Why banks shouldn’t matter (for NGDP.)

This post is a reply to a recent post by Nick Rowe:

If the central bank increases the supply of reserves, and banks respond by increasing the supply of loans, the money supply expands, regardless of whether those borrowing from the bank demand a greater quantity of money. Banks really are like helicopters, in that they can increase the quantity of money held without needing to create a demand to hold more money. For any other asset, if you wanted people to hold more, without getting them to want to hold more, you would have to give the asset away, for free. Throw it out of a helicopter.

Now commercial banks certainly don’t give away money for free. But they can “force” people to hold more money even when people don’t want to hold more money. Each individual borrower accepts money in exchange for his IOU, because he can get rid of that money by buying something. But in aggregate they can’t get rid of the money they don’t want to hold. One person’s spending of money is another person’s receipt of money. But their attempts to get rid of that money are what gets us out of the recession, by eliminating the general glut of other goods.

I very much like the second paragraph as an explanation of the excess cash balance transmission mechanism.  But I would rather apply it to the monetary base (which is usually overwhelmingly cash.)  With no loss of generality we can assume that all reserves are held as vault cash.  Then we can use the more descriptive term “currency” for central bank created base money.  In that case:

1.  The central bank can control the nominal stock of currency, demand deposits and copper.  (Where ‘nominal’ means in dollar terms.)

2.  The central bank has no long run control over the real stock of currency, demand deposits and copper.  (Where ‘real’ means in terms of their purchasing power.)

Of course the ability of the central bank to “force” people to accept more money is not at all special:

1.  The central bank can always sell more currency for other assets, and only in extremely rare cases would the price of cash fall to zero.

2.  Copper mines can always sell more copper, and only in very rare cases would the price of copper fall to zero.

3.  In both cases (i.e. currency and copper) producers could still give the stuff away if the price fell to zero.

So I find the first of the two Nick Rowe paragraphs to be a bit misleading, as it suggests central banks and/or commercial banks have some sort of unique ability to force people to accept more of the stuff they produce.

Now let’s think about why money is special.  If copper mines “force” the public to accept more copper, then the nominal and real price of copper both fall.  If the central bank forces people to accept more currency then the real price of currency falls.  But the nominal price of currency is fixed by convention.  (By that I mean that people follow the convention of pricing all other goods in terms of currency.)  There is only one possible way that the act of creating more money can reduce its real value but not its nominal value—inflation.  That is why currency differs from copper.

Where do banks fit in?  In Nick’s defense, banks produce an asset (demand deposits) that has a fixed nominal price.  But it is misleading to suggest that banks have the ability to increase the amount of this asset in the same sort of way that central banks do.  Yes, if you had a monopoly bank, or a bank cartel, they could get together and move the stock of DDs away from its profit-maximizing value.  In general, however, the market will determine the equilibrium quantity of DDs once the central bank determines the price level.  But can the central bank determine the price level?  After all, DDs are close substitutes for currency.  And they have a fixed nominal price.

Yes, close, but not nearly close enough.  People like to use checks to pay bills through the mail.  They like to hold cash to make small purchases quickly, to buy from people who don’t trust them, and to hide wealth from the IRS and/or their spouse.  So there are actually two very different demands for currency and DDs. 

Here’s why it is important that the two products are not perfect substitutes.    No matter what is happening in the banking system, and no matter what impact those events would have, ceteris paribus, on the price level, the central bank has almost infinite ability to offset those changes with adjustments in the supply of currency.  And furthermore, they do just what I have described all the time.

Suppose banks innovate in some way that allows them to profitably increase the supply of DDs for any given stock or currency.  Nick is right that, other things equal, this would cause the total stock of transactions balances to rise, and it would have exactly the expansionary impact described in his second paragraph.  It would also change interest rates.  But now suppose the central bank is targeting interest rates (or exchange rates, or M2, or the price level, or expected NGDP, or anything except currency.)  In that case the central bank will adjust the supply of currency so that the banking shock has no effect on its intermediate target, and presumably no effect on the price level as well.  And again, this is exactly what modern central banks do all the time. 

There are two exceptions.  One exception is the gold standard.  In that case the stock of currency is somewhat beyond the control of the central bank.  So if banking expands or contracts, it will affect the price level in much the same way it would have if a modern central bank was targeting currency.

The other exception is that the events in the banking industry may somehow disrupt the central bank’s ability to hit its targets and/or policy goals.  In principle this should never happen under a fiat money regime.  For all intents and purposes the central bank has virtually unlimited ability to adjust the size of the currency stock, and can hit any expected NGDP target it likes.  But what if the central bank targets expected NGDP through an intermediate target like the nominal interest rate?  And suppose a banking crisis drives the nominal rate to zero.  And suppose the central bank has no backup plan to deal with the zero rate bound (despite having loudly insisted that it did.)  In that case, problems in banking really could affect the price level.  Fortunately, our modern central banks are much too sophisticated to let that happen. 

[Gold standard-era central bankers would roll around on the floor in laughter if someone told them that modern central banker couldn’t figure out how to create inflation with fiat money.]



18 Responses to “Why banks shouldn’t matter (for NGDP.)”

  1. Gravatar of Pariz Pariz
    7. December 2009 at 08:07

    Hi Scott!

    First time to comment here but could you explain this statement a bit further:

    “The other exception is that the events in the banking industry may somehow disrupt the central bank’s ability to hit its targets and/or policy goals. In principle this should never happen under a fiat money regime.”

    Why should this be the case? Thanks!

  2. Gravatar of Nick Rowe Nick Rowe
    7. December 2009 at 08:36

    Scott: It’s good to see you responding to my post.

    Let me take a simple example.

    Suppose the demand to hold stocks of copper depends only on the price of copper, and nothing else.

    Suppose the demand to hold stocks of currency depends only on the price of currency (1/price of goods), and nothing else.

    Suppose some crazed price-controller fixes the price of copper by law, at its original equilibrium. Now suppose a new copper mine opens, and tries to increase the stock of copper people hold (at the same fixed-by-law price). I say it will fail.

    Suppose the price of currency (1/price of goods) is also fixed (by law, or whatever). Now suppose the central bank prints more currency and tries to increase the stock of currency people hold (at the same fixed price). I say it will succeed.

    The copper mine cannot get people to buy more copper unless it lowers the market price (or throws it out of a helicopter). No individual will buy more copper because he doesn’t want to hold more himself, and knows he won’t be able to sell it to others, because nobody else wants to hold more copper either.

    The central bank can get people to buy more currency even at the existing price, where people don’t want to hold more money. Each individual will accept more currency because he knows he can get rid of it either buy buying more other things or selling less other things. But in aggregate they can’t get rid of it.

  3. Gravatar of JimP JimP
    7. December 2009 at 09:55

    Bernanke is on TV yacking – promising us high unemployment and low inflation for an extended period (apparently proud of this) – and describing all the wonderful ways the Fed will use to make the situation worse should it ever start to get better.

    How can this man get away with making this speech? We have lost our minds as a nation to accept this.

  4. Gravatar of Philo Philo
    7. December 2009 at 10:21

    Currency *is* different from copper, or any single commodity. But Nick’s copper-mine story isn’t really coherent. He imagines that, with the price of copper fixed by law at the old equilibrium level, a new copper mine opens and starts producing excess copper. There is no change in anyone’s demand schedule for copper, so now there comes to be more copper than people want to hold–the price of copper is legally fixed *above the new equilibrium level*. But the excess copper exists, so someone has to hold it, willy-nilly. If nothing else, *the owners of the new mine* will be holding the excess. Since they are “people,” *people* will be holding it.

    Can it be given away? Surely that would violate the law, since it would be, in effect, a price of zero–different from the legally fixed price. If giveaways were legal, a would-be seller of copper could say to a potential customer: “If you’ll take nine tons of my copper at the legally fixed price, I’ll *give you* an extra ton.” That would, in effect, be selling the copper at a 10% discount to the legally fixed price.

    (Of course, in practice such a law can’t be enforced, at least not fully; giveaways–which can take many forms, including tied sales of various kinds–can’t be made entirely illegal.)

    “Fixing the price of copper” means (a) preventing sales/purchases of copper at any price other than the legally specified one. (Let’s ignore barter transactions.) It doesn’t guarantee that a would-be buyer or a would-be seller of copper can buy or sell at that price–he may be unable to find a willing counterparty. The case would be different if (b) the government promulgated a standing offer to buy or sell copper at the fixed price. Then it would find *itself* holding all the excess copper. If Nick considers the government to be “people,” he would have to admit that *people* would be holding the excess copper.

    When the government fixes the price of currency, it usually offers to redeem it for a specified good or goods (case (b)). Then any excess currency it created would be held–*by the government*. It is when the price of currency is “fixed” *a la* (a), whether in terms of a single commodity or a basket, that the special nature of currency comes into play.

  5. Gravatar of ssumner ssumner
    7. December 2009 at 11:27

    Pariz, Strictly speaking this would only apply to targets that could be hit in real time, without a lag. The central bank has almost unlimited ability to increase or decrease the monetary base. This means that as a practical matter they can make nominal aggregates as large or small as they wish. They could create severe deflation, or hyperinflation. Banking doesn’t change that fact.

    Nick, I agree with all that, but I don’t see how it conflicts with what I said. Your example shows an implication of the nominal price of currency being fixed. This means the price level is determined in the money market. Once that price level is determined, then in any other market a change in the nominal price is equivalent to a change in the real price. Hence a nominal price control on copper is also a real price control. Currency is different, as the fixed nominal price does not influence the real value of currency. So nominal price controls on currency are not real price controls. So I think we agree.

    My point was that in both markets the producers (of currency or copper) get people to hold more by lowering the real price of that good. So, absent price controls, both can force their product onto a market that is already in equilibrium.

    I agree that currency is different in the sense that the nominal price is fixed, and that fact underlies the price control on copper example you cite.

    JimP, Yes, everytime I think about it I just shake my head. And Bernanke was a guy that wrote lots of articles about the Great Depression and Japan. BTW, didn’t Japanese unemployment peak at about 6% during their liquidty trap? What a disaster! Glad we avoided that!

    Philo, Your comment raises a stock/flow issue that I hadn’t thought of, and perhaps Nick didn’t consider either. I think his example was meant to show how the flow of new copper from the mines could be slowed or stopped. In contrast, the central bank can always provide a new flow of money.

  6. Gravatar of pushmedia1 pushmedia1
    7. December 2009 at 11:40

    Where does the Central Bank’s role as lender of last resort come in? If the Fed was busy lending of last resorting last Fall (which would explain interest on reserves), they may have become distracted from monetary policy.

    A positive theory of the Fed: suppose the Fed has a NGDP target. When operationalizing this target, they have two instruments, the money base and backstopping the banks (i.e. tweaking the money multiplier). Operationally the best policy would be some mix of tweaks to these instruments. For whatever reason last Fall, the Fed thought tweaking the multiplier would be easier than tweaking the base.

    If this is a good story explaining the Fed’s behavior, the important question is: why did they think increasing the base was hard but helping the banking system wasn’t?

  7. Gravatar of Nick Rowe Nick Rowe
    7. December 2009 at 15:30


    “My point was that in both markets the producers (of currency or copper) get people to hold more by lowering the real price of that good. So, absent price controls, both can force their product onto a market that is already in equilibrium.”

    That’s where we disagree.

    In the case of copper, the copper mine can only increase the quantity it sells by increasing the stock quantity demanded, which means lowering the price of copper in terms of goods.

    In the case of a medium of exchange, the central bank can increase the quantity it sells without increasing the stock quantity demanded. So it can sell more even if the price of money in terms of goods is fixed.

    The difference is that money circulates, and copper doesn’t. If there’s an excess supply of copper, and I don’t want to hold any more, I won’t accept it in exchange. Because I know I can’t get rid of it. If there’s an excess supply of money, I will still accept it in exchange. Because even though I can’t get rid of it by increasing the flow of money out of my pocket, I can always get rid of it by reducing the flow into my pocket.

  8. Gravatar of Mike Sproul Mike Sproul
    7. December 2009 at 17:44

    Scott (and Nick):
    ” But they can “force” people to hold more money even when people don’t want to hold more money.”

    A private bank only ‘forces’ its checking account dollars into circulation by lending them below the market rate, or by over-paying for bonds or other assets. Such a bank will not be in business for long. If that bank could suspend convertibility like 19th century banks, then its money would lose value in proportion as the bank loses assets, but the bank would stay in business. A modern central bank is in the same position as the 19th century bank that has suspended convertibility. Its notes lose value because of a loss of backing, not because there is more money chasing the same goods.

  9. Gravatar of ssumner ssumner
    7. December 2009 at 18:21

    pushmedia1. I think you might have gotten the interest thing backward. The lender of last resort role generally assumed the Fed would charge interest on the reserves it lent out. But the Fed didn’t charge interest, it paid interest, which was contractionary. So that is not a substitute for expanding the base, but rather prevents an expansion of the base from doing any good.

    BTW, the base doubled last fall.

    Nick, You said;

    “In the case of a medium of exchange, the central bank can increase the quantity it sells without increasing the stock quantity demanded. So it can sell more even if the price of money in terms of goods is fixed.”

    I don’t follow (I hope you haven’t become a Klingon.) There might be occasional times when the central bank increases the stock of money at exactly the same time as the stock demand increases, and in that case the price level doesn’t change. But in general any time the central bank increases the stock of currency, the price level will rise. Indeed this is the only way that people are willing to hold the extra currency. Sure, some prices may be sticky in the short run, and the value of currency will then merely fall against the flexible price goods in the short run. The rest of the slack will be picked up as lower nominal rates reduce the op. cost of holding cash. But in the long run the currency is only held because the real value of each currency unit has fallen. And that is exactly like copper.

    I’ve always conceded that the fact that money in the medium of account matters. It means price stickiness makes the money transmission mechanism more complicated. In contrast the real price of copper is not at all sticky. Perhaps that is what we are disputing.

    But in my view monetary analysis must start from the flexible price perspective, to see where things trend in the long run (once prices are flexible), and why they trend in that direction. Then one can start working out the impact of sticky prices. To do this one works back from the long run equilibrium that we fully understand, and start to think what happens if (some) prices are sticky in the short run but we know that prices will rise X% in the long run.

    In any case, sticky prices don’t change my argument that central banks can and normally do fully offset any shocks to banking. Again, with the exceptions of gold standards and brain dead central banks facing a zero rate bound.

    BTW, Thanks for contributing, your questions are always challanging. I wonder if I will ever develop an effective answer.

    Mike, No, because there are no close substitutes for cash the central bank can control the value of cash by controlling its quantity (and perhaps placing legal restrictions on the private production of generally accepted currency and coins-although I don’t know for sure if that is necessary.)

  10. Gravatar of Nick Rowe Nick Rowe
    7. December 2009 at 19:35

    Thanks Scott: I will probably let the argument go here, since I can’t think of any way to make the argument clear enough to convince you. Funnily enough, I only really convinced myself of it the other day, while writing my “bad banks” post. It’s an argument I had heard many times from David Laidler (and a few of his students), and had seen it again in Yeager. But the full force of it didn’t hit until I was writing the bad banks post, and realised that there was much more to the Friedman helicopter metaphor than just a money-financed transfer payment.

  11. Gravatar of Nick Rowe Nick Rowe
    7. December 2009 at 19:46

    I can’t resist one parting shot:

    Suppose some prices are flexible and some are fixed. E.g.
    Suppose asset prices are flexible, and goods prices are fixed.

    In the market where assets exchange for money there is no excess demand or supply for assets against money.

    In the market where goods exchange for money there is excess supply of goods and excess demand for money.

    I would not want to say there is no excess demand for money. There are as many excess demands for money as there are non-money goods.

    BTW: I agree that it matters also that money is the unit of account. That makes it much harder for the price of money to adjust.

  12. Gravatar of Mike Sproul Mike Sproul
    7. December 2009 at 19:55

    “No, because there are no close substitutes for cash the central bank can control the value of cash by controlling its quantity (and perhaps placing legal restrictions on the private production of generally accepted currency and coins-although I don’t know for sure if that is necessary.)”

    Let’s try to settle the issue of private banks before getting to central banks. Nick’s quote said that private banks can force their currency into circulation. I said they can only do that by issuing their money too cheaply–say by issuing 100 checking account dollars in exchange for bonds worth only $99.

    A bank like that will lose money and go under, right? So in what sense can you claim that a private bank (presumably competitive) can force its money into circulation?

  13. Gravatar of ssumner ssumner
    8. December 2009 at 08:44

    Nick, Since you regard that post as important, I took another look. My new post addresses a different issue.

    Mike, You’ll have to ask Nick. I am not sure whether private competitive banks would issue unbacked currency or deposits.

  14. Gravatar of Felix Felix
    8. December 2009 at 13:31

    “For all intents and purposes the central bank has virtually unlimited ability to adjust the size of the currency stock, and can hit any expected NGDP target it likes.”

    No. Suppose you target 20% deflation and we are at the ZLB. Now the central bank wants to expand the money supply. Three points:

    (1) The real return on cash will be higher than on anything else. Thus, the public will hold money as an investment, not as liquidity, so the demand is limited only by the desire to save, not by the desire to hold liquidity. This will mean that the demand for money will explode.

    (2) the central bank has to buy things other than short-term government bonds. Thus it takes on risk and will suffer losses.

    (3) The central bank buying assets with currency at the zero lower bound has the same effect as the government buying those same assets with money raised by issuing short-term government bonds.

    Point (1)-(3) lead me to the conclusion that expanding the base at the zero bound is actually no longer monetary policy. It is a fiscal policy of subsidizing away the risk premia of risky assets in order to make business lend more, thereby raising the real equilibrium interest rate.

    So monetary policy, IMHO, does end at the ZLB. However, it does not end at the ZLB on current interest rates, but at the ZLB on current and future expected rates. So if the current interest rate is at zero we can do monetary policy by changing expectations about the future interest rate, but nothing else. This is my interpretation of Mishkin saying that monetary policy can be highly effective even at the ZLB.

    So we can again describe monetary policy completely as setting a path for future interest rates.

  15. Gravatar of jean jean
    8. December 2009 at 17:01

    Even under a NDGP target scheme, the central would unfortunately have to care about the banking system. Because dealing with NDGP scheme requires the traders to be solvent, the central bank would have to select eligible counterparties,
    solvency which could in turn be dependent to the access to NGDP future market. I am not sure that this circularity can be easily broken.
    On the other hand, maybe that the NGDP scheme would lower the need of bailouts, since the schock of a failure would be offset quickly by the purchase of NGDP futures.

  16. Gravatar of Scott Sumner Scott Sumner
    8. December 2009 at 18:08

    Felix, There are all sorts of things that can be done at the ZLB. Many people have suggested that Japan devalue the yen, that it would be a foolproof way to inflate. Another is to increase the expected rate of inflation, as you said.

    I don’t think a permanent ZLB is even worth thinking about, as it implies the national debt could be paid off by printing money, with no inflationary consequences. And we could buy all the assets in the rest of the world as well, and still not see the dollar lose any value.

    Other schemes would involve the government buying real assets like gold, silver and land. Or common stocks. But I see no reason to even think about these scenarios as they have no relationship to the real world.

    A government that wants to create inflation can always do so rather easily. And it doesn’t need to resort to the extreme measures I just described. But central bankers are very conservative, and don’t want to set price level targets. That is the real problem.

    Real world liquidity traps aren’t barriers to easy money, they are symptoms of tight money.

    Jean, No, banks need not play a role. The margin requirements could be met with cash or T-bills. But obviously in the real world there would be plenty of solvent banks. Even in the worst banking crisis in US history we have 1000s of banks in fine shape. There is no problem finding a place to set up a checking account.

  17. Gravatar of Bill Woolsey Bill Woolsey
    8. December 2009 at 21:55


    You need to focus more on process.

    How is it exactly that an increase in the quantity of currency results in a higher price of other goods?

    It is true that people will only be willing to hold an additional quantity of currency at a higher price level. But God doesn’t note this, and set all prices higher so that people will be in equilibrium.

    Be will accept currency in payment for what they have to sell. They will accept accept checks or POS payments too. They indend to spend the currency, not hold it. Once the check or POS signal clears, and their balance in the checking account rises, they intend to spend it, not hold it.

    Because money is the medium of exchange, people accept it in payment even when they have no plan to hold it. And so, its price doesn’t have to change (or anything else change) so that they are willing to hold it.

    Of course, if they aren’t willing to hold it, they spend it. They accept it. And they spend it. But then people will actually be holding more money than they want. And they will fix that by spending it.

    The unintended consequence of this spending is for prices to rise. (and maybe other things to,) and the the higher prices make them want to hold more money.

    But no one has to know that this will happen. Someone wants to buy my product with currency (or a check or POS signal.) I won’t accept it uniless I believe that the price level will rise enough so that I want to increase my money holdings. No. You accept it because you indend to spend it. And the impact of the price level is unintended. Maybe the slight increase in demand (someone is buying your product) motivates you to raise your price. But it is the increase in the prices of things you might be buying that results in you being willing to hold more money. That such must happen to return to equilibrium has nothing to do with the willingness to accept money.

    As for banks, banks can lend money into axistence without lowering the prices (or raising the yields) on the checkabe deposits they use to fund the loan. People will take the borrowers checks even if they don’t entend to hold more money in their checking accoutns, but intend to spend the money.

    The contraint on banks isn’t profit maximization. It is rather redeemability into base money. It could be reserves only. Or it could be currency.

    The central banker does’t have to worry about that because the money isn’t convertible. There is no constraint. He can lend money into existence without any worry about redeemability.

  18. Gravatar of Scott Sumner Scott Sumner
    10. December 2009 at 11:59

    Bill, I think both things happen. Extra money can drive up prices w/o any Ratex explanation being employed, as you say. The public tries to get rid of the excess cash balances, and that drives up AD. But I think in the real world Ratex also plays a big role. If the future expected price level rises due to a permanent injection of money, then that puts upward pressure on current prices. With wages being sticky, that causes output to rise as firms move out along their supply curves.

Leave a Reply