Primer: Funding Versus Credit Risk

The act of lending involves two fundamental operations: extending financing (funding) ans taking credit risk. These two roles are typically thought of together, which obscures what is happening. By decoupling these concepts, we can better understand the effects of debt issuance. If we can eliminate credit risk, the circular nature of financial flows means that the only limitations on debt issuance are real resource constraints. This understanding helps us better understand the behaviour of economic models (why they depart from real-world behaviour), governmental finance, and banking.

Financial Engineering

Financial engineering has a deservedly bad reputation as a result of sell-side financial engineers devising toxic securities to dump on their clients. However, on the buy side, a lot of "financial engineering" consists of breaking apart traded instruments into their fundamental risk factors. Knowing what risks you are running is the key to good portfolio management.

In this article, I discuss credit financial instruments of all types, including some that might not be thought of as "financial," such as accounts receivable. However, the remarks are not really applicable to equities.

The following factors are the main ones that determine the "fundamental" value of a credit instrument  (although we may need to add some to get an expected market price). I am listing them all for completeness; only the first two are discussed in the rest of the article.

• Funding (or financing). The need to transfer "money" (or assets of some sort) to the borrower.
• Credit risk. Will the borrower pay back the borrowing?
• Term (maturity). 
• Interest rate risk. This can be tied to the maturity (fixed coupon), but not always (floating rate notes).
• Optionality. Can either party have the right modify the structure of payments?

Once again, the latter three factors are critical for pricing instruments, but they can be viewed as secondary to the act of lending.

In order to do such a decomposition, we need instruments to be able to isolate these risks. We can separate the funding/credit risk components by using a credit guarantee or credit default swap (CDS). A bond owner can hedge out the credit risk of a bond by "buying protection" in the CDS market (albeit by gaining a credit risk exposure to the swap counterparty). Meanwhile, seller of CDS protection ended up with the credit risk associated with the loan, but without extending any financing,

My argument is that the first two terms are quite different, but common usage of "financing" causes people to blur the two together. One method to deal with this would be to invent a new term for the act of extending money without worrying about credit risk, but I will stick with "financing/funding" as that is already the phrasing used by many (most?) authors.

Modelling Perspective

Embedding credit analysis within a mathematical economic model is difficult, but funding poses no problems. It is therefore easy to set up models with behaviour that is hard to interpret if we lump together funding and credit.

We can use the Python sfc_models stock-flow consistent (SFC) modelling framework to create an example. It currently is only emulating models from the Godley and Lavoie's Monetary Economics, but the equation structure is slightly different, and it is easy to illustrate the limited constraints on funding if there is no credit risk.

Within sfc_models, each sector has a net financial holding variable, denoted F. Imagine that a non-government entity issues debt, but otherwise undertakes no transactions in an accounting period, the variable F associated with that sector would not change. Instead, it will grow its balance sheet: it would increase the stock of financial liabilities (it increased its debt), and it will invest the proceeds into other financial assets. For the sake of simplicity, label this other financial asset "short-term paper," which is issued by a number of issuers (think of a money market mutual fund).

Where does the funding come from? If all entities in the model issue liabilities that do not have credit risk, there is nothing to distinguish them within portfolios. If all liabilities are short-term instruments, they are all indistinguishable from "money" or "treasury bills." As a result, other entity would increase their holdings of the issuing entity's new debt, and there would have been a reshuffling of other entities' portfolios that would allow the issuer to increase its holdings of "short-term paper." (The simplest case is one other entity reallocating from "short-term paper" into the newly issued debt.)

In summary:

• The target entity issues $X of new debt; it ends up $X in new financial assets ("short-term paper"). Balance sheet $X larger.
• Other entities exchange $X of "short-term paper" for $X of new debt. Balance sheets have same size, just with an allocation change.

Since there are no worries about credit risk, there is no financial reason for there to be any limit to this process. Entities could issue arbitrarily large amounts of debt, and the proceeds invested in the arbitrarily large issuance of liabilities issued by other entities. Since such a result appears nonsensical, the modeller has to impose behavioural rules that limit debt issuance. For example, debt issuance is typically tied to consumption or investing (in real assets). Production capacity constraints exist (although not in the simpler SFC models!), and so arbitrarily large nominal debt flows cannot be matched by arbitrarily large flows of real goods and services. In other words, the increased nominal flows would likely just represent inflation.

The exact mechanism of the transfer is deliberately shrouded in mystery. The sfc_models framework follows the aggregation methodology of standard stock-flow consistent models, and all transactions are assumed to clear simultaneously at the end of the accounting period. We cannot trace the individual transactions, rather all we see is the new portfolio configurations (and flows) at the end of the period. This is somewhat akin to the notion of equilibrium in mainstream economics, which is briefly touched upon this survey paper by Michalis Nikiforos and Gennaro Zezza. Ramanan has a longer discussion of how equilibrium fits in within the post-Keynesian approach. I would note that the notion of (short-run) equilibrium as used here is a well-defined mathematical concept, unlike the ambiguous ideas used in mainstream mathematics.*

This situation is distinct from what might happen in an agent-based model, which theoretically might track all transactions. In such a case, we would explicitly model how debt is funded, and it may be that the developer of the model will put into place behavioural limits on how much debt can be funded.

My argument is that adding such arbitrary limits on debt issuance is perhaps misleading; the private sector will arrange its affairs in such a way so as to eliminate funding constraints. Instead, the real limit is on the willingness to take credit risk. A business cycle is largely defined by the increased willingness of the private sector to take credit risk -- allowing debt issuance to expand -- until that willingness disappears (the financial crisis that has terminated every expansion in recent decades). This should sound roughly familiar; it is just a restatement of Minsky's Financial Instability Hypothesis.

The rest of this article discusses how to translate these observations on mathematical models toward the real world. In my view, the apparent ability of mathematical models to allow unlimited debt expansion is not a deficiency of the model (as a result of the violation of the common sense idea of "loanable funds" limiting debt issuance). Instead, it is the difficulty of modelling credit risk, which is the true brake on debt growth.

Why is Credit Risk Hard to Model?

It is perhaps not obvious why credit risk is hard to model within a macroeconomic model. The reason is that mathematical models follow rules, whereas credit risk often revolves around people and firms not following rules.

We have to remember that a mathematical model is just a statement about sets, and elements of sets. There are no mathematical elves that live inside the model, running production functions, trading, and lending to each other.

Lending decisions are often based on instinct, but they generally credit ratio guidelines, which could be modeled. In the real world, both sides of the transaction will scour business plans and accounting data to determine whether such guidelines are met. Lenders have to factor in the possibility that borrowers are delusional, or lying through their teeth. As a result, the financial sector has an active role in the decision-making process.

In a mathematical model, such negotiation is not going to be easy to represent (outside of an agent-based model, or game-theoretic models that abstract away from the rest of of the economy). The aggregate position of the borrowing sector is presumably known, as are the lending standards. Therefore, the natural set up is to set up a borrowing capacity function, and a willingness to borrow to function. The "decision" within the model is based on those functions, and thus it appears that only the borrowing sector is behind the decision, and the financial sector is entirely passive. (Since this is the natural implementation, I am not a fan of the heterodox complaint about the absence of the financial sectors in models; its role can be implicitly buried in the borrowing behaviour od the real economy sectors.)

In summary, the credit risk component of a macro model is most likely going to look unrealistic.

Not Just Banks

The ability of banks to expand their balance sheets is well known; loans create deposits. There is considerable mysticism about this ability, and the belief that makes them special. (I discuss this topic in "Are Banks Special: Yes and No." As the title suggests, banks do have some privileges, but this can be overstated.)

Non-bank finance also allows the private sector to expand their balance sheets. Firms can issue commercial paper, and invest the proceeds in the commercial paper market. So long as the issuers are willing to hold eachother's commercial paper, this process can continue indefinitely. Certainly, banks are involved in the payments mechanism, but there is no need for any borrowing from banks that is observed at the close of the day. If banks somehow became a bottleneck for the system, arrangements would be made to bypass those constraints.

We are not even confined to the financial sector. Firms routinely sell goods and services on a credit basis: accounts receivable and payable. The seller effectively extends funding in the form of goods or services provided to the buyer; since the seller had to previously finance the inventory that was sold.

Trade finance is an important component of capitalism, even if it gets less attention than banking in popular analysis. Vendor financing was a critical component of the technology bubble, and is often sneered at. However, rediscounting receivables was one of the drivers of the development of the money markets. Merchant banks rediscounted receivables, and this gave the instruments enough credibility to be traded in the money markets.

Furthermore, there is an intense economic force behind vendor financing. By extending financing, you make sales, and thus profit. If regulation clamped down on financing -- such as is suggested by full reserve banking -- commerce would grind to a halt. With apologies to Frank Herbert -- the credit must flow. As a result, attempts to protect the sanctity of "money" -- such as gold backing, or full reserves -- will be bypassed to grease the wheels of commerce. The drive for profit and incomes creates a political coalition that dooms attempts to restrict private credit creation too tightly.

The problem with vendor financing is that the seller needs to correctly judge the creditworthiness of buyer. (This was a key problem in the tech bubble.) There is an incentive for businesses to find a way of outsourcing that credit analysis. Historically, general stores had to keep track of the tabs that they extended to their regular customers, and the determination of the size of the tab was a role that only trusted persons could make. (The fact that publicans acted as financial intermediaries when the Irish banking system was shuttered by a labour dispute in the 1970s is an example of the credit expertise in the real economy -- discussed in this book review.) Credit card companies have taken over the role of credit assessment for retailers, and so it is possible to have junior employees acting as cashiers. In other segments of the economy, intermediaries spring up to fulfil that role (such as receivables factoring, or trade finance companies). For large fixed investments, the bond market exists to absorb the credit risk that is too concentrated for even the banking system.

The fact that financing forms will adjust to allow credit to be advanced explains why I believe that the "equilibrium" description of financing in mathematical models are relatively close to the truth. The system will evolve so that if there is some entity that is willing to take on the credit risk of the borrower, so there will be a way of matching the supply and demand.

The willingness of someone to underwrite the credit risk of a borrower is thus one constraint on borrowing. The other constraint is real: trade finance can only be advanced against existing goods and services. If we attempt to advance more paper claims against resources than there are resources available, inflation would result.

This analysis implies that arguments suggesting various policies are needed to raise savings rates to fund investment are bunkum. There is no need to raise the volume of savings; borrowing is inherently self-financing. The role of the financial sector -- if it is functioning properly -- is to assess credit risks. The only advantage of new financing forms is that they may be able to undertake credit risk that banks naturally shun. Stories about entrepreneurs needing to save to build their business (or people hoarding coconuts on desert islands to "invest") miss the point: they needed to save to invest since no one was willing to lend them money. Since most new businesses fail, there is not a huge lineup of firms to extend such financing. However, existing firms have a track record, and financing is normally available for viable ones.

The Role of Banks

The importance of money creation by banks is overstated. By itself, money creation is a financing operation, and financing itself is not an issue. Instead, the role of banks in credit analysis is what matters. Banks have an inside view of depostors' liquidity position, which is of great value in credit analysis. They also are diversified lenders, which naturally reduces their risk. By contrast, the fate of specialist lenders depends upon a narrow group of clients, who are therefore exposed to sectoral downturns.

The evolution of banks towards the originate-to-distribute model is an abdication of the fundamental responsibilities of banks. As a result, they have lost the technical competence to execute their core business functions properly, which sadly is a feature of many large corporations.

What Happened in Canada

The explosion of Canadian household debt after the disastrous decision by the CMHC to underwrite the credit risk of risk borrowers is very easy to understand from this point of view. They took away the credit risk eliminating the true constraint on borrowing. As one might expect, borrowing accelerated. The only limit on borrowing was finding new buyers that could pass the exceedingly easy barriers against borrowing too much money, not really financing. (Arguably, the Canada Mortgage Bonds were created to allow the necessary intermediation, which would have been difficult if the mortgages remained on the bank balance sheets. The institutional structure of Canada ensures that wealth leaks into non-bank finance -- for example, pension funds -- and it would have been necessary for the banks to increase their already large bond borrowing programmes to recirculate the funding from the shadow banks to the mortgage positions.)

Offering credit guarantees looks like a way to implement government policy "for free," which is attractive in an environment where politicians have a primitive ideological fear of government debt. However, such guarantees eliminate the constraint on credit growth for that class of borrower, and so it is necessary to think whether unchecked growth of a certain type of debt will be a good idea.

Central Governments

A central government (that controls its central bank, and borrows in its own free-floating currency) is essentially free of the risk of involuntary default. (To what extent that claim is controversial, I discuss it at greater length in Understanding Government Finance.)

Therefore, such governments are free of the credit risk aspect of borrowing, and they are thus only engaged in a financing operation. Meanwhile, there are no financial constraints on such an operation, as I discussed earlier. Since government money always trades at par, liability issuance by the government is matched by rising non-government financial asset holdings (often called a circular flow). The only constraint on government borrowing is the real constraint: there has to be someone willing to sell real goods or services (or work) for the nominal dollar amount offered by the government.

As a result, economic models that allow for arbitrarily large amounts of government borrowing appear to be closer to real world behaviour.

Keen observers of Modern Monetary Theory (MMT) would note that I used the dreaded word "financing" with respect to government spending. They do not like to use that word with respect to government debt issuance. They have realistic political concerns about the misuse of words to frame debates. For example, Bill Mitchell discusses the role of language in a recent article on the language used in World Bank publications:

The shift from systemic failure to individual choice – from a lack of jobs to transactional choice – from responsibility to aid all citizens to governance and oversight of ‘taxpayer funds’ – from the responsibility of government under international treaties to provide enough jobs to contract brokering and surveillance – from full employment to full employability – reflected the neo-liberal dominance of public policy.
And the shift in language was an intrinsic part of this policy shift. Of this abandonment of full employment. Of this jettisoning of responsibility.
That is when I became interested in the way language and framing works.

With respect to government borrowing, MMT authors are very keen to differentiate the central government from other borrowers. This is in opposition to hypocritical rambling about "magical money trees" (as discussed by Neil Wilson here). As a result, I have some sympathy for this view. However, they are effectively lumping the act of financing with the act of credit transfer (like everyone else), which I view as awkward. "Financing" or "funding" appear to be the best technical terms for the act of transferring money, without worrying about the transfer of credit risk. Until such a term is established, a discussion similar to the one here is going to be awkward.

Concluding Remarks

We need to decouple the act of funding from the act of taking credit risk. Credit risk and real resource constraints are the true limits on borrowing; trying to find magic debt-to-GDP ratios is a doomed quest.

Footnote:

* The notion of "long-run equilibrium": that there are steady-state ratios towards which variables converge is somewhat harder to define without adding classifications of variables within models if the economy is growing. Different classes of variables will exhibit different growth rates (for example, real growth rates versus nominal), and so badly selected ratios will not converge towards any value, even in a "steady state." These classifications are not in place in the sfc_models framework, and so a formal definition is hard to implement.

(c) Brian Romanchuk 2017