These two first-year textbook models -- the simple money supply multiplier model; and the simple keynesian income-expenditure multiplier model -- are formally identical. Translated into math, or game theory, you can't tell the difference between them. They contain exactly the same important insight: that what is true for the individual bank/household is not true for all banks/households. And they both contain the same flaw: they ignore interest rates.
This is really just a rhetorical device, designed for people who understand and appreciate the keynesian multiplier but not the money supply multiplier. Or vice versa.
Assume a closed economy, with no government. For "bank" read "household" throughout. The other translations are obvious (S=savings, I=investment, Y=income, C=consumption).
Let S be desired reserves. Let Y be deposits. Assume desired reserves are a constant fraction s of deposits: S=sY. Let I be the supply of reserves. In equilibrium the supply of reserves must equal the demand for reserves (desired reserves equal actual reserves), so S=I. In equilibrium therefore, deposits Y must adjust so that sY=I, which means we get the multiplier Y=(1/s)I. Just to complete the model, recognise that accounting reminds us that loans C plus reserves S must equal deposits Y, so that C+S=Y=C+I.
One important insight of this model is that the supply of reserves I determines desired reserves S, not the other way round. Deposits Y adjust to ensure that desired reserves S equals the supply of reserves I.
A second important insight is the difference between the individual bank and the banking system as a whole. An individual bank, one that is small relative to the system as a whole, will take deposits Y as given. The individual bank that increases loans C will suffer a dollar for dollar reduction in reserves S. But for the banking system as a whole, an increase in loans C will cause an equal increase in deposits Y, with no change in reserves S.
If banks are at less than "full employment", they could all be better off if they jointly increased loans C and deposits Y. They don't need extra reserves S to do this. And aggregate reserves S would not decrease if they did this. But this expansion of loans C and deposits Y can never happen, without an increase in the supply of reserves I. That's because any individual bank that increased its own loans C would see no corresponding increase in its own deposits Y, merely a reduction in its reserves S. All the benefits of increased loans C would be an increased level of deposits Y at other banks.
People who miss this important insight, who miss the distinction between desired reserves S and actual reserves I, end up with Say's Law of Banking, which states that there can never be a general shortage of reserves I, so banks must always be at full employment.
Now it is true that one bank's deposits Y are not strictly speaking exogenous to that bank. By offering better terms to its customers, it can increase its deposits Y at the expense of other banks' deposits Y. But that does nothing to increase aggregate deposits Y to the banking system as a whole, and get all banks to full employment. It simply redistributes a fixed amount of deposits Y among the different banks. One bank gets to full employment, but only by reducing employment for other banks. The only way to get to full employment of all banks, without an increase in the supply of reserves I is if all banks decrease their desired reserve ratio s. But no individual bank has any incentive to do that.
So the policy message is clear. The only way to get the total level of deposits Y to that required for full employment of banks is to increase the total supply of reserves I. And then hope that banks don't just increase their desired reserves S by the same amount. Ricardian Equivalence Theory, based on the permanent deposits Y theory of loans C, argues that any transitory increase in the supply of reserves I will indeed be mostly hoarded with a corresponding increase in desired reserves S. So lets make the increased supply of reserves I permanent.
Nick,
As I understand your model, the only thing standing between "Desired Reserves = Zero Excess Reserves" is the expectation that reserves are "permanent". What are the policy implications of that view?
If the Fed makes the current level of total reserves permanent, and if DR=ER0, then loans will expand by more than $10tr. Surely this is not a desirable outcome. That means the Fed must do one of two things: reduce ER (and make the remainder "permanent"), or raise the IOR to control the supply of ER.
To reduce ER, the Fed must either sell MBS or "sell" some sort of s.t. time deposit (reverse repo's etc). Selling MBS is problematic because, away from the money supply, it represents a "tax" on term yields (in the same way that buying MBS was a "subsidy" to term yields). Clearly, this is difficult to achieve in the midst of a fragile recovery in housing. As for time deposits, they are essentially the same as ER's. If reserves are made "permanent", the banks have a choice as to whether to invest them in s.t. risk-less assets, or longer term assets that carry interest rate risk (term Treasuries) or credit risk (loans, securities). The interest rate on Fed Time Deposits would have to be adjusted to raise or lower the opportunity cost of taking on that risk, which is essentially the function of the IOR.
Which leaves us with the IOR as the Fed's liability management tool. I think we can agree that the IOR determines bank reserve demand. But the Fed Funds rate also determines reserve demand in a Fed Funds targeting regime. That is, banks can have ANY amount of reserves they demand at a given Fed Funds target rate (because the Fed must inject reserves to keep the target rate from rising). What is the difference between the banks having $1tr in ER's controlled by the IOR, and the banks having IMMEDIATE ACCESS to $1tr in reserves controlled by the Fed Funds Rate? Nothing (barring operational issues raised by the GSE's). Therefore, making reserves "permanent" and controlling their conversion into required reserves with the IOR is essentially the same thing as having zero ER's and managing reserves with the Fed Funds rate. Am I wrong in thinking that having access to unlimited loanable reserves at a FF rate and having access to $1tr in loanable ER's at an IOR rate is essentially the same thing?
The upshot: for "permanence" to have an impact on reserve demand, the Fed would have to first soak up roughly 70-90% of Excess Reserves (or face a money supply explosion of upwards of $10tr). To do otherwise (manage ER's using the IOR) is effectively the same as doing nothing since an IOR-targeting regime is no different from a Fed-Funds targeting regime.
My question to you is, what level of Reserves should be made permanent, and if the number is less than $1tr, how would you remove the "excess"?
Posted by: David Pearson | February 20, 2010 at 10:53 AM
Nick,
If you are so flexible in translating reserve multiplier model to keynesian multiplier model, why are you so against translating multiplier model where reserves are just the liabilities of central bank to the more realistic model where there is a broader range of reserve assets?
Posted by: TheMoneyDemand blog | February 20, 2010 at 11:14 AM
"they both contain the same flaw: they ignore interest rates."
but now satan enters the picture
i wish you hadn't so gliberlaly sprayed the walls with barro -x ide
otherwise a nice essay in cheap model tricks
and
off the same raw formulaic bones
as produces the zeno paradox too
Posted by: paine | February 20, 2010 at 11:39 AM
Of course, the "multiplier" is even closer to the fundamental proposition of monetary theory. Of course the keynesian expenditure multiplier is just an awkward expression of the fundamental proposition of monetary theory.
Posted by: Bill Woolsey | February 20, 2010 at 06:11 PM
Nick. I don't get this. Say's Law should apply to banks; there should never be any reason for banks to hold more than the desired amount of reserves. If their desired reserve ratio is 1/10, and total reserves are $5000, then total deposits should be $50,000. And this should always be true, as any unwanted reserves could immediately be exchanged for T-bills. What am I missing?
Perhaps this is your argument: If reserves increase, then interest rates fall, and this increases the desired reserve ratio. Hence the money supply rises by less than the multiplier predicts. Is that the argument. I'm sure I missed something here.
The other difference is that in a flexible price world the money multiplier is correct, and the Keynesian multiplier is wrong. And of course the long run is a flexible price world.
Posted by: Scott Sumner | February 20, 2010 at 08:28 PM
NIck --
1) Why do banks desire reserves? Since the banking system is not at "full employment", there are clearly some banks with excess reserves in a regulatory sense, so it can not be a matter of present compliance. Households desire to hold money savings for insurance purposes, to cover mismatches between income an planned consumption. Banks the same? (Banks don't retire, so we can leave that part out.)
2) Never-retiring households rationally save less when they i) are certain of their future mean earnings; and ii) have ready access to credit at reasonable rates to cover transient mismatches. Then do banks desire reserves because i) they doubt their own solvency; or ii) they fear credit will not always be accessible on reasonable terms, despite the existence of a carefully managed reserves market?
There are certainly banks for which (i) applies. Banks of questionable solvency rationally hoard reserves, to increase the option value of their uncertain positions by delaying actual failure. But for a clearly well capitalized bank, it is hard to understand why a bank wouldn't lend reserves, if the rate on the loan more than covers the risk-adjusted return on reserves (in the reserves market or via the central bank). This deprives the lending bank of reserves, and creates deposits elsewhere (as the loan is spent), but the lending bank has no use for reserves as insurance and gains more value from reserves by lending, as long as it is sufficiently well-capitalized and the reserves market sufficiently well-organized that the insurance benefit of holding reserves is minimal.
I agree with you that stories that suggest absolutely no meaningful reserve constraint are overstated. But that (post-Keynesian / endogenous money) story is a better baseline model than one that suggest all banks are reserve constrained, jealous of reserves for insurance purposes like a nervous household. I think the sensible view is to understand there is a continuum of solvency among banks, the post-Keynesian "only capital constraints" matter story applies to the very well capitalized group, while reserve and liquidity-related constraints apply with increasing severity as bank solvency becomes questionable.
Posted by: Steve Waldman | February 20, 2010 at 11:12 PM
Wow, these last 2 posts are a veritable 'clash of the titans'. Interesting ideas, all.
But what about the depositors? I have yet to hear ideas on what happens to depositors when the banks are awash with reserves. Do they still continue keeping their money with the banks, knowing that it nets them no differently from putting their money under the mattress? What of their other alternatives? What if it's causing them to pour more money instead in objects with 'perceived' scarcity? Could this lead to an actual reduction in the multiplier?
Posted by: Rogue | February 21, 2010 at 12:50 AM
Steve: I'm not thrilled to see you here. Banks do NOT lend reserves. I've been through this with you many times, and you do not carry the emotional and psychological baggage that limit Professors like Nick, and you should know better by now. The only reason why solvency concerns might cause banks to hoard reserves is because credit risk seeps into the overnight interbank lending market. In effect, potentially insolvent banks get frozen out of their reserve accounts by their peers. But this has to do with poor discount window design, it has nothing to do with lending being reserve constrained, or the "money multiplier" fallacy at core of Monetarist beliefs.
Doesn't your "kumbaya -- it's all a continuum" cop out at the end nauseate you? Is a little rigor and clear-thinking too much to ask? An insolvent bank has more immediate constraints to extending loans that what kind of a deal it's going to get on ON money.
Nick: You're using analogies now to make your case? What's next -- theological arguments? Banks "multiply money" just like Jesus multiplies fish? The causality you picked in your model is arbitrary and backwards. You should move away from your "fallacy of composition" fallacy because it has banks checking their reserve requirements whenever they cash a check. What does it say about Economics that reality never gets a chance at bat?
"But this expansion of loans C and deposits Y can never happen, without an increase in the supply of reserves I… People who miss this important insight, who miss the distinction between desired reserves S and actual reserves I, end up with Say’s Law of Banking, which states that there can never be a general shortage of reserves I"
In real life, "desired" translates into "required" and they are provided by the Central Bank. PKs understand this, as does anyone who actually manages reserves in a real life bank.
Sumner: I marvel are your hermetic approach to all of this. Please never speak with an actual reserve manager at a real life bank. It will not be good for you (although you will get the opportunity to figure out what you are missing).
Posted by: winterspeak | February 21, 2010 at 03:56 AM
SRW,
“The post-Keynesian "only capital constraints" matter story applies to the very well capitalized group, while reserve and liquidity-related constraints apply with increasing severity as bank solvency becomes questionable”
- Capital constraints apply to all in the sense that excess capital must exist prior to the booking of a new loan and subsequently be allocated to the loan when it is drawn down; there is no corresponding temporal or spatial constraint with reserves
- the “capital constraint” terminology is my own, not originally from PK but used by some regularly now
- the meaning of “capital constraint” parallels the reserve case in the sense of whether there is an ex ante stock requirement for either – there isn’t with reserves; there is with capital
- also, there is no capitalizer of last resort as a normal, regular feature of the monetary architecture; once every 80 years or so there appears to be an extraordinary exception
- also, the central bank supplies the system with newly required reserves, with a regulatory time lag, following the creation of the new deposits that generated that requirement; there is obviously no corresponding lagged supply function with capital, with or without the government
- Reserve and liquidity constraints apply to none in the sense that the Fed is always available as lender of last resort, UNTIL the bank fails via FDIC or otherwise
- liquidity management is obviously a live and important issue in the sense banks want to avoid the window (in normal times) due to moral suasion and as a demonstration of cash and liquidity management competence, but it’s not a constraint in the meaning intended by the word in the PK use
- When the Fed forces an outsized excess reserve position on the system, as now, individual banks only “desire” reserves when the risk free alternatives (e.g. t-bills) earn no more than reserves
– “desire” in this sense means that banks won’t attempt to exit excess reserves delivered to them by clearings if they can’t find a better priced equivalent risk free alternative
- “desire” thus means the absence of a “desire” or reason to switch (risk free) rather than a desire to hold
- banks do not substitute risky assets for risk free assets
- they allocate capital to new risky assets, but only if they have excess capital, and in that case they’ll do that anyway, regardless of the starting risk free asset position
- Saying that banks are not reserve constrained doesn’t mean poorly capitalized banks won’t attempt to accumulate reserves in difficult times, but they certainly won’t accumulate them with the idea that they “multiply” them into new loans and deposits; they’ll accumulate them primarily as a call on settlement balances and protection against liquidity calls on the liability side – not for “multiplication” into asset expansion – that’s the point about rejecting the multiplier concept specifically in the EARLY STAGES of the crisis environment (in addition to rejecting it generally); "desire" for reserves in the CURRENT stage of the crisis has little to do with this precautionary motive and is almost entirely about the absence of better priced zero risk substitutes
- the meaning intended by the absence of a “reserve constraint” equates to the rejection of the multiplier concept as it is normally understood
- In summary, the capital constraint story is correct and the reserve constraint story is incorrect and the continuum story is inaccurate and inadequate when you understand the meaning intended by the word “constraint” in reserve and capital cases
- As Winterspeak notes, banks do not “lend reserves” and they do not require reserves ex ante in order to acquire risky assets; they require capital ex ante
- And banks do not make decisions to allocate capital to risk as a substitute for holding risk free assets; they make those decisions on the strength of the capital case, regardless of their risk free asset portfolio
- The entire explanation of the "desire" for reserves must be couched in the reality that the Fed has forced the current outsized excess reserve position on the system as a whole as a result of its own asset activity, and for reasons unrelated to reserves per se; it is irrational to expect that a forced result has an explanation in the normal sense of the word "desire"
(P.S. winterspeak - exception in this case :) )
Posted by: JKH | February 21, 2010 at 07:02 AM
SRW,
Macro constraints (or their absence) determine micro constraints (or their absence):
The central bank always supplies aggregate reserves sufficient to meet aggregate required reserves and steer the trading range for the policy rate toward the target policy rate. That means that aggregate reserves are always available sufficient for individual banks to meet their individual reserve requirements. That, plus the existence of a last resort facility, means that banks collectively and individually are not constrained in lending or the effect of lending on new deposit creation and new reserve requirements.
Conversely, the central bank and the government in the larger sense have no regular role in supplying aggregate capital sufficient to meet aggregate required capital in the sense of new risk taking. (The extraordinary exception being the 80 year crisis mode) That means that banks collectively and banks individually are always constrained in the sense that they require that their own source of excess capital be in place prior to booking new risky assets requiring fresh capital allocation. Such a constraint does not mean that banks can’t raise new capital. Rather it means excess capital must be on the books prior to new risk lending. This is not the case with reserves.
It is actually the central bank, rather than the commercial banks, that is reserve constrained:
a) It must supply adequate reserves in response to the generation of new reserve requirements by commercial bank deposit expansion, in order to control the upper trading level for the policy rate.
b) With asset expansion for non-reserve reasons (e.g. “credit easing”) it must at least consider the use of reserves as a liability management option.
c) If it exercises the option in b), it must pay interest on reserves in order to control the lower trading level for the policy rate.
The “multiplier” concept ignores the real world process by which the central bank provides required reserves following the creation of the requirement. It confuses ex ante reserve requirements for the capital requirements that are the real world prerequisite for the addition of new risk assets. Those assets are the source of the creation of new deposits and the knock on creation of new reserve requirements.
Posted by: JKH | February 21, 2010 at 09:29 AM
Does this analogy really hold in the case of the Canadian banking system?
"Reserves" aka settlement balances are almost completely netted out everyday from 6-6:30pm, after all transfers have gone through. Overnight balances are almost never borrowed/deposited at the BOC
It always nets out because the Banks know that for every dollar they are short, someone alse is necessarily a dollar over. Since you can lend settlement balances at a better rate than the deposit rate at the BOC, and borrow settlement balances at a better rate than the Bank Rate at the BOC,the overnight rate at which settlement balances are netted out is very very close to the target rate, exactly halfway between the deposit rate and the bank rate. Unless there is some sort of disturbance, or extremely skewed distribution of excess reserves,the BOC target is usually hit by market participants.
Banks in Canada can extend loans during the day without worrying about this "paradox of thrift", because they know that at 6pm they will ALWAYS have the opportunity to borrow any settlement balances that they are short, and at a better rate than that offered by the central bank.
Posted by: bob | February 21, 2010 at 10:43 AM
"there can never be a general shortage of reserves I, so banks must always be at full employment."
In Canada this is true...
I don't think your model can really tell us much with no interbank lending.
Posted by: bob | February 21, 2010 at 10:51 AM
Driving by in my new laptop and disappointed to see that there appears to have been little convergence in my absence. The balance sheet expansion production function with capital and reserves as factors of production made most sense to me, and that was last year!
JKH, I think you are too dogmatic about capital being a prerequisite for a bank to acquire risk assets. Banks are seldom right on the margin of their regulatory capital requirements - the ease with which banks were able to game capital requirements was one factor that contributed to the financial crisis. There must be sufficient flexibility for a bank to opportunistically acquire assets and subsequently increase capital. Why do you say that the Fed has "forced" reserves onto the banking system? They pay a market determined price.
Winterspeak, is it not possible that a bank might find itself with excess reserves from a chance surge in deposits, and then seek to spend them on assets such as loans?
Posted by: RebelEconomist | February 21, 2010 at 10:56 AM
I'm in agreement with the comments by Winterspeak and JKH.
There is one additional aspect not mentioned yet, though bob's comments are on the same track, which is assuming the pre-Lehman case with the target rate set above the rate paid on reserve balances, the qty of excess reserve balances desired at the target rate is mostly a function of the context within the cb achieves its target rate on a daily basis.
In the US, for instance, the Fed sets a very high penalty on non-collateralized overnight overdrafts (400bp + the day's fed funds rate), the settlement system is very decentralized, and the Fed has very little ability to perfectly offset net autonomous changes to its balance sheet on a daily basis (Tsy's account, float, currency demand, etc.). These contributes significantly to the qty of desired ER.
In Canada, it's just the opposite. The BOC has virtual perfect ability to offset net changes to its balance sheet by the end of the day, while individual banks have virtual perfect certainty they can clear any positive or negative balance in their reserve account at the target rate given the special settlement period, BOC's ability to offset autonomous changes, and fairly centralized settlement; even if there was a large penalty on uncollateralized overnight overdrafts as in the US, they would almost never be a material concern toindividual banks. Not surprisingly, Canadian banks desire no ER at the target rate.
Posted by: Scott Fullwiler | February 21, 2010 at 11:16 AM
bob – always nice to see some real world knowledge injected into these discussions
Scott – always real world knowledge from you, of course; nice detail on the comparison; I’m rusty on the details of the Canadian system - is the fact that Canada has paid interest on surplus balances for some time (I believe; inc pre-Lehman) also a contributing factor to a normally razor thin excess setting?
Posted by: JKH | February 21, 2010 at 11:29 AM
Rebel writes: is it not possible that a bank might find itself with excess reserves from a chance surge in deposits, and then seek to spend them on assets such as loans?
There are 2 answers to this:
1. A bank with undesired ER will attempt to lend them in the fed funds market. This is the only market in which (using the US case) reserves are actually directly loaned, though depending on settlement method used, this could also settle any number of other money market or asset purchases. This is assumed already by PK and fully recognized by JKH and Winterspeak (see some discussions here from last year).
2. On creating a new loan to a customer, a new loan creates a new liability for the bank. You're missing the tenor of the PK view here, which is to point out that the bank that has more ER doesn't have any more OPERATIONAL ability to create a new loan or purchase an asset than one that has less. The additional deposits can absolutely affect the profitability (net interest margin) of the loan or asset purchase vs. not having the ER, but this is always recognized by PK (and Winterspeak and JKH). In short, regardless what a bank with more ER does or doesn't do, the PK argument is quite different from the point you are making.
All this assumes capital is sufficient, too. And the fact that regulators were lax on capital requirements isn't an argument against JKH's points, BTW.
Posted by: Scott Fullwiler | February 21, 2010 at 11:35 AM
JKH . . . The theoretical models done by William Whitesell (see his FEDS papers at the BOG's site) suggest this is true. A related point is that if there is uncertainty about end-of-day balances by banks and on the part of the CB in hitting the exact desired qty of excess rbs at the target (as in the US but not in Canada, again) then averaging with reserve requirements may be necessary to flatten the demand even with a "channel" system.
You may find the PP here by Marc Lavoie of interest (slides 13-14 in particular): http://aix1.uottawa.ca/~robinson/Lavoie/Presentations/en/DR12.ppt
Posted by: Scott Fullwiler | February 21, 2010 at 11:46 AM
Returning after a weekend in Toronto, trying to catch up...
David Pearson: the bit about "permanent reserves" was really a bit of an afterthought. It's not central to the main topic of the post. I was seeing how far I could push the analogy. It was inspired by a comment that Josh had made on my earlier post http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/02/fallacies-of-composition-and-decomposition-the-supply-of-money-and-reserves.html?cid=6a00d83451688169e2012877afb573970c#comment-6a00d83451688169e2012877afb573970c
I think it's important, but haven't really thought through the policy implications. An increase in reserves that were expected to be permanent would have a bigger effect on the supply of money that one that were expected to be temporary.
But I get really confused when you talk about "excess reserves". Does it mean "excess desired reserves", or "excess required reserves"? The former matters. The latter is irrelevant, for a given value of the former.
Posted by: Nick Rowe | February 21, 2010 at 07:00 PM
TMDb: "Nick,
If you are so flexible in translating reserve multiplier model to keynesian multiplier model, why are you so against translating multiplier model where reserves are just the liabilities of central bank to the more realistic model where there is a broader range of reserve assets?"
Good question. I think that Bill Woolsey's somewhat obscure comment here is the answer:
"Of course, the "multiplier" is even closer to the fundamental proposition of monetary theory. Of course the keynesian expenditure multiplier is just an awkward expression of the fundamental proposition of monetary theory."
"An excess of desired savings over desired investment" in the Income Expenditure model means a excess demand for any good except newly-produced goods. It *ought* to mean an excess demand for the medium of exchange. There is no Paradox of Thrift. There is a Paradox of Hoarding Money. The Income Expenditure model is close to the truth, but misses, because it casts its net too broadly. (I have argued that point elsewhere on this blog).
The simple money multiplier model is best when it casts its net narrowly too, and considers only an excess demand for those assets that are media of exchange between banks (i.e. "reserves"). (I have not argued this point before. I only saw this point with Bill's help. I wish I could see it more clearly.)
Posted by: Nick Rowe | February 21, 2010 at 07:16 PM
This comment below is from Steve Waldman via Email. He tried to post it, but for some unknown reason TypePad wouldn't accept it. (I sometimes get the same problem, and have to conpy my comment, log out, log back in, paste my comment, make a tiny change to my comment, then it seems to work.) Sorry Steve. Very glad to have you here.
"1) I think (from comments here and elsewhere) that our host at this
blog is interested in considering the degree to which fallacies of
composition lead us to adopt an overly simplistic view of banking. I
think he's right to be concerned about this. The post-Keynesian story
is very good, much better than the money multiplier story, at
describing the banking system in aggregate. But it misses details
about the behavior of individual banks that are somewhat captured by
more "orthodox" stories.
2) Banks really do manage their liquidity. They incur costs to do so.
In a world in which governments opened Winterspeak's metaphorical
umbrella, they would not do so. They would borrow without friction
from the central bank or the interbank market at will, and liquidity
would be utterly superfluous. That is not the world we live in. Except
during a general crisis, central bank borrowing is stigmatized and
requires collateral. Interbank loans are bilateral and unsecured. Some
banks, by virtue of strong balance sheets to implicit government
guarantees, can borrow reserves frictionlessly. (Canada, which is
dominated by a few large banks, is probably more perfectly post- Keynesian than the US.) Many of the seven-ish thousand US banks cannot
be certain that any reserve shortfall will be covered by borrowing at
the headline Federal Funds rate. At best they face a credit spread.
The US central bank just raised its direct lending "discount rate",
and explicitly stated they were doing so to discourage, to
restigmatize, relying upon direct borrowing of reserves. This might be
poor policy. I might be persuaded that we should "rationalize" banking
along the lines that Winterspeak frequently proposes. But it is the
world as it is.
3) No one is claiming that all holdings of excess reserves are
precautionary. (I should caveat that... Free Exchange and TBI recently
ran posts with that rather idiotic implication.) It is obviously true
that in aggregate, the quantity of reserves is determined by the
central bank, and the central bank can force excess reserves into the
aggregate system in whatever quantity it desires and which the system
must accommodate. The point is thatsome banks (smaller, less
flamboyantly well-capitalized) do desire precautionary reserves, so
some bank lending is in part reserve constrained. Not aggregate
lending, which the central bank will accommodate. But built into the
cost of a loan from Podunk State Bank is a fee for the drain on
Podunk's liquidity, given that it will have to pay a credit spread or
prematurely liquidate securities if it runs shy of liquidity. In a
frictionless system with perfect information, that would just mean
that Podunk couldn't lend competitively and all lending would migrate
to poorly-capitalized-but-guaranteed Citibank. In the real world,
customers face search costs and may already do business with Podunk,
and Podunk may be able to underwrite the loan more efficiently by
virtue of its relationship and information about Jane Shopowner.
4) All of this is perfectly consistent with the broad post-Keynesian
story of how the banking system behaves in aggregate. To acknowledge
micro-level imperfections, and that they might matter, is not to take
some wishy-washy middle ground as a political position.
5) Note that these microlevel imperfections can matter, even to
behavior in aggregate. It is possible that Jane Shopkeeper is a risky
a credit for Citi, which has frictionless access to reserves but does
not babysit her children, but not for Podunk, which does do child care
but must demand a spread to cover the cost of liquidity risk. The net
effect might be that Jane doesn't take a loan from anyone: Citi
demands a large credit spread, Podunk demands a smaller credit spread
but also liquidity spread, which together add up to the same large
spread. If Podunk had a larger quantity of precautionary reserves, it
would have required a smaller spread for liquidity risk, and Jane
would have taken the loan.
6) Understanding this should be good news for post-Keynesians, because
it explains the empirical literature which does find that
contractionary monetary policy disproportionately affects lending by
smaller, less creditworthy banks. In a perfectly horizontal world,
monetary policy would uniformly affect the hurdle rate for loans to
small banks. So an overdogmatic horizontalist can't explain observed
variations. The core accuracy of the post-Keynesian description of the
banking system in aggregate can be much better defended if the
contours of microlevel variation are acknowledged, understood, and
explained.
7) The upshot is that, yes, reserves policy can matter under present
institutional arrangements, but only at the margins of the banking
system, where small banks meet small customers who are informationally
bound together. However, "expansionary" reserves policy cannot
possibly be a reasonable description of excess reserves at the present
scale, largely held by large reserve-unconstrained banks. That is an
outgrowth of the Fed's asset-side policy choices, not a matter of bank
accommodation. Withdrawing reserves (putting aside indirect effects in
asset markets) won't be meaningfully contractionary, because the
aggregate banking system is not meaningfully reserve constrained. But
in a world with institutional frictions, when excess reserves are
pulled from huge to near-zero (we've a long way to go), it may
actually affect lending by smaller, less well-known and well- capitalized banks, as it has in the past.
JKH -- a minor note: your contributions to these debates have been
extraordinary. but the notion of a capital constraint in banking is
perfectly orthodox among academics. if you take a banking class, the
first thing you will learn is "two constraints, capital and
liquidity". your (and the post-Keynesian) contribution has been to
point out, very accurately, that given a reserves price-making,
quantity-taking central bank, the liquidity constraint only binds at
the margins and capital is overwhelmingly the more important concern."
The comment above is from Steve Waldman, via Email
Posted by: Nick Rowe | February 21, 2010 at 07:33 PM
Scott: "Nick. I don't get this. Say's Law should apply to banks; there should never be any reason for banks to hold more than the desired amount of reserves. If their desired reserve ratio is 1/10, and total reserves are $5000, then total deposits should be $50,000. And this should always be true, as any unwanted reserves could immediately be exchanged for T-bills. What am I missing?"
Say's Law is (sometimes) false, in a monetary exchange economy (one with a medium of exchange, rather than a single Walrasian market in which all goods exchange for all goods simultaneously). Say's Law is false when there is an excess demand for the medium of exchange. When the real money supply is "too small", real GDP will also be "too small".
Say's Law of Banks (I just made up that concept) is equally false when there is an excess demand for reserves, which function as a medium of exchange between banks. When real reserves are "too small", the real size of the banking system will also be "too small".
Yes, when the price level is perfectly flexible (in the LR), the real money supply, and real reserves, will adjust endogenously through changes in P. So real money and real reserves cannot be too small in the LR.
The individual bank (or person) can always get rid of excess reserves (or money), or get more reserves (or money). But banks (people) in aggregate cannot get rid of excess reserves unless the central bank (banking system) reduces the total supply of reserves (money).
Posted by: Nick Rowe | February 21, 2010 at 07:55 PM
Scott: "The other difference is that in a flexible price world the money multiplier is correct, and the Keynesian multiplier is wrong. And of course the long run is a flexible price world."
Fair enough. Good point. The supply of reserves 'I' in the money multiplier is a nominal variable. Investment 'I' in the Keynesian multiplier is a real variable. So the price level can always adjust to make the money multiplier true, but can't do the same for the keynesian multiplier.
OK. Let me duck/cheat my way out of this one, by re-defining my question: *In the Short Run, when prices are fixed*, how could someone like one multiplier but not like the other multiplier?".
Posted by: Nick Rowe | February 21, 2010 at 08:08 PM
Steve @Feb 20, 11.12
"NIck --
1) Why do banks desire reserves? Since the banking system is not at "full employment", there are clearly some banks with excess reserves in a regulatory sense, so it can not be a matter of present compliance. Households desire to hold money savings for insurance purposes, to cover mismatches between income an planned consumption. Banks the same?"
Damn good question. And "yes" to your suggested answer. But let's be very precise here. The comparison should be between households' demand to hold a stock of *money*, the medium of exchange used by households, and banks' demand to hold a stock of reserves, the medium of exchange used by banks. It's not all forms of household savings we are talking about. It's about the demand for money. We use money in all transactions, including when we buy and sell bonds and shares, not just when we buy consumption goods.
In principle, I could do the same flow of income and expenditure, using money, while holding on average a vanishingly small stock of money. But I don't. I like to keep about $100 in my wallet, and about $1,000 in my chequing account (or whatever). My desired velocity of circulation of money is finite.
So in my mind it's the same for banks, and the monetary base. In principle they could do the same flow of transactions holding a vanishingly small stock of currency and other reserve balances. But they don't. Their desired velocity of circulation of reserves is finite.
I'm not sure if what you say here is right:
"I think the sensible view is to understand there is a continuum of solvency among banks, the post-Keynesian "only capital constraints" matter story applies to the very well capitalized group, while reserve and liquidity-related constraints apply with increasing severity as bank solvency becomes questionable."
First, the banking system as a whole can always get more capital (at a price) by issuing new shares. (If existing banks are insolvent, they can't do this of course, but new banks could). But the banking system as a whole can only get more base money (Bank of Canada liabilities) if the Bank of Canada chooses to let them have it.
Second, I would have thought that capital constraints become more relevant for banks that are *badly*-capitalised? Was that a typo above in your quote?
Posted by: Nick Rowe | February 21, 2010 at 08:31 PM
Rogue: Yes, what depositors want is totally missing from the banking multiplier story. As I said in my earlier post: I think that's a feature, not a bug. Unlike all other assets, it's possible to get more money into circulation even if people don't want to hold more money. Each person accepts it, because each believes he can get rid of it to someone else. When I willingly sell my car for $1,000 cash, that doesn't mean I would rather *hold* $1,000 cash than hold my car. It means I would rather hold a motorbike I can buy with $1,000 than hold my car. New money is always dropped by helicopter, in effect. (Subject for a later post).
Posted by: Nick Rowe | February 21, 2010 at 08:39 PM
Winterspeak: Steve is only here trying to reform us professorial streetwalkers; I'm sure his intentions are entirely honourable! ;-)
Yes, I sometimes argue by analogy. It's legit. Do you like the Keynesian income-expenditure model? If so, why is it good but the money multiplier bad? OK, you believe the supply of reserves (I) is perfectly interest elastic. Would you reject the income expenditure model if investment (I) were perfectly interest-elastic?
Posted by: Nick Rowe | February 21, 2010 at 08:49 PM
Steve,
“The notion of a capital constraint in banking is perfectly orthodox among academics.”
That doesn’t surprise me, although I see virtually no evidence of it from some of the monetary economists in the vicinity. It’s certainly a point that’s understood by those who work in banking, especially those who work in the related functional areas. And I have no reason to doubt that’s it is well understood by PK’s. What I actually noted was that I hadn’t encountered that particular phrase in reading the PK’s in the same way I had encountered the parallel phrase of “reserve constraint”.
To say that banks are not reserve constrained is not to suggest that they don’t manage their liquidity. All banks operate under constraints of regulatory and/or self-imposed liquidity policy.
But that in no way means they hold stocks of reserves as a regular matter of preparing for balance sheet expansion over time. That’s the PK point that banks are not “reserve constrained”.
Liquidity in general is an actual or potential claim on settlement balances. Liquid assets and the capability to issue new liabilities are typical categories of liquidity.
The PK point is that banks do not expand their balance sheets on the basis of working from a reservoir of settlement balances. There is no fallacy of composition issue here. It’s true for the system, and it’s true for individual banks.
The Canadian system demonstrates this with zero reserves of course, at both the systemic and individual bank level. Scott F. has referenced this in some detail above. The Canadian architecture is a graphic demonstration of the PK point on reserves.
And I’m surprised by your point on small banks. I had thought the US regional banks as a group were generally sellers of fed funds to the money center banks. I don’t have that backwards do I? If not, it proves the point further. Fed funds sales represent liquid access to settlement balances; not settlement balances per se. It’s good liquidity management – not stockpiling of settlement balances.
The management of liquidity per se, with respect to the issue of access to settlement balances, is quite apart from the false issue of the multiplier. Banks manage liquidity in order to access settlement balances when they need them. That’s another way of saying that there’s a better way of managing liquidity requirements than by simply holding stockpiles of settlement balances.
And if you want to relate balance sheet management and balance sheet growth to liquidity policy, that’s an entirely different scope and point than the PK observation about the nature of reserves per se.
And there is no fallacy of composition issue here as it relates to the PK observation on the role of reserves. The stockpiling of liquid assets as a matter of liquidity policy has nothing to do with the issue of “reserve constraints”. PK’s are quite aware that banks manage their liquidity profile and that they heed “moral suasion” rules about window borrowing where and when it is appropriate to do so. The PK point has to do with the erroneous multiplier concept. That’s not at all the same issue as the functionality that liquidity management provides in terms of access to settlement balances.
The very point of liquidity is NOT to have to rely on stockpile reserves for purposes of settlement balance access. Access to settlement balances is obviously required in running a bank, whether it’s through prudent liquidity management or LLR. But NO bank systemically stores settlement balances in order to grow its balance sheet according to a “multiplier”.
These well understood issues of individual bank liquidity management having nothing to do with the fallacy of composition, at least not in the way you diagnose it with respect to the PK theme on reserve constraints.
There are two reasons for the outsized excess reserves that central banks have produced during the credit crisis. The first was to increase real time access to settlement balances in an environment where banks were supremely cautious about credit risk emanating from other banks. The second was the central bank choice to use excess reserves as an indefinite liability management tool in absorbing the immediate monetary effect of asset expansion for credit easing purposes. Neither of these relates to a multiplier mechanism.
Posted by: JKH | February 21, 2010 at 08:50 PM
A couple of general comments, because I can't keep up with all the comments:
1. Only a few comments have addressed my main challenge in this post: what is the difference between the keynesian multiplier and the money multiplier models? How can you like one but not the other?
2. Horizontalists insist that the supply of reserves is perfectly interest-elastic.
a) suppose it weren't. Suppose the Bank of Canada made it perfectly inelastic. Would you reject horizontalism and embrace the money multiplier model?
b) Long run elasticities are not always equal to short run elasticities. The Bank of Canada (normally) keeps the interest rate target fixed for 6-weeks only. If you consider a longer time-horizon than 6 weeks, can you still say that the supply of reserves is perfectly interest-elastic? How can you be certain that the supply of reserves is not perfectly interest-inelastic over some longer period?
Posted by: Nick Rowe | February 21, 2010 at 10:17 PM
Nick: "I think that's a feature, not a bug. Unlike all other assets, it's possible to get more money into circulation even if people don't want to hold more money."
But I think flooding commercial banks with reserves without regard to its effect on depositors' rates is more a feature for Wall Street investment firms looking to capitalize on depositor disenchantment. Even if they get only those disenchanted enough at the margin, that could still be enough to explain for the return of risk among investment funds.
Posted by: Rogue | February 21, 2010 at 10:24 PM
Steve,
Banks do not lend on the basis of available real time settlement balances. They use liquidity management where necessary to attract the settlement balance flow necessary to square their reserve position as it is affected by new lending. The management of the settlement balance effect of lending is a RESPONSE to that lending. This is what it means to say banks are not reserve constrained at the micro level. At the macro level, banks are not reserve constrained because the central bank provides reserves in RESPONSE to the requirements generated by deposit growth. Where reserve requirements exist, banks maintain the required settlement balances in RESPONSE to the central bank’s supply of same. There is no ex ante reserve multiplier.
Banks lend on the basis of available capital. They must have excess capital in place in ANTICIPATION of a new loan being booked. At the time of booking, they allocate required capital from that excess position. That pre-existing requirement is what it means to say a bank is capital constrained. It does NOT means that banks can’t raise new capital. It DOES MEAN that banks MUST stockpile excess capital prior to lending. Banks do NOT source required capital after the fact of making a loan in the same way that they source settlement balances after the fact through liquidity management. And banks do not “trade” capital positions with each other the way they buy and sell fed funds. Capital is a stock requirement for banks before they lend.
Posted by: JKH | February 21, 2010 at 10:27 PM
"2. Horizontalists insist that the supply of reserves is perfectly interest-elastic."
Not true. They insist that the cb supplies reserves at its target rate such that the target is achieved, while allowing the quantity supplied to float (while recognizing the valid points regarding liquidity mgmt Steve and JKH have been making--there's no inconsistency there, as JKH explained). This is very different from your point, Nick, as it is in no way inconsistent with your points regarding observed or otherwise estimated short-run or long-run elasticities.
"a) suppose it weren't. Suppose the Bank of Canada made it perfectly inelastic. Would you reject horizontalism and embrace the money multiplier model?"
Assuming you set the vertical qty at a point different from what banks desire to hold at the target rate, what would happen is you would have either (a) an increase in the overnight rate until it hit the central bank's penalty rate, at which rate the CB would then lend in order to avoid a massive crisis in the payments system, or (b) the overnight rate would fall to the remuneration rate (or zero if there is none). It's impossible NOT to set an interest rate somewhere, and CBs have learned over time (the 1979-1982 period in the US, for instance) that it is far better for the functioning of money markets and financial markets in general to avoid too much day-to-day volatility in the rate. Ulrich Bindseil at the ECB (whose book I highly recommend, particularly to neoclassicals who don't want to take our word for it that this is how cb operations actually work) noted this as well in his more neoclassical models of central bank operations.
FWIW, these last two threads are filled with misinterpretations of PK, Chartalism, etc., and no member of any of these groups has found any of the critiques here or in the previous one the slighted bit persuasive (and some have been paying attention). Whether we critique you or vice versa, everyone ends up talking past each other, unfortunately. Don't really know how to fix that, but I wish I did.
Best,
Scott
Posted by: Scott Fullwiler | February 21, 2010 at 10:46 PM
Nick,
Team Canada just lost to the USA in olympic hockey, I'm having a hard time seeing any good in the world so I can't really pay attention to the econ arguments.
If anything can get me through tomorrow it is my belief that Canada did in fact play better with the exception of Brodeur who played like a blind and deaf three legged dog. So perhaps things are not as bad as they seem.
In any event I've always thought it was better for a team to struggle and overcome then cruise through the prelimary round too easily but it would certainly have beeen better to overcome and win this game. Now USA really has confidence.
It's getting close to 4am here and I'm drinking beer to help me at least try to see a bright side and get some sleep, good luck with the arguments.
Posted by: Adam P | February 21, 2010 at 10:53 PM
I should clarify my point above. Under a gold standard or a currency board, you do have a situation where the supply curve is generally inelastic. In those circumstances, the money multiplier is essentially valid (with some technical qualifiers), and chartalists point this out all the time. And that's why you get payment crises eventually in those systems.
Posted by: Scott Fullwiler | February 21, 2010 at 10:53 PM
Scott,
I don't think the money multiplier is valid even in a gold-standard context. One way to check this is to look at the U.S. data in the gold standard era. There is not a fixed relationship between the quantity of gold and the quantity of loans. The level of gold serves as a (soft) ceiling, but not a floor. And because the ceiling is soft, you can get to high multiples prior to the payment crisis, at which point you have the same dynamics as you saw in Japan; shrinking balance of loans outstanding, deflation, zero short term rates and falling long term rates, falling ratios of loans to reserves, etc. Loans create deposits even in a gold-standard world, but you get more and sharper banking crises in the process.
Posted by: RSJ | February 22, 2010 at 12:45 AM
Hi RSJ . . . agree (that was essentially the "technical qualifier" I was referring to--loans still create deposits as a matter of accounting).
Posted by: Scott Fullwiler | February 22, 2010 at 01:08 AM
Nick:"Only a few comments have addressed my main challenge in this post: what is the difference between the keynesian multiplier and the money multiplier models? How can you like one but not the other?"
I think that most obvious difference is that Keynesian multiplier is about demand side, while money multiplier is about supply side. A country can increase its output Y by Keynesian multiplier, until it hits the ceiling Y*. On the other hand, money multiplier defines the ceiling in itself. A country can increase money supply by money multiplier, but if there isn't enough demand for loan, imbalance between supply and demand ensues. And as supply and demand must equal ex post, the ceiling becomes lower, and reserves not used for money creation become excess reserves.
And...
Nick@20:39"Unlike all other assets, it's possible to get more money into circulation even if people don't want to hold more money."
Isn't this Say's law (of money)?
Posted by: himaginary | February 22, 2010 at 11:10 AM
Nick, I think I see your point here, but let me restate it in my own language, and see if it sounds anything like what you are thinking. Suppose the nominal aggregates in the banking system are 20% too low to provide for full employment in the economy. I.e. you have $800 billion reserves and $8 trillion in deposits, but you really need $1 trillion of reserves and $10 trillion in deposits to get AD up to a full employment level. So there is unemployment in the economy. I think what originally threw me off is that the banking system still might be in nominal long run equilibrium. Banks are holding the 10% reserve ratio I assume they want to hold in the long run. So in nominal terms the banking system isn't too small, it doesn't need to get bigger to reach equilibrium.
But I have also assumed the general economy is in recession because the money supply is 20% too small (and prices are assumed sticky.) In that case you need to either boost reserves by 20% (actually 25%) or you need to reduce prices 20% to reach the LRAS. If you think about the 20% reduction in the price level option, it has the effect of increasing every real aggregate in the banking system by 25%. So even though the banking system was already in long run nominal equilibrium, in real terms it was doing too little banking, too little real intermediation. And that problem can only be fixed by either a smaller price level or a bigger reserve level. Does this make any sense?
Posted by: Scott Sumner | February 22, 2010 at 09:39 PM
Nick,
really, why is this so hard to understand?
Let's take Sumner's example above where the minimum desired reserve ratio is 10% but add a desired minimum capital ratio that is also 10%. Suppose we beign with the case $800bil of reserves and $8 trillion of deposits and $800 billion of capital (all aggregate system wide levels).
Now suppose you PERMANENTLY inject another $200 billion of reserves, do deposits increase by $2trillion? NOT IF THE BANKS DON'T SIMULTANEOUSLY RAISE MORE CAPITAL. With $800bil in capital any new loans, even fully reserve backed, take the capital ratio below the desired minimum.
Posted by: Adam P | February 23, 2010 at 05:15 AM
continued...
The question you should now be asking is, why does this only appear as a binding constraint when short-rates hit the zero bound? Because banks hold long-term fixed rate (but risky) assets and short-term floating-rate liabilities (continously rolled over). Thus, when the fed lowers the short rate the bank makes a capital gain on its assets but suffers no capital loss on its liabilities (a continoulsy rolling FRN is always worth par). LOWERING SHORT-TERM RATES ITSELF INCREASES THE BANKS CAPITAL. An equivalent way to say it is that what the fed is doing is raising the risk-premium that banks earn on their assets, thus raising their profitablity and increasing the value of their capital.
Notice, lowering interest rates can increase the supply of credit but increasing the supply of reserves can't. The quantity of money, provided it is enough to keep the system liquid, has nothing to do with anything. It is all about interest rates (and other rates of return/risk premia).
Posted by: Adam P | February 23, 2010 at 05:16 AM
I do not have experience in a bank treasury, but to me it seems too dogmatic to say that capital is generally the binding constraint rather than reserves - although actually I prefer the term "restraint" because I think it expresses the idea of something more elastic than a sharp constraint. My point about the somewhat subjective nature of capital revealed by the financial crisis is that it contributes to this elasticity, rather than being a dig at lax regulation. To some extent, I think the disagreement here is semantic. Although JKH says that banks are capital constrained, he also says that banks hold "excess" capital in anticipation of lending opportunities. If one redefines this excess as being part of the bank's (own rather than regulatory) required capital, then capital is not such a binding constraint. I would expect that banks similarly hold a slightly higher level of reserves than the statutory requirement for precautionary reasons, although not much more if they can generally rely on borrowing reserves late in the day. In an ideal world, one imagines some operational researcher calculating the bank's desired level of both capital and reserves according to some expected cost / benefit analysis.
I was sorry to read of Adam P's hockey anguish, but I see Canada is going to get another chance in the final.
Posted by: RebelEconomist | February 23, 2010 at 06:25 AM
Rebel,
yeah, I think we all agree that banks need both liquidity (meaning reserves) and capital. My point above was simply that if the capital constraint is binding then increasng the supply of base money won't increase credit yet Nick and Sumner deny this. If the problem is lack of bank capital then it has nothing to do with either supply and/or demand for base money or the permanence of the reserve injection, it is the disintermediation that is killing us.
Further, it would appear obvious that right now, if any constraint is binding it is the capital one. Of course these go hand in hand, if a banks capatlization is questionable it can find itself without liquidity (that is it gets run).
Posted by: Adam P | February 23, 2010 at 06:55 AM
PS: and I'd imagine JKH's point is that if banks ever did find their liquidity (aka reserve) constraint binding then that would show up as a rise in the inter-bank lending rate (fed funds in the US) and so if the CB is conducting policy by setting an inter-bank lending rate then keeping the rate at the chosen level REQUIRES the CB to supply reserves whenever the constraint begins to bind. So as a practical matter it is not the constraint that matters.
Posted by: Adam P | February 23, 2010 at 07:03 AM
" suppose it weren't. Suppose the Bank of Canada made it perfectly inelastic. Would you reject horizontalism and embrace the money multiplier model? "
No. I could design a crappy plane that doesn't fly properly by making up my own laws of physics. Even if Boeing built the plane, it wouldn't make those quack theories correct, it would just provide a useful demonstration of why they are wrong, as happened with the attempts at implementing monetarism.
Posted by: bob | February 23, 2010 at 08:34 AM
"I think we all agree that banks need both liquidity (meaning reserves) and capital."
OK, help me out with something here.
In Canada, the BOC runs what they call a "zero reserve system": not 'no required reserves', zero reserves.
Intraday liquidity in the LVTS is guaranteed by the Bank of Canada. They stand behind all transactions, and will clear them even in the case of a participant failing. In exchange for this, the BOC requires a pledge of collateral from all participants in proportion to the potential size of the intraday settlement imbalance that could be realized on their account. This collateral is capital on the bank's balance sheet, right? Does that replace the role of reserves in terms of liquidity management?
Is there really any place at all for the concept of reserves in the Canadian system, or is it really entirely about capital constraints on pretty much every level?
Posted by: bob | February 23, 2010 at 09:39 AM
Hi bob,
Collateral is an asset, like a Treasury security, not capital. In theory, it has to be something the lender can sell in the event of non-payment. For the cb, of course, this isn't a problem for its own financial standing as the cb creates reserves in the first place, so collateral requirements are more for incentivizing discipline in bank behavior.
Incidentally, the Fed's LVTS, Fedwire, has much the same arrangement with banks. In Regulation J, it guarantees all payments sent via Fedwire will be made whole whether or not the sender's account has a positive balance. So, why do US banks hold ER but Canada's banks don't (under pre-Lehman operating procedures)? As I noted previously above, the combo of decentralized payment settlement, inability of the Fed to perfectly offset changes to its balance sheet, and large penalty on overnight overdrafts (uncollateralized--not to mention additional frown costs of collateralized overdrafts at the primary lending rate) lead banks in the aggregate to desire positive ER, whereas the opposite is the case in Canada.
Posted by: Scott Fullwiler | February 23, 2010 at 10:25 AM
I didn't know about that feature about the Canadian system but I'd say that it's about capital only.
Posted by: Adam P | February 23, 2010 at 10:25 AM
thanks Scott
"Collateral is an asset, like a Treasury security, not capital. In theory, it has to be something the lender can sell in the event of non-payment."
Of course.
So if a bank is required to post more collateral, it would need to increase its assets by acquiring more treasuries, etc. for that purpose. That would require them to raise more capital, right?
Posted by: bob | February 23, 2010 at 11:14 AM
So, the BOC gets an ownership stake in a bank if it fails to pay back the overdraft? That seems very strange, but I'm not familiar enough with settlement regulations in Canada to go further than I have.
In the US, I know for sure that the collateral requirement is that the bank submit some of its financial assets from a pre-approved list.
Posted by: Scott Fullwiler | February 23, 2010 at 11:19 AM
No no, it's the same thing here: government bonds, munis etc. They have a list of acceptable collateral, it's not bank capital. I agree, and you were right to correct that
What I'm asking is whether or not, in order to acquire that additional collateral, the banks would need to raise capital by issuing shares, etc.?
Posted by: bob | February 23, 2010 at 11:23 AM
Hi bob
It would require them to acquire more of the particular financial assets that can be used for collateral if they didn't already have them. There are several ways to do that, but yes, raising capital would enable that. Note, however, that most every bank would have more than sufficient appropriate collateral as a matter of normal operating procedures. Also, in the US, at least, the collateral requirements for intraday overdrafts aren't particularly difficult to meet anyway. Finally, again in the US at least, not having sufficient collateral just means you get a bigger penalty--I haven't seen anywhere that the Fed rejects the payment, though it proposed doing so several years ago and changed its mind after it received the predictable response during the comment period.
Posted by: Scott Fullwiler | February 23, 2010 at 11:29 AM
bob, the Bank of England also used to have a zero statutory reserve requirement, but in aggregate, banks still held a small positive reserves balance. Central banks typically levy penal overdraft charges, so it pays banks in terms of avoiding costs to aim at a slightly positive reserve balance, even where the required balance is zero.
I disagree with Scott Fullwiler when he says that central banks do not really need collateral. Central banks are typically highly risk averse, so insist on high quality collateral to avoid credit losses, even if they do not need it to raise their own funds. Insolvency matters for a central bank, because it would require state recapitalisation, with potential loss of independence.
As I am sure you know, in principle a bank's capital or equity is the value of its assets less the value of its liabilities, but since the value of risky assets tends to be volatile, regulators insist that a certain amount of this equity is held in the form of safe assets. Safe assets like government bonds, the entire value of which counts towards capital, tend to be held to meet this capital requirement. Since the central bank is risk averse, such assets also tend to be required as collateral for borrowing reserves, so they are a close substitute for reserves as liquidity (note that collateral pledged to the central bank remains on the bank's balance sheet, so it continues to contribute to the bank's capital). Note also that, as a virtually risk-free asset, a reserves balance is also a capital asset. I hope you find that this explanation helps rather than confuses - both "capital" and "reserves" are used with many meanings!
Posted by: RebelEconomist | February 23, 2010 at 11:37 AM
thanks Scott and Rebeleconomist, very helpful
Let me reframe the question just one more time. I'm kind of thinking in terms of "at-the-limit":
If a bank were to be right up against its capital ratio requirements (has the bare minimum of capital), and currently has pledged all of its low-risk assets as collateral, and then the BOC decides that the rest of their assets (consumer loans, etc.) have become more risky and demands more collateral - is the only way of satisfying that by raising more capital to purchase low risk assets? If they just borrowed those assets, wouldn't that put them under the required capital ratio?
Posted by: bob | February 23, 2010 at 11:54 AM
Yes it would put them under the required capital ratio bob, because borrowing, say, treasuries for, say, ABS would increase the size of the bank's balance sheet and increase their capital shortfall. As per my previous comment, the loaned ABS rather than the borrowed treasuries would still count as part of the bank's capital.
Posted by: RebelEconomist | February 23, 2010 at 12:09 PM
Hi Rebel,
"Central banks are typically highly risk averse, so insist on high quality collateral to avoid credit losses, even if they do not need it to raise their own funds. Insolvency matters for a central bank, because it would require state recapitalisation, with potential loss of independence."
I agree that there are political issues for central banks related to its capital, defaults on central bank loans, etc., but there's no operational need for collateral. That was my point.
Also, the only reason there's any question about the size of the Fed's capital is that the Fed has been legally required to send about 95% of its profits for the past 95+ years to the Treasury. I looked this up for all the Fed income statements availabe online going back to the early 1990s, and this amounted to well over $300B just in that period. If the Fed had retained its earnings like any other private bank is allowed to do (not that I'm advocating this), there would be virtually no issue regarding state recapitalization--indeed, it's the height of hypocrisy to not allow the central bank to retain its earnings and then potentially withhold recapitalization if that were to be necessary. And again, operationally, there's no need for recapitalizataion of the currency issuer in the first place.
I agree with your 3rd paragraph mostly, though I would say that a reserve balance is a bank asset that doesn't use up any of the bank's capital, not that it is capital itself. I see where you are going with that, but 99% of people (economists, in particular) would get confused (more than they already are) if we start conflating reserves and capital.
Best,
Scott
Posted by: Scott Fullwiler | February 23, 2010 at 12:15 PM
bob at 11:23 . . . I was responding to Adam's point there. All cleared up now. Thanks!
Posted by: Scott Fullwiler | February 23, 2010 at 12:16 PM
Note that the bank would not have to issue capital securities (eg shares) to raise capital, however. If the bank sold ABS outright and bought treasuries with the proceeds, that would increase its regulatory (as opposed to economic) capital, because the treasuries get a greater weighting. This is, of course, one reason why treasury yields are lower despite massive government borrowing - a bank can effectively deleverage without actually decreasing the size of its balance sheet.
Posted by: RebelEconomist | February 23, 2010 at 12:18 PM
I agree Scott. I used to call myself a "reserve manager", and that was yet another type of reserves......lets not go there!
Posted by: RebelEconomist | February 23, 2010 at 12:25 PM
Scott, at 11:23, if you were responding to me you've cearly not understood what I was saying.
I was saying that the constraints are only on capital, there is virtually no reserve constraint (though the requirement to have some assets that are acceptable for collateral I guess is a type of liquidity constraint).
I was not implying what you seem to have understood and it's quite myserious how you could have thought that.
Posted by: Adam P | February 23, 2010 at 01:15 PM
correction, should be "Scott at 12:16"
Posted by: Adam P | February 23, 2010 at 01:16 PM
Adam . . .sorry. When you said "it's about capital only" I interpreted the "it" as collateral which now I understand you obviously weren't intending. My misinterpretation was because bob had asked if collateral was from capital and I thought you were responding to that. Sorry again.
Posted by: Scott Fullwiler | February 23, 2010 at 01:33 PM
Rebeleconomist - good point @ 12:18
Posted by: bob | February 23, 2010 at 03:26 PM
would it be possible for the BOC to make the bank rate equal to the deposit rate @ 0.25%? In such a situation the BOC would borrow all settlement balances at 0.25% and lend them at 0.25%
Could that work? Would it be desirable?
Posted by: bob | February 23, 2010 at 03:42 PM
bob . . . MMT'ers would say "yes." Warren Mosler has proposed precisely this frequently. (Note that this would also put the target rate at 0.25%, so if you want the target rate higher you need the other two higher,too.) Charles Goodhart--a non-MMT'er--has also proposed this, by the way, though his proposal would raise the spread between the two after individual banks reach a particular threshold of balances.
Posted by: Scott Fullwiler | February 23, 2010 at 03:46 PM
Yes bob, I think it would work, but since it would probably mean that the central bank became the intermediary for all reserves loans, I would expect the central bank to prefer to maintain a spread between their lending and deposit rates.
By the way, you might appreciate a post that I wrote to explain quantitative easing:
http://reservedplace.blogspot.com/2009/04/easing-understanding.html
Posted by: RebelEconomist | February 23, 2010 at 03:54 PM
"since it would probably mean that the central bank became the intermediary for all reserves loans, I would expect the central bank to prefer to maintain a spread between their lending and deposit rates."
Yes, I can see that they wouldn't want to do anything too radical, but why not really? The BOC would become the intermediary, but is that really a problem? It seems like it would be more efficient than keeping an interbank market for settlement balances where they are constantly intervening to drive the rate down to 0.25% yet leaving the bank rate at 0.50% (under the current post-crisis policy).
Posted by: bob | February 23, 2010 at 04:22 PM
that should read "intervening to drive the overnight rate down"
Posted by: bob | February 23, 2010 at 04:23 PM
bob . . .exactly. you might find this of interest: http://www.newyorkfed.org/research/staff_reports/sr337.pdf
Posted by: Scott Fullwiler | February 23, 2010 at 05:06 PM
thanks Scott!
very very interesting paper. I also checked out one of Martin's other papers on the reason why intraday liquidity is so cheap and overnight liquidity so expensive... good stuff.
It seems like the interbank market for overnight balances may actually be pretty useless, just a big waste of time really.
Posted by: bob | February 24, 2010 at 07:41 AM
I'm going to take one last crack at this thread, then leave it.
I agree with Scott Fulwiler, at least on this point, ;-)
" Whether we critique you or vice versa, everyone ends up talking past each other, unfortunately. Don't really know how to fix that, but I wish I did."
Yes, it's sort of depressing, isn't it. I don't know how to fix it either.
But, on the bright side: I now think I understand and appreciate the other side's (or it should be "other sides'", because there's more than one or two) positions now, even if I don't agree. And, most importantly, *I* understand *my own* position better now than I did before this debate. (I know that's sort of selfish, but so what?)
I'm going to make just one last point before leaving this topic. It may help clarify a source of disagreement, or (more likely) it may fail totally. But I'm going to make it anyway.
Banks (either individually, or in total) need a lot of different things if they want to expand. Reserves, capital, yes. But also loan officers, computers, tellers, etc. Not to mention people and firms who want to lend or borrow from banks. Why should I "privilege" just one of those many things (reserves)? Why should others "privilege" capital? Aren't the others on the list equally important?
Here's my answer: I privilege reserves because I am interested in the public policy question. The Bank of Canada controls one aspect of public policy (monetary policy), and the Bank of Canada controls the supply of reserves to the banking system. (Again, you can read "the supply of reserves" as either an interest rate (price) or a quantity, but better yet as a functional relationship.) The Bank of Canada does not (except when it bails out banks) directly control the supply of capital, reserve officers, tellers, or anything else.
Anyway, thanks all for the argument!
Posted by: Nick Rowe | February 24, 2010 at 12:44 PM
Nick,
your last comment has inspired me to write a post called "Loan officer theory of helicopter drop monetary policy":
http://themoneydemand.blogspot.com/2010/02/loan-officer-theory-of-helicopter-drop.html
Posted by: TheMoneyDemand blog | February 24, 2010 at 01:54 PM
Nick, it's such a cop-out, very disappointing.
The basic question is: will a bank with a binding capital constraint make new loans no matter now large or permanent is the reserve injection?
You have no comment?
Posted by: Adam P | February 24, 2010 at 02:01 PM
What about this scenario, due to the recent economical crisis, a lot of banks went bust. But what about a bank going bust due to the panic reaction of their clients. People have become very jumpy when it comes to having money on the bank.
Just a little bad news of a bank can cause the clients to massively take their money out of the bank, which result in a downward spiral that the bank doesn't recover off.
If the banks clients wouldn't pull their capital out of the bank it would not go bust. Theire are no safe assets that the bank can have if a large part of capital is pulled out of the bank. Even the most profitable bank won't recover from that.
Posted by: Don Miat | February 24, 2010 at 03:02 PM
Adam P: OK then!
"The basic question is: will a bank with a binding capital constraint make new loans no matter now large or permanent is the reserve injection?"
No, but:
Or yes, but:
1. It could get more capital, by issuing new shares, or preferred shares, if the price is right.
2. If (say) bonds have a lower required/desired capital ratio than loans, it can buy bonds, sell loans (or not renew them when they get repaid), thereby changing the mix of assets on the asset side of its balance sheet, to relax the capital constraint, and then expand both assets and deposits. Again, the simple multiplier model is a model of deposits, not loans. Loans are just one part of banks' assets. And banks aren't the only source of loans.
Posted by: Nick Rowe | February 24, 2010 at 04:35 PM
Fine Nick, but then what does all of this have to do with an "excess demand for the medium of exchange"?
Posted by: Adam P | February 25, 2010 at 05:51 AM
Adam: chequable demand deposits at the commercial banks are media of exchange. An increased supply of reserves at the BoC causes the commercial banks to buy something (loans, bonds, antique furniture), creating demand deposits to pay for them, which reduces the excess demand for media of exchange.
(Glad to see a very good Canada/Russia game last night.)
Posted by: Nick Rowe | February 25, 2010 at 06:01 AM
surely reserves and govie bonds, both being government debt, have similar capital charges. How did the reserve injection help them do something they couldn't already do?
Posted by: Adam P | February 25, 2010 at 07:07 AM
It was a fantastic game:). left me feeling much better.
Posted by: Adam P | February 25, 2010 at 07:08 AM