Banks maximize profits. Central banks maximize social utility. That’s why banks don’t matter.
This post was inspired by an excellent recent post by Nick Rowe. Unfortunately he won’t entirely agree with this one.
Is it more useful to define money as the monetary base, or as M1 (cash plus checking account balances?) Nick would say M1. I prefer the base. So let’s think about what makes the base different from checking account balances. The basic difference is that changes in the base are neutral; in the long run they affect all nominal variables proportionately. If the Fed wants to increase all nominal variables by 100 fold, it simply increases the base 100 fold. Because banks are profit-maximizing institutions, the DD/MB ratio is not affected (in the long run) by changes in the monetary base. It’s a real variable. In nominal terms the Fed is the dog and banks are the tail.
Banks can increase M1 by making it more attractive to hold demand deposits. They can pay interest on checking accounts, for example, or offer free services. That changes a real variable; DD/MB. But they cannot make the Fed print more money. That’s the key difference. The Fed can get the banks to create more deposits, but the banks can’t get the Fed to print more. The banks maximize profits while the Fed maximizes social welfare.
What about the fact that bank money is also a media of exchange? Doesn’t that make banks special? Not really. Monetary theory is symmetrical. There is essentially no difference between an increase in the supply of money and a decrease in the demand for money. Any sector of the economy that has a major impact on the supply or the demand for money can influence nominal spending as long as there is no monetary offset. If drugs were legalized and less currency was used for drug smuggling then there would be a drop in the demand for base money. Typically we don’t worry about this problem because of monetary offset. But that’s equally true of a change in the supply of media of exchange originating in the banking system. Banks don’t matter because the Fed won’t let them matter.
There are situations where a change in the private demand for base money does matter. The bank failures of the 1930s were important because we were on the gold standard and that limited the ability of the Fed to do monetary offset. If prohibition had increased the demand for currency during the early 1930s that would have also been deflationary. So banks did matter under the gold standard because of the difficulties in doing monetary offset. So did prohibition. Today, however, banks are not particularly special or important.
There is one possible exception to this claim. If the inflation rate is already relatively low, or more precisely if the nominal GDP growth rate is relatively low, then a major banking crisis could drive the nominal interest rate to zero. At that point monetary offset requires unconventional policies that central banks seem reluctant to engage in.
To summarize, Nick thinks banks are special because they create a significant fraction of transactions balances, a.k.a. M1. I don’t think banks are very important because of monetary offset. The Fed drives the nominal economy because it’s not a profit-maximizing institution. That gives it the ability to target nominal aggregates such as inflation or nominal GDP growth.
PS. This is my first post done with “Dragon dictate for Mac.” I find it difficult to get used to “writing” this way. Hopefully I will get used to it over time.
PPS. At least Nick and I agree that loans don’t matter.
Tags:
30. September 2013 at 19:30
I don’t see why Nick would disagree with any of this.
30. September 2013 at 20:15
Scott, you said;
“If the inflation rate is already relatively low, or more precisely if the nominal GDP growth rate is relatively low, then a major banking crisis could drive the nominal interest rate to zero. At that point monetary offset requires unconventional policies that central banks seem reluctant to engage in.”
I’m surprised you say this! It does not require unconventional policies. It requires exactly the same policy NGDPLT, and the same tool…OMP (exchanging money for financial assets).
30. September 2013 at 21:15
“There is one possible exception to this claim. If the inflation rate is already relatively low, or more precisely if the nominal GDP growth rate is relatively low, then a major banking crisis could drive the nominal interest rate to zero. At that point monetary offset requires unconventional policies that central banks seem reluctant to engage in.”–Scott Sumner.
Exactly!
And that’s why a 2 percent inflation target (and that means ceiling, and that means central bankers get nervous at 1.5 percent inflation) is probably too low.
An economy can sink into ZLB-recession land, while central bankers dawdle, pettifog about the perils of inflation, and talk about being prudent in the use of limited QE. That is what happened after 2008, for example.
30. September 2013 at 22:28
What’s the DD/MB ratio?
30. September 2013 at 22:35
“The Fed can get the banks to create more deposits, but the banks can’t get the Fed to print more.”
If the banks create more deposits they will demand more reserves due to reserve requirements and payments demand. Therefore they will push the fed funds rate above target and they force the fed to print fed funds. So the banks are forcing the fed to print more.
30. September 2013 at 22:51
“The basic difference is that changes in the base are neutral; in the long run they affect all nominal variables proportionately. If the Fed wants to increase all nominal variables by 100 fold, it simply increases the base 100 fold. ”
The empirical data doesnt seem to support this.
30. September 2013 at 23:57
The extent to which banks matter from a macro-economical standpoint is a measure of central bank incompetence.
“The extent to which X matters from a macro-economical standpoint is a measure of central bank incompetence” is valid for X in { fiscal policy, banks, shadow banking system, who the president is, leverage, whether there is meaning in life once you grow past the age of 30 and do not belong to any clique or party }.
1. October 2013 at 01:06
I think two papers from the Bank for International Settlements describe the role of reserves quite clearly:
BIS (2003): The role of central bank money in payment systems
“Thus, while central bank money plays a pivotal role in the economy, no central bank sees the use of central bank money as an end in itself. An increase in the amount of central bank money used for payments does not necessarily indicate greater economic welfare. Rather, central banks’ interest lies primarily in the use of central bank money at the apex of large-value payment systems, as a complement to the use of commercial bank money in such systems.”
BIS, Borio & Disyatat (2009): Unconventional monetary policies: an appraisal
“In fact, the level of reserves hardly figures in banks’ lending decisions. The amount of credit outstanding is determined by banks’ willingness to supply loans, based on perceived risk-return trade-offs, and by the demand for those loans. The aggregate availability of bank reserves does not constrain the expansion directly. The reason is simple: in order to avoid extreme volatility in the interest rate, central banks supply reserves as demanded by the system.
By the same token, an expansion of reserves in excess of any requirement does not give banks more resources to expand lending. It only changes the composition of liquid assets of the banking system.”
1. October 2013 at 02:49
The empirical data doesnt seem to support this.
That’s because it’s wrong.
1. October 2013 at 03:33
Scott: thanks!
I’m not 100% sure yet, but I think I might agree. Whether or not commercial banks matter for macro depends on what the central bank does. If the central bank does the right thing, they don’t matter. The central bank’s job is to try to ensure that nothing (including banks) matters for AD (curve).
(Commercial banks matter for all sorts of micro reasons of course, plus for monetary economics in the sense that sometimes cheques are more convenient than cash, and we do want a convenient monetary system, because trading frictions are very real, otherwise we wouldn’t use money at all.)
1. October 2013 at 03:39
And that banks create money Nick. Though you seem to find that inconvenient.
1. October 2013 at 03:58
Saturos, That’s good to hear.
dtoh, Conventional is usually described as interest rate targeting.
Kristian, Demand deposits/monetary base.
Mike, the central bank would only print more money if necessary to hit their nominal target. The Fed can change the price level. Banks cannot. Regarding your second point, there is a mountain of empirical evidence suggesting that money is neutral in the long run.
JCM, That sounds like the basic MMT fallacy, which I’ve addressed many times. So has Nick. It confuses short run base endogeniety (interest rate targeting) with long run base endogeniety and (inflation targeting.) The central bank moves the base as necessary to hit a short run interest-rate target, which is adjusted as necessary to hit its inflation target.
Nick, Thanks. I am curious as to your reaction to my symmetry argument. Banks can affect the supply of transactions balances but lots of other institutions affect the demand for transactions balances. Since monetary economics is symmetrical there is really nothing special about supply changes. That fact, combined with monetary offset, suggests that banks are not particularly special.
1. October 2013 at 04:12
Only currency is a relatively minor part of the money supply while reserves don’t form part of the money supply at all. The money supply is overwhelmingly bank liabilities. But banks aren’t special. Apparently.
1. October 2013 at 04:23
Scott: if by “special” you mean “special because banks can affect AD and other things can’t (given some non-optimal CB policy)” then you are right. Banks are not special in that sense. But maybe that’s not the only sense of “special”. They are “special” in the sense that they create money.
(I also wonder about how Banks respond to shocks, because if a shock hits banks will (generally) respond by changing both the supply and the demand for their money, by changing the rate of interest they pay on deposits. This is more of a Bill Woolsey question.)
sdfc: “And that banks create money Nick. Though you seem to find that inconvenient.”
Why should I find that inconvenient? It’s what I have been teaching first year students for the last 35 years. And what I was taught for a few years before that. God only knows where you get your ideas from.
1. October 2013 at 04:28
JCM: plus, that whole Neo-Wicksellian analysis of monetary policy is falling apart in front of your eyes. See for example my recent post: http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/09/two-neo-wicksellian-indeterminacies.html
1. October 2013 at 04:45
ssummers
” The Fed can change the price level. Banks cannot.”
If the fed can change the price level how is inflation below the fed’s target? It must be becuase they cant change the price level and they are hoping for banks to help them? Both banks and the fed influence the price level IMO.
“there is a mountain of empirical evidence suggesting that money is neutral in the long run.”
According to what I see broad money is neutral in the long run but not base.
1. October 2013 at 04:48
sdfc, That’s a silly non-argument. If I claim money should be defined as the base, saying it’s currently defined as M1 is not an argument, it’s merely a definition. The question is which definition is most useful? You haven’t responded to the post.
1. October 2013 at 04:55
How can central banks engage in monetary offset if the banking system is insolvent? (and there is a difference between insolvent and illiquid… During the great depression, most banks were illiquid. Spain today: most banks are insolvent)
Perhaps you think that by maintaining NGDP growth on trend, banks wouldn’t become insolvent in the first place?
However, I think that would be an oversimplification. There are tons of reasons why banks can become insolvent, and many aren’t related to monetary policy or stable nominal demand (although I agree that falling NGDP makes things worse).
1. October 2013 at 05:18
I think JN’s point is best re-stated this way: assume a worst case scenario, where a meteorite destroys all commercial banks’ assets (or the real assets backing those financial assets), and assume it takes some time before new banks can get started. We could still use central bank currency as the MOE (though the central bank would need to expand the supply a lot to match the increased demand). But in some cases bank deposits are a much more convenient medium of exchange than currency, so real trading volume would fall because of the practical difficulties of using currency as the MOE.
That brings the MOE back into the picture, in the right way. Barter is an incredible hassle, and lots of trades wouldn’t get made at all if we had to use barter. Using currency for all trades is much less of a hassle than barter, but it’s still a hassle.
1. October 2013 at 05:43
Thanks for the ‘restatement’ Nick.
My point is: assuming your scenario, how does the central bank expand the money supply with no banks as intermediary in practice? Does everybody has to queue up at their national central bank?..
Will the central bank replace all deposits with hard currency in such circumstance, and wouldn’t that be massively inflationary once a new banking system is in place and all this new base money is deposited back in the system (and consequently multiplied)?
It looks to me that the “banks don’t matter” point of view is a little extreme…
1. October 2013 at 05:52
Scott, even if you disagree with Stein on how monetary policy works, don’t you think that the fact that it impacts term premia needs to go into your model somewhere as something to consider when setting policy?
1. October 2013 at 07:30
The meteorite scenario has a real world counterpart when Irish banks were indefinitely shuttered in a 1970 strike. Pubs stepped in to facilitate transactions.
But the Fed doesn’t necessarily need a healthy banking system to move the price level up. All it has to do is sufficiently reduce the expected return on reserves to the point that all the unhealthy banks simultaneously spend away their reserve balances on assets, thereby pushing the price level higher.
1. October 2013 at 08:55
JP,
When banks are worried that asset prices will collapse or that counterparties (retail, corporates, or other banks) will default, they don’t spend away their reserves, even if the yield on reserves is zero. The simple reason is that 0% is better than -10%. (unless you meant a negative yield on reserve, in which case I guess that it would push banks towards safe securities instead)
Also, unhealthy banks are more at risk of runs. Thereby they may well wish to maintain their current level of reserves or even increase it.
1. October 2013 at 10:17
JN, You said;
“However, I think that would be an oversimplification. There are tons of reasons why banks can become insolvent, and many aren’t related to monetary policy or stable nominal demand (although I agree that falling NGDP makes things worse).”
I agree.
A collapse of the banking system would be a disaster, just as the collapse of the electricity generating sector would be a disaster. But neither has any bearing on our ability to raise NGDP. They are AS shocks. Would Zimbabwe have been unable to raise NGDP if their banking sytem was dysfunctional?
Saturos, No, it doesn’t need to be in my model, as my model doesn’t rely on interest rates. However it does make me revise my estimates of the strength of the liquidity effect. I always knew there could be a liquidity effect on long term rates, but that estimated effect is larger than I anticipated.
JP, I agree.
1. October 2013 at 11:56
“A collapse of the banking system would be a disaster, just as the collapse of the electricity generating sector would be a disaster. But neither has any bearing on our ability to raise NGDP. They are AS shocks.”
So would this increase the price level ceteris paribus?
1. October 2013 at 13:47
I am a little surprised you did not invoke one of David Glasner’s favourite papers: James Tobin’s “Commercial banks as creators of money”.
http://cowles.econ.yale.edu/P/cm/m21/m21-01.pdf
If asset prices are based on income expectations, then Steve Waldman’s point that credit expansion leads to asset price inflation still leads us back to … NGDP!
http://www.interfluidity.com/v2/4522.html
Is that was you would argue?
1. October 2013 at 14:40
[…] Sumner reiterated his “banks don’t matter” claim. For those who don’t know him, he is the Market Monetarism guru, a new school of thought […]
1. October 2013 at 17:31
“Banks can increase M1 by making it more attractive to hold demand deposits.”
No, No, No! Banks create money when they lend, not when they take in deposits.
1. October 2013 at 18:19
“Banks maximize profits. Central banks maximize social utility. That’s why banks don’t matter.”
Central banks maximize the utility of a subset of the population, not everyone. The proof of this is that its own existence rests on coercion, not consent. Coercion is required precisely because it doesn’t maximize everyone’s utility. If it did, no coercion would be necessary.
2. October 2013 at 01:21
“If the Fed wants to increase all nominal variables by 100 fold, it simply increases the base 100 fold.”
If we are at some kind of equilibrium now, then I can see that there is another equilibrium with 100 times the quantity of base money and all nominal quantities 100 times greater.
What I find harder to see is whether the economy can easily move from one equilibrium to the other. If base money were the only financial asset, it would presumably involve no more than a revision of the price of all goods. In the real world, of course, we have a complex network of financial assets and liabilities, most of which are denominated in nominal terms. The structure of these has a big impact on what happens in the economy.
Adjusting to the new equilibrium would therefore seem to require substantial financial flows and I don’t know how realistic it is to assume these would take place. Would banks make new loans to distressed borrowers to maintain the real value of their debt, for example? My instinct here is that the economy would end up in some different equilibrium, which may be a better or worse one than before. Is this something you have looked at in other posts?
2. October 2013 at 03:57
Nick
If you have been teaching that banks create money you’d better clear up what you mean by the statement.
Whether or not commercial banks matter for macro depends on what the central bank does. If the central bank does the right thing, they don’t matter.
Scott
sdfc, That’s a silly non-argument. If I claim money should be defined as the base, saying it’s currently defined as M1 is not an argument, it’s merely a definition.
I don’t see how it’s a silly non-argument when M1 etc is the money that drives economic activity, unlike bank reserves at the CB, which in normal times are a function of bank money creation and are currently merely a byproduct of Fed asset purchases.
2. October 2013 at 05:53
vidk, It would increase the price level if the Fed was targeting nominal GDP. It would not increase the price level if the Fed was targeting inflation.
Lorenzo, yes, it’s very hard to see how QE could increase asset prices without increasing expected nominal GDP growth.
Doug, I can’t tell if you’re joking. I presume you’ve read the Nick Rowe posts that show how it’s a simultaneous system. It makes no sense to talk about deposits causing loans or loans causing deposits.
Nick, I’m talking about long-run effects. If the change were unexpected, then the short run effects would be very messy as you indicated.
sdfc, In my view changes in the base drive changes in nominal GDP. Of course you also have to account for changes in the demand for base money, especially when nominal interest rates fall posted zero. Reserves play important role in the banking system, especially when there are reserve requirements. When interest rates are positive an increase in bank reserves leads to a large increase in bank deposits.
2. October 2013 at 06:18
I am not convinced 🙂
Scott, you said:
“But neither has any bearing on our ability to raise NGDP. They are AS shocks. Would Zimbabwe have been unable to raise NGDP if their banking sytem was dysfunctional?”
Agreed, I see how theoretically the central bank can drive NGDP by purchasing anything it wishes (although I would tend to call this fiscal policy).
My question is: what about the real world?
Central banks deal through the banking system right now so are limited in their ability to purchase assets in practice.
2. October 2013 at 06:58
Scott,
I was also thinking about the long run. Put another way, it seems to me that there must be various equilibria consistent with a 100-fold increase in base money. If all real variables stay the same, then presumably there is only one true equilibrium, which is the one with all nominal values increased 100 times. However, it is not obvious to me why the real variables should not be path-dependant and therefore why they must necessarily eventually be the same. Maybe someone has proved this must be the case – but I don’t think it’s obviously so.
3. October 2013 at 06:20
JN, You said;
“Central banks deal through the banking system right now so are limited in their ability to purchase assets in practice.”
This statement is false. The Fed buys bonds from the bond market.
Nick, I agree. That’s why I mentioned a 100 fold increase. That sort of increase would completely dominate any likely changes in real variables. So the increase in prices would be roughly 100 fold, although not necessarily exactly, as you said.