Why can't all banks be as safe from insolvency as the Bank of Canada?
I put a question on my final exam:
"What is a bank? Should banks have legally required minimum reserve ratios? What about 100% reserve ratios? Should banks have legally required minimum capital ratios? What about 100% capital ratios?"
I wanted the students to talk about illiquidity and insolvency. I only threw in the bits about 100% ratios to get them to think about trade-offs, by thinking about the two opposite extremes.
There is a long literature on banks with 100% reserve ratios. Forget that.
There is a literature on capital ratios. But sometimes it helps us get our thoughts clear by imagining an extreme case. What about banks with 100% capital ratios?
What would a bank with a 100% capital ratio look like?
I can think of two very different answers:
1. The bank has (say) $100 in chequable demand deposits on the liability side. And $200 in (say) loans on the asset side. So its net worth (shareholders' equity) is $100, and equal to 100% of its demand deposits. Basically, the shareholders give $100 to the bank in return for shares, and the bank lends out that $100, plus another $100, and creates $100 in demand deposits. But that bank (unlike the 100% reserve bank I told you to forget about) can still become insolvent. If its loan portfolio loses more than 50% of its value, that bank would become insolvent. It would need an infinite capital ratio to completely eliminate the risk of insolvency. (Or capital equal to 100% of its assets, which means it has no deposits at all, and isn't really a bank.)
2. The bank has $100 in loans, and $100 in chequable demand deposits, but those demand deposits are themselves shares in the bank. They do not have a fixed dollar redemption value. So if you want to say they are not really deposits, OK. But they are chequable. If you write a cheque for $20, the bank debits your chequing account by S shares, where S = $20 divided by the current price of shares. So the shares are not the medium of account (Bank of Canada currency remains the medium of account), but the shares can be used as a medium of exchange, provided those shares are "deposited" in the same bank whose shares they are.
That second 100% capital ratio bank cannot become insolvent, unless its loans lose 100% of their value. Because its own shares are its only liability.
There is a risk that cheques might bounce if the share price dropped too much in the short delay between looking up the share price to see what's in your account, writing the cheque, and the cheque clearing. But electronic debit cards could help resolve this problem, by shortening that delay.
(It's a bit like writing cheques on a closed-end mutual fund, where the mutual fund owns shares that are not themselves traded. I vaguely remember reading some monetary or finance economist writing about something like this. Was it Fischer Black? Or Fama? Both my memory and Googling skills have failed me. [Update: Bill Woolsey says "It's Fama. But Greenfield and Yeager's first pass at the BFH (for Black-Fama-Hall) emphasized purely mutual fund banking."])
[Just as an aside: what would happen if those banks bought put options on their own shares, to make their shares safer?? Do companies ever do that? It's too weird to think about.]
You can think of the Bank of Canada as a bank like that second case. Owning a $20 note is like owning 20 shares in the Bank of Canada. They are worth whatever the market thinks they are worth. Except they are non-voting shares. And they don't pay dividends. (They do pay dividends if you deposit them in your chequing account at the Bank of Canada, but only commercial banks and the government are allowed to do that). And the management turns over most of its profits from its loan portfolio to the government, the voting shareholder. The management issues new shares, and buys back old shares, to try to make capital gains on the shares stay at roughly minus 2% per year. In a worst case scenario, the Bank of Canada might fail to prevent the shares depreciating at only 2% per year. But the Bank of Canada can never go insolvent. Because it has a 100% capital ratio. Because its own shares are its only liability. It's a bank whose own shares are used as money.
Why can't all banks be as perfectly safe from insolvency as the Bank of Canada? They could be, if we imposed 100% required capital ratios on all banks, interpreted in that second sense.
(BTW, I didn't expect my students to give that answer.)
[Update: I now realise this post wasn't clear enough, so I am going to add this, from one of my comments below, to make it clearer how my 100% capital version 2 works:
Suppose I bank at BMO, and I have 100 BMO shares in my chequing account at BMO. The bank is like my stockbroker.
Suppose I write a cheque for $20 to buy a bike from someone who banks at TD. That cheque is an instruction to my bank to sell $20 worth of my BMO shares and transfer the cash proceeds to TD bank, which buys $20 worth of TD shares and deposits those shares in the bike seller's account.
Presumably other people who bank at TD are writing cheques to buy things from people who bank at BMO. So there is a central clearing house which is transferring reserves between banks (just like today), but is also buying and selling bank shares, just like the stock market does today.
If I want to withdraw $20 in currency from my account at BMO, the teller would give me $20, send an instruction to their broker to sell $20 worth of BMO shares on the stock market on my behalf, and debit my account the appropriate number of shares, depending on the current market price.
If the stock market were closed, because it's a weekend, so that the current price of BMO shares was not observable, BMO might need to impose a haircut on immediate withdrawals. You can only withdraw 80% of what is in your account based on Friday's share price, just in case BMO shares drop by 20% when the market reopens Monday morning. I think that is the only case where a delay comes in, and it's only a delay in withdrawing the full 100% of what is in your account.]
Worthwhile Canadian Initiative 
I thought that the (non-bank) UK lender Wonga is an example of a lending company that has a 100% capital ratio in that it is 100% funded by venture capital and retained earnings. P2P lenders such as Funding Circle I guess are lenders with zero capital but 100% maturity matched loans and liabilities.
I guess M-pesa in Kenya and Paypal are examples of payment systems that are not also lenders. Those (non-bank) payment systems sort of have 100% reserve backing.
Banks are fragile IMO because they amalgamate the lending function and the payment system function and we use bank liabilities as a medium of exchange. We end up bailing out irresponsible lenders so as to preserve the function of the payment system but they needn’t be connected. The payment system could be just like m-pesa or paypal IMO.
I think the fragility of our lending system comes from maturity transformation. If all lenders only issued debt that was of the same time to maturity as the loans they made, then a “bank run” would not imperil the bank. It would simply reduce the amount that the debt could be resold for in the secondary market. The risk would be born by the bond holders not by the lending company.
JKH @2.32: Agreed. Banks can make loans and create money in much the same way they do now. And loans could get repaid, or not get repaid, in much the same way they do now. The difference is in who bears the losses if the loans do go bad.
stone: “Banks are fragile IMO because they amalgamate the lending function and the payment system function and we use bank liabilities as a medium of exchange. We end up bailing out irresponsible lenders so as to preserve the function of the payment system but they needn’t be connected.”
We are on the same page there.
“I think the fragility of our lending system comes from maturity transformation.”
Maturity transformation mostly causes liquidity problems. (It can cause solvency problems too if interest rates rise.)
If liquidity problems were the only problems, I wouldn’t worry much. A competent lender of last resort central bank can handle those, with no losses (and maybe even profits) to the taxpayer. It’s the bank solvency problems that worry me.
Hmmm. Suppose we lived in a world where all banks were like my 100% capital version 2 banks.
Now suppose some economist came along and wanted to advocate a change, in which all banks changed to become regular banks, like we see in today’s world. What would that imaginary economist say? He would say:
“All banks should be forced to give their shareholders (at least, those who deposit their shares at the bank for safekeeping) a put option, that can be exercised at any time, to sell their shares back to the bank at a fixed price of $1 each. But if you get that put option, you don’t get dividends.”
It would sound crazy.
That is Miller-Modigliani, isn’t it?
There’s more risk left in the shares that aren’t put back.
Because capital and share value have a higher probability of going to zero for a given asset risk level.
While those left with deposits have a higher probability of incurring no losses at all compared to the case of their previous share holdings.
JKH,
“He’d service the loan just as people do now from normal deposits.”
Sure, that is the theoretical case that we could also have now. But loans go bad and that is the reason banks have capital requirements/loan-loss reserves. I don’t see in Nick’s case 2 where those would be coming from. Who would hold the loan and how would it be accounted for? I think we agree that when a loan gets bad the depositors would bear the loss through a corresponding reduction in the share price. Translating it to today’s world: The depositors at a bank would bear any losses the bank would make through failed loans. How that would make for a better/more stable banking system than we have now is not clear to me. In fact, whether a bank issues its own shares or $ would not make much difference; the system of loan losses could be changed with both. Just deduct an appropriate amount from each deposit account once a loan fails. It’s like going back before the FDIC. I think your’s and Frances’ concern about bank runs in that model are very true.
Odie,
I’m just trying to interpret ideas that are being floated.
These are pretty radical ideas but fun to play with sometimes.
One point I’d add though is that I think these kinds of proposals such as the Chicago Plan that tend to carve out tranches of risk to different institutions (e.g. deposit banks; equity banks; whatever) tend to overlook the way in which actual banks deal with such risk tranches internally. For example, the idea that banks shouldn’t be able to fund risky loans with demand deposits is already considered internally in banks through the way they manage liquidity and interest rate risk – within determined limits for exposure. The Chicago Plan includes matched maturity funding for loans. That’s already considered in actual banks – considered and acted on, – although probably not intended with the same 100 per cent degree of “perfect” matching. The fact that the financial system blew up in the financial crisis makes these kinds of proposals more attractive, but some banks were reasonably well run through the crisis as well.
If the lenders got funding by selling maturity matched convertable bonds that converted to equity if the lending company became insolvent, then wouldn’t everything be rock solid? The lending company could have a certain amount of equity that would be first in line to take losses. If that was exhausted, then the bond holders would become the new owners. If say a retail saver held the bonds via an ETF then presumably the ETF administrator would sell the equity whenever convertable bonds converted. Presumably that happens now with “total bond market” ETFs doesn’t it?
I had a go posting about this earlier this year
To my mind debt finance needs to cost what it really costs as a self supporting system. Then we wouldn’t get economic distortions due to over-leveraging etc.
Why use the equity of lending companies as a medium of exchange as in version 2 ?
To make a silly example, it would be no more awkward to use the equity of say utility companies as a medium of exchange. Utility company equity might actually be less unsuitable since that is typically less risky. To my mind the whole issue stems from our dubious practice of amalgamating lending companies and payment administrating companies. We could have a payment administrator such as paypal or m-pesa that was totally seperate from the lending companies.
The medium of exchange used by such a pay-pal / m-pesa system could be stable and the same as or fixed convertable to the medium of account.
Nick said: “JKH @2.32: Agreed. Banks can make loans and create money in much the same way they do now. And loans could get repaid, or not get repaid, in much the same way they do now. The difference is in who bears the losses if the loans do go bad.”
If the capital requirement is 100%, I’m pretty sure banks can’t create “money” overall (with emphasis on the overall part).
Nick’s post said: “stone: “I think the fragility of our lending system comes from maturity transformation.”
Maturity transformation mostly causes liquidity problems. (It can cause solvency problems too if interest rates rise.)
If liquidity problems were the only problems, I wouldn’t worry much. A competent lender of last resort central bank can handle those, with no losses (and maybe even profits) to the taxpayer. It’s the bank solvency problems that worry me.”
I agree with Nick there. Think about this scenario. A bank makes all 3 year car loans (bank assets) with a 10% capital requirement. It has 90% of assets in 3 year CD’s. There is no liquidity risk. At the end of the three years, assets have fallen by 20%. There is a solvency risk. The CD holders won’t get all their “money” back.
Heavens, what a lot of comments since my last appearance!
Nick,
Calling it a “funding gap” is bank-speak. You can call it a “risk-bearing gap” if you like. The point is that you will have an increased need for people willing to contribute risk capital to the bank. They might not.
Like JKH, I had missed the “market sale” part. I had assumed that you were simply converting deposits to shares, but what you are actually doing is investing the deposits in shares, which is slightly different and does require market transactions. I agree this removes the risk of bank runs, but heavens it’s complicated. All this just to force depositors to take risk?
Which brings me back to the point I made about risk aversion (or perhaps more correctly, loss aversion). I’m with you on the sheer illogicality of pretending that banks can provide a safe home for the savings of people who really can’t or won’t take risk. The right place for those savings is the government.
As you’ve probably realized by now, both JKH and I looked in some detail at the IMF’s Chicago Plan Revisited and independently concluded that it was unworkable as a private sector banking model. I think your model is too, for essentially similar reasons. I don’t necessarily think that’s a bad thing, though. My analysis is here: http://coppolacomment.blogspot.co.uk/2012/10/the-imf-proposes-death-of-banking.html
Re your MoE: continually investing deposits in shares only results in the shares themselves becoming a medium of exchange if the transactions are always netted out instead of settled. If they are settled via reserve accounts then the MoE is currency, not shares.
JKH,
Yes, the clearing arrangements in the IMF plan were problematic. We ended up with reserve pre-funding of loan agreements, which as the IMF also restricted the types of loans that banks could make effectively meant that the central bank had to approve every loan. Mind you we are fast heading down that road already with the level of intrusive regulation being put in place.
TMF,
Money would be created in much the same way as at present. This is because of the way the accounting works. Loans always expand both sides of the balance sheet, so a new loan asset must be balanced by (we assume) new share issuance. It is the expansion of the RH side of the balance sheet that is new money creation: exactly what form that expansion takes doesn’t matter. However you look at it, money is created, because there is a new loan which consists of a completely new asset-liability balanced entry.
Frances Coppola said: “Money would be created in much the same way as at present. This is because of the way the accounting works. Loans always expand both sides of the balance sheet, so a new loan asset must be balanced by (we assume) new share issuance.”
Let’s stick to demand deposits here. Loans create demand deposits (“money”). JKH has told me new share issuance destroys demand deposits (“money”). Net change overall is zero with a 100% capital requirement.
I don’t believe the shares should be considered “money”.
TMF
I don’t think it is correct to say “destroyed”. If the conversion process involves a market transaction, then the money has simply moved to the seller of the shares. That would apply even if the seller was the bank itself – as in a rights issue, for example. If the conversion does not involve a market transaction, then what we have is something more akin to a scrip issue, but in this case it would be correct to regard the shares as money.
Too Much Fed’s response to Frances just above can be put more strongly. Too Much Fed says “I don’t believe the shares should be considered “money”. I’d ask the question: if shares in banks (or to be more accurate, the mutual fund section of the banking industry) are in danger of becoming money, why isn’t every share quoted on the stock exchange counted as, or used as money?
Re Frances’s reply to TMF (just above) and the question as to how money is created, there is more than one way of working this under the system contemplated here. Those who want to ban private money creation (like Positive Money) wouldn’t let the mutual fund section of the banking industry issue cheque books, plastic cards or any other method of transferring money. Only the “100% safe money” or full reserve section can do that.
Also, the mutual fund section would not be allowed to extend loans unless it had first received funds to match from the sale of new shares (as assumed by Frances just above). That system would give the issuer of base money (i.e. government / central bank machine) much tighter control over aggregate demand than at present.
An alternative would be to let the mutual fund section of the banking industry lend money into existence, regardless of whether there were shareholder funds to match. That would be more like the existing system. And doubtless, as under the existing system, we’d have rules on minimum capital ratios.
Hope that makes sense.
Nick,
Thinking about this more, I believe the way you described the checking transaction in your comment and the way I (and probably Frances) originally interpreted it are compatible. However, they are descriptions of two different things, both of which are (potentially) involved in the process:
E.g.
You write a cheque on your BMO equity account payable to me.
I bank at TD.
I deposit the check in my TD equity account and I get a TD equity credit in that account.
TD clears the check back to BMO.
BMO pays TD with reserves.
BMO debits your BMO equity account for the amount of the check.
No sale.
Now that’s the way I originally interpreted it.
And I now believe that is correct.
What you said is that BMO sells BMO equity shares in order to cover the check obligation.
That’s also correct.
But ONLY IF the BMO reserve manager chooses to do that – and he has a choice.
So let’s follow that through.
Suppose the BMO reserve manager decides to sell BMO equity shares to cover the outflow of funds from the check obligation.
Suppose he sells them to Nick E., who now happens to bank with TD also.
What happens is essentially the reverse of the original checking transaction.
I.e. Nick E. writes a check on TD, deposits it at BMO, gets equity credit at BMO, and BMO clears the check back to TD, and TD gets paid in reserves.
The net result is an inflow of reserves to BMO that offsets the original reserve outflow from your check to me.
So what I described in the first part is the actual checking transaction.
What you described in the second part is the decision of the BMO reserve manager to offset the reserve and equity effect of the first transaction.
So what you described is not the original checking transaction, but a funding transaction designed to offset the original checking transaction.
That sort of thing is a discretionary decision on the part of the BMO reserve manager.
For example, if the BMO reserve manager already has sufficient reserves in his account to cover your check to me, he would not react due to that transaction alone. BMO would debit your equity account and the central bank would debit BMO’s reserve account. The BMO balance sheet shrinks as a result.
And the same thing and the same choice happens in reverse at the TD end. The TD reserve manager has a choice as to whether to simply leave the reserve impact of the new equity inflow in place, or to react – meaning to deploy reserves – meaning in this case to buy TD equity shares.
The interesting thing about that last sort of decision is that the balance sheet effect is the same as occurs in real life with a share buyback – the reserve account is debited (i.e. assuming the shares are bought by people banking at other banks) and the same amount of TD equity disappears. And the purpose is analogous in an interesting way. A normal share buyback is designed to improve earnings per share as the bank believes there’s too much high cost equity in the capital structure. In this case, it’s already an all equity capital structure, but the Treasurer may feel he should shrink the balance sheet rather than put on some other asset against the new equity. So he gets rid of his reserves by buying back equity.
In fact reserve management involves continuous action and reaction by the reserve manager to deal with myriad flows in and out.
So I believe the original checking transaction is as I originally assumed – i.e. BMO equity is debited but shares are not sold.
And what you described in terms of a BMO share sale is actually a funding transaction that replaces the lost equity and the lost reserves – which is a follow up transaction (possibly) but a separate transaction.
In summary, the full sequence in your case of a BMO share sale is:
Debit BMO equity and reserves for the original check
Credit BMO equity and reserves for the share sale
And if you assume Nick E. pays for his new shares with a check, what you have is:
Debit BMO equity and reserves for your check to me
Credit BMO equity and reserves for Nick’s check to BMO
So you can see clearly there that what you described is essentially a reversing transaction designed to cover the outflow of funds from the first checking transaction – which is separate from the first checking transaction itself.
(I’ll leave the medium of exchange question up to you.)
Hope that makes sense.
TMF,
In the real world banking system, newly issued equity is paid for by debiting bank deposits. From the full system perspective, equity increases, deposits decrease. If I used the word “destroy” somewhere to describe the debiting of deposits, that’s what I meant.
In Nick’s system, the banking system is 100 per cent equity funded. There are no deposits in this system. So the system can only expand its equity by increasing assets – i.e. loans create equity in this system in the same way that loans create deposits in the real world system.
I think any proposal like this would need to allow for the sort of process JKH describes where the bank can decide whether to vary its issued share capital in response to variations in the demand for holdings. It would also probably make sense to allow banks to deal with excess liquidity by purchasing shares in other banks (a sort of inter-bank market), and for them to have tranched equity, with the deposit substitutes representing a form of preferred share rather than true equity shares. All of this sort of thing would help stabilise their value.
The various proposals for safe banking all have their flaws, but rather than dismiss the idea out of hand, I think it’s more useful to think about how those flaws might be addressed and to work out what would be the best way to structure such a system. Personally I’m still on the fence on this one, but I think there’s enough merit in the idea for it to be taken seriously. If nothing else, thinking about how one would structure things might throw up ideas that would improve the existing system.
JKH: “That sort of thing is a discretionary decision on the part of the BMO reserve manager.”
Yep. I now think that is key.
In the limit, as the BMO reserve manager always buys back shares or issues new shares, to keep the market share price always equal to $1, and is seen as having a de facto obligation to do so, we end up back with banks very much like the banks we have today. And the bank goes “insolvent” when he is unable to buy back shares at $1. The bank shares “break the buck”.
In the other limit, as the BMO reserve manager never does this, we get the sort of bank I originally had in mind when I wrote this post.
Nick,
Regarding your point on maintaining the stock price, I see it a bit differently:
Again, the type of sale transaction you described is really a follow up funding transaction designed to cover the reserve and equity outflow due to a check written on BMO.
The BMO manager would issue stock or buy back stock “in the limit” as you say if he were following a strict rule to offset all transactions at the micro level. There’s some logic to that, as that preserves the balance sheet size and the level of reserve balances, other things equal.
Although it wouldn’t apply in the case of “loans create equity”, since that on its own is an intended balance sheet expansion. If the borrower subsequently wrote a check on his equity, then it would apply, but that is a separate discrete transaction.
Conversely, if the BMO manager “never” responds to either equity inflows or outflows, then you would have a wildly fluctuating balance sheet – a balance sheet that is essentially unmanaged, because there’s no strategy at all for reacting to equity flows and reserve flows. The reserve position is essentially not being managed in this case.
Anyway, my point really is that none of this would have any predictable effect on the stock price. The stock is determined by the market, according to the market’s perception of macro stock market risk and risk specific to BMO.
In order to target the stock price, the manager would have to offer to buy or sell in unlimited quantities as necessary on both sides the market. The balance sheet (i.e. quantity) would also fluctuate wildly in that scenario, but according to the price objective for the stock. However, this objective seems irrational IMO, because there’s no reason to set up a 100 per cent equity funded bank in the first place if you intend on fixing the stock price. The market needs to determine the risk and the pricing of equity.
In the conventional sense, I imagine there’s evidence to show that banks tend to buy back their stock when the going is good and they have surplus capital and the stock is relatively expensive, and they tend to issue stock when the going is tough and they are short on capital and the stock is cheap (which is not good for issuance), all that on a cyclical basis. There’s a certain amount of shooting in foot I think.
JKH, if the capital requirement/ratio is 100%, does that mean $ borrowed = $ saved so that banks can’t create “money” overall? Does that also mean banks can’t create more “money” than they destroy?
Frances Coppola said: “TMF
I don’t think it is correct to say “destroyed”. If the conversion process involves a market transaction, then the money has simply moved to the seller of the shares. That would apply even if the seller was the bank itself – as in a rights issue, for example. If the conversion does not involve a market transaction, then what we have is something more akin to a scrip issue, but in this case it would be correct to regard the shares as money.”
JKH can correct this if it is wrong.
When new equity is issued by a bank, the demand deposits move and are then destroyed. When new equity is issued by say a retailer, the demand deposits move and are not destroyed.
JKH said: “In Nick’s system, the banking system is 100 per cent equity funded. There are no deposits in this system. So the system can only expand its equity by increasing assets – i.e. loans create equity in this system in the same way that loans create deposits in the real world system.”
JKH, I can’t tell if Nick’s “banks” are 0% capital requirement or 100% capital requirement.
Case 1 0% capital requirement
$1,000 in loans with $1,000 in demand deposits
The demand deposits then trade like shares without 1 to 1 convertibility to currency.
Case 2 100% capital requirement
$1,000 in loans with $1,000 in shares
Ralph Musgrave said: “Also, the mutual fund section would not be allowed to extend loans unless it had first received funds to match from the sale of new shares (as assumed by Frances just above).
That is what I think of with a 100% capital requirement.
Do Nick’s “banks” have a 0% or 100% capital requirement?
Ralph Musgrave said: “I’d ask the question: if shares in banks (or to be more accurate, the mutual fund section of the banking industry) are in danger of becoming money, why isn’t every share quoted on the stock exchange counted as, or used as money?”
The shares should not be counted as “money”. They are not 1 to 1 convertible.
zyxzyxooxwo