning of period #1 then:
Agent #1 – Spending – Capital Goods: $1.80 Consumer Goods: $1.20
Agent #2 – Spending – Capital Goods: $0.00 Consumer Goods: $0.00
Agent #3 – Spending – Capital Goods: $0.00 Consumer Goods: $0.00
Agent #1 – Income – Capital Goods: $1.20 Consumer Goods: $0.00
Agent #2 – Income – Capital Goods: $0.60 Consumer Goods: $0.60
Agent #3 – Income – Capital Goods: $0.00 Consumer Goods: $0.60
Period #2:
Agent #1 – Spending – Capital Goods: $0.72 Consumer Goods: $0.48
Agent #2 – Spending – Capital Goods: $0.96 Consumer Goods: $0.24
Agent #3 – Spending – Capital Goods: $0.00 Consumer Goods: $0.60
Agent #1 – Income – Capital Goods: $1.12 Consumer Goods: $0.00
Agent #2 – Income – Capital Goods: $0.56 Consumer Goods: $0.60
Agent #3 – Income – Capital Goods: $0.00 Consumer Goods: $0.88
I think it is apparent that eventually all agents will get back to spending and receiving $1.00.
If a third good is added, is it conceivable that given an uneven distribution of money at the outset, a good or agent is either eliminated or faces returns that decline over an infinite horizon? The obvious would be single good agents (all income and spending dedicated to one good), and single agent goods (all consumption and income for a good are the result of one agent).
But are there any non-obvious situations where an uneven initial injection of money causes a permanent change in relative prices?
LikeLike
Sorry ’bout that, I didn’t mean to single you out for fierceness, the adjective was intended as a modifier of the response of the commenters on Scott’s blog.
LikeLike
I don’t care if someone benefits first from QE. Everything government does benefits someone first, from handing out local trash contracts to national defense to the Fed.
Does the action also benefit the whole?
LikeLike
Benjamin,
Do permanent changes in relative prices benefit the whole?
LikeLike
Marcus, Thanks. I hope so.
Bill, Thanks for the link to Horwitz’s comment, and I pretty much agree with your take on it. My point, about the inability of banks to affect interest rates was primarily aimed at the longer end of the yield curve. The central bank can control the overnight rate and perhaps other rates up to say 90 days. But beyond that, banks can affect longer rates only insofar as they can alter expectations. Expectations of low short-term rates far into the future are possible only if real rates are expected to be low. Otherwise, the attempt of the central bank to keep short-term rates low far into the future must be self-defeating. I don’t think that it necessarily matters whether the central bank tries to implement its policy by pegging an interest rate or by an increase in the quantity of money. The question is whether the central bank can manage expectations.
PeterP, When I wrote that newly printed money enters the economy at “specific points,” I was quoting Sheldon Richman who was providing the Austrian take on monetary policy, so you should raise your point with him, not me.
Sorry for the tardy replies to comments. Will try to have further replies tomorrow.
LikeLike
Nick, The way that I am thinking about it, the short-run price in a stock-flow model is determined by the intersection of the stock demand with the fixed (in the short run) stock. Thus, prices fluctuate in response to demand changes in the short run even though a given technology may imply that the price will revert to a particular level in the long run. The slope of the PPF determines the rate at which supply adjusts to restore the long-run equilibrium price. The point is that the existence of a large stock relative to the flow implies a different price adjustment path from the one in which the flow is large relative to the stock. That is very different from saying that price is determined strictly determined by technology which implies that flows are (or can be) very large compared to the existing stock.
I agree that Austrians believe that prices are demand determined which rules out an equilibrium determined entirely by technological parameters. My point about Cantillon effects is that you need to specify a mechanism whereby there are predictable relative price effects associated with the injection of money. The Austrian view was that bank lending to entrepreneurs would shift the relative price of investment goods relative to consumption goods, or goods in process relative to final output.
On what do you base your assertion about Robinson’s confusion about the PPF?
Ritwik, Is it your view that commodities prices overshoot and undershoot in response to monetary shocks, or just that they tend to adjust to the shocks very rapidly. I gather that it is the former, but if so, you need to explain to me what accounts for this tendency to overshoot and undershoot, which it seems to me is not quite the same as a Cantillon effect.
I am generally on board with the assertion that an open-market operation does not necessarily have a systematic effect on the asset purchased, but that is a very tricky argument in any case because it is very sensitive to the effect of policy on expectations, which can hardly be reduced to a simple functional form. Beyond that I must admit that I don’t have a well-thought out view on how to distinguish between fiscal and monetary policy. There is obviously some room for semantic confusion there.
Becky, Thanks for your comment. You are very kind to describe what I do on this blog in those terms.
Nick, I totally agree.
Greg, As I have pointed out in some previous posts about Hayek’s monetary theory, his grasp of the process of money creation was not entirely free from confusion, following too closely the Currency School paradigm that was handed down to him from von Mises. The endogenous creation of shadow money and money substitutes is not necessarily destabilizing, but it can certainly amplify the monetary collapse associated with a deflationary disturbance even if it is not the trigger for that disturbance.
Where was Hayek’s Cambridge lecture published?
Greg, I am glad to know that you approved of this post.
dajeeps, Very astute observation. In principle, Cantillon effects ought to be symmetric with respect to expansion or contraction, but I am unaware of any discussion of the Cantillon effects associated with monetary tightening.
David, As indicated in my reply to Bob, I tried to clarify the reference to Bob in my post.
Frank, I think that Scott and Nick have tried to answer that question in the posts on the subject. A transitory wealth transfer can in principle have lasting effects on the distribution of income and wealth and thereby on relative prices under appropriate assumptions about individual preferences. But as Nick points out it is hard to come up with a case in which those wealth effects are large enough to make a big difference.
Benjamin, You have correctly identified the question that matters.
Frank, The question is whether there are effects other than a change in relative prices.
LikeLike
I recognize that this thread is incredibly old (for the internet), but I have some comments.
Someone asked why Cantillon Effects are never discussed in terms of declines in the money supply. First, I am not sure that that is true. Is not the Austrian cycle theory premised on the fact that eventually the money supply must (or does?) stop increasing, and then a whole set of new effects are unleashed. Maybe I missed something.
Second, Cantillon was living in an environment where the money supply was largely metallic, and the treasure fleets reached Spain and Portugal almost every year. The expansion of the money supply was a fact of life for Europeans for over three hundred years. We read the 18th century economists very stupidly when we forget to consider the institutional context. Moreover, Cantillon was living in an environment where nation-states were trying very hard to increase their share of that constantly but unpredictably increasing supply of money.
Unless you want to argue (as at least one comment seems to do) that Cantillon Effects eventually wash out in general equilibrium (whenever that nirvana arrives), I do not understand the argument that because it is obvious that some benefit and some are hurt, at least until general equilibrium is achieved, disparate impact is therefore trivial. Does that not depend on who is hurt and how severely? For example, if some starve to death because of a Cantillon Effect, they at least are likely to view it as far from trivial. Theoretical obviousness does not entail empirical triviality.
As for the issue that appears to have started this thread, that one does strike me as trivial. It appears fairly obvious that the immediate effect of a purchase of securities by the Fed is to raise the price of securities and thereby lower the rate of return on securities and that therefore the nominal wealth of those who are net investors is increased. That of course is the whole point of the exercise. But surely that is the shortest of short term views because eventually someone liquidates securities (or issues securities or incurs a debt) and purchases real goods and services. If no one does so, the increased value of the securities has no real effect; they are “paper” gains. It is the more distant effects that are of interest. My knowledge of Austrian interest theory is very superficial, but it seems to me that that the key part of their theory is that expansion of the money supply increases only demand for capital goods. If that were empirically true, then you can argue that an expansionary monetary regime benefits only those who are net investors, but it is obviously not true empirically.
LikeLike