Thursday, April 29, 2010

The Use of Currency

There is an interesting piece a here on new cell-phone technologies, that allow individuals to make small decentralized transactions without currency. I like to tell students stories about why we still need some currency in circulation. These stories often involve small hot-dog vendors who won't take credit cards, but clearly that could become a thing of the past, along with the problem of splitting up the dinner bill among 12 people. Certainly the experience of us relatively wealthy people is that currency is much less useful than in the past - few retailers will turn down a credit or debit card, and even the short trip to the ATM may not be worth the bother. However, if we look at the data, the use of currency in the world is remarkably persistent. I didn't take the effort to give you references to these studies (I'll let you find them yourself), but the numbers I remember are the following (check for accuracy if you want). One hard number is the following (it's in my textbook). The quantity of US currency in circulation in the world was $2776 per US resident in April 2009. That's a lot of currency - the stock held at any point in time is about 6% of US annual GDP. In case you think that's all held overseas, a study by the Bank of Canada (in the Bank of Canada Review - look it up) shows that Canadian currency outstanding is about 3% of annual Canadian GDP, and most of that has to be in Canada. That's still a huge amount of currency, and that quantity does not appear to have been falling recently. In some other countries people hold even more currency, for example the Japanese hold a quantity of currency which is about 15% of annual GDP. Fernando Alvarez (again, look it up) at the University of Chicago documents a persistently large stock of currency held in the world.

How do we reconcile these aggregate observations with what we observe at an individual level? First, it appears that the denomination structure of the currency in circulation has changed, with less low-denomination notes and more high-denomination notes. We all know what that is about. Currency is the principle means of payment for drug dealers and for people who want to evade taxation. Second, survey evidence for the US (e.g. the Federal Reserve's Survey of Consumer Finances) tells us that a significant fraction of the population (mainly in poor urban areas) have no relationship with a bank. These people are executing all their transactions using currency.

What implications does this have for monetary policy? First, currency is still important. After all, in normal times, most of the Fed's liabilities consist of currency - the stuff is financing the central bank's portfolio. We want to understand what motivates people to hold currency, why they want to hold so much of it, and what the prospects are for currency-holding in the future. All these things matter for how monetary policy works. Second, in terms of optimal policy, there may come a day when it seems appropriate that the central bank withdraw all currency from circulation. Though currency is the medium of exchange for the poor, central banks may ultimately decide that currency is not worth the bother. We have to pay the costs of thwarting counterfeiters by designing the currency appropriately, we have to print new currency to replace worn-out notes, and for what? To provide a medium of exchange for criminals?

Now, what happens if we get rid of currency? Does this mean that monetary policy has nothing left to do? Absolutely not. The central bank still has a role as the lender of last resort and in most countries it runs an interbank payments system - Fedwire in the US. The liabilities of the Fed - reserves - are the medium of exchange in the payment system. Outside money (reserves) can finance a central bank portfolio of assets (just as reserves are now financing more than half of the Fed's portfolio), and open market operations in the usual sense can matter, even without currency in circulation.


  1. Steve: I think you said in an earlier post that the price level is determined by the supply and demand for currency. How do you see it being determined if the supply and demand for currency go to zero? And if reserves go to zero too?

  2. If there is no currency, we could have one of two things: (i) positive reserves held overnight; (ii) no reserves held overnight. What's different about overnight of course, is that the reserves are not being traded - they just sit. During the day, reserves are essentially the transactions accounts of the banks, and are used in interbank transactions. In the daytime, the central bank is performing an intermediation service - at any given time some participants in the payments system have positive balances and some are negative. Overnight, if there are positive reserve balances, then the central bank can be intermediating government debt. If so, what is it doing that is useful? Maybe it is just intermediating across maturities. Whether there are positive reserves overnight or zero, I think you can determine the price level - outside money trades for other assets (bank deposits for example) that are traded for goods and services, so no problem. I'm just writing words here of course. No one has written this down in a coherent model (yet).

  3. I gather from my German class that throughout much of Germany, currency is much preferred to debit and credit cards; many shopkeepers and restaurants will not take cards at all. What are the figures on outstanding German currency?

  4. When I run the Patinkinesque thought-experiment through my head, I can't see any way that it can be written down, to generate a determinate price level. It seems to leave the central bank without a nominal anchor -- some nominal variable (non-zero) that it can hold fixed. There's nothing akin to any sort of "real balance effect" that gives the central bank any leverage. All behavioural equations are homogenous of degree 1 in nominal variables, and the nominal variable the central bank can hold fixed (currency plus reserves) is zero in equilibrium (by assumption).

    On the other hand, in a sticky-price world, and with an activist central bank adjusting the stock of reserves (from negative, to positive levels, as needed) more quickly than the price level adjusts, it seems feasible to have a determinate price level.

  5. Drag out the call option analogy again. A private bank's checking account dollar is a call option on one of the Fed's green paper dollars. If the fed replaces paper dollars with its own checking account dollars, then the private bank's checking account dollar is now a call option on the Fed's checking account dollar. Nothing important changes. Heck, the private bank might even start issuing paper dollars, which are call options on the Fed's checking account dollars.

    Compare this to calls on actual stock. Merrill Lynch issues calls, which exist only as computer blips, and which give their owner a claim to 1 share of (paper) GM stock. GM then replaces its paper shares with shares that also exist only as computer blips. Nothing important changes.

    Nobody would argue against the above story about GM stock, because everyone recognizes that the backing theory is true of GM stock. That is, the value of GM stock is equal to the value of the assets backing that stock.

    It's a different story with money, because most economists (e.g., Nick Rowe et. al.) deny that the backing theory applies to modern paper dollars. This leaves economists formulating unlikely theories in an attempt to explain how so-called fiat money can have value, and what would happen if the physical form of that money changes.

    Anyone who has spent any time studying monetary theory knows that it takes a lot of tortured reasoning to explain how fiat money has value. The backing theory, on the other hand, requires no logical contortions at all.

  6. Nick,

    Don't run "Patinkinesque thought-experiments" :)

    Only kidding. I assume that you are going down the path of a Wicksell-Woodford pure credit economy, so let me turn the question around. If there is no currency or reserves, how does one settle debt? The primary concern in the cashless approach is how to determine the price level. However, I think we first need to consider what a pure credit economy would look like and how it would operate. Thus far, I haven't been persuaded to take this view seriously.

    I largely agree with Stephen's point. I can envision a literally "cashless" economy, but this doesn't rule out a role for monetary policy because of the existence of reserves.

  7. Mike: I agree that it makes absolutely no difference if we replace the paper in paper currency with electronic book entries. The "physical form" of money is irrelevant (except for how heavy it is to carry in your pocket, etc.). So let me re-phrase my question: what happens if the net quantity of the Fed's "checking account dollars" goes to zero?

    Josh: ordinary people and firms will settle debts between themselves on the books of the commercial banks. Commercial banks will settle debt between themselves on the books of the central bank. What happens if the balances of the commercial banks on the books of the central bank net to zero?

  8. Nick,

    Let's back up a step. I don't mean to be pedantic, but why does this circumstance arise? Or, what are the conditions under which the cashless approach emerges?

  9. Nick: I can see that you are not a New Monetarist. You won't find the answers to these questions in "Money, Interest, and Prices," or in "Interest and Prices." I'm thinking about how to write this down in a well-articulated model. Two important points: (i) Think of the central bank as just another financial intermediary. (i) Fiscal policy is important, and may play an important role in price-level determination. The fiscal authority issues claims for which the payoffs are defined in the unit of account, and the central bank can intermediate these claims, issuing reserves as liabilities.

  10. For anyone who is interested in comparative statistics on means of payment in different countries, go to this Bureau for International Settlements (BIS) publication:

  11. Thank you Stephen. This seems to highlight where I am going with my questioning. If you look at currency as a percent of GDP, it seems relatively constant across the time period and across countries.

    I am not trying to be difficult in prodding Nick about why we should believe in the "cashless" approach (luckily I think he knows me well enough to know that I am not being difficult for the sake of being difficult). However, I do think it is important to consider the conditions under which a pure credit economy would emerge -- and if those conditions are likely to arise -- before assessing price level determinacy.

    BTW, I am not a New Monetarist either -- although I am likely to be more agreeable on many topics. I very much enjoyed the critique of the New Keynesian approach in the papers outlining New Monetarism. Nonetheless, I have yet to come around to search theory as the best way of modeling money. (I did find Aruoba and Schorfheide's NK model with monetary search theory interesting.)

  12. Nick:

    "what happens if the net quantity of the Fed's "checking account dollars" goes to zero?"

    That's easy to answer from a backing theory view. Let's say the fed's checking account dollars were backed, and convertible into an ounce of silver, a square foot of land, etc. As people demand fewer and fewer fed dollars, the fed buys them back with its assets and retires the dollars received. Backing falls in step with the number of fed dollars, so they hold their value.

    Even though I don't think legal reserve requirements matter, suppose they do. Once the number of fed dollars is barely enough to provide legal bank reserves, the dollars will stop returning to the fed, and there things remain.

    Suppose there are no legal reserve requirements, as would be the case if Americans all traded with eurodollars. When the number of fed dollars is reduced to 1, in the form of a computer blip on the fed's balance sheet, we can imagine that people will occasionally redeem their private bank dollars for that computer blip, only to pass it on to someone else in the next instant. But of course long before this happened, private banks would have changed the denomination of their dollars into oz. of silver, square feet of land, etc. Changing the denomination of money doesn't affect anything that matters.

    Now I'd go on to say that what's true of convertible currency is also true of inconvertible currency, for the same reason that convertible (American) calls sell for the same price as inconvertible (European) calls, but that's for another thread.

    But I think that your belief in fiat money obliges you think that as the number of fed dollars approaches zero, the price level becomes indeterminate and all kinds of other weird things start to happen. This is where I'd have to ask you to just consider the possibility that fiat money is no more real than phlogiston, ether, caloric, etc. Those imaginary substances also led scientists to all kinds of crazy conclusions.

    Why do you believe in fiat money anyway? Probably because, as most textbooks say, "the fed does not pay out gold for dollars, so the dollar is unbacked." But a better explanation is that the dollar is backed but inconvertible.

  13. New Monetarism isn't necessarily about search - that's not key to the approach. Search is sometimes convenient in modeling transactions, but most of what we want to do we can accomplish with competitive pricing. Indeed, bargaining and various approaches to splitting up the surplus in trade can be distracting, and does not have a lot to do with monetary policy.

  14. Steve: From reading you, I think of New Monetarism more as a research agenda. One that takes monetary exchange seriously, and wants to model it properly. I will applaud that research agenda from the stands. Does that make me a "New Monetarist"? (About as much as it makes me an Ottawa Senator, I expect!)

    Sure, the answer to this question is not in Patinkin, at least not explicitly, because he never really did model monetary exchange. But I think the answer may nevertheless be implicit in Patinkin, in the same way that it is implicit in David Hume. Whatever the system of equations that defines your model, unless at least one of those equations has some sort of nominal anchor, the price level will be indeterminate. That insight is also implicit in Hume, even though he never mentions equations. So what I'm doing is using that insight to make an educated guess at what one result of your research agenda will be, when you do "..write this down in a well-articulated model" (which I applaud your wanting to do).

    And I think your own intuitive conjecture, that fiscal policy may turn out to be important, is probably correct. If G were fixed in nominal terms, and T fixed in real terms, for example, that should make P determinate. (But if G were fixed in real, and T in nominal, P would be determinate but unstable?). You would get some version of the Fiscal Theory of the Price level.

    Mike: closed-end vs open-end mutual funds would be a more familiar metaphor of inconvertible vs convertible paper money. And closed end funds *do* often trade at a significant premium or discount to Net Asset Value. And one theory of why this might be so would refer to the liquidity or illiquidity of closed-end funds. (There are other theories, of course). Just as off-the-run bonds trade at a discount to on-the-run bonds (even though both have the exact same backing and fundamental value), presumably because they are less liquid. Think of money as the limiting case, because it is the most liquid of all assets. (Steve will complain that I am throwing around the word "liquidity" without saying what it means or modelling it, and he will be absolutely right to complain.)

    According to the "backing" theory, the value of money is determined by M/(Real Backing). In the limit, as both M and RB converge to 0 at the same rate, P is still determinate. (But if the Bank of Canada misplaces one ounce of RB, P might suddenly jump tenfold). But at the limit, when both are 0, God only knows what determines P!

    Josh: a fair question. Personally, I don't see the *demand* for currency disappearing. The most likely case is where the government abolishes currency so it can monitor and control people better. (Either to stop illegal trade, or for some other political purpose -- Margaret Atwood Handmaid's Tale?)

    But perhaps my raising the question of the determinacy of the price level in a currencyless economy is more of a stalking horse. I see the existence of positive stocks of currency and reserves as having very little relevance for the central bank's ability to control P. I see the central bank's ability to actively control the supply of medium of exchange as being important, and it makes little difference whether the medium of exchange is mostly outside or inside money. And it can control the supply of inside money provided there is asymmetric redeemability of inside for outside money (the commercial banks promise to convert their money into central bank money, but the central bank does not promise to convert its into theirs).

    Sorry Steve, I'm wandering off your topic, and leading people astray. But this makes such a nice change from arguing with "Modern Monetary Theorists" and minimum wage supporters!

  15. One thing I'll point is that in Canada which no longer has central bank reserve requirments under normal conditions the amount reserves kept at the BoC by the LTVS clearing banks is 25 million CAD which is less than the real estate value of the Bank of Canada's headquarters in Ottawa. Even this 25 million the Bank of Canada has admitted could be in theory stripped down to even a lower number.

    One symptom of Canada's no reserve policy is the federal government has had to increase its reserves dramatically at the BoC to help execute monetary policy which kind of negates the whole idea of an independent Bank of Canada. Other interesting kink could be the future privatisation of the Royal Canada Mint. After privatisation would the mint still have a monopoly or would there be competing coinage mints. Also the Mint which is run on for profit basis by the federal government also makes Canadian Tire's notorious "private coins" under contract to Canadian Tire. The Mint has also been given the right to make 1$ and 2$ coins by the federal govt largely because they could make more seignorage profit for the govt than the BoC.

    Another point I would make is that Canadian T-Bills now serve a role similar to pre 1935 Dominion notes in collaterallising interbank financial and derivatives transactions. As I have pointed out before almost a third of the current federal deficit sits as govt reserves at the BoC. It will be intersting next week though to see how the BoC balance sheet has changed.

  16. Nick:

    "when both are 0, God only knows what determines P!"

    You could say the same thing about stock or any other financial security, but nobody denies the validity of the backing theory in those cases. If the value of the fed dollar becomes indeterminate when there are no fed dollars, I'd say that's not much of a problem. And if private banks have denominated their currencies in those fed dollars, they only have to denominate their currencies in something else as the fed dollars disappear.

    As for mutual funds, if they ever sold for more than their asset value, then rival funds would issue their own funds to take advantage of the free lunch. And if some country issued money that sold for more than its backing, rival money issuers would issue their own money until that free lunch was gone. Remember that money crosses borders pretty easily.

  17. Nick: Do you mean Ottawa Senators like the hockey team, or the old guys on Parliament Hill?