In this article I assume that the reader is somewhat familiar with balance sheet concepts, and some of the basics for bank accounting. There is a series of tutorials on the subject by Eric Tymoigne. It has been awhile since I read those tutorials, and I am unsure how my discussion here relates to his arguments. My comments here may or may not fit into any accepted school of economic thought.
I have mainly been concentrating on my inflation-linked report, so this article was somewhat rushed. I am not able to cover all of the points that I probably should. (I might do a primer on banking in the future. The working title is: "Fractional Reserve Banking Basics: Why Fractional Reserve Banking Is Inseparable From Industrial Capitalism, So Stop Slagging It Off, Eh?")
Customers Limit ExpansionFor any business, its ability to expand is limited by its ability to attract customers that can be profitably serviced. (Some firms may decide to put profitability concerns to a side and just go for growth, but that is not sustainable.) One could argue that the resource extraction industry is in a somewhat different boat, but even there, if nobody can afford to purchase their output, the industry will have to curtail production.
Banking is no different. Banks are not in the "money creation" business, they are in a lending business. Imagine that you walk into a bank, looking for a loan. You will sit down in an office with a loan officer (several loan officers if the loan request is large). There will be no discussion whatsoever about the bank's ability to extend the loan, instead the entire conversation will revolve around your plans. How much do you want to borrow? What will you do with the loan? How will you pay it back? Do you have collateral? Etc.
The decision whether the loan will be extended depends upon the following criteria.
- Do you want to borrow money in the first place? (Why you would walk into the bank otherwise is not entirely clear.)
- Does the loan officer have the authority to make a loan of the given size? (In most cases, this would just require setting up a new meeting with a more senior lending officer. However, it is possible that even if you bump all the way up the hierarchy, the bank cannot give you the loan due to size issues.)
- Would the bank be willing to lend you the money at any interest rate for this particular loan?
- Are there any other banks willing to lend to you at a lower interest rate (in which case you presumably go there)?
- Are you willing and able to borrow at the offered rate of interest?
Only point #2 (does the bank officer have the authority to make the loan?) reflects an internal constraint on bank lending; every other point either depends upon the customer, or the banks competitive position versus other banks.
We can then ask ourselves: does this constraint ever show up in practice? During the Financial Crisis, banks were unwilling to lend. Otherwise, I cannot remember any other non-crisis period where we saw banks tell prospective clients with reasonably-sized loans that they are unwilling to lend to anyone at any interest rate. In other words, the only effective constraint on growth during normal circumstances is the demand for loans (by creditworthy customers at the market interest rate).
The remainder of this article discusses why a loan officer would not have the authority to make a loan, and why this is generally rare.
Concentration LimitsOn top of capital constraints, there are concentration limits imposed on well-run banks. These limits might not exist in some countries, but they are an important feature of modern banking systems in most of the developed countries.
The reasoning is straightforward. Banks run their loan books on a portfolio basis; they want to have a lot of small loans, so they can use semi-actuarial principles to estimate loan losses. They do not use mark-to-market accounting on their loan books; they come up with a statistical estimate for potential losses on current loans, and have a loan loss reserve. (Once a loan is no longer current - the customer missed required payments - they have to categorise the loan differently. The objective is to then mitigate losses.)
For mark-to-market purists, the ability to keep loans at book value (modulo the loan loss reserve) is "mark to fantasy" accounting. However, it allows the banking system some ability to weather small crises. The well-known problem is that if too many bad loans pile up, the bank has to slowly crawl itself out of a known problem. You end up with what is often called a "zombie banking system," which is unable to finance growth properly.
Loan concentration destroys the principle of running a relatively diversified portfolio. If you have a single loan that is larger than bank capital, that one borrower defaulting blows out the whole bank. ("If you owe the bank $100,000, it's your problem. If you owe the bank $100 million, it's the bank's problem.")
Some potential loans are too big to extended by a single bank. For example, the amounts needed for a large project, such as a new railway or canal. Meanwhile, running to 100 banks is hardly a solution -- what happens if you only get partial acceptance? You locked yourself into borrowing money, but not enough to build the project. (Admittedly, there are syndicated loans, but see the point below.)
This is why the bond market exists. It allows investors (that are willing to take more risks than bank loan officers) to finance large projects. (Syndicated loans are technically not bonds, but they often end up being sold to qualified bond investors.)
The relationship between the bond market and banks is often underestimated. Many analysts draw a sharp distinction between the two: banks create loans flexibly, whereas bond markets allegedly allocate existing funds. That misses the joy of leverage in the fixed income markets.
Rather than lending to the firm building the railroad, the banks lend to the bond investors. This allows for "flexibility" in the lending market, beyond pre-existing "loanable funds." From a top-down perspective, the bank is actually lending most of the funds used to buy the bond. The down payment by the bond investor inserts a new sliver of equity behind that piece of the loan: the bank has a claim for the loan against the bond investor, on top of bond, which will be collateral for a loan. (In the modern world, this is normally done through repo financing, not outright bank loans.)
This means that "large" debts will always be funded by the bond markets, and that the formal banking system will never extend the credit.This cannot be captured by simplified models that aggregate the banking system into a single bank (plus a central bank). If there were a single private bank, it would presumably be big enough to ignore concentration limits (except for lending to governments).
The extension of concentration risk is industry risk. In practice, we tend to see capital investment concentrated in a few popular sectors. Even if an individual borrower is below concentration limits, banks could easily be highly exposed to a particular sector. (This was a common feature of earlier banking crises in the formal banking system.) The modern practice in well-run banking systems is to vent this risk onto the fixed income markets.
This modern practice makes some assertions about the inevitability of repeats of the Financial Crisis suspect. One of the reasons for complacency going into 2008 is that the system worked in the telecom crisis. Although many casual observers were captivated by the bubble in the dot-com shares, the real danger was in the debt issued by the telecom industry to pay for the 3G licenses and capital expenditures. However, that risk was almost entirely borne by bond investors, and the formal banking system was not even scratched. The assumption was that this performance would be repeated gong forward. Of course, extreme risk taking behaviour overwhelmed those institutions, but one may note that the problems were started outside the formal banking system in 2007-2008.
Interest Rate RiskThis is a related issue to concentration risk. Quaint old textbooks describe how banks allegedly operate as maturity-transformation vehicles: the borrow at a floating rate, and lend at a fixed rate.
Historically, this was the case. The Savings and Loan Crisis was started off by Chairman Volcker of the Federal Reserve scrapping the social contract. By raising the short rate in a deranged fashion, we destroyed the basis of the 3-6-3 model of the savings and loan industry. ("Pay 3% of deposits, lend at 6%, on the golf course by 3.") The disinflation was not just a magical costless shift in time preferences to consumption; Volcker blew up the banking system as a means to drive economic activity to a halt. Bank balance sheets were not ready to absorb the higher interest rates.
Advances in digital computing since the 1990s has allowed both banks and regulators to easily measure interest rate risk. In fact, interest rate risk is the most easily hedged risk on the planet. And regulators and banks keep bank interest rate risk on a very tight leash. Any fixed lending is matched by fixed borrowing, with swaps filling in the gaps. (Bank treasury departments take proprietary interest rate risk, but those risk limits are small versus the banks' balance sheets.)
Unfortunately, economic textbooks as well as economist commentators have not caught up to structural changes that happened after the 1994 bond bear market, and it is easy to find assertions that banks engage in maturity transformation. (They do engage in liquidity transformation, but without taking interest rate risk. That is, they convert illiquid loans into liquid deposits. This is why central banks need to backstop the system as a lender-of-last-resort.)
In practice, corporations want to fund large capital expenditures at a fixed rate. The only way that the bank can hedge such a risk in practice is to issue long-dated bonds (or find very large swap counter-parties). The problem is that the bank will incur long-duration debts matched against a concentrated pool of large borrowers. If any of those borrowers fail, their interest rate exposure is suddenly unbalanced (on top of the credit losses). As a result, interest rate risk exposure compounds the problems created by concentration risk, and so banks are not natural funders of long-dated fixed investment.
Since the deposit is a liability of the bank, the bank has two options to keep its balance sheet in balance: either "sell" an asset (or transfer an asset in the settlement system) or issue another short-term liability to replace the deposit. (The proceeds of the liability issuance is used to fund the transfer.) Such short-term liability issuance could be borrowing in inter-bank market, using the repo market to borrow against existing security holdings, or issuing commercial paper.
Although the usual reaction is to focus on the transfer and assume that the bank will shrink its balance sheet, a more likely reaction is to find another funding source. Banks normally do not want to keep running down their liquidity buffer. This explains why the interbank lending market is an extremely important part of the financial system. (UPDATE: I would like to thank Sean Geary for pointing out the initial draft of the above explanation was confusing.)
I covered this in the chapter (number 13) "No, Banks Do Not Lend Reserves" in Abolish Money (From Economics)!
This factor is only a short-term issue. Banks will issue bonds or long-term deposits (e.g., Certificates of Deposit in the United States, Guaranteed Investment Certificates in Canada) to rebuild that buffer if needed. In other words, the banking system could not grow by 50% overnight, but is certainly capable of growing its balance sheet in nominal terms quite rapidly if needed. (For example, bank balance sheets as a percentage of GDP do not go to zero in eras of high nominal GDP growth rates.)
CapitalOn top of liquidity requirements, banks need to keep a certain amount of capital on their balance sheet. However, this capital is not just common equity, it can also include preferred shares, and some types of subordinated debt. (The exact allowed make-up is set in the bank regulations; the reader is free to find the appropriate regulations for their jurisdiction.)
Profits that are reinvested in the bank is one way to grow capital; this is obviously what the regulators love to see. However, common equity can also be increased by the issuance of new equity or rights issues. (Rights issues are essentially a public auction of call options on new equity; these are more popular in Europe than in North America.) However, capital can also be increased by issuing other subordinated debt instruments (and preferred shares); this is what equity holders tend to like.
Once again, capital constraints are only a short-term constraint. If the bank is growing profitably, it will be very easy to issue new capital instruments at acceptable prices.
SecuritisationSecuritisation is another way for a bank to deal with the above issues. It originates the loan, and then sells it off its balance sheet in a securitisation. The sale of the security raises cash (raising the liquidity buffer), and by reducing the risk assets on the balance sheet, reduces the capital required.
(Getting credit guarantees is another way to alleviate capital needs. For example, Canadian "high risk" mortgages require mortgage insurance, most of which is issued by the CMHC -- a full faith and credit subsidiary of the Federal Government of Canada. This means that those mortgages are effectively claims on the Federal Government, and thus have a low risk weight. This means that less capital is required.)
PricingThe above points showed that issuance of new instruments is how a bank can grow its loan book, even if loses deposits. However, it has to pay a market rate. If it is growing too fast, sensible credit analysts should be unhappy with lending more money to that bank in the bond or money markets. The bank will need to issue debt with higher spreads, which eliminates its ability to offer loans at a competitive rate.
The above argument applies to a single bank, that is growing faster than the banking system in aggregate. However, the constraint shows up on an aggregated basis. Bond investors do not like allocating too much risk to one sector, and so bank debt spreads on average could become wider versus industrial corporate bonds. This makes the banking system in aggregate less competitive versus other sources of funding, reducing the growth of bank loans. That said, they would probably just dump the securitisations on the bond market, which is a different risk bucket for bond investors.
Concluding RemarksThere are a number of factors that prevent the banking system extending very large loans in a short period of time. This means that unrealistic thought experiments about bank lending are technically correct: yes, banks cannot extend arbitrarily large loans right now. However, this tells us absolutely nothing about how banking works in practice: banks have the capacity to extend loans to match plausible nominal growth rates. (Hyperinflation, as always, might require a separate analysis.) It is very easy for banks to alleviate the short-term constraints on lending growth by issuance of securities, particularly securitisations.
The ability of the formal banking system to grow (and lend) is partly determined by the relative size of the non-bank financial sector (also known as "shadow banks"). If everyone is allocating their cash to money market assets, it means that bank deposits are a correspondingly diminished source of financing. It would then be unsurprising that the formal banking system will shrink relative to non-bank financial entities. That said, the use of securitisations allows banks to move the financing role to the shadow banking system, and once again, there are few practical limits on plausible nominal growth rates for lending.
(c) Brian Romanchuk 2018