But this ability of banks to create money ex nihilo is not limited to loans -- it extends to every transaction they do where they are acting as a principal! Check out JKH's comments on this thread:
Me: Loans generating deposits are totally different to this as they expand balance sheets.Note that the money creating nature of bank spending (not bank credit extension) and the money destroying nature of bank revenues exactly parallels with money creating nature of Government spending and the money destroying nature of Govt taxation.
If the bank purchases something from the private sector, then it's a transaction, and you debit some cash account and credit an asset. The seller debits some asset and credits their cash account. You've rearranged assets in the private sector, as there is a transaction, but no new money has been created, and balance sheets stay the same size. This is why I asked Nick about the IBM server vs. the IBM bond (or whatever other privately held asset he wants to ponder)
JKH: Winterspeak,
"Is there something different when a bank is the buyer?"
Yes. It is completely different. But no different than a loan. I gave the overview above.
Try again.
So we assume the server is accounted for as an investment.
That means it stays on the balance sheet.
So:
Bank cuts cheque “from nothing”.
Bank pays for server.
IBM deposits money in bank (assume same bank; it doesn’t matter).
This is an increase in money supply.
The change in the balance sheet is that the bank has a new asset (real investment) and a new liability (deposit).
There is no change to the bank’s equity account.
Very simple.
No difference between a new loan and the acquisition of any other new asset, financial or real, in terms of new money created, provided that the counter party is not a bank.
(And expenditure has exactly the same effect. This is much more counterintuitive, but very worthwhile to understand if you really want to understand money creation and destruction. If Scott F. is reading, this is just starting to touch on Steve Keen's circuit modeling area of interest.)
Me: Ye Gads!
Wow!
Overall, the thread was mixed though. I learned something new, which is great, yet Nick (the poor academic we were ganging up on) remains as wrong as ever about how money, banking, and the economy works. And we got the best post ever on how banks actually work from JKH:
Nick,
A thought provoking post, thanks. It prompted me to flesh out a model of my own. Apologies for the length of this “comment”; it will require someone to have an unusual level of interest in order to read it.
I’ve responded to some of your points specifically near the end, but first here’s my model:
CAPITAL AND RESERVE MODEL (JKH)
Let the world consist of two banks, A and B, and a central bank CB.
The employees of Bank A are a reserve manager, a capital manager, a loan manager, and a deposit manager.
Bank B has a similar staff.
The CB just has one employee – its own reserve manager. The 3 reserve managers’ jobs interface with each other in a way to be described.
Banks A and B start out the day “in balance” from all perspectives. Their assets equal their liabilities plus equity. And their world resembles Canada in that required reserves are zero. Both banks are at zero reserves.
There is one transaction to start. Bank A’s loan manager wants to make a loan of $ 10 million to customer X. (Think of X as a large blue chip corporation, or perhaps a wildly leveraged but fully prescient Canadian university professor betting the farm and all of his family’s and friends’ collateral on gold.) The loan manager does the credit analysis and approves it. Suppose the capital manager has an existing surplus capital position due to prior retained earnings that have not yet been allocated to risk assets. Assume that surplus capital is temporarily invested in treasury bills, which have a zero risk weight for capital purposes. The capital manager decides on the amount of capital required to support the loan risk, and allocates it accordingly as capital underpinning for the loan, in the event the loan transaction is completed. He allocates $ 1 million.
The reserve manager is notified of the pending loan drawdown. He assumes an outflow of reserves when the loan is draw down. He wants to attract an offsetting reserve inflow somehow in order to square his reserve account by the end of the day. Assume he has treasury bills in a liquid asset portfolio that represent the investment of capital funds that have so far been in surplus – i.e. not previously used to support risk assets. He plans to sell those bills for cash and redirect that internal capital to the new risk asset of $ 10 million. He needs additional funding of $ 9 million. He advises the deposit manager than he will require that funding.
The loan manager prices the loan. His inputs are the cost of capital, which he obtains from the capital manager, the benchmark cost of deposits, the cost required to cover expected credit losses, and other expenses assumed in pricing the loan such as related administrative salaries, etc. All of these costs can then be translated to an equivalent all in credit spread over a benchmark cost of funds. The loan is priced at that cost plus the credit spread.
I use “benchmark cost” here because the actual source of new funds in a “universal” bank can be wholesale or retail. New retail funds are “sold” internally into a central collection point at such a benchmark rate, providing retail bankers with a deposit spread to cover their own administrative and other costs. In the case of this simplified example, I’ll just assume now that the benchmark cost is the same as the actual wholesale cost of funds that the bank will end up paying in the market to fund this loan.
The loan manager goes to the deposit manager for a quote on the cost of funds – i.e. the expected wholesale deposit rate. The deposit manager may build in an additional small spread to allow for risk that the market price may “move” while the transaction is in progress.
The loan manager advises the customer of the all in cost based on the quoted market rate plus the credit spread. The customer accepts.
The loan manager completes the loan and advises the reserve manager and the deposit manager jointly.
Knowing that the lending transaction has been priced and accepted, the reserve manager then sells $ 1 million in treasury bills (previously funded by excess capital). The deposit manager puts out a bid for $ 9 million in additional funds.
Meanwhile, customer X has drawn down the $ 10 million in funds (in the form of a cheque drawn on A) and places them on deposit with his bank B.
Similar communications starting happening in bank B, and the reserve manager there is soon informed of the inflow of funds from bank A. He will now have an excess reserve position that he doesn’t want. At the same time he knows from market gathered information that bank A is looking for funds. He’s been informed by his loan manager that bank A continues to be a good credit risk. And his capital manager is comfortable in allocating capital to an interbank deposit transaction.
Away from the two transacting banks, the CB reserve manager observes that the market interest rate for interbank funds is quoted slightly above the interest rate he pays to banks for excess reserves. He’s satisfied with market conditions and leaves the system excess reserve setting alone.
The deposit manager in bank A has a bid out for $ 9 million in funds. Bank B’s reserve manager accepts that bid and places a $ 9 million interbank deposit with bank A.
Bank B also buys the $ 1 million in treasury bills sold by A through an investment dealer.
After the transaction, reserve accounts are flat again.
Had bank A been unable to attract funds from bank B for any reason in this example, A’s reserve manager could have gone to the central bank to borrow funds.
Additional internal activity within Bank B:
The balance sheet has increased with X’s $ 10 million deposit, $ 1 million in treasury bill assets, and the $ 9 million interbank loan to bank A. Bank B’s capital manager allocates $ 250,000 from his bank’s own pre-existing surplus capital to the interbank loan. This is a lower proportionate amount than Bank A allocated for X’s loan, because Bank A is itself a higher quality credit than A’s customer X. Again, B was in a surplus capital position prior to placing money with A. The assets in which this capital was previously invested (e.g. existing treasury bills) are now in a sense funded instead by $ 250,000 from the new deposit funding just raised. This is just internal book keeping reconciliation in order to keep track of the new allocation of risk capital and the corresponding depletion in surplus capital.
We assumed the CB reserve requirement was zero, so the new deposits for both banks have no impact on the CB’s strategy for the system reserve setting. In a system with positive reserve requirements, the CB would have supplied new reserves, perhaps through system repos with non bank dealers who took on new collateral purchased from non banks. The two banks would have competed for new deposits that had been created in conjunction with that reserve injection sequence, in order to attract their share of additional newly required reserves. And the process would compound from there. This is actually the textbook multiplier working in reverse, whereby the central bank responds to system deposit expansion by supplying any reserve requirement that follows from that. This is the actual mechanism when there are positive reserve requirements. The textbook description with the reverse causality is wrong. (See further discussion below.)
In a system with positive reserve requirements, banks can factor in the opportunity cost of holding zero or low interest paying reserves, if necessary, as part of their asset-liability pricing methodology; e.g. the opportunity cost of zero interest could be factored into the loan spread as a reflection of additional funding required to support the additional reserve requirement associated with the primary funding. However, factoring in the opportunity cost in the case of a rate of interest paid on reserves that corresponds closely to the short term risk free rate (or short term policy rate) is a bit trickier, because that sort of rate is more or less already justified by virtue of the fact that reserves are a risk free asset.
Anyway, that’s the complete set of transactions in my model.
Both A and B were subject to capital constraints. There seems to be some sensitivity evident in several recent comments, surrounding this term “constraint”. Minimum capital requirements are specified as a ratio of risk weighted assets. Banks must hold at least that amount of capital in order to be capital fit from a regulatory perspective. The regulatory requirement, as well as any self-imposed capital requirement that may be stronger than the regulatory standard, constitute an effective lower bound limit for capital held against an existing position. I call that lower bound limit a constraint. Banks may hold actual capital in excess of that lower bound, in which case they are said to have excess capital. Whether or not that sort of excess capital position is viewed as evidence of a non-binding capital constraint is a matter of preference. The minimum capital constraint is non-binding in the sense that the bank has ready access to additional internal capital not yet allocated to risk. The minimum capital constraint is binding in the sense that the level of aggregate minimum capital for the bank will actually shift higher when a new risk asset is added. As well, it is binding in the sense that the loan officer must still get approval from the capital manager in order to receive an allocation of capital for the risk he wants to take. Under either interpretation, the minimum capital constraint is a fact in the form of a lower bound limit for capital required against total risk assumed. That said, the terminology “constraint” is also borrowed from it’s converse use in the case of reserves, where MMT’ers (Modern Monetary Theory advocates) often say that banks are not reserve constrained in lending. Finally, substituting the word “restraint” in the case of capital seems a bit soft on the real meaning of capital requirements, because there is definitely a lower bound hard limit on required capital, corresponding to a given risk asset position.
In this case, although both A and B were subject to capital constraints, those capital constraints were dealt with smoothly by the presence of pre-existing excess capital in both banks. However, the transaction meant that the amount of capital actually allocated to risk taking, for each bank and for the system as a whole, needed to increase. That amount of additional capital had to be ready prior to the act of assuming the risk. And it was ready, in the form of an excess capital cushion.
A prudently run bank will tend to have surplus capital as a stock (unallocated) and expect new surplus capital as a flow (incoming retained earnings). That puts the bank in a position to acquire new risk assets without having to go to the new issue equity market for every additional transaction. The incoming flow of retained earnings is normally enough to satisfy new capital requirements over time. This recent credit crisis has been extraordinary in that sense; e.g. Canadian banks were very active in tapping the new issue market for equity capital, which is not normal. (They also tapped the market for qualifying debt capital, which is a more normal and a more regular occurrence in one form or another).
Both A and B had to notify their respective capital managers of a pending new utilization of capital for risk allocation. The stock of utilized capital had to increase, due to the lower bound limit set by regulatory and/or internal capital limits. Unutilized capital had to be available before the transaction could be approved. That’s what it means to be capital constrained. Thus, lending insofar as capital was concerned was an issue of both availability and pricing. The required capital was available because the capital manager had a surplus capital position and approved the transaction on the basis of the risk assessment and the capital it would require.
The case of reserves is very different. What happened in the model transactions is consistent with the MMT observation that banks are not reserve constrained in lending. This means simply that the central bank always provides sufficient reserves to the system in order for banks to square their required positions. In the example, and in Canada’s case, this means meeting a requirement of zero. Under normal market conditions, sufficient reserves should be available for nearly all if not all banks to meet their requirement through various transactions with customers and/or between the banks themselves. Under volatile markets conditions, which can be accompanied by relatively volatile reserve distributions among banks, some banks may need to access borrowing from the central bank. Banks with good collateral will be able to borrow and bring their reserve accounts back to requirement, which is zero in this case.
It is important to note that “absence of a reserve constraint” is intended to characterize the reserve effect, other things equal. This assumes that banks are of good enough credit and liquidity quality to able to source funds and attract the reserves they require in the normal course. In particular, it assumes that banks have met their capital requirements and are perceived by the market to have met them. If not, they may fail to meet their reserve requirements as well. This is the case with a run on the bank. And if a bank is unable to square its central bank reserve position because neither the market nor the central bank is willing to provide it with funds, that bank must go into some form of wind up such as the FDIC process, a situation ultimately attributable to an inadequate level of capital. In the model example of banks A and B, both banks were normally healthy and therefore not reserve constrained in the intended meaning of that phrase.
Bank A’s deposit manager did not have to assume anything analogous to an additional requirement for capital, as in the availability of a pre-existing “unused” stock of reserves, either for his own bank or for the system as a whole. That’s because he knew to expect that the funding necessary to attract reserves to return to bank A would be available in the normal course. Even if market conditions had been particularly “choppy” on that particular day, A could have assumed the risk of having to borrow from the central bank on a temporary basis. And most importantly perhaps, A’s deposit manager knew that A’s capital condition was such that A was perceived to be a good credit risk itself in the market place, and that therefore A should have no problem attracting deposits in the normal course. Provided A’s own capital constraints are not contravened, A should have no problem attracting required reserves in what is a closed system of reserves supplied by the central bank. Any problem in squaring reserve positions can only be attributed logically to market perceptions of capital inadequacy, which is generally how bank runs start. And even then, the bank can borrow from the central bank with good collateral. With adequate capital, the only issue for reserve management is pricing, not availability of reserves. Bank A was not reserve constrained in lending because the reserves required to square A’s position at the end of the day had to be available from somewhere. The normal function of a liquid market, including the central bank’s management of the short term interest rate, will ensure that regular transactions such as the interbank loan from B to A will accomplish the required rebalancing of reserves. The lending exercise insofar as reserves were concerned was only an issue of pricing. The required reserves were assumed to be available and in fact were available somewhere in the system, given the closed nature of the reserve system, and given bank A’s strong capital position and therefore its good credit rating.
Thus, the critical idea in the notion that banks are not reserve constrained is the willingness of the central bank to supply reserves to banks systemically and to banks specifically, sufficient to meet their required reserve levels systemically and individually, provided they are in sufficiently good capital health. Central banks supply required reserves on a daily basis. There is obviously no corresponding idea underlying the notion of capital constraints. Central banks may or may not supply required capital about every 80 years or so, in the midst of extraordinary financial crises and depression type risk environments. Otherwise, there is a constraint placed on the private sector commercial banking system to produce a net stock of capital at least equal to its minimum capital requirements, individually and therefore collectively. That collective capital stock must increase as risk assets increase. That is a constraint on the banking system. It must confirm additional unused capital availability or issue new capital in order to expand its risk assets and it must source this capital in the normal course without central bank or government provision of same.
Moreover, the causality sequence in the case of capital demonstrates the power of the constraint in comparison to the case of reserves. Banks must have required capital in place prior to the moment in time when their risk assets increase by lending or some other risk taking activity. The capital requirement precedes the risk expansion. This is opposite to the causality with respect to reserves, where the reserve acquisition that is required to square offside positions due to asset expansion follows the asset expansion in order of time sequence.
There is an enormous irony here around the issue of causality. MMT observes correctly that the central bank supplies required reserves in response to the banking system’s creation of new loan and deposits, most obviously in those systems that have positive rather than zero reserve requirements. This correct description positions the traditional textbook described “multiplier” causality as literally backwards. The irony is that the textbook causality direction applies correctly to capital, not reserves. Not only does a significant proportion of the economics profession not yet understand that the textbook model is wrong, but they don’t yet understand that the correct version of the same causality applies to capital rather than reserves, and they really don’t understand that they have effectively been confusing reserves with capital all along. Economists in general are typically weak on the subject of bank capital. That’s probably why they got the role of reserves wrong in the first place. The MMT group has distinguished itself in getting it right. Macroeconomic theory that directs itself toward an understanding of the financial system in general and the banking system in particular needs to do a MAJOR REBALANCING of conceptual thinking around the dual subjects of capital and reserves.
In sum, banks are not reserve constrained in lending because they source required reserves from each other’s existing positions and/or from the central bank, and they do so after the fact of new loan and deposit creation. Banks are capital constrained in lending because they must source new capital for risk allocation either internally or externally, and they can’t get it from the government (normally), and they must have this source of risk capital in place before the act of lending.
So my conclusion as always is that bank lending (as well as other forms of risk taking) is capital constrained but not reserve constrained.
ADDTIONAL NOTES ON NICK’S POST
The following points roughly follow the order presented in Nick’s post:
As discussed, but to address Nick’s first example, the textbook theory of the multiplier is wrong. The main reason it is wrong is the aspect of causality. Banks do not require reserves before lending. The central bank supplies the reserves required by banks as a whole based on the deposits created by bank lending and the statutory requirement pertaining thereto. The reason the central bank supplies required reserves is to control the upper bound for the target fed funds rate or range. Otherwise, banks would bid the actual rate above target should the aggregate reserve supply be inadequate according to the requirement created by deposits. The reserve requirement in Canada is zero. That is different than saying there is no requirement. The requirement is zero because the central bank expects banks to target for a flat reserve position over time.
Nick’s “Loan Officer Theory of Money Supply” is interesting. It initially assumes perfectly inelastic “inputs”. But then Nick seems to assume the desired conclusion, which is that inputs of assumed influence on loan supply become influential if their supply becomes elastic. The conclusion is presupposed; e.g. reserves influence loan supply when their supply becomes elastic. Nick proves it by assuming that’s what happens when supply changes from inelastic to elastic. But that’s not what happens in fact. Banks do not depend on an injection of reserves before the fact in order to lend, as explained earlier.
Nick’s “Bank Capital Theory of Money Supply” starts out roughly OK in terms of the idea of a required capital ratio. But it veers off course because it ignores the difference between risk and nominal assets. The purpose of capital is to absorb unexpected losses. Capital is allocated based on the risk of unexpected loss. That is why nominal asset amounts are “risk weighted” when determining a required capital allocation. The risk weighting for a (Canadian) residential mortgage is far less than the risk weighting for an equivalent nominal amount of junk bonds. Treasury bills have zero risk weighting. Central bank reserves have zero risk weighting as a commercial bank asset. So Nick is not accurate when he says that a capital model fails merely because a (nominal) loan/capital ratio varies. The nominal ratio varies constantly based on changes in the risk weighted composition of the nominal asset base over time.
Then we get into the notion of the supply elasticity of capital. I’m not sure the idea is that relevant. The important idea is that the cost of capital at any time is a major input into loan pricing, as already described above. Banks are capital constrained in lending, as I’ve defined it, which effectively means that banks must source required capital, internally or externally, normally from non government sources, and necessarily before the fact of risk taking, i.e. before they can book a net new risk asset. I’ve said that banks typically run surplus capital positions as buffers in good times, even while generating capital internally from retained earnings. They don’t take on new risk and then go find some external capital to support the risk. For starters, they’d be outside of regulatory capital guidelines in taking on that incremental risk without capital support in place. And they’d be fools to assuming such a strategic pricing risk on the most expensive and riskiest form of funding there is, even if there were no capital regulations, but in the latter case we’re into a contradiction about why they’d need to go get more capital anyway. In any event, the type of unconstrained approach that would be reckless in the case of capital is a matter of fact way of dealing with reserves in a normal and prudent fashion. Banks need only square reserve positions after making a new loan. The operative temporal causality is the reverse of that of capital. Banks require capital before the fact to lend. Banks don’t require reserves before the fact, because the system (including the ultimate lender of last resort – the central bank) will always be willing to supply reserves from somewhere to a capital-healthy bank in response to its need for reserves. And that is because the central bank is the monopoly supplier of total reserves at its chosen price, and because it will ensure that the market with respect to healthy banks will clear at that price. There is no such central pricing mechanism with respect to either the availability or the pricing of bank capital. That blunt difference is in part what we mean by saying that banks are capital constrained but not reserve constrained.
With respect to loan supply, neither the interest rate on reserves nor the marginal cost of capital affects the supply of loans, provided that prior to the lending transaction the bank has an identified source of new capital as required, either internally or externally. Capital costs obviously affect the pricing of loans, which I described above as part of my model, and pricing obviously affects demand. In the case where new risk capital supply is not available internally, and the bank perceives the cost of externally sourced capital to be too high, it may shut down the supply of loans. But this is due to the bank’s judgement that the demand for loans will be insufficient at such a market price for new capital, given the associated high cost of externally sourced capital that must be factored into incremental loan pricing. This is not due to an inability to supply loans at that cost of capital, assuming external supply can be sourced at that cost. It is due to an expected insufficient demand at that pricing point. Supply potential is definitely affected by the availability of unused or surplus capital; that’s in part what it means for lending to be capital constrained.
Bank capital is always important, Nick – not just in this recession. What is important in this recession is that capital levels have been under pressure in proportion to the unusual severity of the recession.
Reserves are only an “input” to the production of loans and money according to their pricing, which is determined by the policy rate set and controlled by the central bank. And even then, reserve pricing may only be a benchmark reference pricing point for the market rates that are applicable to new asset liability transactions with the customers of those banks, transactions which have the effect of transferring reserves from or to those banks as a consequence. In any event, reserves are not an input in terms of their availability, because the central bank is the monopoly supplier of reserves to the system at its chosen price, and capital healthy banks will have no problem attracting their required share of them based on that price, ultimately from other banks or from the central bank.
The marginal cost of reserves affects the pricing of loans but not the availability of loans. It is the risk assessment of loans and the availability of capital to allocate to that risk that affects the availability of loans. Note that banks have complex credit policies that may restrict their risk exposure in aggregate across all sorts of dimensions – e.g. industry, geography, loan type, etc. etc. This credit analysis, which is inextricably linked to capital availability, has nothing to do with any question about the availability of reserves. It has to do with the availability of capital to support that risk category under consideration.
As far as the supply of loan officers is concerned, it seems reasonable to assume that supply will affect both pricing and supply of loans, given the labour requirement to do risk analysis prior to supplying.
Regarding your closing paragraph, I agree in particular with two points made by Declan in an earlier comment:
“... capital requirements are directly influenced by policymakers as well. The question might better be framed as why economists concentrate so much on central banks as opposed to financial regulators ... the two constraints on lending are capital and the availability of willing/capable borrowers (in some sense these two constraints are one and the same if capital requirements are set accurately - basically banks can lend as much money as they want as long as they don't threaten their own solvency in so doing).”
There’s another discussion in the comments concerning the difference between loans and money and the meaning of capital constraints in this regard. This is a good place to emphasize that the purpose of capital is to absorb unexpected financial losses due to any kind of risk. Capital allocation is inextricably linked with risk assessment, and capital must be allocated to all assessed risks. So far we’ve been talking mostly about a simplified model that addresses loans and corresponding credit risks. But there are additional kinds of risks to which banks allocate capital – especially market risks such as interest rate risk, foreign exchange risk, and equity portfolio risk, which are analytically separable from credit risk, but frequently in combination with it in the form of counterparty credit risk contingent on market risk. And there are market related risks that are relevant for structural asset liability interest rate mismatches. All of these risks and more require allocations of capital. There is even a complex category called “operational risk”, in which activities just about anywhere on the balance sheet can attract additional risk capital requirements, including the deposit gathering function, depending on the risk assessment for operational disruption scenarios.
CONCLUDING COMMENT
Much of what I’ve said should be consistent with standard post Keynesian interpretation of banking system reserve dynamics, MMT in particular. As far as I can discern, MMT tends not to focus so much on capital directly, but more generally on the idea of creditworthiness as being the driver of lending supply rather than reserve availability. It should amount to the same thing as my capital interpretation. I don’t know if MMT experts would agree with everything I say about capital. But I wouldn’t expect enormous objection to it.
I would recommend again the following blogs for source MMT material:
Kansas City (Scott Fullwiler, Randall Wray, and others)
http://neweconomicperspectives.blogspot.com/
Winterspeak
http://www.winterspeak.com/
Warren Mosler
http://www.moslereconomics.com/mandatory-readings/soft-currency-economics/
Bill Mitchell (an interesting Australian banking comparison for Canadians)
http://bilbo.economicoutlook.net/blog/?page_id=1667
No comments:
Post a Comment