Large bank corporations now tend to have both traditional lending divisions as well as securities market divisions. This was not always the case; regulators used to keep financial firms locked to specialisations — this was referred to as “the pillar system” in Canada. However, ongoing deregulation eroded the pillars — I discuss part of the economic logic below. It is possible to find banks that stick to a traditional loan/deposit structure (particularly in the United States, with a highly fragmented banking system), but those banks tend to be smaller.Editorial note: This is possibly the last of my series of articles on the theme “how do banks work?” These articles would end up as a chapter in my banking primer. The “elevator pitch” for my book is that I want to pour cold water on loopy fractional reserve banking stories — which requires some discussion of how banks operate in the real world. However, I am certainly not aiming to write a manual on how to operate a bank, or even how to be a bank analyst. Instead, the focus is how banks fit into the macroeconomy, which requires some knowledge of the basic strategies banks follow.
Bank Interactions With Securities Markets
In this text, I will look at three broad ways banks interact with the securities and wholesale funding markets. (I am lumping currency trading with securities trading, which might be a misnomer.) I divide them based on banks’ counterparties point of view, and not necessarily how banks think about their organisational structures.
- Banks need to manage their liquidity position.
- Banks have securities dealing operations. This includes brokerage operations, and things like offering derivative and foreign currency transactions to customers. I would also lump in “financial technology”: new ways for people to gamble (err, invest).
- Investment banking — issuance of securities in the primary markets.
Banks may also own asset managers and insurance divisions. My experience is that those subsidiaries would be regulated on a stand alone basis, and assets are presumably segregated. As such, they are not really the topic of discussion here.
The three categories of securities interactions implies the need for the bank to execute trades in order to operate its business. For example, one day it might need to buy Treasury bills to put a cash inflow to work, then it might sell them the next day. That execution would require having a “desk” of traders to implement the trades.
For many people who wag their fingers at banks’ risk taking, that is all they should be doing. However, if the traders on the desk just execute trades to meet needs of the bank, they are just a cost centre for the bank. Meanwhile, reliable traders tend to command high salaries. This creates the economic incentive for those desks to also trade on their own behalf to generate enough profits to justify their salaries (and pay bonuses) — proprietary trading (“prop trading”).
All of this trading activity is now managed using the same principles as other fixed income portfolio managers. The desk manages one or more “portfolios” which will have a liability structure versus their actual assets. The liabilities might be accounting liabilities with other firms, they might be internal funding liabilities, or a benchmark (which is what “real money” bond investors manage against, seems less likely for a bank). The team manages its assets in a way that is supposed to put limits on the risk of their liabilities increasing in value versus their assets. The easy conceptual way to do this is to imagine holding a portfolio of assets that matches liabilities exactly, then one adds “overlay” trades to that naïve matching portfolio to end up with the actual portfolio held. The risk of the positions versus the liabilities equals the risks of the overlay trades.
This conceptual structure means that fixed income managers can consume the same bond investment strategy advice despite the wide differences in portfolio structures. (Note that some strategies will not be available to some managers — e.g., money market funds cannot trade 30-year bonds.)
1. Liquidity Management
We can imagine a theoretical financial system where all transactions are intermediated by banks. (This is fairly common in economics textbooks.) This intermediation is accomplished by banks transferring funds to one another via transferring deposits on a senior bank — either the government-owned central bank, or a private correspondent bank. In practice, it might be possible for some banks to operate in this way. However, it is difficult for the banking system as a whole to do this.
The reason is that bank customers will allocate their portfolios towards non-bank deposit financial assets — which includes central government bonds and bills. If bank customers in aggregate allocate deposit funds either into or out of those instruments, that implies a net funding flow out of/into the banking system (the flip side of the customer transactions).
Although an academic might be able to come up with a scheme to isolate bank funding from the wholesale funding markets (e.g., bank customers can only interact with bank customers, keeping net flows to the banking system at zero), this isolation would likely cause recurring crises due to circular liquidity flows being broken. As such, at least part of the banking system needs to use the wholesale funding markets for liquidity management.
2. Securities Dealing
Other than proprietary trading, securities dealing for divisions that are not asset managers is going to be a variant of acting as a broker or dealer. They either broker trades between their clients and clients of other brokers (e.g., retail stock trading), or they trade directly with their clients (dealer relationship).
As a broker, they are not supposed to be taking risk (unless something goes horribly wrong in the settlement process). Acting as a dealer, they want to find counterparties that will allow them to exit their risk position. The ideal situation is that they find the matching counterparty immediately, so that they book an immediate profit that is the difference between bid-/offer-spreads. Realistically, they will hold the position for a period of time — “warehousing the risk” — and trade out of it.
Aside: Dealer Balance Sheet Growth
I left finance in 2013. At the time, there was an increase in trading fixed income products on screens, but in Canada at least, large-scale trading was done on a dealer basis over the phone. In a dealer market, large clients create a need for large dealers.
As a client that needs to make a large trade, you have two broad options. (You change the option for different large trades to keep everyone else on their toes.)
- Break the trade into small chunks, and try to execute them over a period of time.
- Sound out dealers, and find one that can absorb the entire trade.
Critics of big dealers would just say “choose option one” and leave it at that. But that is not a good position for clients to be in. If you always choose option #1, as soon as you execute a trade, your counter-parties will assume that there are more chunks to come, and they will move the price ahead of them arriving. (Some people would call this behaviour “front-running,” but that is a misnomer — front-running is an illegal act by a broker defined in laws, but is not directly applicable to dealer transactions.)
Option #2 is obviously trickier to negotiate, but once it is negotiated, both parties have an alignment of interest. The customer needs to unload a large risk position quietly. The dealer then is stuck with the opposite position, that they need to unload. As such, they do not want to signal to anyone else that the transaction happened.
The growth of balance sheets of asset managers — pension funds, insurance companies, etc. — implies a need for similarly large dealers to intermediate their transactions. One might quaintly hope for a return to a community of small dealers like in the 1950s, but they would be winnowed out by economic forces rather rapidly.
3. Investment Banking
Investment bankers structure new instruments for sale. (The sale of a new instrument is called the primary market, trading of existing instruments is called the secondary market.) Investment banking is the one area of fixed income that is qualitatively different from other branches, since the other branches are in the secondary market.
In an ideal world, the investment bankers structure the deal, pocket the fees, and their sales team dumps the new bonds on clients. This ends the bank’s risks, allowing them to go on to the next deal. In practice, the main risk is funding the assets when on the bank’s balance sheet.
Investment banking used to be a separate pillar of the financial system, but the Financial Crisis has meant that many of those firms turned into banks.
As a final aside, common usage in the United Kingdom seems to refer to investment bankers as “bankers,” whereas in North America, “investment bankers” were viewed as distinct from employees of traditional banks.
From a macro perspective, securities market risk events show up in three categories.
- Idiosyncratic risk: trading teams lose money due to typical market movements, mistaken execution (“fat finger error”).
- Rogue trader risk.
- Systemic meltdown!
My split here is not completely arbitrary, they are related to two underlying issues.
The difference between the first and second category (idiosyncratic losses versus rogue trading) is that quantitative risk management systems generally do a job of identifying the risk characteristics of the first. Rogue trading — by definition — means that a trader (allegedly single-handedly) bypasses the risk control systems.
For an asset manager, being incompetent is not a competitive advantage, and flushing investor capital down the toilet can easily be a firm-ending risk. For a somewhat traditional bank, trading activity is superimposed upon their core businesses, and so risk limits are supposed to be set so that typical trading losses are not significant to the “normal” profits generated by other business lines. The numbers might sound scary to an individual, some people might lose their jobs — including entire teams — but the bank just takes a hit to earnings and stumbles along. (An investment bank presumably needs to look at least somewhat competent to outsiders.)
Rogue trading situations are unfortunately common, with most cases being the result of firms whose core competency is elsewhere ends up with a “star trader” who eventually subverts whatever risk limits exist. This can happen to industrial firms with interest rate/forex hedging operations, or commodity producers with commodity trading. Barings is the classic example of a bank getting dragged down by a poorly-supervised employee in what was seen as a peripheral division (until too late).
One might point to rogue trading “incidents” at large banks that one might expect to have strong trading controls. The question in those cases is to what extent the trader was a lone wolf. Nevertheless, rogue trade risk is on the radar screen of regulators and risk management.
A sound banker, alas, is not one who foresees danger and avoids it, but one who, when he is ruined, is ruined in a conventional way along with his fellows, so that no one can really blame him. – John Maynard Keynes.
From the perspective of discussion of banks, the main risk of interest is systemic risk. As the Keynes quote above points out, banks are notorious for all going off the cliff at the same time. And if we look at fixed income markets, banks are hardly alone in this behaviour — pretty much the entire fixed income complex somehow ends up on the wrong side of the trade. Although that sounds theoretically impossible, credit losses can turn fixed income into a negative sum game. Given the disastrous effects on the real economy, authorities have no choice but to intervene in the case of systemic risk. Saying that “investors ought to take the losses” is a silly stance when we see how rapidly real activity contracted as trade finance dried up in 2008-2009. The collapse of incomes makes previously prudent lending “imprudent,” so activity spirals lower unless checked.
That said, we need to be cautious about the meaning of “systemic risk” and how it relates to bank behaviour. It does not mean that an individual bank can do whatever it wants — an interpretation that an anonymous critic incorrectly accused me of offering. Although there might be smaller countries with exceptions, no shaky bank is “too big too fail” on a stand-alone basis: the rest of the financial system can feast on the remains of a failed bank. (If a bank has a hidden weakness suddenly pop into view — e.g., some core accounting scandal — there would be no way to organise a carve up, so one might expect a punitive short-term intervention.)
This means that banks cannot laugh off doing stupid things: they can only “get away with it” if pretty much everyone else is doing the same thing. In other words, they will shy away from the obvious cliff until they see the other banks also running towards it.
I think it is a safe bet that most investors in bank bonds (or people with deposits beyond insurance limits) are too complacent about bank failure risk on the basis of “too big to fail” arguments. However, the industry is heavily regulated to reflect that situation already.
Can Systemic Risk Events be Prevented?
My belief that periodic systemic risk events are largely inevitable is a view that is designed to annoy progressives, fans of efficient markets, and technocrats. Part of it reflects my curmudgeon attitude, but it is also based on my reading of Hyman Minsky’s work.
Regulators and the banks themselves prefer that core banking be as safe as possible liquidity and credit analysis flow service business. As such, they want to dump as much of the risk into non-bank financial markets — particularly credit and interest rate risks.
One can point to alleged safety of traditional banking (implicitly backed by the central bank’s lender-of-last-resort operations) versus the loony behaviour of securities markets. However, this is exactly what the system was “designed” to do. Attempting to “insulate” banks from the risk markets just means that the risk is stuck on balance sheets. Yes, regulators should keep banks out of the stupid corners of financial markets (that the banks’ traders want to get into), but the big blow ups are invariably in the core funding markets that actually intermediate financial activity (unlike equity derivatives. cryptocurrency and other collectibles, etc.).
Industrial capitalism has the side effect that people use leverage to finance speculative activity — which raises the assets prices of the objects of speculation, validating the investment. This works until it doesn’t. When the process goes into reverse, there will be credit losses that will cause funding markets to enter nervous collapse. The price behaviour has to surpass previous observed results, almost by definition (as otherwise, the speculative bubble would not pop). This guarantees that quantitative risk management will fail.
(Editorial note: I seem to recall writing a very similar passage, but want to repeat it here given its importance. I will need to get my manuscript assembled to weed out the repetition.)
The banking system in aggregate has little choice but to interact with financial markets (although a small individual bank can avoid them). In the short run, that is where most of the risk might come from. Otherwise, traditional banking is a flow activity that generates fees, and so it takes time for a bank’s performance to deteriorate. A recession will cause losses on its loan book, but the loss recognition process also takes time.