Good monetarism, bad monetarism

This is a sort of response to a recent Nick Rowe post.

Good monetarism uses the excess cash balance mechanism.  It relies on the thought experiment that if you double the supply of base money in the economy, and the demand for base money is unchanged, then nominal expenditure levels must also double in order for money supply and demand to reach equilibrium. And that in the long run monetary shocks don’t have real effects, so any change in NGDP is attributable to prices, not output.

Bad monetarism is focused on the banking system.  Like hydraulic Keynesianism it is obsessed with all sorts of mechanical transmission mechanisms.  Bad monetarism sees monetary policy affecting the economy by first impacting the monetary aggregates through a “multiplier process.”  In fact, the aggregates respond endogenously to the overall macro environment, which is determined by expectations of future changes in the supply and demand for base money, and hence NGDP. 

Let’s quickly review that mechanism.  The monetary base is normally more than 90% currency.  Indeed it would usually be 99% currency and coins except for the foolish system of reserve requirements.  When the Fed changes monetary policy it adjusts the expected future path of the monetary base.  It may signal its intentions by adjusting its target interest rate, but that target rate plays no important role in the transmission mechanism.  Indeed market rates often move in the opposite direction from unanticipated changes in the target rate.  Monetary policy is primarily about changing the number of dollars in people’s wallets.

So let’s suppose the Fed permanently and unexpectedly raises the base 20% above the previously expected level.  What happens?  Because of sticky wages and prices there is a gradual adjustment in output and prices, not an instantaneous 20% increase in P.  Let’s suppose nominal income rises 8% in year one, 6% in year two, and 6% in year three.  What will happen to the monetary aggregates?  It is hard to know for sure, but it is quite possible that the demand for M2 might change in proportion to the change in national income.  In that case M2 would rise 8%, then 6% then 6%.  So why doesn’t the money multiplier work in the short run?  Why doesn’t M2 immediately rise by 20%?  Actually, we don’t know exactly how much M2 would rise, but the point is that it would reflect changes in demand for M2, which reflects changes in interest rates, national income, and all sorts of other variables.  It is determined by the public’s demand for M2 as a share of NGDP.

Here’s the problem with bad monetarism.  It implicitly treats the money multiplier as something reliable, while the velocity of circulation is a bit unreliable.  (Yes, that’s right, monetarists don’t assume V is constant.)  So monetarists reason this way:  A 20% rise in the money supply will raise M2 by 20%, and in the long run NGDP will also rise 20%.  In the short run velocity can be a bit unstable.  But the causation runs from changes in M2 to changes in NGDP

The problem with this approach is that when velocity is unreliable, the money multiplier is equally unreliable.  In the US during the early 1930s velocity fell sharply.  So did the multiplier.  In the 1990s in Japan velocity fell sharply.  So did the money multiplier.  In the US in 2008 velocity fell sharply, so did the money multiplier.  Do you notice a pattern?

It is velocity that should be called “the multiplier” as V isn’t really the average number of times cash turns over in transactions, it’s just the NGDP/M ratio.  (V isn’t really “velocity” because most transactions do not involve final output.)  So there are as many “money multipliers” as there are nominal aggregates.  Each money multiplier is simply the ratio of the aggregate in question, and the monetary base.  And here is my key point, none of those multipliers add anything to the NGDP/MB multiplier.  Yes, that multiplier is also unstable, but so are all of the others.

So when people start asking me about the banking system and reserves and loans my eyes just glaze over.  Most of the base is normally cash, and monetary policy consists of changing the supply of cash relative to demand.  The nominal size of the entire banking system and all its components; capital, loans, reserves, deposits, etc., is determined endogenously, just like the nominal size of the plastic surgery industry, or nominal size of the ice cream industry.  Normally, a permanent 20% increase in the base would be expected to increase the nominal size of the banking, plastic surgery, and ice cream industries by 20%.  But other things are often not equal.

Banking is only special a few cases.  For instance, government regulation of banks might create a large and time varying demand for base money.  Or the public may hoard cash because they fear a banking collapse.  Otherwise, banking is of no interest to monetary economics.  If the Fed abolished reserve requirements, insured bank deposits, and targeted NGDP growth expectations at 5%, then you might as well drop banking out of monetary textbooks.

Is this endogenous M2 theory the same argument that the Post-Keynesians are making?  I don’t think so, as the Fed still controls the expected path of NGDP by controlling the monetary base (relative to the demand for base money.)  The base may appear endogenous as well, as the Fed often uses a short term interest rate target.  But in practice the Fed is merely using fed funds rate changes to signal an intention to change the MB path relative to changes in the expected future demand for base money.  So the fundamental tool has been control of the base.

Of course it is also possible to target nominal aggregates by controlling the demand for the medium of account, i.e. the base.  Interest payments on reserves is one way of doing this.  And perhaps the Fed will go that way in the future.  But that still wouldn’t make the money multiplier any more interesting.

Does this approach fit with Nick’s views?  I’m not sure.


Tags:

 
 
 

31 Responses to “Good monetarism, bad monetarism”

  1. Gravatar of Bill Woolsey Bill Woolsey
    1. December 2009 at 17:07

    You need to use the term “currency” to refer to paper notes and coins. The term “cash” is used among monetary economists to refer to the medium of exchange. For example, the cash balance approach is no about just spending currency, but rather currency and deposits. In finance and business generally, they use cash even more inclusively. For example, selling stocks and buying T-bills means gowing from stocks into “cash.”

    Cash is a horrible.

    By the way, what is up with M2? Why would anyone count a CD with a blance of 99,000 as money, but a CD with a balance of 101,000 as not money?

    MZM at least has zero maturity. M1 is supposed to include transactions accounts, but because of sweep accounts, it doesn’t count them.

    I realize that your argument says that doesn’t matter, but still.

    I don’t believe that hand-to-hand currency plays a key role in the monetary order. If no one used it tomorrow and everyone used checks and electronic payments for everything, it wouldn’t make a bit of differnce in how I see things. In my view, treating what you name “cash” as central to monetary economics is a mistake. You have the tail wagging the dog.

  2. Gravatar of StatsGuy StatsGuy
    1. December 2009 at 19:31

    “If the Fed abolished reserve requirements, insured bank deposits, and targeted NGDP growth expectations at 5%, then you might as well drop banking out of monetary textbooks.”

    This is a perfect storm for moral hazard. A privatized money system where government owns the downside. Or am I misreading? Did you mean, by abolish reserve requirements, going to 0%, or to 100%?

    I have to admit, I find this post very confusing and difficult to reconcile with your past support of endogenous money perspectives. I suspect there is a missing element, which relates to peoples’ beliefs about how much wealth they own. This belief changes with a nominal wealth increase that is not immediately rendered into a price shock.

    One key issue that I think this perspective overlooks (at least in the short term) is that creating money and getting it into peoples’ hands are not necessarily one and the same. If the primary mechanism for injection is banks (via the credit channel), then if that channel was closed due to an expectations-equillibrium-trap, would we not expect (at least in the short run) for monetary action to be handicapped? I expectation of falling prices, banks will cling to reserves (not just to cover loan losses, but because they might be able to get a better deal in the future with their money).

    Or, in Nick Rowe’s terminology, the creation of a large mass of reserves simply creates the potential for the consensus-equillibrium to shift. But since many equillibria are semi-stable, it’s impossible to predict what events might shift it. You could argue that the existence of a prior distribution over those events (combined with a large supply of reserves that widens the distribution) creates a prior distribution over the equillibria, and that by itself creates a self-reinforcing shift _out_ of a low-level stable equillibria.

    In a sense, this sort of dynamic would explain the high-volatility, random-walk-like environment we now have. Strong policy guidance seems like it could coordinate expectations on an ideal focal point solution.

    I think what you might be saying is that IF we were really really good at committing to level targeting, then banks would be nigh irrelevant. But, since we’re pretty lousy at it, banks are still relevant.

  3. Gravatar of ssumner ssumner
    1. December 2009 at 19:36

    Bill, I think the tail does wag the dog. I believe Fed created money (which is normally mostly currency, and would be virtually all currency w/o reserve requirements) drives NGDP growth. Everything else is endogenous. M1, MZM, M2 and M3? They are all endogenous.

    I realize people call T-bills cash, but it is sloppy terminology. M1 is composed of cash held by the public plus DDs. The base is vault cash plus cash held by the public plus member bank deposits at the Fed. People need to learn the correct terminology if they want to talk about monetary policy. But I did use the term ‘currency’ in my explanation of the base in paragraph 3, for exactly the reason you mention–I thought people might be confused.

    People confuse targets and instruments. The monetary base is an instrument, whereas the broader aggregates are targets, just as NGDP is a target. The question is why we would want to target M2. Why not just target NGDP?

  4. Gravatar of Nick Rowe Nick Rowe
    1. December 2009 at 21:32

    Wow! There I was, taking a short break from valiantly defending my little model against the Post-Keynesian/Neo-Chartalist hordes attacking from one side; I turn around and find Scott launching an attack from the other side!

    It’s gonna take me a little time to re-position my defenders.

    Main points though:

    1. To my way of thinking, the fact of monetary exchange is central to understanding macroeconomic disequilibrium; we do not have one centralised Walrasian market, but n-1 markets for n goods, and the nth good, the medium of exchange appears in all those n-1 markets. So “money” means “medium of exchange”. The unit of account is important, because of sticky prices/wages, but the medium of exchange matters more. And demand deposits are media of exchange. And I want to at least imagine a monetarist model in which currency disappears, but demand deposits remain. So banks matter.

    2. The money/banking model is a partial equilibrium model, not a general equilibrium model. Its job is to explain how shifts in the central bank’s supply curve of reserves can shift commercial banks’ supply curve of demand deposits. It needs to be bolted into a general (dis)equilibrium framework in order to work out the full implications of changes in central bank policy.

    But it’s late, and I need to reflect further.

  5. Gravatar of Simon K Simon K
    2. December 2009 at 00:04

    Scott – This is the first thing you’ve written that I’ve had trouble buying into, because it doesn’t make sense to me. Previously you’ve persuaded me to dig out my very faded undergrad memory of the quantity equation and look at it rather more intelligently. But I’m really finding this article hard to square with the picture of monetary policy I’ve been building up, largely thanks to you. There must be something wrong either with my mental picture built up from what you’ve written before, or my understanding of this article. Hopefully you can tell me where it is.

    Most of the monetary base is currency – okay. But most transactions are not conducted in currency. Most transactions are conducted by clearing current account balances from one bank to another without ay reference to currency. The little bit that really needs to be cleared at the end of the day is cleared through the banks reserve balances, not driven accross town in a truck full of dollar bills. And most money holdings are not currency either – they’re demand deposits. Currency mostly sits in ATMs waiting to be taken out, spent, and put back in another ATM.

    Given that most cash holdings aren’t really currency and most transactions aren’t really conducted in currency, if the base is 90% currency, how is the size of the base a serious constraint on economic activity? Banks can increase the supply of the higher aggregates pretty much at will, and that money is regarded as interchangeable with base money. If two things are essentially perfect substitutes, isn’t is basic microeconomics that their prices will be the same, and if the supply of one is much more elastic than the supply of the other, that’s going to be the bulk of the supply, and its that supply curve that determines the price and quantity – the base is irrelevant.

    This would all seem to indicate than NGDP targetting, or any other kind of targetting, is actually impossible because the instrument the fed really constrols (the base) has no effect on what actually determines NGDP (some higher aggregate). But I don’t buy this, so there must be a flaw in the logic.

    I’m not sure where the flaw is, but I suspect that in fact the fed has more control over the higher aggregates than is implied above. I’m not sure how, though. Scott, anyone, care to point out where I went wrong?

  6. Gravatar of Giles Giles
    2. December 2009 at 00:07

    Sorry to pose a naive question (it seems that they are the only ones I have) but:

    “Good monetarism uses the excess cash balance mechanism. It relies on the thought experiment that if you double the supply of base money in the economy, and the demand for base money is unchanged, then nominal expenditure levels must also double in order for money supply and demand to reach equilibrium.”

    Isn’t all the magic somehow contained in those words “if . . .. demand for base money is unchanged”? What I mean, in brief, is: why would any of this change the mind of a strong Keynesian like Brad DeLong of “Simple Keynesianism for Monetarists” – and leave people like him arguing that the demand for base money becomes entirely floppy in the presence of other real problems/expectations issues, like, say, a sudden detorioration in the banking channel – and uncertainty about the central bank’s real objectives?

    What bothers me is that the quoted section seems to say, to me: ‘good monetarism is that which enables my theory that the Fed can choose whatever NGDP it wants, to be true”. But does it really work like that, in 2008-10?

    thanks for the great blog, you and Nick.

  7. Gravatar of woupiestek woupiestek
    2. December 2009 at 01:00

    When the FED buys securities, it doesn’t actually ship of a load of banknotes to the seller, does it? Do deposits at the FED count as currency? Or is it more like: on the long run all FED deposits would be withdrawn, if there were no reserve requirements?

    Reserve requirements are indeed a very clumsy way of providing deposit insurance.

  8. Gravatar of Bill Woolsey Bill Woolsey
    2. December 2009 at 06:19

    Currency, not cash.

    Cash is a terrible term to use.

    It is bad enough that banks use “vault cash” to refer to currency held at the bank, but don’t add to the damange by calling “currency held by the public,” “cash held by the public.”

    When everyone else uses the cash balance approach they are not refering to the “currency balance approach.”

    And the use of the term “cash” to refer to all liquid assets, including T-bills, may be stupid, but why cause more confusion.

    Vault cash == currency held by the bank

    Currency == banknotes and coins.

    I don’t deny that bank deposits used as money are endogeneous. But that doesn’t mean they aren’t important. Excess supplies and demands for money, whether created by the Fed or by banks, impact nominal expenditure through what the rest of us call the “cash balance approach.” And “cash” includes all money, not just currency.

  9. Gravatar of Bill Woolsey Bill Woolsey
    2. December 2009 at 06:29

    The connection between base money (currency or not) and bank deposits used as money is redeemability.

    The price of zero maturity deposits cannot be less than face value because of arbitrage. They could be greater, but not if the private issuers maintain convertilility (accept currency deposits directly or indirectly.)

    Scott is right that the quantity of all of the deposits is endogeous–demand determined– _in equilibrium_. That is because the price can’t vary. But the process by which equilibrium changes very much involves the quantity of the media of exchange more broadly. Most of it is created by banks (depending on how it is measured) and most of the spending that is generated the price and output chagnes that bring about equilibrium is done by these deposits and not currency.

    Too many economists, in my opinion, focus too much on comparative statics and not enough on process. How does a doubling of the quantity of base money result in a doubling of the price level?

    Imagining a world were currency is the sole medium of exchange is far removed from reality.

    Noting that if nothing else changes, then the price level must double for there to be a new equilibrium, tells us nothing about how we get there.

    Switching the meaning of cash, between currency, as in “vault cash is currency held by the banks” and money, as in the “cash balance approach to monetary equilibirum” will hardly do.

    Much of my study of monetary economics was in the context of trying to understand Greenfield and Yeager’s monetary system. They called it Black-Fama-Hall. Hall is a good monetary economist. Black and Fama are finance guys. I think they are way to focused on comparative static equilibrium.

  10. Gravatar of Scott Sumner Scott Sumner
    2. December 2009 at 06:46

    Statsguy, I agree about moral hazard. I am just saying that we have deposit insurance it would make banking especially irrelevant to monetary policy. And by the way, FDIC has happened, and will continue to happen despite my opposition. So banking will continue to be irrelevant on that count. Of course there may be other reasons banking is relevant, as we’ll see in my reply to Nick.

    This post discusses “good monetarism,” not my own views. I agree that if the Fed targets NGDP, the base becomes endogenous. And that’s what I’d like to see. I’m simply trying to explain the parts of monetarism that I think make sense as analytical tools.

    In general, I think I agree with all your points. I think a huge overhang of reserves might affect NGDP growth expectations. But that would be a case where the monetarists’ treasured “multiplier process” broke down. My post is directed against those who see causation going from MB to M2 to NGDP. I see it going from MB to future expected NGDP, to current NGDP, to current demand for M2, to current levels of M2.

    Nick, I think you might be reading too much into this post. I am not taking an anti-medium of exchange view here. I am essentially saying:

    “Look, the Fed actually produces the MB. In normal times with no reserve requirements that would be 99% currency. So if we model the supply and demand for currency we can model the price level and NGDP. And M1 or M2 don’t really add anything. Yes, there are often fairly stable relationships between MB and the aggregates, but there are also fairly stable relationships between MB and NGDP. And when the MB/NGDP relationship is unstable, the multiplier often breaks down as well.”

    As I said to statsguy, I also don’t like the MB to M2 to NGDP transmission mechanism. It’s pretty obvious to me that recent changes in M2 are an endogenous response to NGDP, which in turn was caused by the expected path of MB being less than required to produce adequate NGDP growth.

    But my whole story is consistent with the MB being viewed as a medium of exchange, and also being the only medium of exchange directly controlled by the central bank. So your medium of exchange view can explain why a 20% increase in the MB causes future NGDP to be expected to rise 20%. But then that expectation feeds back and affects current NGDP, which then affects the demand for M1, M2, etc., that is, the amounts of each of those aggregates that the public prefers to hold at the current NGDP level. The multiplier doesn’t force M1 and M2 to rise mechanically with the increased MB. And current changes in M1 and M2 don’t cause current changes in NGDP, even if your medium of exchange story is correct in explaining the long run impact of monetary policy.

    My basic argument is that there is a redundancy. We can either tell the story without the aggregates and with the MB, or else the story can’t be told at all from a monetarist perspective, because both the multiplier and velocity will be too unstable. I agree that the multiplier is a partial equilibrium story, but it is inappropriately bolted onto the GE story by assuming that M2 is exogenous, and causes changes in NGDP.

    I completely agree that banking would be extremely important if we went to a cashless economy. But up until 2008 we were an economy where the base was 90% currency. That’s why I think M&B courses should study wallets, not banks.

    Simon K,

    I’m sure the fault is mine, as everyone seems just as confused as you are. You said:

    “Given that most cash holdings aren’t really currency and most transactions aren’t really conducted in currency, if the base is 90% currency, how is the size of the base a serious constraint on economic activity? Banks can increase the supply of the higher aggregates pretty much at will, and that money is regarded as interchangeable with base money. If two things are essentially perfect substitutes, isn’t is basic microeconomics that their prices will be the same, and if the supply of one is much more elastic than the supply of the other, that’s going to be the bulk of the supply, and its that supply curve that determines the price and quantity – the base is irrelevant.”

    I don’t think cash and other media of exchange are perfect substitutes. But let’s say you are right, that would make my critique even stronger. Why? Because the whole money multiplier literature only makes sense if they are not perfect substitutes. If they were, then the C/D (currency deposit ratio) would be unstable, and so would the money multiplier. So your comment is more anti-monetarist than mine.

    In your view the Fed could reduce the monetary base (in normal times, when it is mostly cash) and it wouldn’t hurt the economy because banks would just substitute other media of exchange. I think a reduction in the base would be deflationary, and indeed I think even those who disagree with me on my post (like Nick Rowe) would agree with me on that point. Cash and DDs aren’t perfect substitutes, but if they are, then the money multiplier has even bigger problems than I thought.

    To conclude, you went wrong at the very beginning with your assumption about perfect substitutes. Once you made that assumption, the remaining logic of your argument was correct.

    Giles, As I mentioned above, this is “good monetarism”, but I’d like to add other elements to the picture, like level targeting, and targeting expectations. DeLong is right that a big increase in the MB may not boost spending if it is expected to be temporary. So you need some sort of assumption about expectations. I think later in the post I mentioned a change in MB that was expected to be permanent. I should have mentioned that right upfront.

    The other argument is that the Fed has almost infinite ability to increase MB, so there is some increase that would get the job done.

    woupiestek, Believe it or not, there were no bank deposits at the Fed until 1914. As far as I know, prior to that date when a bank bought a bond from another bank they paid in cash. There used to be $100,000 bills in circulation for this purpose. Yes, the Fed now initially credits a banks’ account at the Fed when it buys a security, but it doesn’t have to be that way.

    Furthermore, soon after this occurs 90% of the new money goes out into circulation as cash. It would be 99% if we got rid or reserve requirements.

  11. Gravatar of StatsGuy StatsGuy
    2. December 2009 at 07:21

    ssumner:

    “My post is directed against those who see causation going from MB to M2 to NGDP. I see it going from MB to future expected NGDP, to current NGDP, to current demand for M2, to current levels of M2.”

    This helps. I think where everyone is scratching their heads is this: How, exactly, do we get from MB to expected future NGDP without going through M2? This would require incredible faith both in the technical capability of the Fed, and the willingness of the Fed to use all tools at its disposal (and the independence of the Fed from meddlers and political pressure). The argument on the other side is that the Fed takes time to act (ye olde long and variable lags), and there may be skepticism in the technical ability of the Fed to inject/withdraw liquidity rapidly. (Hence the reason the Fed is “testing” liquidity withdrawal mechanisms – this is the equivalent of a naval demonstration in international waters).

    I would anticipate your argument to be that, even if the Fed couldn’t get it perfect in the short run, a truly credible long term commitment would cause financial firms to arbitrage rapidly toward the target.

    But I would like to see the MB ==> expected NGDP mechanism spelled out and defended a little better…

  12. Gravatar of Jon Jon
    2. December 2009 at 08:12

    “If the Fed abolished reserve requirements”

    Because of sweep accounts there are no meaningful reserve requirements in the US. Clearing balances are what matter–which is why a country like Canada with a so-called “0%” reserve-ratio, still has a functional monetary policy.

    “Base” is a transmission mechanism. But credit substitutes. Therefore the banking system does matter. Further, any form of draft matters. They substitute for and circulate as money. Notes are the primary means of payment in international trade with developing nations for instance.

    I think your equation of exchange has the wrong variables…
    1) NGDP does not appear in the equation of exchange, transactions do
    2) The monetary base does not appear in the equation of exchange, broad money does.

    Yes you can compute ‘k’ using other variables, but its a meaningless statistic which is one reason that it appears to be erratic. … and yes, my definition is the ‘classical’ one.

  13. Gravatar of Doc Merlin Doc Merlin
    2. December 2009 at 09:16

    @Statsguy
    “This helps. I think where everyone is scratching their heads is this: How, exactly, do we get from MB to expected future NGDP without going through M2.”
    A top down fed-like entity couldn’t do effective targeting because socialist planning inevitably fails. A Bill Woolsey style free banking system with per capita NGDP convertibility could do very effective targeting, however.

  14. Gravatar of MikeDC MikeDC
    2. December 2009 at 09:53

    I echo Simon’s post almost word for word.

    To me, most private money seems to be privately created (and destroyed) as a function of expected economic activity and this is largely (though not completely) independent of the government created base.

    If I get a loan from my bank in the fractional reserve system, money is being “created” as it gets deposited and re-loaned in a way roughly proportional to our expected future value added (If I reported to the bank I expected no future income, no one post 2006- would lend me money). Current money stocks get privately destroyed every day when we realize some of the plans we’ve made, and the values we’d put to paper in our accounts are unsustainable moving forward. Suddenly someone’s not willing to pay the NPV of my stream of mortgage payments because it’s unlikely I’ll make them all. That’s destruction of money just as surely as if I set fire to a stack of hundred dollar bills.

    I picture the money supply as a bathtub with a drain and a faucet at both ends. On one end is the government, and at the other what, I guess, we’d call the market result. If the market result is the drain wide open, then yes, the government ought to open its faucet to compensate. But it ought to consider whether its actions caused the private sector drain to be open or stay open, and how much water you can put in the tub.

  15. Gravatar of david glasner david glasner
    2. December 2009 at 19:02

    Scott, This was an incredibly good post. Aside from Bill Woolsey’s terminological quibbles about cash and currency, which are correct, I agree with everything you wrote with one exception. The theoretically correct version of Monetarism (which is really Tobin’s classic article on Commercial Banks as Creators of Money, which was not really a Monetarist favorite) would not, as you assert, eliminate banking from the money/macro textbooks. The reason is that, as you point out, currency (plus reserves at the Fed) is not a perfect substitute in demand for deposits. Since deposits earn interest and currency does not, obviously currency provides services that deposits don’t. However, banks, by committing to convert deposits on demand into currency, make deposits and currency perfect substitutes in supply so that they are perfectly exchangeable at par (except at ATM machines). Therefore the terms on which banks supply deposits strongly influences the demand for currency. The Fed imposes reserve requirements because if banks could create deposits without paying any tax, the demand for currency would go down and the Fed would lose seingorage. That’s also an important reason why we don’t allow banks to issue banknotes. All the seingorage would then accrue to banks, or to deposit holders if banks competed away the profit from issuing banknotes by paying higher interest on deposits.

    As for Bill’s argument about needing deposits to explain how a doubling of the monetary base would double the price level, I would simply ask: do we need deposits to explain how a doubling of the stock of gold would (approximately) double the price level under a gold standard? I don’t think so. Gold plays (approximately) the same role in a gold standard model that currency plays in a fiat money model. The demand for and the supply of deposits can affect the price level only by affecting the demand for gold or the demand for currency. But the banking system simply provides whatever quantity of deposits the public demands at whatever price level is determined in the market for gold or for currency. The market for deposits is equilibrated not by the rate of interest but by the interest that banks pay on deposits. This point also has implications for the coherence of natural rate theories based on banks lending at less than the natural rate of interest, but that is a whole other story.

  16. Gravatar of Nick Rowe Nick Rowe
    2. December 2009 at 19:50

    Scott (and David Glasner):

    People buy currency and DDs not because they necessarily want to hold currency and DDs, but because in a monetary exchange economy, if you want to sell apples, and buy bananas, you must necessarily buy currency or DDs and then sell currency or DDs. M1 is fundamentally different from the non-M1 components in M2.

    You can increase M1 even when there is no increase in the quantity demanded to hold M1. You cannot increase the non-M1 components of M2 without persuading people that they want to hold the non-M1 components of M2.

    When I accept a cheque in exchange for my labour, I might ask about the wage rate, but I do not ask about the interest rate on my chequing account.

    When a bank makes a loan, and does so by crediting the borrower’s DD, the borrower may ask about the interest on the loan, but does not ask about the interest rate on the DD. He does not intend to hold the DD, but to spend it.

    The quantity of DDs (and currency) is supply-determined, in a way the quantity of other assets is not.

    Suppose, just suppose, that the demand to hold currencies and DDs were perfectly interest-inelastic. Hold income and the price level constant in the very short run. Would it be impossible for the Fed (and banks) to increase the quantity of currency and DDs held? of course not. But for any other asset, it would be impossible.

    I think I’m channeling Yeager, contra Tobin, amn’t I?

  17. Gravatar of Nick Rowe Nick Rowe
    2. December 2009 at 20:04

    And that’s why banks, and the “banking multiplier”, matter. Banks are important for monetary policy in a way that other financial institutions are not. That’s because some of banks’ liabilities are themselves media of exchange. And it is possible to “force” people in aggregate to hold more medium of exchange than they want to hold, because they accept it willingly even if they don’t want to hold it. And it is their attempts to get rid of it that create the consequences across the economy, and which constitute the monetary policy transmission mechanism.

    And if the “banking multiplier” is broken, because, for example, banks are capital-constrained, and so will not extend loans and create deposits, then one very important part of the monetary transmission mechanism is broken. It’s not the loans per se that matter, because not only banks can create loans. It’s the DDs that matter. If banks cannot create loans, they cannot “force” people to hold more DDs than they want to hold. That’s what matters.

  18. Gravatar of Nick Rowe Nick Rowe
    2. December 2009 at 20:22

    “Good monetarism” is not some sub-section of portfolio-choice theory, with a fetish for a particular class of financial assets. Good monetarism is about monetary exchange economies. And the medium of exchange in a monetary exchange economy is fundamentally different from all other goods. It is the only good that does not have a market of its own. Its market is every market. And this would be true even if it were not the medium of account. If people were better at math, and could multiply and divide in their heads, the medium of account would not matter. But the medium of exchange would still matter.

    In modern economies, monetary policy works by influencing banks to supply more or less demand deposits. And banks do this regardless of whether people want to hold more or less demand deposits. And it’s because banks have this power, and the Fed exploits their power second-hand, that monetary policy works. Currency just comes along for the ride, nowadays.

    Currency is demand-determined; DD’s are supply-determined.

  19. Gravatar of Scott Sumner Scott Sumner
    3. December 2009 at 07:51

    statsguy; You asked:

    “This helps. I think where everyone is scratching their heads is this: How, exactly, do we get from MB to expected future NGDP without going through M2? This would require incredible faith both in the technical capability of the Fed, and the willingness of the Fed to use all tools at its disposal (and the independence of the Fed from meddlers and political pressure). The argument on the other side is that the Fed takes time to act (ye olde long and variable lags), and there may be skepticism in the technical ability of the Fed to inject/withdraw liquidity rapidly. (Hence the reason the Fed is “testing” liquidity withdrawal mechanisms – this is the equivalent of a naval demonstration in international waters).”

    Two answers:

    1. There is no problem getting from the base to NGDP forecasts, as people are always (at least implicitly) making NGDP forecasts.)

    2. If you mean how does the Fed go about changing NGDP forecasts, the best way is with an explicit NGDP target, level targeting. But even an inflation target will imply some sort of associated NGDP forecast, depending on the public’s views of AS. When I say M2 adds little to the MB, I don’t mean to suggest the Fed should target the MB, obviously I don’t favor that policy. But they implement monetary targeting regimes by changing the MB. That is literally what they do. M2 also often changes, but so do lots of other nominal aggregates, like the nominal size of the ice cream industry. But M2 isn’t causing the change in NGDP, otherwise we wouldn’t have had a severe recession this year, after all, M2 didn’t fall.

    The mechanism is the excess cash balance story. When people or banks have more MB then they want, they try to get rid of it. But they (collectively) can’t get rid of it. Instead AD rises. All the other monetary variables (interest rates, exchange rates, M2) endogenously respond to the NGDP growth expectations produced by the expected future path of MB, and the expected future demand for MB.

    Jon, The reason I put NGDP into the equation of exchange is because I am interested in NGDP. But I don’t really use the equation for analysis anyway.

    I don’t understand your statement that there are no effective reserve requirements. I am told that (in normal times when rates are positive on T-bills) banks only hold large reserve balances because they are forced to.

    MikeDC, I’ll just give the same answer. I think he is wrong, cash and other aggregates are not close substitutes. And if he is right it strengthens my critique of bad monetarism.

    Thanks David, I completely agree with everything you said, but with one minor quibble:

    “Therefore the terms on which banks supply deposits strongly influences the demand for currency.”

    Sure, you can make this argument. But income taxes and drug laws usually create a far larger demand for base money than does the banking system. So why not include those variables in money and banking texts? The answer is that the Fed can easily offset shifts on drug lord demand for cash with open market operations. But isn’t that also true of shifts in bank demand for base money? Can’t those shifts be just as easily offset with OMOs?

    Nick, You almost snuck this by me. 🙂

    “You can increase M1 even when there is no increase in the quantity demanded to hold M1. You cannot increase the non-M1 components of M2 without persuading people that they want to hold the non-M1 components of M2.”

    Here is what is symmetrical. You cannot increase the non-cash component of either without cooperation from the public. And you can increase both total m1 and total m2 without cooperation from the public. So it’s exactly symmetrical.

    you said:

    “Suppose, just suppose, that the demand to hold currencies and DDs were perfectly interest-inelastic. Hold income and the price level constant in the very short run. Would it be impossible for the Fed (and banks) to increase the quantity of currency and DDs held? of course not. But for any other asset, it would be impossible.”

    I don’t follow. Is this a liquidity trap argument? If so, my answer is that liquidity traps don’t prevent the Fed from increasing NGDP as much as they want, and hence increasing the nominal amounts of any parts of NGDP, or the nominal amounts of any asset.

    Nick#2, I don’t think banks are unimportant because they can’t influence the demand for base money (I agree they can) but because the central bank can offset those effects just as easily as they offset the effects of drug dealers on base demand.

    I’m not saying the Fed should target the MB, far from it. I am saying the MB is the only lever they need to target P or NGDP or whatever their goal is. And all they need to do is focus on expectations of NGDP and use the MB as the lever. There is no role for banks. But sure, banks do influence the level of NGDP, holding the MB constant, as do drug dealers.

    Nick#3, You said:

    “In modern economies, monetary policy works by influencing banks to supply more or less demand deposits.”

    I think that in modern economies monetary policy works by influencing the expected path of NGDP over time. And that impacts current relative prices. And what changes future expected NGDP? Certainly not interest rates, as they aren’t impacted by policy in the long run. I think it is the expectation of the excess cash balance mechanism playing out. And that mechanism can be explained without recourse to banks, just looking at the amount of base money that people and firms (including banks) want to hold, vs the supply of currency. Anything else is redundant. I know of no macroeconomic problem in the real world that cannot be explained with this striped down approach, but which can be explained with your approach.

    But this debate will probably never be resolved, as I can’t think of any reason why you shouldn’t look at things your way. Ultimately it is an aesthetic judgment on my part. The real debates are what aggregates should be targeted, and how should we target them? And it is not clear that our debate here has any implications for that broader debate. Or am I wrong.

  20. Gravatar of Nick Rowe Nick Rowe
    3. December 2009 at 09:44

    Scott: “I don’t follow. Is this a liquidity trap argument?”

    No. The exact opposite. Imagine a vertical money demand curve. In fact, suppose the money demand curve doesn’t have any arguments that change at all. Everyone wants to hold $100 on average, because their pants fall down if they hold more.

    Is it possible to increase the money supply? If it were any good other than money, the answer would be “no”. You can force people to buy more land if they don’t want to hold it. But with money, you can force people to hold more than they want. You just buy something from them. They don’t *plan* to hold the extra money, they plan to get rid of it. But in aggregate they can’t.

  21. Gravatar of Nick Rowe Nick Rowe
    3. December 2009 at 09:45

    Damn. Typo. Should be: “You can’t force people to buy more land if they don’t want to hold it.

  22. Gravatar of Carl Lumma Carl Lumma
    3. December 2009 at 10:57

    >Let’s suppose nominal income rises 8% in year one, 6% in
    >year two, and 6% in year three.

    Uh, that wouldn’t result in a 20% increase.

  23. Gravatar of david glasner david glasner
    3. December 2009 at 13:10

    Scott, I totally agree that taxes are relevant to any discussion of fiat money because if the government did not accept money in payment of taxes, fiat currency would have no value. Hehe, you fell into my trap! As for drug dealers, my puritanical sensibilities make me recoil in horror at the mere thought of mentioning any such thing, but I certainly do think that it would be perfectly fine to point out that one of the incentives for holding currency instead of deposits is that the former is preferable to the latter if one is engaging transactions that, for whatever reason, one would prefer not to be transparent.

    Nick, I agree that people are stuck with currency, but I absolutely deny that people are stuck with deposits. Please reread Tobin’s essay and tell me what you disagree with in his argument. I can’t find a thing. He deserved the Nobel Prize just for that one paper. There is an economic mechanism for destroying as well as creating deposits, and profit seeking banks and utility maximizing customers will mutually arrive at an interest rate on deposits and a quantity of deposits at which banks have no incentive to increase the amount of deposits given the current levels of interest rates at which banks can lend and borrow and bank customers are holding exactly the amount that they wish to hold given the interest rate banks are paying on deposits and the rates being paid on alternative instruments.

  24. Gravatar of ssumner ssumner
    3. December 2009 at 13:55

    Nick, Sorry, I’m working too fast and misread your point. Sure, I agree that monetary policy is highly effective in that case. I guess I was thinking of other assets in nominal terms. If the Fed increases the money supply from $100 to $200 per capita, and the demand for cash and DDs is interest inelastic, then the nominal size of cash and DDs doubles. But the nominal value of stocks, real estate and commodities also doubles. Or am I still missing the point?

    (Also, see my second response to David.)

    Carl, I don’t have time to fiddle around with compounding. Assume first differences of logs.

    David, Then we agree. Let’s give equal time to banks and drug dealers in our M&B texts.

    On your response to Nick, I’d say the government can’t force us to hold either currency of reserves, just base money. But as a practical matter, in the absense of RRs, most cash would be held by the public, and bank demand for reserves (say vault cash) would be analogous to Walmart’s demand for base money. Nick will say that banks take those reserves and create media of exchange. And media of exchange drive AD. And then we are back to that debate-is it important that the medium of account is also the medium of exchange. I say no and Nick says yes.

    But I still can’t find a policy implication of our debate–especially as we move (hopefully) into futuristic regimes such as using OMOs to target the forecast. His thought experiments are so abstract I can’t figure out how they relate to the real world. But he probably feels that way about my thought experiments.

  25. Gravatar of Burt Bushmaker Burt Bushmaker
    15. March 2010 at 17:59

    Nick,

    Are you sure the monetary base is normally more than 90% currency? I thought I read somewhere else (I believe it was a magazine) that it was above 75%. I may be wrong though, and sorry if I am!

    Burt

  26. Gravatar of scott sumner scott sumner
    16. March 2010 at 10:54

    Burt, You can google the St Louis “Fred” website, they have all the currency and base data in an easy to access format. I think reserves were about 8% to 10% until September 2008.

  27. Gravatar of GANESH GANESH
    19. September 2010 at 09:09

    Burt, You can google the St Louis “Fred” website, they have all the currency and base data in an easy to access format. I think reserves were about 8% to 10% until September 2008.

  28. Gravatar of bocce balls bocce balls
    4. October 2010 at 00:04

    On your response to Nick, I’d say the government can’t force us to hold either currency of reserves, just base money

  29. Gravatar of atlanta ac services atlanta ac services
    29. December 2010 at 18:09

    Burt, You can google the St Louis “Fred” website, they have all the currency and base data in an easy to access format. I think reserves were about 8% to 10% until September 2008.it is great

  30. Gravatar of Josh Josh
    12. January 2011 at 00:53

    The answer is no. Nick Rowe doesn’t know what he’s talking about. The monetary system isn’t as complicated as many make it out to be. Bank capital is “irrelevant”. It isn’t irrelevant to the bank, Nick. Retake your approach and go back to the textbooks. Banks don’t invest how we think they should, they just secure our money and try to make more off of it.

    The only thing i agree with Nick on is the loan officer theory. I think it’s still applicable… if not more necessary than it was before our recent economic crisis.

  31. Gravatar of casas en uruguay casas en uruguay
    16. February 2011 at 01:48

    I have to admit, I find this post very confusing and difficult to reconcile with your past support of endogenous money perspectives. I suspect there is a missing element, which relates to peoples’ beliefs about how much wealth they own. This belief changes with a nominal wealth increase that is not immediately rendered into a price shock.

Leave a Reply