This archived content originally appeared at, the predecessor site to, and does not carry the sponsorship of the Cato Institute.

Sumner v. Cantillon

(For Don.)

After trying a couple months back to defend the Austrian-School thesis that excessively rapid monetary expansion might give rise to what I termed "Intermediate Spending Booms," I promised myself that I'd keep out of the recent, related exchange between Scott Sumner and Sheldon Richman (among others) concerning so-called "Cantillon" effects. My inclination was, I daresay, only natural: after seeing commentators misrepresent my humble suggestion that a Fed policy involving a negative FFR target might perhaps have contributed just a wee bit to the subprime boom as (1) the simian proposition that that boom was entirely the Fed's fault and (2) the no-less absurd claim that the post-2008 collapse of nominal spending itself did no harm, I decided that this time I'd resist supplying raw material for more such libels by (for once) keeping my opinions to myself. (NB: Scott himself wasn't among the libelers.)

A couple nights ago, though, my good buddy Don Boudreaux wrote me asking, in effect, whether in denying Cantillon effects Scott had perhaps fallen off his rocker, and I promised to review the exchanges in question and to then give him my answer. And now, having done that, I just can't help sharing my conclusions. So much for resolutions.

The specific claim to which Scott objects is Sheldon's assertion that Fed open-market operations benefit those directly involved in them more than others because “early recipients—banks, securities dealers, government contractors—have the benefit of increased purchasing power before prices rise.” On its face that assertion appears, to me at least, as incontestable as the claim that, if I choose to do my Christmas shopping at Macy's rather than at Bloomingdale's, Macy's gains more from my display of Christmas spirit than Bloomingdale's does.

Nor can I see why it should matter whether the spending in question involves newly created money. Were I a counterfeiter whose products are perfectly indistinguishable from the real things, Macy's would still profit more from my spending than Bloomie's, provided it succeeds in fobbing the notes off just as easily as I do. The suppliers that Macy deals with are likely to profit as well, as may others who experience an increased nominal demand for their goods at stages of the "circular flow" not far removed from the fake notes' point-of-entry. Eventually, though, the notes will have worked their influence on prices generally, so that increased revenues, rather than going hand-in-hand with enhanced profits, merely compensate their recipients for a heightened cost either of living or of doing business.

How, then, does Scott attempt to refute Sheldon's argument? He does so mainly by resorting to two counterarguments, to wit: that Sheldon wrongly assumes that the Fed gives away new money instead of selling it, and that he confuses the effects of monetary policy strictly understood with those of what is properly regarded as fiscal policy.

"Richman seems to be assuming," Scott observes,"that OMOs are gifts of purchasing power from the Fed to the recipients.” Because the Fed actually sells new base money, Scott claims, initial buyers gain no more from it than anyone else, because they must part with other assets that are worth as much as the money they receive.

Much as I'm tempted to observe that the distinction between selling money and giving it away can get pretty darn blurry in practice (QE1, anyone?), it seems to me that Scott's position is unsound even putting that observation aside, for unless I'm missing something Scott here appears to neglect the basic truth that voluntary economic exchange is not a zero but a positive sum game. From that it follows that Primary Dealers, for instance, gain (or at least expect to gain) from their dealings with the New York Fed's Open Market desk no less than Macy's expects to gain from my doing my shopping there rather than somewhere else. They gain, moreover, even assuming that competitive bidding enforces a "zero-profit" condition, for that doesn't mean that such bidding rules out normal profits. And if anyone doubts that this is so, they would presumably have to insist that Primary Dealers, for starters, attach no particular importance to their status, and might indeed prefer to forgo it and let others go through the bother of selling stuff to the Fed since they would gain no less from its OMOs by waiting for new money to trickle its way toward them, and would do so notwithstanding the risk that the same money might in the meantime have raised the prices of their inputs.

But of course Primary Dealers do prefer dealing directly with the Fed to waiting along with everyone else for their share of enhanced Aggregate Demand. In suggesting that they shoudn't Scott appears (to invoke the terminology of Roman law) to overlook the crucial distinction between lucrum emergens and damnum cessans. Fed insiders alone experience the former, whereas the latter is the paltry reward typically granted by the Fed to hoi polloi.* (The reward is, of course, greater when the initial equilibrium involves a shortage of money–but Scott never claims that his arguments refer only to monetary expansions undertaken during a state of deflationary recession.)**

As for Scott's second counterargument, it seems to me to amount to nothing more than wordplay. Monetary expansion, Scott insists, is really "fiscal expansion" when it can be understood to involve an element of government largesse. Even a Friedman-style helicopter drop, according to this view, is really a form of fiscal rather than strictly monetary stimulus; monetary expansion in the strict sense of the term must take the form of conventional open-market purchases.

I've never had much use for a definition of "fiscal" policy that would have us believe that there's nothing "fiscal" about the Fed supporting the market for U.S. government securities. But in this case I fear the problem is more serious even than usual, for Scott comes perilously close to defining as "fiscal" any monetary operation that might have distribution effects of the kind Sheldon (and Cantillon) insist upon. Semantics aside, the real question isn't whether "pure" monetary expansions involve distribution effects, but whether most real-world monetary expansions have such effects, however pure or impure those monetary expansions may be.

As I conclude these remarks I already anticipate someone declaring in reply to them that I apparently believe that Cantillon effects are the only effects of monetary policy, or that Sheldon Richman is a better monetary economist than Scott Sumner, or that water flows uphill. For whoever does so, may the lamb of God stir his hoof through the roof of heaven, and kick you in the arse.

*Upon further reflection I concede that my emphasis on Primary Dealers is misleading, for when the assets that the Fed purchases are being actively traded in an organized market–as is the case with U.S. government securities, though it wasn't with MBSs back in 2008–all holders of such securities, and not merely those who deal directly with the Fed, gain from the Fed's purchases of them and their consequent appreciation. Here the analogy with shopping at Macy's also proves deficient, because an increase in demand for shirts at Macy's doesn't serve to immediately bid up the value of shirts everywhere. But the increase in the relative price of government securities, and corresponding reduction in their yield, ceteris paribus, is nonetheless a Cantillon effect, stemming from the initial injection of new money into a particular market; indeed, it is the most important sort of Cantillon effect in modern monetary arrangements, and one the existence of which is presupposed whenever the Fed employs an intermediate interest-rate target. (Note added on 12/9 at 8:45PM.)

**Upon still further reflection I conclude that most important determinant of whether or not Cantillon effects arise is not, as Scott maintains, whether or not the Fed gives money away, but precisely whether the economy is or isn't suffering from a deficiency of aggregate demand, with P > P,* when monetary expansion takes place. (Added 12/10 and 11:10AM.)


  1. I haven't followed the exchange so perhaps it's already been suggested that if the Fed issues a steady stream of new money, fully anticipated, there will be no Cantillon effect as prices will have been bid up to anticipate the new money. Only an unanticipated money supply shock would generate the effect. — Warren Gibson

  2. Here's a reply by Scott Sumner to Bob Murphy:

    "Bob, When the government buys lots of wheat the real price of wheat rises. When the government buys lots of bonds the real price of bonds usually falls (from inflation). So that’s not a good analogy."

    Inflation has to be a lot higher than it is now for this statement to be true. Inflation (or the purchasing power of the dollar) is not the only factor determining the real prices of either wheat or bonds. Supply and demand matter too; in the short run and in a low-inflation economy such as we have now (certainly compared to the 1970s), government purchases will raise the real prices of these c.p. Inflation has to be much higher for their real prices to decline when government is buying them.

    In another post he claimed that Austrians don't understand ratex. Someone else gave a great rebuttal. Why ratex remains part of the economic canon is a mystery to me. Maybe it's because it's a bit of a canon, with all that implies.

    All I can say about Scott's comment is that it's clueless, and that's being charitable. He really needs to learn about the real world, starting with financial markets.

  3. Thank you for weighing in on this. I too have a hard time understanding how anyone can insist that supporting the market for govt debt isn't "fiscal"!

    On an unrelated note: I wonder if you had any proposals for reforming the international monetary system (there don't seem to have been any posts here about it). I assume you oppose any moves toward expanding the IMF's role of lender of last resort or a greater use of SDR's as international reserves. But what alternative exists to using the dollar as the de facto international reserve currency (and the unbalanced stockpiling of dollar reserves by OPEC nations and China/Japan)? Also, wouldn't expanding the money supply, Sumner style, have potentially destabilizing effects on the int'l monetary system?

    What books would you recommend for understanding the int'l monetary system? I'm working through Eichengreen's "Globalizing Capital," and I plan to read his "Exorbitant Privilege." Any other good ones you know (including textbooks)?

    Any comments or recommendations would be highly appreciated.

  4. Suppose the Fed announced that it will increase the quantity of money 10% and prices instantly rise 10%.

    The Fed then actually purchases some bonds. It seems plausible that the primary dealers market is not perfectly competitive, and the particular primary dealer the Fed chooses benefits from the greater business. (In perfect competition, no seller wants to sell any more at the competitive price.)

    Note, however, that this benefit has nothing to do with getting the new money first before prices have risen. Prices have already risen before less than perfectly competitive primary security dealers earn their added profits.

    The more competitive are the primary securities dealers, the more those particular firms dealing with the Fed trade with the Fed, the less their traders have time to trade with others. What they gain on the bid-ask spread on the T-bill transactions, they come close to losing on other securities transactions. The further they are from perfectly transactions, the less that is true. That is, they have plenty of excess capacity to handle these extra transactions. Great, more business.

    If Macys and Bloomingdales are perfectly competitive, when the Fed shows up at Macys, customers see how crowded it is and instead go to Bloomingdales. Of course, high end department stores (like most real businesses) aren't perfectly competitive and they can handle more business and make more profit from it.

    Sumner said that these benefits to the primary dealers exist, and they are trivial.

    Most importantly, they have nothing to do with getting the money first before it works its way through the circular flow. They would still exist if prices rose before the primary security dealers actually received any money.

    Would they say, no, I don't want to deal with the Fed. The price level already increased. Too late. No.

    That the Fed has actual transactions costs from buying and selling securities, and deals with security dealers rather than operating its own retail (or even wholesale) securities operations, and that because of imperfect competition, the securities dealers make money on this, has nothing to do with money creation.

    Suppose the Fed had a policy of selling and buying equal amounts of securities every day, leaving the quantity of money same. (Not a fanciful notion, is it?) If the securities market is imperfectly competitive, the security dealers would find this policy very attractive, even though there is no net money creation. (Do you want to claim that those dealers that buy from the Fed lose money because they give up the money first?

    In fact, consider money contractions. The Fed sells to primary security firms. The security dealers give up money in exchange for securities which they are buying on the market. Do they lose money because they give up the money first? Or do they make some minimal profit from providing the same service to the Fed they provide to everyone else buying and selling securities?

    This framing of Austrian economics–some people get the new money first and benefit because they get the money before it loses value–creates confusion.

    Rejecting that framing doesn't mean that there are no injection effects.

    Anyway, Sumner's actual argument would be the following: The Chinese decide to buy an additional 10,000 tons of American wheat. Do the particular wheat famers that sell their wheat to the Chinese benefit more than the other American wheat farmer? Do those who continue to sell wheat to Japan benefit less? Do those who continue to sell to Americans benefit less?

    When the Fed buys Treasury bills it is like the Chinese buying wheat. The particular individual that sells the T-bills to the Fed benefits no more or no less than anyone else who is selling their T-bills to other buyers at the same time.

    Sumner admits that under some conditions the prices of T-bills might rise when the Fed buys them, though he also makes lots of caveats and points out that it doesn't necessarily happen. But if it does happen, those who happen to sell their T-bills to the Fed do no better than those who sell T-bills at the same time to other people.

    Suppose the Fed announces that it will buy a bunch of T-bills. The result is everyone expects the price to rise. The offer prices rise, and the Fed buys some of them with "new money" and other people buy some with "old money." Do the sellers receiving the "new money" from the Fed do better than those sellers who sold T-bills to other people for "old money?"

    In my view, it is time to drop this way of describing the impact of money creation on interest rates and the demands for different goods.

    1. I think it's generally recognized, Bill, that the effects of anticipated monetary expansion aren't the same as those of unanticipated expansion, and that the former may indeed, under certain conditions, be nil. But I take Scott's argument to be the more ambitious one to the effect that Cantillon effects are not relevant even for the case of monetary expansion that isn't fully anticipated. It is that more sweeping claim to which I intend to reply here.

      1. My point is that with an anticipated expansion, the effects you describe, not only occur, and only a trivial extra effect could exist because prices have failed to completely rise to their new equilibrium.

        Because of realistic institutional assumption, Sumner calls those effects "fiscal policy." He assumes the government captures the seignorage, so what the government does with it is fiscal policy.

        To the degree prices have failed to rise to their new equilibrium level, there may well be some differential impact on demand. But how likely is it that they much like the effects you identify, effects that have nothing to do with disequilibrium?

        Sumner said that the particular person selling T-bills to the Fed doesn't especially benefit by receiving "new money" relative to others selling T-bills at the same time (for "old" money.)

        If the Fed gave me money, as a gift, I would benefit more than people obtaining "old money" by selling things.

        But the Fed doesn't give money as a gift when it does an open market operation.

  5. P.S.

    The U.S. runs a budget deficit. The Fed sells T-bills for the Treasury to the primary securities dealers. The Primary Security dealers now have all of these securities which they sell, earning money on the bid ask spread. They sell some of them the Fed. Why does it make any difference to them that they sell them to the Fed rather than some other buyer?

    Suppose we had no deficit, but a national debt as old bonds came due, new ones had to be purchased. The Treasury is constantly selling new bonds to the primary security dealers, who then turn around and resell them to investors. Do those primary securities dealers who resell to the Fed earn more profit than those who resell to other investors?

    So what is the assumption for the benefits from the primary security dealers? Everyone is holding the T-bills they want. The open market operation is new trading that wouldn't otherwise occur.

  6. "In suggesting that they shoudn't Scott appears (to invoke the terminology of Roman law) to overlook the crucial distinction between lucrum emergens and damnum cessans."

    Not to be pedantic, but the terms are mixed up here–they are lucrum cessans and damnum emergens. In international law these are two ways of measuring damages for expropriation–whether the compensation should be paid for lost profits or gains (lucrum cessans) or merely for the fair market value of the tangible property taken (actual loss, or damnum emergens). E.g. the LIAMCO case goes into these (Libyan American Oil Company (LIAMCO) v The Libyan Arab Republic, ). Though I have always been suspect of the validity of this positive law classification and distinction. For a taking of property, of course full, adequate and prompt compensation should mean accounting for lost profits. Not that expropriating states like this idea.

    1. They aren't mixed-up, Stephan: although you refer to the more common expressions, the circumstances I refer to call for their opposites. Lucrum emergens means newfound gain; damnum cessans means losses extinguished.

        1. Well, the formulations were mine for only as long as it took for me to affirm their actual occurrence in the legal literature. Would that it could be easier for me to be truly original!

  7. "Suppose the Fed announced that it will increase the quantity of money 10% and prices instantly rise 10%."

    This assumes the impossible, so Cantillon effects don't exist. That's one approach, but in the real world, the one Cantillon described, money spreads throughout the economy over time and the first recipients spend it before later recipients. The relative prices the former pay are lower than those paid by the latter.

    1. How is it impossible?

      I didn't say it was likely.

      My point is that if it did happen, and the Fed was buying retail goods, then it would still be the case that Bloomingdales would benefit marginally if the Fed purchased from them rather than Macy's.

      The money enters at a certain place and those receiving the new money first benefit because they receive the money before prices rise is confusing. I think it is because it balls up a bunch of different things.

  8. Those who deplore the effects of a credit money bubble bust – should, logically, oppose the creation of the credit money bubble in the first place.

    Being upset about (for example) the bust of 1921 without being upset about the First World War credit money expansion – makes no sense.

    Nor does being upset about the 1929 credit money bust – without being opposed to Benjamin Strong's backing of the monetary expansionin the late 1920s that caused the boom-bust.

    Nor does it make sense to be upset about the current crises (which has hardly started yet) without being opposed to the policy of Alan Greenspan of backing (again and again – refusing to allow it to bust, each time making the bubble bigger) the credit money expansion that created the crises.

    As for money lenders – whether individual money lenders or banks…..

    If they lend out real savings (whether their own – or those of other people entrusted to them) they can not create a mass boom-bust event.

    They can lend out the savings very foolishly (for example to finance consuption, such as house buying from people who can not really afford the houses they are buying, rather than productive investment) in which case they are likely to go bankrupt – but they CAN NOT create a mass boom-bust event.

    As Richard Cantillion pointed out – only by monetary expansion (NOT just by lending out real savings) can a boom-bust event be created.

    What should be done in the case of a bust?

    Nothing interventionist.

    The banks and enterprises dependent upon them should be allowed to go bankrupt (i.e. no "suspension of cash payments" or "efforts to fight the deflation") i.e. close their doors.

    Wages (along with prices) must be allowed to crash – so that markets can clear (credit money shrink back down to the monetary base), and people have to then rebuild economic activity from the ground up. It was, of course, the refusal of Herbert "The Forgotten Progressive" Hoover to allow wages to crash in the wake of the bust of 1929 (he was a victim of the "demand" fallacy) that turned the bust of 1929 into the Great Depression.

    The only difference between the situation in 2008 and (say) the situation in 1921 was one of SCALE.

    All the above is, of course, obvious.

    What is not obvious is what (if anything) should be done to prevent the "boom" in the first place – the only way of really preventing the "bust" (the slump).

    Obviously such things as the Federal Resever should not exist (they should be closed – at once, today), and no support (of any kind – whether bailouts or corrupt court judgements) be given to a bank in trouble.

    But how does this prevent all bubbles? It does NOT.

    After all there were boom-bust events (bubbles) before the creation of the Federal Reseve in 1913.

    An end to government backed "suspension of cash payments" (which go back even before the Civil War – with insolvant banks being declared "not insolvant" by government courts) and an end to government bans on "discounting" the various forms of favoured bank(not all banks were favoured – only the politically connected New York "National" banks were favoured) debt paper…..

    This might have a effect of making bankers think twice before setting out to build credit bubbles (i.e. lendng out "money" that no one really saved).

    As might a total seperation of "banks and state" – as Martin Van Buren (as far as I know the only professional banker to become President of the United States) suppported. No national bank – and no Andrew Jackson style "pet" State banks either.

    No tax money (at any time) going to banks. For them to "look after for awhile" (i.e. build a vast inverted pyramid of debt upon).

    People paying their taxes – in money.

    And the government storing the money (itself) and paying it out as it spent it.

    No subsidy for bankers at any time – nothing for them to build their inverted pyramids of debt upon.

    However, bankers tend to be very intellgent (although, at the same time, very unwise) and will doubtless find other ways to build credit bubbles.

    Just as bankers (with ease) found ways round the British Prime Minister Sir Robert Peel's Act of 1844 – and found new ways to loan out "money" that no one had ever really saved (thus creating boom-busts).

    It is not, I believe, within the wit of man to prevent this folly totally.

    All one can do is to limit its SIZE – by the government refusing to subsidize or support this activity (in any-way-shape-or-form) and, when it collapses into bust, just standing aside from it – and allowing the market to clear (by the crash of prices and wages) and economic development (real economic development – based upon real savings) to rebuild.

    If the credit-money bubble is small – then the crash will be small.

    If it the bubble is vast – then the crash will be vast.

  9. It is amazing how complicated it can be to understand or explain the effects something so simple as the central bank buying some bonds.

    IMHO the real problem with our monetary system is that it requires central bankers influenced by politicians and the therefore the median vote to guide the supply of money. If even macro-economists can disagree, meaning at some are getting it wrong, the politicians and voters can very easily push in the wrong direction. In free banking even the bankers do not need to understand how the system as a whole works, just their little part.

  10. I have no problem with the argument that the Fed's newly created money directly and disproportionately affects certain prices (i.e. treasury yields, assuming the Fed adheres to a "treasuries-only" policy). I imagine Dr. Sumner concedes that much. My impression, however, was that he was arguing so long as the money creation was necessary to keep prices stable or meet the Fed's 2-3% inflation target it would not have the unwarranted "ripple affect" Cantillon and others refer to and it would be justified to prevent a decline in NGDP.

    Comparing this to a Free Banking ideal, we would not oppose it if banks were to uniformly increase the supply of money in response to an increase in the demand to hold money (fall in velocity). That new credit would no doubt push interest rates down in whatever market the banks lent the new money into, but we wouldn't say that's an inherently bad redistribution of purchasing power. Instead, we'd argue it was a necessary response to maintain a steady flow of overall spending, and to the extent rates were influenced they were influenced by a market process that allowed banks to allocate credit where they saw most appropriate(banks surely wouldn't adhere to any sort of "treasuries only" policy).

    This is why I tend not to criticize the Fed's normal open market operations for its Cantillon Effects unless its actions result in higher than anticipated inflation. Yes, the types of assets the Fed purchases have a direct affect on certain prices, particularly treasuries, that wouldn't occur in a free banking system, but so long as we have a monopoly central bank such effects would be necessary for preventing the far greater redistribution associated with a decline in nominal spending.

    Please correct me where I am mistaken. And thanks for the informative post and discussion.

Comments are closed.