Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Tuesday, September 13, 2011

What makes a central bank special?

Nick Rowe over at WCI has an interesting post asking what makes a central bank special; see Currency, Interest, and Redeemability.

According to Nick, what makes a central bank special is not that it can borrow and lend, create money, and set interest rates. All of us can do these things in principle; and in practice, many large private financial institutions do do these things. The difference, as Nick explains, is what he calls "asymmetric redeemability." The Bank of Montreal, for example, issues money redeemable in BoC liabilities; but the reverse is not true.

I think that's right; but let me expand with a few related thoughts of my own.

If one looks at history, a recurring property of monetary economies is the emergence of a "base money" that serves as the redemption object for "broad money" objects. Frequently, the broad money is made redeemable, on demand, and at par, with the base money. (We might even call base money "central" money, as it serves as the focal point for all other monies.)

In modern economies, the demandable liabilities created by chartered banks (M1, say) constitute broad money made redeemable, on demand, and at par, with government cash (small denomination paper). In the antebellum US, private banknotes were made redeemable on demand at par for specie (gold and silver coin). And so on throughout much of recent monetary history.

Now, there are a lot of interesting questions that arise here that are not the main focus of my post here. For example, why do banks embed their liabilities with what amounts to be an American put option (liabilities made redeemable on demand at a fixed strike price)? Is it the byproduct of a legal restriction, or is it an equilibrium phenomenon?

Either way, it seems obvious that the agency in control of the supply of base money (the object of redemption) is going to be in a position to implement "monetary policy." Whether this is a good or bad thing is the subject of much debate of course (let's not get into that here). And so, I have to agree with Nick's conclusion:

There is something very seriously wrong with any approach to monetary theory which says we can assume central banks set interest rates and ignore currency. It is precisely those irredeemable monetary liabilities of the central bank (whether they take the physical form of paper, coin, electrons, does not matter) that give central banks their special power.

(He is, however, not quite right about private agencies not being able to issue irredeemable objects and get them to be used them as money; see Bitcoin).

Scott Sumner, who runs a very nice blog himself, has an interesting take on Nick's post; see: The Bank of Canada is Important Precisely Because it is not a Bank.  Here is Scott:

Now let’s consider a different monetary system.  Imagine a gold standard and a monopoly producer of gold.  The gold mine company would reduce short term interest rates by increasing the supply of gold.  Like currency, gold is irredeemable.  But no one would call this gold mining company a “bank” because it possesses no bank-like qualities.  Banks don’t create irredeemable assets, gold mines do.

Scott is trying to tell us that a central bank is not a bank; it is the monopoly supplier of base money. He also says that "banking has nothing to do with monetary policy."

I find it difficult to evaluate this view because he does not define "bank" or "monetary policy" here (although I'm sure he does elsewhere). Permit me once again to give my 2 cents worth.

A financial intermediary is an asset transformer. An insurance company, for example, takes "deposits" (premiums), purchases assets, and creates a set of state-contingent liabilities backed by these assets. A pension fund, as another example, takes "deposits" (contributions), purchases assets, and creates a set of time-contingent liabilities backed by these assets. A bank, finally, takes "deposits", purchases (or finances) assets, and creates a set of demandable liabilities backed by these assets.

So a bank is a special kind of intermediary. It is special because the demandable liabilities created by banks are used widely as payment instruments; i.e., money (and the demandable property of these liabilities probably goes a long way to enhancing their acceptability as money, but that's another story).

When a bank accepts your land as collateral for a money loan, it performs an asset swap. It is transforming your illiquid land into a liquid asset (the liability created by the bank). Banks are in the business of transforming illiquid assets into liquid assets. This leads us to ask what the Fed is doing when it purchases (say) MBS or UST assets? In my view it is transforming (relatively) illiquid assets into a very liquid asset (Fed cash). QE is banking; and it falls under the category of monetary policy, in my books at least.

So in some sense, all banks (private and public) are engaged in a form of  "monetary policy." But it still remains true that a central bank with legislated monopoly control over the economy's base money object is "special" precisely for this reason. And any theoretical framework that is to have any hope of ever understanding the role of money and banking in society is going to have to model these objects explicitly; the way this guy does, for example. 


  1. David: thanks! Yep. Basically agree with most.

    Bitcoin. I don't see anything impossible about private irredeemable monies getting off the ground. "Hippy monies" like LETS are another example, that sort of work. But somehow, if the Bank of Montreal said it was offering its customers the opportunity to swap their existing demand deposits for irredeemable new BMO liabilities, I just don't see it working. Otherwise, insolvent & illiquid banks could keep on running fine. Which they usually can't.

    As to *why* the Bank of Canada can get people to accept as money its irredeemable liabilities, and BMO can't...yep, that's a whole other question. Like you, I see focal points as having a lot to do with it. We expect people to use whatever they used in the past, so we choose to use it too.

    And it's good that people like Steve are trying to model all this explicitly.

    (Paul Krugman says he believes you can't do macro without "frictions". Steve is the guy trying to model some of those same "frictions". Sometimes I think those two should stop fighting and get a room. ;) )

    Suppose the Bank of Canada threw away all the assets on its balance sheet. And just kept the printing press. Would it still be a bank? Not really. Would it still be able to do monetary policy? Yes. The only limitation is that it wouldn't be able to reduce the stock of currency by an Open Market sale of bonds. It could still decide how much to increase the stock of currency. (It would find it harder being Lender of Last Resort).

  2. It really is important to understand the mechanisms and the accounting here, and any first year Canadian macro textbook should be explaining the details of this. It's a shame so much of the web doesn't bother reading basic explanations... I had thought a simple white paper on basic Canadian banking laws and regulations would be something the Bank of Canada could have available on its website.

    The government has a large account at the Bank of Canada as well as chartered banks. One of its primary operations is to transfer funds between the Bank and chartered banks, and only it can do that. Charter banks are also required to keep some deposits on hand at the Bank of Canada. As a poster on Nick's blog stated, it matters that the Bank of Canada is the government's primary bank and the government is the Bank's primary customer.

  3. "Suppose the Bank of Canada threw away all the assets on its balance sheet"

    The accountants would want a word if this were the case. The accounting identity must always hold.

  4. jesse: are you telling me it is logically impossible for the Bank of Canada to give stuff away? Because this would violate an accounting identity?

    The BoC (and all the other banks) needs to post this on its door:

    "Sorry. We cannot make charitable donations. Our accountants tell us this would be logically impossible, because we would be violating an accounting identity if we did this."

    This is how accounting misleads people. It gets people to say things that contradict all common sense.

  5. Nick:

    Actually, it's the concept of fiat money that contradicts common sense. It starts by denying that liabilities are liabilities, then denies that assets matter to the value of those liabilities, then denies the no-free-lunch principle by asserting that the issuer of 100 paper dollars gets a $100 free lunch. All of this because people are misled into thinking that inconvertible=unbacked.

  6. @Mike:

    Fiat money does not contradict common sense if you think about it as a record-keeping device. Your personal credit history is "fiat" in the sense that it has no intrinsic value. And yet, companies will devote resources to collecting such information. Does this not seem to be common sense? And if a monetary token embeds within it information relating to a person's past contributions, why should such an object also not have value? (Having said this, I agree with the last thing you said; i.e., that people commonly confuse convertibility with backing.)

    @Nick: You made me chuckle with that image of Paul and Steve in a room together. Yes, even without the ability to engage in asset swaps or lending operations, the central bank could still undertake a form of monetary policy. But monetary policy, the way I define it, may take on more than one dimension. The ability to control the outstanding supply of base money is certainly one of these dimensions. But not the only one, in my view.

  7. David:

    Maybe I'm not following you. People follow my credit history because it tells them whether I pay my bills, and of course that information is valuable. If some token says "Mike harvested a bushel of wheat yesterday." then that token has value only if it entitles its bearer to a bushel.

  8. Mike:

    You are not following me, but something tells me that you are close.

    Your last sentence is a proposition that requires proof; it is inadequate to simply assume it, or to point to examples in history.

    As a matter of a theoretical proposition, I can show you a mathematical model economy that contradicts your last sentence. Ergo, it is logically possible for tokens to have value. The theory suggests that the value is closely tied to the value that a credit history possesses; it constitutes information.

    When I buy my morning coffee, I hand over a token ($) and someone goes through the effort of brewing me coffee. Presumably, I earned that $ somehow (I sacrificed my time delivering lectures, for example). Now the coffee brewer has the $. That $ (according to the interpretation of the theory I am explaining) is now a record of their own sacrifice. They may go use that $ to buy shoes; and so on.

    In this way, the token $ can represent information relating to personal contributions (acts of production) for members of society.

    Note: I am not saying that this is the only way that monetary objects can possess value. Surely, objects of intrinsic value will possess value relating to their backing. But when liquidity is scarce (this is a well-defined notion in our models), then even asset-backed monies will exchange at a "liquidity premium." The premium reflects a scarcity of good collateral assets; or more fundamentally, a scarcity of commitment. It is difficult to get people to make good on their promises.

  9. The money as memory idea is an interesting one. IIRC it goes back at least to Hicks.

    But I'm a bit dubious about using that idea to explain the *value* of fiat money. It sounds to be like it commits the "functionalist fallacy". "It would be a good and valuable thing for society if fiat money existed. Therefore each individual will value fiat money". Not sure. (That probably wasn't clear, anyway.)

  10. So, you might be interested in this:

  11. David,

    I think there is something different about a central bank exposed to a "capital call". Irredeemability would imply that a CB has no liquidity risk. In fact, the function of a CB is partly to extinguish liquidity risk held by the private sector when that risk spikes.

    However, even its liabilities are irredeemable, that doesn't mean a CB has no solvency risk. A CB balance sheet made up primarily of interest-paying reserves and risk assets might experience a negative net worth. Nick Rowe argues this should not happen given the NPV of future seigniorage profits. I assume he means seigniorage on currency. However, when interest paying reserves are large relative to currency, the NPV of the CB may be negative.

    So what are the implications of CB negative net worth? I would say that the CB has a call on Treasury for its equity. Thus, any activity that increases CB non-liquidity risk actually just transfers that risk to the Treasury, and back to the private sector (through taxes). Therefore, a CB cannot change the private sector's aggregate duration or credit risk -- only liquidity risk.

    In short, a CB is unique in part because it can never be illiquid. It can be insolvent, but only if it engages in large QE programs. Implications?

  12. BTW, my conclusion from the above is that Central Bank buying of risk assets should much not affect their price except when liquidity premia are abnormally high. The CB is just shifting ownership of a set of cashflows from one group (investors) to another (taxpayers). How does this shift add/subtract value to those cash flows?

    This assumes the CB creation of reserves adds zero value; i.e., they wind up in Excess Reserves.

  13. Nick: When I asked Joe Ostroy about it, he claimed to have come up with the "money is memory" idea independently (he was not aware of anyone else having done it). In any case, I have no idea what you are saying (haha). A "functionalist fallacy"? People are risk-averse and have idiosyncratic shocks. Society would value insurance. Therefore insurance will be valued. Is this a functionalist fallacy too?

    David Pearson:

    In short, a CB is unique in part because it can never be illiquid.

    I like that. But I wonder whether it is, strictly speaking, true?

    The liquidity of an asset depends on how easily people accept it as payment. If people begin to distrust central bank money, they will discount it, just like any other asset they distrust. It may become illiquid as people flock to currency substitutes. Of course, if there are laws preventing currency competition, then what you say sounds true. But then, this is just saying that monopoly control over the base money is what makes central banks special. No?

  14. David (and Nick):

    I expect those math models of yours assume that rival moneys are partly or completely suppressed. Otherwise any token whose value exceeds its backing would attract rival tokens until value=backing. So maybe gold, silver, tobacco, etc. can have a small monetary premium (<20%?). But that premium would attract rival moneys like salt, cows, etc. until the premium was small. When you throw in the fact that people will trade with IOU's that also serve as rival moneys, it's hard to see how any monetary premium could be more than 1% or so.

    I'm with Nick on that "money is memory" thing. It sounds to me like a sunk cost fallacy.

    I didn't know you knew Joe Ostroy. I see him a few times a year at UCLA.

  15. Mike:

    Evidently, you have a rational expectation. It is true that factors that compete with fiat currency are suppressed. In particular, an implicit assumption (as it appears to me, but some may disagree) is that a uniform government money is less susceptible to counterfeiting (relative to say, thousands of competing currencies circulating side-by-side.)

    But if we free up private money in these models, it remains the case that private monies trade at a liquidity premium, if liquidity (commitment) is scarce. that I think about it...a role for government money (or debt) remains in these models even if private monies are allowed to circulate (the economy may be dynamically inefficient, for example).

    Joe is a great guy and a great economist. Please say hello to him for me!

  16. David:

    "if liquidity (commitment) is scarce"

    Your "if" is a great peace maker. What if liquidity is not scarce? What if people can easily trade with computer blips, bits of paper, wooden tallies (like in the middle ages), clay tablets (ancient Babylon), etc.? Then wouldn't your models imply zero monetary premium?

  17. Mike,


    But do you really believe that liquidity premia are zero? Think of the shares of large cap stocks relative to small cap stocks. I think that the large volumes on GM shares may confer GM shares a liquidity premium, though how large, I do not know.

  18. David:

    Credit card companies charge about 3% for the liquidity services they provide, so call that a good first guess of the liquidity premium for any kind of money. That means that the Fed, for example, could issue a dollar that trades on the street for 1 oz. of silver, while the fed only holds assets worth .97 oz. That .03 oz premium might be just low enough that it's not worth anyone's trouble to introduce a rival money of some kind (checks, credit cards, gift certificates, eurodollars, foreign currency, etc.)

    If that 3% figure is right, then the dollar is 3% fiat and 97% backed. So economists should be focusing 97% of their attention on the backing theory and 3% on the quantity theory, instead of the current levels of 99.999% on the quantity theory and .001% on the backing theory.

  19. David,

    My point was that the risk of insolvency might interfere with the benefit of irredeemability. CB's add value through the latter, they subtract through the former. There is something about paying interest on large swaths of balance sheet liabilities that makes a CB more of an ordinary bank, and less "special".

  20. David:

    Let's say the liquidity premium is 3%. That seems small enough that it might not pay rivals to introduce competing moneys like credit cards, checks, gift certificates, eurodollars, foreign currency, etc. That means money is 3% fiat and 97% backed, so economists' emphasis on fiat money is misplaced. (A post I made an hour ago might have been dropped, BTW.)

  21. Mike:

    You claim that liquidity premia are generally small; at least in financially developed economies like the U.S. You may be correct (I have no idea, though I wish I did). But even if these premia are small, it doesn't mean that they are not important. Ricardo Lagos (NYU), for example, that asset pricing models that incorporate liquidity premia are better able to explain asset price movements.

    You seem to be suggesting that if the liquidity premium on an asset is large enough, that competition will bid it away. This is a proposition that needs to be proved, not assumed. If commitment is in short supply (say because people cannot be trusted to make good on their promises), it does not magically follow that competition will supply that commitment. It may try, but commitment is a scarce resource. The constraint may bind.

    One interpretation of the emergence of MBS in repo up to the crisis is that it represented the private sector's response to high Treasury prices (low interest rates). It is true that the AAA rated tranches of MBS seemed for a while to displace Treasuries as collateral for repo, but the crisis revealed just how difficult it is to manufacture good collateral objects.

    Finally, yes, a previous post of yours was sent to the spam folder. It is now recovered. Sorry about that. Still don't know why that happens!

  22. David:

    As an example, the British pound of 1796 was convertible into about .25 oz. of gold. Nobody would have called it fiat money, and if the paper pound had had a liquidity premium, so that it traded on the street at .26 oz, then arbitragers would have pounced.

    Convertibility was suspended in 1797, and the value of the pound stayed the same for a few years, only gradually losing about 10% (maybe 20%, I forget) of its value by 1810. Where was the liquidity premium? The pound certainly had none when it was convertible, and since the B of E's asset to liability ratio was not much changed from 1796 to 1797, it seems that no liquidity premium developed during the suspension.
    The idea of fiat money developed during this time, mostly due to Thornton and Ricardo, who clearly committed the error of thinking that inconvertible=unbacked. When convertibility was restored in 1821, their own logic would have said that the pound had switched back again from fiat to backed, but the obvious explanation is that the Bank's assets were there all along, and the pound was backed by the Bank's assets all along, with no trace of a liquidity premium.

  23. @ David,

    There has always been an active repo market for agency MBS (which have always been implicitly guaranteed by the government, and retain that mysterious status post-crisis). What I think you're referring to is the repo market for structured credit, inc. non-agency MBS, CDOs, ABS, etc.

  24. David: "People are risk-averse and have idiosyncratic shocks. Society would value insurance. Therefore insurance will be valued. Is this a functionalist fallacy too?"

    Yes, I think so. We can imagine societies where people are risk averse, and have idiosyncratic shocks, but insurance isn't valued. We need somebody to come up with the idea of insurance, and start selling it. Or something like that.

    There are things that would be good and valuable if we all did them, but that doesn't necessarily give each individual an incentive to do them.

    For example, if I know that a note is counterfeit, and so provides a "false memory", I might still accept it if it's in my interest to do so, even though it makes the value of money less, both in the normal sense of "price", and its value as memory in helping the economy work.

    The thing I like about the Menger/von Mises story of money, and fiat money, is that it tries to explain everything from rational *individual* behaviour.

  25. @Mike Sproul: OK, but I'd rather sit down and discuss this with you over a pint. :)

    @Anonymous: Yes, you are correct; that is what I was referring to.

    @Nick: I see what you mean now. The only thing I'll say is that in the "money is memory" explanation, it is individually rational for agents to use money for its record-keeping properties. Anyway, no big deal here. I want to discuss your recent posts at WCI, but am struggling to find the time...

  26. David:

    If I'm ever in St. Louis, or if you're in LA, I'll hold you to that.

  27. Nick: "are you telling me it is logically impossible for the Bank of Canada to give stuff away? Because this would violate an accounting identity?"

    Well if I were an accountant, yes it would be. Maybe I don't understand how things work, but are the Bank of Canada's books unbalanced? I don't think they are, provided you attach value the paper they hold for the currency they produce.

    I am pointing out a difference between running "printing presses" and what the Bank of Canada actually does. It could logically leave a wad of newly-printed $20 bills on a bench on Capital Hill but it doesn't do that.

  28. I need to do more research on central bank. My friends used to work for it and they all were telling me about their mixed experiences in this bank. Thank you David for this article. By reading your article I'm convinced that central bank is surely one of the best banks.

    HCBL Bank - Tathastu