Recent Posts

Friday, April 29, 2022

Banking Debate Woes: Part III

This article finishes off a sequence of articles on an ancient debate in economics: what limits bank lending? (Links to Part I, Part II.) My argument is that the usual ways of looking at this question are highly misleading. The mainstream Economics 101 description of banking is hopeless — and then heterodox critics got bogged down in attacking a hopeless description of banking. The key to understanding banking is that they are a financial firm that borrows to invest in assets (a “levered financial entity”) — and they have some similarities to other financial firms, albeit that they have some special privileges.

What makes banks different is that they are in the liquidity management business: they are expected to be able to make large transfers of short-term financial assets reacting to transactions of their clients. Their structure and privileges are built around this need to provide liquidity. Furthermore, the non-bank financial system is built around liquidity provision by banks, such as the backing of commercial paper facilities with credit lines.

In the previous article, I argued that from a bank’s perspective, it faces the “constraint” of keeping its financials within accepted norms for banks, which then allows it to expand its balance sheet by extending loans and credit lines. Unlike Economics 101 descriptions of banks, it is not concerned about the liquidity drains associated with the proceeds of a loan being transferred to another bank: so long as it retains access to markets, it expects to be able to navigate customer transactions. The question is: why should a bank expect to be able to do this? What about central bank policy (or whatever)?

Funding Flows Circular

We can decompose a credit instrument into two parts: funding, and credit. The funding component is the face value of the transfer, and the credit component is the associated credit and liquidity risk. If we could nullify the credit component, we end up with the equivalent of the transfer of a risk-free instrument (like a central government bond).

If we just look at funding, we see that the flows associated with lending are circular. The size of funding transactions has limits — nobody has the capacity to extend a loan for seven quadrillion dollars. However, if we accept that funding transactions occur in a somewhat incremental fashion, the system can theoretically support a very high rate of lending growth — we cannot point to any level of debt that cannot be readily surpassed. (I justify this assertion in sections below.) Nevertheless, when we look at the real world, actual credit growth is below this theoretical maximum.

What throttles credit growth below its maximum is the need to accommodate credit risk. How much credit risk a lending instrument represents depends upon the current beliefs of market participants. The key property of credit risk is that it is not uniform, and thus dollar amounts do not tell us about the associated credit risk. It is often easier for a large investor to extend a billion dollars in a funding transaction than to buy $100 million in distressed debt (where “ease” is defined in terms of risk budget and analyst work load).

From the perspective of banks, this explains why they worry about their credit standing and not necessarily where funding flows come from: as key liquidity providers, so long as they are trusted, they will be able to tap into the funding flows within the system. For example, take the incredible growth of Canadian mortgage debt since 1999 — when the Canada Mortgage and Housing Corporation (CMHC) relaxed its credit standards on mortgage insurance. The Canadian regulatory system forces all mortgages with a loan-to-value ratio over 80% to get mortgage insurance — which means that the Canadian Federal Government backstops the credit risk on mortgages (the CMHC is a Crown Corporation that is a full faith and credit issuer under the Federal Government’s umbrella). Canadian banks have zero problem funding the mortgages — they just securitise them as needed. The effective constraint on mortgage growth is the ability and willingness of Canadian households to take on insane levels of debt. One could try to argue that the Bank of Canada can influence the level of mortgage borrowing via interest rate policy, but it is hard to see how the level might be directly “controlled.”

Analogy to Functional Finance?

I think there is an analogy between what I allege about credit growth and one non-standard way of describing Functional Finance (related to Modern Monetary Theory’s discussion of governmental finance). There are institutional norms that act as a limit on how much debt the central government can issue in a single day. (These norms vary from jurisdiction to jurisdiction.) Although these limits block though experiments of spending quadrillions of dollars on a single project, they still allow extremely high growth rates of government debt.

In practice, we do not see usually see actual government debt growth near those theoretical limits — since open-ended spending of that magnitude would likely create inflation that is politically unacceptable. (Admittedly, hyperinflations do happen, but they are more complex than financial market folklore suggests.) The funding flows for governments are effectively unbounded, inflation acts as the constraint on debt growth. For private debt, credit risk (and consolidated governmental macroeconomic intervention) limit growth.

Out of Thin Air!

One of the facts about bank lending that causes widespread distress is the fact that bank deposits can “appear out of thin air.” That is, when a bank loan is extended (or a credit line drawn), there is an instantaneous balance sheet growth: the bank simultaneously increased its deposits (a liability) and its loan book (an asset). This apparently is not how things are supposed to work.

My guess is that this based on an analogy to either gold coins or banknotes: they are normally transferred among people and firms, but the stock is supposed to increase only because of coinage/money printing. Since bank deposits have been declared to be “money,” they apparently are supposed to follow this “rule.” (To what extent my diagnosis of the logic is true, this is an example of how mysticism about “money” leads to bad economic theory.)

The issue with such logic is that every lending transaction creates a new financial asset (the loan), which means that financial assets tend to grow in a growing economy. (Paying down loans reverses the transaction, causing balance sheet shrinkage, but the usual tendency is for expansion.) Examples include the following.

  1. If someone hands another person an “I owe you” of $20 in exchange for a $20 bill, the person who issued the IOU grows their balance sheet by $20 (got a $20 bill asset, $20 liability), while the counterparty swaps an existing asset for another (same size balance sheet).

  2. When a firm purchases a good using an accounts payable, the buyer grows assets and liabilities (the good is a new asset, the account payable is a new financial liability), and the seller swaps a non-financial asset for a new financial asset (the account receivable). Accounts receivable are key financial asset — they can be refactored, or financed using trade finance (the origin of “merchant banking”). The terms by which credit is extended to customers is an important part of sales and business strategy. (Note that if a service is financed using an account payable, there are income effects of buying/selling the service, so equity would be adjusted.)

  3. When a bond is issued, the effect is similar to the IOU above: the bond issuer will get a cash inflow to grow its balance sheet, while the bond buyers exchange assets (cash exchanged for the new bond).

  4. A transfer from a bank account to money market fund is similar: the customer swaps financial assets, while the money market fund gets a cash inflow and an increased unit holder liability/equity stake.

  5. Retail customers buy goods using credit cards, creating intermediated accounts payable/receivable (the credit card company manages collection and payment), which grows the household and credit card balance sheets, while the selling firm exchanges a real asset for the financial claim on the credit card company.

In the transactions above, we can see that the transactions are in a sense “self-financing”: aggregate balance sheets grow without an influx of funding from elsewhere. (In the cases where an existing financial asset changes hands — #1, #3, #4 — the financial asset had to be on the balance sheet of one party before the transaction, which might have required earlier financing.) As such, the fact that bank loans are “self financing” is not something worth spending too much time worrying about. They are just one financial asset among many within a modern industrial economy.

Banks and the Payments System

Let us now turn to one of the main peculiarities of the banking system — the payments system. The payments system accommodates the high volume of transactions between private banks and the central bank. So long as nothing really bad happens, nobody defaults and we do not need to worry about the exact plumbing of payments (which varies from country to country). What we are normally interested in is how (central) bank balances are affected by the payment flows.

My view is that the default descriptions of payment systems and banks is distorted by economist beliefs about money (surprise, surprise). In my book Understanding Government Finance, I provide an alternative “model” for banking that is much cleaner — and eliminates what I see as misconceptions about “money.” The model is an idealised version of actual Canadian banking system practice after the abolition of bank reserves and before the 2020 imposition of “Quantitative Easing.” In this system, private banks always drive their balance with the payments system at $0 at the end of the business day (which implies that that central bank is also forced to keep a $0 balance, since balance sheets need to balance).

In the book, I describe how the “Simplified Framework for Government Finance” works from the government’s perspective (since that is what the book was about). But we can then ask: what about a bank’s finances?

I will turn to bank finances after first making two parenthetical observations.

  1. These payment settlement balances are effectively equivalent to “bank reserves” (excess balances for private banks imply balances held at the central bank). Under this system, those are zero. This means that the “monetary base” consists solely of banknotes and coins. The economic importance of banknotes and coins is greatly overstated in market folklore (“We’re going to withdraw money from banks to create a banking run!” is one of the funniest crypto bro memes floating around.) Banknotes are just used to grease the wheels of everyday retail and the underground economy, and are a separate sub-system from the wholesale payments system. As such, they have no real impact on the bank funding discussion that I am discussing herein.

  2. Understanding Government Finance explains how we can add non-zero bank reserves to the story. I do not have space here to discuss this; I will let the reader read the book to pursue this topic (the book is a great value, I may add).

Circular Funding Flows

Once we look at the simplified framework, we can easily see what I mean about “circular flows.” The book described the circular flows from the central government’s point of view, but we can just as easily look at any particular bank.

Once again, the rule of the framework is that every bank (including the central bank) starts and ends the day with a $0 balance with the payments system. The only way this rule is violated is that somebody defaults. However, if we look at a country with a sensible banking system like Canada, defaults do not happen very often. (I need to verify this, but I believe that no Schedule I bank (large domestic bank) has failed since Confederation in 1867 — foreign banks and some community banks have gone bust.) This is not an accident: the central bank expects banks to hit the $0 target, and will penalise deviations. (During the Financial Crisis, banks left positive balances in the payments system anyway, and now with QE, balances are positive.)

Which means that if a bank avoids defaulting, if it does a funding transaction, it must always end up with that flow being balanced out by the end of the day — somehow.

For example, imagine that a bank issues a bond to shore up its liquidity position. The buyers of the bond (that do not bank with the issuer) will wire funds to the bank, generating a net inflow in the payments system. Since the balance is expected to be zero by the end of the day, if no other major net outflows happen, the bank needs to “get rid of” the excess balance — typically by buying short-term debt issues, or other short-term lending. (In fact, it could end up sending financing to the bond buyers via the repo market.) Since we assume that the bond is being used to purchase short-term financial assets, the bond is injecting funding into the system equal to the amount of the bond issuance proceeds (not counting flows that end up being purely internal to the bank).

That is, the system did not require “excess funding” in order for the deal to get done; it just needed to be sized in such a fashion so that it fit the funding constraints of buyers. These constraints are not infinite, but are still large enough to allow quite high theoretical debt growth.

From the bank’s perspective, the concern is not so much the availability of funding (as I noted earlier, otherwise the bank is doomed), rather the price of funding: what is the spread on the bank’s funding?

The example given seems unusual: is not the worry of a bank that it is draining liquidity? As explained in Part II, a bank will draw down its liquid asset buffer to meet outflows. It will periodically need to top up its liquidity buffer. It does this by issuing long-term securities (bonds, securitisations) in the market. The question was: why does the bank assume that the funding exists to issue the long-term paper? The reason is that the circular flow implies that the payments system inflow from the bond issuance is matched by the bank entering into short-term funding instruments — implying that the bank supplies the financial system with an amount of funding that matches the funding it draws.

Credit Constraint, Not A Funding Constraint!

What limits a bank’s ability to expand its balance sheet is the need to meet expected credit norms. Too risky a loan book raises the potential for credit losses, and too much wholesale short-term borrowing poses liquidity risks. Otherwise, they should be able to find the funding — they do not need to wait for a depositor to walk in the door to extend a new loan.

Concluding Remarks

All of the remarks I have made in these articles are straightforward observations of fact about the operation of levered financial institutions within current institutional norms — there never should have been any controversy about this subject in the first place. What seems to have happened that economists started in the wrong place — “money” is a “fixed endowment” that is “traded” — and never managed to get back to the right place.

Email subscription: Go to 

(c) Brian Romanchuk 2022


  1. "they do need to wait for a depositor"

    They don't?

  2. This comment has been removed by a blog administrator.

  3. This comment has been removed by a blog administrator.

  4. This comment has been removed by a blog administrator.

  5. This comment has been removed by a blog administrator.

  6. This comment has been removed by a blog administrator.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.