Wednesday, December 5, 2012

Richard Cantillon on Cantillon Effects

There's a dustup between market monetarists and Austrians over Cantillon effects. See Nick Rowe, Scott Sumner, Bill Woolsey, and Bob Murphy. What are Cantillon effects? One definition is the effect that a change in the money supply has on the real economy due to where money is injected. Rereading Cantillon, I think its better to define the effect he is writing about as the influence that a change in the money supply has given that people are incapable of anticipating that change.

Cantillon wrote in a world in which huge discoveries of gold in the Americas had steadily increased the price level. We know that if people perfectly anticipate the arrival of new gold, all prices will immediately rise. Cantillon thought somewhat differently. According to him, the initial discovery of gold would go unnoticed by people:
It is also usually the case that the increase or decrease of money in a state is not perceived because it comes into a state from foreign countries by such imperceptible means and proportions that it is impossible to know exactly the quantity which enters or leaves the state.
Il arrive aussi d'ordinaire qu'on ne s'apperçoit pas de l'augmentation ou de la diminution de l'argent effectif dans un Etat, parcequ'il s'écoule chez l'Etranger, ou qu'il est introduit dans l'Etat, par des voies & des proportions si insensibles, qu'il est impossible de savoir au juste la quantité qui entre dans l'Etat, ni celle qui en sort.
In his paper on Richard Cantillon, Michael Bordo echoes this:
In Cantillon’s work, the dynamic path of adjustment of relative prices, output, the interest rate, and specie flows depends on the expectations of agents in the various markets. This emphasis on expectations presages much of modern monetary theory. It is unclear exactly how expectations are formed in his scheme but the frequent examples of agents catching on slowly suggests that they are formed adaptively. Moreover, the repeated examples of people being fooled suggests that the availability and cost of information is an important aspect of Cantillon’s scheme. Such an emphasis antecedes modern macro theories of disequilibrium.
Cantillon then goes on to describe how unanticipated gold inflows would first be spent on food, forcing up food prices and the earnings of farmers. Farmers in turn employ more land, forcing up land prices. While all prices have now adjusted to the change in the money supply, during the adjustment period landowners are relatively disadvantaged since the price of their product is the last to increase.

While we don't have to agree with Cantillon's ordering of effects, it seems uncontroversial to assume that if expectations only adapt slowly, then there will be some sort of distributional effect during the adjustment period to an unanticipated change in the money supply. There can certainly be debate over the size and consistency of this effect. Austrians, for instance, build a business cycle theory out of it. Others consider the effect to be ephemeral.

On the other hand, if rational expectations are assumed from the start, then the location of gold's injection point is moot since everyone perfectly anticipates the repercussions and adjusts. In talking about injection points under rational expectations, it seems to me that market monetarists are having a totally different conversation than Austrians, who are interested in injection points under imperfect expectations. Is this just a debate over the nature of expectations? I see that Bryan Caplan has made the same point.


  1. The rational expectations view goes a step further and says that in any given period, people will either overestimate or underestimate how much new gold will arrive. Thus prices will not react to new gold in any systematic or predictable way.

    But when the focus changes to paper money, the creation of new asset-backed money won't affect prices anyway, so Cantillon effects take on a whole new insignificance.

    1. "the creation of new asset-backed money won't affect prices"

      What happens if the new paper money is being used to overpay/underpay for assets, and people are slow to realize this?

  2. Since people will make unpredictable errors in guessing whether/how much the issuing bank underpays/overpays, I wouldn't expect any Cantillon effects.

    1. Let's say that a clerk at the central bank sets half its bonds on fire and they're gone forever. Does the market value of the central bank's liabilities immediately fall? What happens if the clerk keeps it secret but just tells one friend? The clerk and the friend immediately sell their notes. Then the friend tells everyone else about what happened and their price plummets. The Cantillon effect is that the clerk and the friend have enjoyed a relative gain.

    2. If people know, then the market value of the liabilities immediately falls. If it's secret, nothing happens. (Except that people don't put their money into such banks in the first place) But given the conditions you state, there is a Cantillon effect.

  3. JP:

    The effects of a change in monetary policy will depend on whether or not it is anticipated. Yep, that's uncontroversial.

    The effects of a change in fiscal policy (e.g. a switch from buying one good to buying another) will also depend on whether or not it is anticipated.

    The anticipated/unanticipated distinction seems to me to be orthogonal to the question of exactly where the new money enters the economy.

    If "Cantillon effects" really means "the effects of unanticipated vs anticipated money", count me in as a strong believer.

    1. Nick, it could be that over the years "Cantillon effects" have come to mean something different from the effects-that-Cantillon-wrote-about.

      It seems to me that Market monetarists (Scott Sumner for sure) employ near rational expectations when it comes to financial markets and slower expectations when it comes to goods markets. Austrians seem to employ slow expectations even to financial markets so that they can motivate the interest rate distortion that they like to talk about. Do you think that's a fair characterization? Still trying to dehomogenize everything.

    2. Even if I expect the Fed to create $100 trillion of new money next year, which will be injected specifically into the primary dealer's bank accounts, there is very little I can do to offset the depreciated money I will soon own. It's not like I or my employer or his customers have the physical ability to raise our selling prices (for labor, goods, etc) sufficiently to offset it. There will be relative gains and losses from the inflation.

      Expectations only deals with the specifics of who will relative gain and lose, and when. It cannot eliminate the category "relative gains and losses" from taking place.

    3. Major Freedeom, if certain prices are sticky then I can see relative gains and losses.