To paraphrase Churchill, accountants and economists are divided by a common language. We seem to be using the same words to talk about the same things, but we don't understand what the other is saying. This is my attempt to provide an economist's perspective on the relation between accounting and economics.
What I say here will not I think be new or controversial for economists (until I start talking about money). Accountants may find it hilariously wrong, like locals hearing tales from a traveller who can't speak the local dialect and gets everything muddled. Let's see.
Like most economists, I don't think about accounting much. Normally I only think about it when I teach the ECON 1000 bits on national income accounting and money and banking, and try to remember if assets go on the left or right side of a balance sheet (I rely on my students to remind me). I decided to write this after an interesting exchange with Winterspeak, both here and on his blog.
So, here's my take on accounting:
There are two fundamental accounting identities. Here's the first:
1. stuff bought = stuff sold
If some people bought 100 apples, some (other) people must have sold 100 apples. The "stuff" in question can be a flow or a stock, measured in monetary or real units, or whatever. If it's a flow we call it income accounting. If it's an accumulated flow, or stock, we call it balance sheet accounting. The stuff could be real goods and services, or it could be financial stuff. All accounting consists of dividing and subdividing stuff into different categories, trying to keep your head straight when doing it, so that the RHS and LHS both capture mutually exclusive and jointly exhaustive ways of dividing the same stuff. Don't miss anything out, and don't double-count.
This is an identity; not just an equation. it's true by definition of what we mean by "bought" and "sold". It's not something we could go out and test empirically. If we did test it empirically, and found the numbers weren't equal, we would figure we must have made some sort of mistake in the test, like adding things up wrong, forgetting something, defining "stuff" inconsistently, or somebody lied to us, or whatever. And because of that, it doesn't tell us anything about the world, only about how we must use words in a logically consistent manner if we are to think about the world without internal contradictions.
The equivalent sentence in economics has three different variants:
2a. stuff demanded = stuff bought = stuff sold = stuff supplied
is the complete version, though it's often shortened to:
2b. stuff demanded = stuff traded = stuff supplied
And sometimes we shorten it further still, to:
2c. stuff demanded = stuff supplied
First, 2a and 2b are logically equivalent, since the middle "=" is an identity, so that "stuff bought" and "stuff sold" are just two ways of saying the same thing.
Second, "stuff demanded" means the amount of stuff buyers want to buy, and would buy if it were available to buy. It doesn't mean "stuff bought". Same with "stuff supplied".
Third, the other two "=" are equalities, not identities. They can be empirically false, and are empirically false when there is excess demand or supply. They are only both true in full market-clearing equilibrium.
Fourth, 2b reveals that there are really two "=" in market equilibrium, where buyers get to buy what they want to buy, and sellers get to sell what they want to sell. Normally we impose the "short side rule" so that quantity traded is whichever is less, quantity demanded or quantity supplied. This means that if 2c is true, both sides of 2b will be true as well. But the short side rule is an empirical fact about markets: that exchange is voluntary. I can at least imagine cases where 2c is true, but both buyers and sellers are forced to trade either more or less than they want, so 2b is false.
By common stereotype, accountants are careful people; economists are usually sloppy people. We often get really sloppy in speaking to distinguish between 1 and 2, and between 2a or 2b, and 2c. That might be the problem. Or it might be that accountants just don't understand the difference between demand and buy, and supply and sell.
Let's take a specific example of 1. In national income accounting, we define "stuff" to be a flow of newly-produced goods. And we divide the stuff up according to who does the buying: households, firms, government, or foreigners (though we aren't altogether consistent here, because we often fudge investment, and imports are a weird category – more a subtraction from C, I, and G).
3. C+I+G+X-M = Y
(Newly-produced) goods bought = (newly-produced) goods sold.
We don't have to divide the stuff up in this way; we could divide it between goods and services; or stuff made on Mondays, Tuesdays, etc. We divide it this way if we think (or some economists think) it's useful to divide it up this way, because, for example, households, firms, and foreigners are influenced by very different things in deciding how much they want to buy. And it's economists, not accountants, who must ultimately decide what categories are useful to us. But already I'm starting to talk about stuff demanded, and economics, not stuff bought, and accounting.
The important point is this: economists can and should decide on their own ways to define the "stuff" that is bought and sold, and how to divide it into sub-categories of "stuff". We decide on a conceptual scheme that is useful to us, not to accountants. And it is useful to us if it matches our behavioural theories of how the world works. We don't have to follow the accountants' conceptual scheme, and shouldn't follow the accountants, if we have a different scheme that is more useful to us. Ultimately, we economists have to be our own accountants.
The second fundamental accounting identity is this:
4. The value of the stuff I buy = the value of the stuff I sell.
We have to be really careful with this one.
Think about a barter exchange: I swap my 10 apples for someone's 5 bananas. The price of a bananas is 2 apples. So I sell 10 apples and buy 10 apples' worth of bananas in exchange. Those "values" are the prices at which the transactions are made, even if the bananas turn out to be rotten. Unlike the first fundamental accounting identity, which can be expressed in physical units if you want, this one can't; it must be expressed in values. (Maybe this is why accountants have so much difficulty dealing with inflation? Dunno. Maybe not.)
What about monetary exchange? If I sell 10 apples at $1 each I "buy" $10 in the medium of exchange – money. That's what causes misunderstandings, because we don't normally talk about "buying" or "selling" money. But we have to in this case.
Now let's consider the economist's version of 4:
5. The value of the stuff I demand = the value of the stuff I supply.
In other words, when I go into any particular transaction, at any price, deciding how much I want to transact at that price, I get the maths right. I don't demand $20 and supply 10 apples at a price of $1 per apple. That makes sense. The only thing that could make it false would be if people made math mistakes. It's close to an identity, but isn't quite, because empirically people do make math mistakes.
Now, what happens when I aggregate up over all transactions I make? You get: the sum of the values of my excess demands must equal zero. Aggregate over all people (and firms and governments): you get Walras' Law. The sum of the values of excess demands must equal zero. [Update for clarity: an equivalent statement of Walras' Law is: the sum of the values of goods demanded must equal the sum of the values of the goods supplied. Excess demand=demand minus supply.] If you forget to include money as one of the goods, you get (one version of) Say's Law (and it's not a version necessarily believed by Say himself, but let that pass).
Now, many (most?) economists believe in Walras' Law (or think they do). But I don't. I think it's fine to aggregate over all transactions in the accountant's version 4 (the sum of the value of excess purchases is identically equal to zero). But I don't think it's fine to aggregate over all transactions in the economists' version 5, even if everyone has perfect math, and is perfectly rational.
If we make all our purchases and sales at the same time in one big market for all goods, then Walras' Law would be right (unless somebody got the math wrong). But in a monetary exchange economy, with n-1 [typo fixed] goods plus money, there are n-1 markets. In each of those n-1 markets we make a separate decision to transact in that market, subject to any constraints we expect we might face on how much we are actually able to buy and sell in each of the other n-2 markets, if any of those markets are in disequilibrium. Clower/Leijonhuvfud/Bennassy etc..
That means n-1 separate maximisation problems, in principle. Each one of those maximisation problems will satisfy Walras' Law, but the only way we can add them all up consistently is if we recognise that there are n-1 separate excess demands for money, coming from each of those separate decisions.
So the only generally correct aggregate economist's version of the second fundamental principle of accounting is this:
6. The sum of the values of the n-1 excess demands for non-money goods, plus the sum of the n-1 excess demands for money, is identically equal to zero, provided the people doing the demanding and supplying can do math.
I have no idea what the accountants will make of 6.
Worthwhile Canadian Initiative 
Nick,
Regarding spreads – I think you’ve missed the point.
The point is that banks require capital to undertake risky lending – not reserves.
The interest rate spread is a reflection of the rate of return required on capital at risk.
The fact that Thoma ignored the spread reflects the fact that he ignores the dependence of risk lending on capital, while erroneously relating it to the availability of reserves.
Adam P.,
Re gold and capital:
The MMT overview is that a fiat system is not financially constrained because the currency issuer operates in a floating rate system rather than a fixed rate one such as the gold standard.
At the circumference of that system, banks are financially constrained.
They are constrained by capital requirements, not by reserve requirements.
So bank capital imposes something akin to a fixed rate financial constraint (i.e. required capital ratios), but only within the endogenous banking system, which is separate from the exogenous operation of the central bank.
The government and central bank of course are not constrained by gold or by capital requirements.
In fact, the GCB is able to operate with “negative equity”. This is the case from an accounting perspective when balance sheet equity appears on the left hand side of the balance sheet rather than the right hand side. The offset to left hand negative equity is a right hand net liability profile. This consolidated position again is a reflection of the fact that the government is a supplier of net saving as desired by the non government sector, and that the government sector operates without financial constraints. It also reflects the idea that the government’s power to create liquidity dominates any technical concern about its solvency (in a fiat system). Fiat systems are able to supersede the issue of solvency risk because of their unique ability to produce liquidity for the non government sector.
(If you think the idea of “negative equity” is a little bonkers, I hope winterspeak can chime in, based on an interesting initial conversation he and I had once upon a time.)
OK, lots of stuff here.
First, JKH is completely correct about capital vs. rbs and that the distinction b/n needing capital to lend but NOT needing rbs to lend hasn’t been addressed or demonstrated to be understood by those critiquing Bill. Nick’s latest post here goes at capital, but not rbs.
Second, I think Adam P’s latest points are pretty good. The difference for a gold standard, though, is that gold really would constrain lending (as Adam noted) but is also what banks can return to the bank to withdraw in order to draw down deposits, unlike capital. So, a gold standard model is one in which gold is the reserve, but it also does constrain lending in the sense that banks must be concerned that deposits will be drawn down. Similarly, to apply JKH’s points, if a bank made a bad loan with its notes that could not be paid back, this would not require it to draw down or otherwise write down its gold reserves.
Third, regarding the discussion on Thoma’s site about real interest rates, the people who actually study how indivdual real world firms make investment decisions in the field of corporate finance will tell you that decisions are based upon criteria such as NPV, IRR, or MIRR. For those measures, there are two drivers: anticipated free cash flows and the discount rate. You can lower the discount rate (real, nominal, or whatever . . . though most analyses suggest the nominal discount rate is more important, albeit with an assumed change in prices paid and sold factored in to the cash flows), but if anticipated cash flows have fallen, the criteria can still be saying don’t invest. This tends to be the result in a recession, the opposite (rising rates and rising cash flows, but the criteria on balance saying to do the investment) tends to be the case in an expansion. This was Bill’s point (which is, again, completely consistent with research in corporate finance) that Thoma and Nick are taking issue with for some reason. Any basic senstitivity analysis will usually show you that the criteria are much more sensitive to a change in the anticipated cash flows than they are to a change in the discount rate. Further, if we add in growing application of options theory to investment decisions, then the discount rate becomes still less important relative to analysis of cash flows (anticipated and now volatility thereof). Further further, the insistence on a downward sloping IS is puzzling . . . if you raise interest rates to 20% in Japan (which would also be a significant increase in the real rate there), given the size of the national debt, it would be hard to say you don’t on balance raise spending there as interest payments would generate a considerable increase in income (which then the spnending might stimulate investment . . . might).
4. A point I haven’t seen addressed is that rbs are generally not used for liquidity risk, either. This is because cbs provide them at a stated penalty rate, and if there are enough having to get them at the penalty rate that the overnight rate rises, the cb increases the qty outright via repos. That’s what it MEANS to be setting an interest rate target. That banks hold any rbs beyond what’s needed to satisfy reserve requirements (where the latter actually apply) indicates a concern with having to go to the cb at the penalty rate (in the US, the penalty for an uncollateralized overdraft if 400bp above the day’s fed funds rate, and collateralizing prior to August 2007 reduced the penalty to 100bp above the target rate). In Canada, under normal circumstances, banks have perfect certainty that any overdrafts can be settled by the end of business without incurring any penalty. Not so in the US. Thus, the desired buffer has to do with how the cb sets the interest rate target, not liquidity risk. When you add the problems of counterparty risk in the US after Lehman’s failure, desire for buffers likely increased considerably (though I doubt the desired agg buffer was anywhere near the actual qty of rbs the Fed left circulating mostly because it didn’t believe it had any way to drain them). As Warren Mosler pointed out, one thing that would have helped immensely would have been for the Fed to set both the remuneration rate and the primary lending rate at the targeted funds rate (granted, they weren’t granted authority to pay interest on rbs for a few weeks after and they then left some institutions w/ reserve accounts out in earning interest), as that’s the sort of thing you do if you are trying to hit an interest rate target . . . the rate can only vary as wide as the corridor between those two rates.
JKH and I were writing at the same time, so didn’t see the last three posts before mine. Will look those over later . . . gotta make breakfast!
Scott,
That’s a fair point on liquidity risk. I knew I was venturing into potentially dangerous territory there. The central bank assures adequate provision of systemic and individual bank liquidity through its reserve and discount window operations.
The point I’ve made is that it is possible for individual banks to seek to acquire excess reserves as liquidity protection IF as part of their liquidity management strategy they also seek to avoid discount window borrowing. You will recall that the Fed created the auction facility as a response to the crisis in order to attempt to remove the stigma usually associated with discount window borrowing.
I agree that the fact of ultimate liquidity provision is undeniable, and Warren M.’s proposals are consistent with absolutely extinguishing any notion of stigma associated with borrowing reserves from the central bank.
So when I referenced liquidity protection, it was only in the sense of the self-imposed constraint of stigma avoidance that is still the norm in the existing system, ex credit crisis conditions.
JKH . . . I think your point at 959 and mine are compatible. I was actually going after a more typical understanding of reserve management, liquidity shocks, etc., than I was disagreeing with your description, as you were primarily focused on describing capital. My overall point is that Thoma’s basically wrong about capital AND liquidity.
I would say we are like the gold standard, plus alchemy; the central bank can print as much gold as it likes. (This analogy doesn’t work at the international level, of course.)
This means the government/central bank can make the nominal budget constraint whatever it wants it to be. Same as under the gold standard, if the government owned a very very large gold mine, where gold could be mined at 1cent per ounce.
JKH: “Nick, Regarding spreads – I think you’ve missed the point.
The point is that banks require capital to undertake risky lending – not reserves.
The interest rate spread is a reflection of the rate of return required on capital at risk.”
No, I understand that. But unless you argue that the aggregate supply curve of capital to banks is PERFECTLY inelastic, AND that there is absolutely fixed coefficients between loans and capital, it doesn’t make any difference to the argument that lowering the interest rate on reserves will lead to an increased supply (offer) of loans.
You need trucks to take apples to market. But this doesn’t make any difference to the fact that the supply curve for apples slopes up. Unless the supply curve of trucks is perfectly inelastic AND there are fixed coefficients technology between apples and trucks.
Bill Mitchell, Stock-Flow Consistent Macro Models
Monetary Economics: An Integrated Approach to Credit, Money, Income, Production and Wealth by Wynne Godley and Marc Lavoie, Palgrave Macmillan (2007)
Monetary Economics: An Integrated Approach to Credit, Money, Income, Production and Wealth reviewed by Pablo Gabriel Bortz
Wynne Godley Home
Pedagogical Approaches to Theories of Endogenous versus Exogenous-Money Pluralism in Action See in particular the section on Godley and Lavoie’s contribution.
[I just rescued this comment from the spam filter. NR]
Nick,
It sounds like you’re making an argument about the elasticity of loan supply assuming some sort of elasticity of capital supply. If I’ve mangled the language, you get the drift.
I’m not sure I have as much of a problem with that, although I’m slightly confused about the directionality of the elasticity you’re assuming according to an interest rate increase or decrease.
Where I have a problem though is that to the degree your argument is valid (and I don’t know whether it is or isn’t), it is a function of the fed funds rate (in the case of the US example) – not the interest rate on reserves.
Reserves and the payment of interest on reserves are irrelevant to your argument for reasons already given at length. The interest rate paid on reserves is a function of a coincidental supply of excess reserves and a requirement to set a floor for the fed funds rate under those conditions. Lending has nothing to do with reserves and nothing to do directly with the interest paid on reserves. The pricing of that lending may reference the risk free rate directly or indirectly as a basis – i.e. fed funds rate. The reserve interest rate is merely a coincidental architectural requirement. It is not the true pricing reference point. So your argument has nothing to do with reserves, while being dependent on a reasonable assumption about capital supply elasticity, which is entirely consistent with the notion of capital constraints, which is the main point of the discussion.
BTW, the idea that banks are reserve constrained in lending doesn’t necessarily mean they can’t get capital when they need it. I’ve never argued that. It means that if the constraint is binding, they must go get more capital before they can lend.
sorry – I meant above “the idea that banks are capital constrained in lending doesn’t mean …”
JKH said: “The point is that banks require capital to undertake risky lending – not reserves.”
And, “They are constrained by capital requirements, not by reserve requirements.
So bank capital imposes something akin to a fixed rate financial constraint (i.e. required capital ratios), but only within the endogenous banking system, which is separate from the exogenous operation of the central bank.
The government and central bank of course are not constrained by gold or by capital requirements.”
If a bank is politically connected, too big to fail, and willing to do what the fed wants (lend to prevent price deflation and asset price deflation to benefit the rich), will it do risky lending if the fed will swap garbage debt for bank reserves (IMO fed debt) and/or treasury debt so it is possible to have no capital requirement?
Scott,
Never mind breakfast (or maybe lunch by now) – you are neglecting your own blog post! I have a few comments stacked up there. It might help discussion develop if you could drop the approval of comments. But perhaps you MMTers are afraid of getting some of the abuse that you dish out to others – I was minded to write on Nick’s post about scientists and economists that scientists usually comprehensively dismantle their opponent’s argument before they dismiss them as wrong.
Here is a question.
What is/are the difference(s) between a gold standard with fractional reserve lending economy and a fiat currency with fractional reserve lending economy?
Hi Rebel . . yes, sorry for the delay responding to you and a few others. Regarding comments, the person who does that hasn’t rejected anything since June, I think, which I heard was rather profanity laden and quite inappropriate for publication. Will get to them ASAP. And we’ve published TONS of stuff dismantling before dismissing . . . don’t think that our blog posts are the sum total of what we’ve done . . .very small % actually. Best, Scott
Sorry. I just can’t keep up with replying to all the comments. (Sounds like poor Scott is in the same boat!)
No worries, Nick. That you provide the forum is more than good enough.
Scott Fullwiler said: “but if anticipated cash flows have fallen, the criteria can still be saying don’t invest.”
And, “Any basic senstitivity analysis will usually show you that the criteria are much more sensitive to a change in the anticipated cash flows than they are to a change in the discount rate.”
How about the firm says we don’t need any currency denominated debt (even at zero percent) to “bring forward” from the future more supply ; we have enough now?
Or, if the firm expands with even zero percent currency denominated debt, the firm probably has to pay more in labor and might get lower prices for their products? That would help the lower and middle class to get closer to real earnings growth. We the firm are the real earnings growth earners and excess savers here NOT the lower and middle class. The fed has said so.
Adam P: Yes, macro economists (such as Thoma, and Rowe) functionally think the US still runs on a gold standard. But if you describe their position as “gold standard thinking” they get mad, because they think you’ve just called them Austrians! Unless you are a libertarian computer nerd, it is unfashionable to be Austrian, who are looked upon as sort of idiot throw-back yokels.
MMT sits even lower on the totem pole than the Austrians, which is why it’s met with befuddlement as opposed to outright hostility and ridicule. Being right probably helps a little, but not as much as one would think.
And bank capital absolutely plays the role of gold if one wants to push that analogy. It makes reserves PROFOUNDLY mysterious, though, and I don’t know how to think about reserves without doing basic double entry accounting. And accountants aren’t even allowed NEAR the totem pole.
NICK: Bank capital is a hard constraint in that, if banks fall below their capital requirements, the Govt is supposed to technically shut them down. Every loan they make counts against their capital, but does not deplete their capital. Bank capital sits in the equity part of the balance sheet, and bank loans are made by enlarging both the asset and liability side of the balance sheet at the same time. The money to make loans is created ex nihilo, as the asset (receivable) and corresponding liability (deposit) are created at the same instant.
So, how many liabilities is a regulator willing to have sit on that equity (capital) base? That’s what people are talking about leverage, and what is a safe amount vs an unsafe amount. Capital plays the role of insurance against loans going bad here, as JKH says, because a bad loan involves writing down an asset, and writing down capital (equity) on the liability side.
Scott,
“Any basic sensitivity analysis will usually show you that the criteria are much more sensitive to a change in the anticipated cash flows than they are to a change in the discount rate … “
This part was quite jarring in its common sense and fundamental importance … like a second wave of reality training, following required accounting indoctrination.
Adam P., winterspeak,
Accounting is the means to the end – which is logic in problem solving. It’s method forces you to analyze things in correct sequence.
I just checked the spam filter, and tried to rescue a real comment that had lots of links to books, including Marc Lavoie’s, IIRC. But something screwed up, and the comment disappeared. Sorry.
So much dogma, so little open-mindedness.
It seems reasonable to me that a bank’s balance sheet expansion is restrained (as opposed to constrained) by both capital and reserves (as in current account balances at the central bank, rather than as in retained earnings or loan loss reserves – the difference may account for some of the cross-talking). Both capital and reserves are normally readily available, but at a cost. One imagines a classic microeconomic optimisation problem with capital and reserves as factors of production, but this would abstract from some interesting details. I presume that reserves are a slightly more immediate restraint, because the reserve requirement comes into effect from the next reserves maintenance period. While capital regulations apply continuously, I am sceptical that banks run so close to their limits that capital is normally a short-term constraint on balance sheet expansion – banks apparently had more than adequate regulatory levels of capital at the onset of the financial crisis, for example. However, I guess that raising new capital is a more lumpy process than acquiring reserves. Whether capital or reserves is the more significant restraint will depend on the size of reserve requirements (if any) and the prevailing level of market interest rates relative to the return on reserves (if any), and the equity return premium. Equity capital is essentially the (hopefully positive) value of assets minus liabilities, but as capital adequacy regulations discount assets according to risk and increased reserves may be needed for balance sheet expansion anyway depending on how it is funded, an obvious asset in which to invest incremental capital is in (zero risk-weighted) reserves.
The immediate effect of advancing a loan is to increase a customer deposit at the lending bank, so it is true to say that loans create deposits, but banks can seek and will take deposits if they can find a profitable use for them. Taking a deposit increases the bank’s balance in its reserve account, and since this leaves the bank with surplus reserves, it will probably seek to switch some of its reserves into more profitable assets, which may include loans (eg lending its surplus reserves to another bank in the short-term). Since this will involve (assuming that the loan is drawn down) transferring reserves to another bank, this reduces the bank’s reserves balance, but does not extinguish reserves across the banking system. When a new deposit is made in the banking system, such as when banknotes are paid in by a retailer, the money multiplier begins to operate, albeit with a multiplying factor that varies according to market conditions.
How about that for a mediatory synthesis?
Rebel,
So much “trapped in the reserve box” thinking.
Consider two dimensions: systemic versus specific; and capital versus reserves.
The purpose of capital is to absorb financial losses. Banks cannot acquire risk assets unless they have required capital underpinning.
(Banks may restrain themselves in approaching the boundaries set by required capital ratios – i.e. they may restrain themselves from utilizing their entire capital position in terms of risk allocation, in which case they have surplus capital by definition. They can restrain themselves from full utilization of permissible regulatory and/or self-imposed limits. But the limits as defined are a capital constraint.)
Risk taking in the system as a whole can also be constrained by capital adequacy if there is unusual uncertainty about risk or about the adequacy of capital against that risk. That has been an operative dynamic in the extreme during the credit crisis. Obviously this applies to individual banks as well.
In normal times, prudently run banks tend to generate surplus capital and tend to keep some surplus capital as a buffer against risk. Surplus capital by definition is capital that is not internally allocated to risk taking. Internal systems ensure that such excess capital is directed toward partial funding of essentially risk free assets such as treasury bills.
Reserves are not a systemic constraint for the banking system expansion of risk assets. And that’s the only issue that really matters here. That’s the issue that neoclassical economists such as Thoma are getting profoundly wrong in their analysis of current monetary conditions.
In the pre-September 2008 US environment, the Fed’s supply of reserves could be an occasional systemic restraint only in the sense that a supply squeeze would force interest rates up against “animal spirits” that wanted to drive them lower. In the current environment of structural excess reserves, with the payment of interest rates on balances, no such supply squeeze mechanism is required to reset the policy rate higher.
But in no instance is the supply of reserves a systemic constraint on risk lending. That is the role of capital.
Is a CBs resetting of the policy rate something that bankers consider as an input to risk lending? Obviously they do. But that’s got nothing to do with reserve management. It has to do with the outlook for interest rates and the effect of that on risk analysis.
Reserves are nothing but a scoreboard for the interbank balancing of assets and liabilities. And if central banks want to change their policy rate, they use some aspect of the systemic reserve architecture to do that (reserve supply; interest on reserves.) That’s all reserves are good for.
Capital on the other hand is the stuff that supports financial risk. By comparison to the nothing-burger of reserves, capital is almost everything (including deposit insurance of course).
Feel free to weigh in, Scott.
Rebel: ” One imagines a classic microeconomic optimisation problem with capital and reserves as factors of production, but this would abstract from some interesting details.”.
That’s the model (for an individual bank) at the back of my mind too. So you can look at the firm’s profit maximising output decision on either of those margins. In each case, the value of the marginal product of that factor will equal the real rental rate. A decrease in the rental cost of capital would cause an individual bank to want to expand loans. A decrease in the rental cost of reserves would likewise cause an individual bank to want to expand loans.
And those first order conditions for individual banks are not made invalid by the fact that if all banks expanded loans together, there would be no loss of reserves (absent a currency drain etc.). And they are not made invalid if desired reserves were zero, or even negative, in equilibrium.
Equivalently, we could assume a fixed proportions production function, and lump reserves+capital together into a joint factor of production. In which case the output of new loans depends on the combined cost of capital+reserves.
What about another analogy: reserves are advertising. If one firm/bank wants to expand sales/loans relative to other firms/banks, it must advertise to take customers away from other firms. But if all expand sales together, none need advertise. Nah. Maybe just confuses things further.
too bad, Nick
reserves as advertising is a cornerstone of MMT
🙂
JKH: “But in no instance is the supply of reserves a systemic constraint on risk lending. That is the role of capital.”
But I deliberately said that reserves are a restraint, not a constraint. Other banks, and, if necessary the central bank, will practically always lend reserves (which is why reserves do not represent a hard constraint), but at a price. If the return on reserves is lower than this price, as it usually is – ie zero – surely this must represent a restraint on balance sheet expansion. As Nick puts it, “a decrease in the rental cost of reserves would likewise cause an individual bank to want to expand loans”. If you disagree with that, please explain why.
Guys. I have decided to do a post on this (reserves, capital), very shortly. When that post goes up, perhaps we should continue this conversation under that new post.
NIck, I sent that post that apparently got eaten by gremlins. I’ll repost it on a new thread. Thanks for letting me know what happened.
So you agree that reserves are not a constraint on risk lending. Nor are they a restraint, to use your word.
I assume the rental cost of reserves is a fancy name for the fed funds rate, in the case of the US.
What matters in loan pricing is the spread. A spread is set against a benchmark risk free or near risk free rate according to expected losses and expenses, and capital is allocated as a cushion for unexpected losses.
Similarly on the liability side, spreads are set according to operating costs.
Banks set the benchmark pricing curve for all maturities and all loans and deposits in this way. Loan spreads are obviously positive spreads; liability spreads can range from flat to positive (some wholesale rates) to negative (retail rates).
In setting the benchmark pricing curve, banks tend to use some sort of variation of or amalgamation of the fed funds curve, the Libor curve, or the swap curve. The ON fed funds rate may or may not be the front end benchmark. It doesn’t matter. Fed funds will exhibit an average basis relationship with the defined front end benchmark rate over time, which is all that is necessary to determine the relationship between fed funds rate and risk lending rates.
The benchmark curves for assets and liabilities net out in determining the overall net interest margin for the bank. I.e. this basically nets out internal transfer pricing in determining the contribution of both sides of the balance sheet to bank profitability.
This mean among other things that the general level of the fed funds rate is not what determines either the lending decision or the profitability of risk asset lending. Because the combination of the spread and the capital requirement fully account for the expected profitability of the loan, the fed funds rate has nothing to do with the determination of whether the bank makes that risky loan.
The fed funds rate matters practically to the market for fed funds, which is a clearing market for reserves, and a factor in the management of bank liquidity positions more generally.
You can’t possibly think that banks make strategic decisions about capital allocation to credit risk according to the momentary bid offer spread on fed funds. And again, the main issue here is the non-role of reserves in bank credit risk management and associated lending.
For a bit more flavour about pricing, see my comment at November 27, 2009 at 09:47 PM
which refers to point # 3 by Scott Fullwiler | November 27, 2009 at 09:41 AM
Here’s something I happened to post on Bill’s blog, in response to a specific question:
You borrow from bank A.
A writes you a cheque.
You deposit the cheque at your bank B.
Bank A transfers reserves to your bank B through the interbank reserve clearings.
You buy stock in your bank B.
If it’s an issue of new stock, B debits your deposit account and sends you a stock certificate (or you get electronic credit for the shares).
B now has a book entry for new paid in equity on the right hand side of its balance sheet, instead of your deposit.
With respect to reserves, your bank B originally received reserves from A in the interbank clearing, after you borrowed from A and deposited the proceeds (e.g. cheque drawn on A) with B. Other things equal, bank B is in a surplus reserve position due to your original transfer of funds to it. In addition to bolstering its equity capital position, bank B by issuing new shares will expect to attract reserves from the rest of the banking system, although it’s unlikely the size of the share issue will be matched in full by an equivalent transfer of reserves. That’s because some of the buyers of the new share issue will be existing depositor with B. Their transactions will be reflected as an effective swap of their existing deposits for equity shares – the same as the second half of your transaction. In a relatively concentrated banking system with big players like Australia’s (I think), major banks doing large share issues will actually do an ex ante estimate of the likely net reserve effect of a new share issue based on netting out their own pro rata share of deposits in the system. That assists with short term cash and liquidity management.
As to what happens next to the net reserves transferred, they will soon get absorbed into the ongoing asset liability management of the recipient bank, although there will be a noticeable “bump” in its reserves when the share issue settles on day 1. That’s easy to take care of. The bank will just deploy those reserves by acquiring short term low risk money market assets such as treasury bills, which will reverse the reserve inflow that came in from the share issue. The interbank clearing will tend to “push” the reserve distribution back to the other banks. From that starting point, the bank is in a position to expand its acquisition of risk assets over time based on its strengthened equity capital position. That’s usually done over a longer term horizon, unless the share issue was done to offset some large strategic acquisition on a one off basis or to make up for an unexpected prior loss in equity capital.
This shows the difference between liquidity and capital. The strategic effect of the share issue has been to increase the bank’s equity capital position. That’s what allows it to take on more risk in the future – e.g. taking on more credit risk by increasing its loan book. In terms of liquidity, it has for the time being improved its position, in the sense that it has additional resources to offset a future reserve outflow that might be associated with future loan book expansion (e.g. by selling those just acquired liquid assets). But the longer the time horizon for new capital deployment, the more the liquidity effect of the share issue gets rolled into the day to day operational management of the bank.
This is all consistent with MMT. Banks need appropriate capital levels to lend. They don’t need existing reserves to lend. But they want to be able to square their reserve positions operationally when they do lend or when they engage in any transaction. They do this in the normal course by asset liability adjustments that offset any reserve outflow due to such transactions. MMT says that the central bank will always make such reserves available to the system as a whole, one way or another, even if it’s through discount window lending. An individual bank that lends just finds those reserves as necessary in order to square its account with the central bank. It can find them by calling on its own excess reserves if it has them, selling/maturing short term liquid assets if it has them, raising new liabilities in the market if it needs to, or borrowing from the central bank. Banks are not reserve constrained in risk lending, because one or more of the first three options are normally available and the final option is always available.
P.S.
“MMT says that the central bank will always make such reserves available …”
ALWAYS.
But the central bank/ government MAY make capital available only about every 70 years or so, when there’s a near depression melt-down of the financial system.
That’s the essence of the no constraint/constraint difference
Nick,
forget your new post
we’re just getting going here
🙂
JKH said: “P.S.
“MMT says that the central bank will always make such reserves available …”
ALWAYS.”
What about if the labor market is tight enough so that workers have real earnings growth and the product market is tight enough too? Will the central bank make reserves available if it produces wage inflation and price inflation INSTEAD OF asset price inflation (stocks, housing, and maybe even currency denominated debt itself)?
Too Much Fed,
Yes, always. That’s what it means to set an interest rate target. But the rate it provides the reserves at can and does change as the cb deems necessary.
JKH said: “P.S.
“MMT says that the central bank will always make such reserves available …”
ALWAYS.”
Alternatively, can the central bank make such reserves available because the labor market is oversupplied (with probable negative real earnings growth for most people) and the product market is oversupplied (both mostly due to positive productivity and cheap labor outsourcing, cheap labor legal immigration, and cheap labor illegal immigration?
Scott, so does that explain the 1970’s?
OK. All comments on banks, reserves, capital, etc., now please switch them to my new post on the subject. I’m trying to coordinate a new focal point equilibrium on where we carry on this discussion. A total of 237 comments on this post! Must be a record.
Just read your previous comments JKH. Wow, you sure are rolling here. Sorry to suggest a party location, but don’t let that stop the party.
JKH,
You seem to be forgetting that, as far as the banks are concerned, the Fed funds rate is lost on a fraction of the balance sheet expansion (assuming, the sake of simplicity, that reserves bear no interest at all). Reserves are effective a tax which drives a wedge between the return on loans and the cost of funding, and a tax on balance sheet expansion ought to reduce it.
I wanted to reply quickly to your comment, JKH, to continue the discussion, but since you have not yet replied today, I obviously would have had time to relate my reply more precisely to what you wrote.
You wrote: “What matters in loan pricing is the spread. A spread is set against a benchmark risk free or near risk free rate according to expected losses and expenses, and capital is allocated as a cushion for unexpected losses. Similarly on the liability side, spreads are set according to operating costs……..The ON fed funds rate may or may not be the front end benchmark. It doesn’t matter……The benchmark curves for assets and liabilities net out in determining the overall net interest margin for the bank. I.e. this basically nets out internal transfer pricing in determining the contribution of both sides of the balance sheet to bank profitability. This mean among other things that the general level of the fed funds rate is not what determines either the lending decision or the profitability of risk asset lending.”.
What I saying is that the opportunity cost of having to hold a certain proportion of a deposit liability in low interest reserves represents an expense or cost which is reflected in either the asset or the liability spread off the benchmark, or a bit of both. I agree that the influence of Fed funds as a benchmark nets out, but not the influence of Fed funds as a cost.
My argument is meant to be general rather than specific to current conditions, but clearly in a situation like the present where there is an interest rate paid on reserves that is not much lower after due adjustment for risk than other interest rates the need to hold a certain fraction of deposit liabilities as reserves is less restraining, while at a time when risk capital is more expensive capital is more restraining. Similarly, the relative importance of reserves depends on the prudential or regulatory reserve requirement, which is why some countries such as China use variations in the reserve requirement to regulate banks’ balance sheet expansion.
Arguably the best comments section ever.