Recent Posts

Wednesday, April 17, 2024

Asset Allocation And Banking

Note: This article would hopefully be worked into my banking manuscript. I think it overlaps other article(s), but I wanted to see how this line of argument looks. Needless to say, I have no put the articles into a single document…

One of the difficulties with understanding banking is that one needs to use relatively complex macro models to see how the formal banking system interacts with the non-bank financial system. Analysis based on looking at the motivations of a single bank or based on models where only the formal banking system exists will be misleading. Stock-flow consistent (SFC) models are one of the few attempts at such a modelling framework.

Asset Allocation the Result of Portfolio Choices

In a discrete time SFC model — or a mainstream model — all variables within a time period are determined simultaneously as the result of the model equations. The term “equilibrium” is poorly defined in mainstream macro, but one often sees phrasing that could be interpreted as the determination of the solution is some form of “equilibrium.” Heterodox economists have a bee in their bonnet about the term “equilibrium,” but if we do want to have a mathematical model of the economy, it needs to have a solution in order for it to be of any use.

(One alternative would be to have a model that attempts to model the economy transaction-by-transaction. Although such models might have some theoretical interest, they will have too many free parameters to be useful for understanding the economy.)

Although the equations are solved simultaneously, in practice, we can think of the solution happening in two steps.

  1. We determine all the real economy variables — e.g., aggregate wages, consumption, etc. — along with pricing variables that effect the real economy (such as interest rates in some models).

  2. Given the net cash flows created by the economy and pricing, the private sector allocates its portfolio among the model’s financial assets based upon some portfolio allocation rule.

Although the portfolio allocation may appear to influence the pricing variables, this may not always be the case. For example, the central bank may use a so-called “Taylor Rule” for setting the policy rate, and the allocation between money/bonds will be the one consistent with that rule-determined policy rate.

Simplest Model: Government Liabilities Only

Most standard SFC/mainstream economic models just have government liabilities as the financial assets: “money” and “bonds.” (The bonds are typically 1-period bills). Normally, “money” has a 0% interest rate, which makes them distinguishable from the bills that pay the policy rate. This corresponds to a fiat currency (i.e., currency is not pegged to an external instrument).

Since there are no other financial assets, the government cannot issue liabilities to purchase them. As such, the growth in government liabilities is entirely the result of the fiscal deficit (including interest payments). The private sector allocates between “money” and “bonds” based on a money demand function. (If the objective is to maximise returns, the private sector would just hold bills. Since all transactions occur simultaneously, there is no need for “money” for transactions within the model logic.)

Unless some arbitrary rule is inserted into the model to force a default, the fiscal deficit is self-financing: the liabilities issued as a consequence of the deficit are automatically held by private sector. Since there is no peg, government liabilities cannot be exchanged for anything else. All the private sector can do is swap back and forth between the liabilities issued by the government.

Extension: Formal Banking Only

We can then extend the model to include a formal banking system. There is the government liability allocation, and on top of that, banks issue loans and gather deposits. This model quite often shows up in heterodox discussions of banking, as the balance sheets and transactions are easy to describe.

The flows for government liabilities are not affected by this. We just add banks to the list of actors in the model that hold government liabilities. Those liabilities still only can be exchanged for other government liabilities when dealing with the government sector (although private sector actors can trade government liabilities for private sector financial assets).

Unlike government liabilities, bank deposits can be exchanged for the liability of a non-bank: customers can exchange deposits for government-issued notes and coins. If there was a wave of such redemptions, we get the spectre of bank runs — people lining up to get cash out of the banks. Although this scenario fires the popular imagination, it is not that plausible a scenario for large banks. (If non-banks were allowed to bank at the central bank, we could then get a large scale bank run on private sector banks.)

Deposits are created by bank lending and deposits of governmental liabilities by bank clients. By extension, transactions in the opposite direction (repaying loans, transfers to the government from clients) reduce deposits. Inter-bank transfers between clients will result in a transfer of deposit assets, with the involved banks needing to transfer cash through the payments system.

For simplicity, let us assume that there are no reserve requirements. All banks start and end the business day with a zero balance with the payments system/central bank. This means that all inter-bank transfers have to sum to zero (unless there is a bank default that unbalances the system). The implication is that if a bank experiences net outflows from the action of its customers, it must make offsetting transactions to gain the lost cash flows back. Conveniently, for every dollar a deficit bank loses, there is an offsetting surplus of a dollar somewhere else in the banking system. In the case of this model, the banks would either lend in the interbank lending market or buy/sell government securities to allow the offsetting transactions.

This creates the self-financing property of pure formal banking systems: in the absence of capital constraints, the banking system can deliver whatever nominal loan growth the economy can support. Although this idea raises the hackles of some of the critics of heterodox banking analysis, it is an obviously needed property of the banking system. If nominal constraints on loan growth existed, high inflation would result in bank loans effectively disappearing as a percentage of GDP. And saying that the bank system need deposits to fund the loan growth ignores that the deposits are created by the loan extension.

Adding reserve requirements appears to add a brake on loan growth — but that would require the central bank to be willing to allow the interbank rate spike far above a target rate if there were shortages of reserves. Even during the “Monetarist Experiment” in the late 1970s-early 1980s, central bank behaviour was not that extreme. They instead let interbank rates rise, but in a loosely controlled manner. The hope was that the higher interest rates would slow money growth in the future.

The alleged weakness of the “self-financing bank lending” story is that individual private banks cannot allow its lending grow so fast that it loses all its liquid assets used to cover redemptions. Although that is true, but all this implies that banks cannot grow much faster than the average bank growth rate — faster growing banks will tend to lose cash flows to the slower growing ones. (The recipient of a loan will tend to transfer the proceeds to others, who may bank with other banks. This means that loans are expected generate cash outflows. However, if other banks are growing their loan books at the same time, a bank should expect to have inflows from those other banks’ customers.) However, bankers tend to run in a herd, and so overall system can generate ever-faster loan growth by each bank taking turns increasing their own loan book growth. Banks that refuse to relax lending standards in line with competitors will end up losing market share and thus increasingly economically irrelevant. Although the banking system cannot generate “infinite” growth, it can still keep up with whatever the nominal GDP growth rate is.

The logical problem with this model is that the non-bank asset allocation looks somewhat problematic if all deposits are demand deposits. Why would anyone hold large amounts of bank deposits that pay 0% if there are government bills/bonds that pay a positive interest rate? As such, some of the deposits would have to be interest-bearing in order to generate a sensible asset mix for the household sector and the non-bank business sector.

Add a Non-Bank Financial Sector

The final modelling stage is to add a non-bank financial sector. For our purposes, that just refers to bond and money market trading (and not whatever the current silliness is going on in the financial sector). These non-bank instruments can be viewed as extensions of the banking system (hence “shadow banks”), but their behaviour is different.

What these instruments do is add new instruments for the nonfinancial sector’s asset allocation decision. Unlike the previous cases, they can flee government liabilities as well as bank deposits. This creates cash flows from bank customer actions that need to compensated for.

In some cases, the buyer may pay with a bank deposit, and the seller/issuer may put the proceeds into a bank deposit — which leaves the amount of deposits in aggregate untouched. But this will not always be case. In particular, banks issue bonds and money market instruments. The customer will lose a bank deposit asset and replaces it with a bank’s non-deposit liability.

This explains why banks in the real world have to diversify their funding sources: individual banks are going to be losing deposits to the bond and money markets, and so they need to tap into those markets themselves in order meet those outflows. Bank bonds and money market instruments are more attractive portfolio assets than bank deposits, and so bank liability issuance has to match up with the desired asset structure of bond/money market investors. Although a small bank could finance itself with just equity and (term) deposits, this is not going to be an option for the overall banking system in a country that has developed private bond and money markets of any size.

The structure of bond issuance seems to suggest that they are not self-financing in the same way as bank lending. The issuer expands their balance sheet by issuing a new bond liability, and it gets a corresponding cash inflow. That seems to imply that there had to be pre-existing “money” to pay for the bond. However, things are more complex.

  • The buyers of the bond may be using non-bank credit sources to pay for the bond.

  • The issuer of the bond may immediately reinvest the proceeds of the bond issuance into money market or bond market instruments.

The result is that we could theoretically see bond issuance just growing balance sheets outside the formal banking system, even if payments are routed intraday through the banking system. In turn, this means that the non-bank financial system is also able to grow through nominal size thresholds.

The growth of mortgage-backed securities outstanding across the developed world was the result of these considerations. If mortgages were left on bank balance sheets, the balance sheets would have been increasingly strained, as bond investors — flush with retirement savings — would have balked at increasing their concentration risks in bank bonds. By pushing the mortgages of their balance sheet, bond investors were presented with what are supposed to be safe investments that are shielded from the business risks of banking. Oldsters leaving the housing market pushed the proceeds into fixed income funds that indirectly financed the purchases by youngsters. The banks were just in the middle to service the mortgages and (theoretically) assess credit risks.


The wacky New Keynesian central banker practice of buying large quantities of government bonds — Quantitative Easing (QE) — unbalances private sector asset allocations. To the extent that private banks sell their government bond holdings to the central bank, this is just a swap of one government liability (a bond) for another (deposit at the central bank). However, private bank holdings of government bonds are not large enough to meet the deranged level of purchases seen, so the ultimate sellers would be bond holders who used primary dealers as intermediaries. Those bond holders would end up with bank deposits as an asset — and their bank would have a corresponding deposit at the central bank (“reserves”) to compensate for this. Although most institutional bond holders would do compensating trades so as to not end up with bank deposits, somebody else has to end up holding the deposits (to the extent that they are not replaced by other bank liabilities, or bank loans paid down).

Although bank deposits are great for meeting potential transaction needs, they are a terrible portfolio asset. The yield on them is administered by the bank. And unless the investor sets up deposit relations with multiple banks, they will end up with concentrated credit risk. The situation is slightly better for the banks: they end up with a lot of cheap deposit funding. However, there is no reason to believe that such deposits will be sticky, since they are in excess of what was seen as being needed to cover transaction needs. The bloating of bank balance sheets could stress some regulatory ratios.

It also does little to contain banking system stress. The risk is a bank becomes insolvent — its equity is reduced to zero. Bloating its balance sheet with deposits and reserves does not help this — and in fact increased the subordination of other classes of liabilities that rank behind deposits in a liquidation scenario. Deposits would flee to safer banks.

The main effect of QE was psychological — market participants were convinced that it was “money printing,” and used central bank balance sheets as a story to “explain” risk market gyrations.

External Sector

Some people (particularly emerging market analysts) have a bee up their bonnet about foreign investment. Apparently, foreign bond market vigilantes have the ability to unbalance a domestic financial system.

However, these concerns are invariably over-played for fiat currencies (not counting the euro as a true fiat currency). Each currency creates its own closed accounting system. If we replace domestic bond holders with foreign private sector bond holders, it is unclear why their portfolio allocation properties would be wildly different. The exception is foreign official currency reserves — as governments may switch their reserve allocations for political reasons. Nevertheless, the large size of many reserve positions limits their flexibility, and the domestic central bank can easily offset their transactions.

(Making the currency convertible — such as gold convertibility — creates a way to destroy financial assets in one currency and replace them with another. As seen historically, this can easily be destabilising.)

Savings and Growth

The possibility that financial markets are effectively self-financing is deeply unpopular. Many financial market participants want to believe that they are doing something useful for society, and politicians want to justify handing over power to financial markets. The typical argument is that by boosting financial markets, we will deliver stronger growth.

Properly understood, the role of financial markets is to price risks, and hopefully end up with an allocation that leads to sustainable investment and therefore growth. They also allow circular cash flows to be reinvested in this more sensible pattern. The problem with under-developed financial markets is that lending gets concentrated in a handful of badly managed banks, badly managed direct investment in firms or real estate, or else leaks into overseas markets. Once the dysfunctions are under control, the financial markets only have a limited effect on growing the pie, rather they just make sure the pie is somewhat edible.

It is possible to generate strong growth with a minimal non-bank financial system, as Germany and Japan demonstrated after World War II. The problem with a bank-dominated system is the herding behaviour of banks. Lending against fixed investment is a lot easier when you have the American consumer hoovering up all your exports. But once the easy lending opportunities are gone, banks have a tendency to all go over the credit cliff at the same time. Offloading concentrated credits on the portfolios that can absorb those losses allows the banking system to concentrate on whatever its alleged skills are.

Concluding Remarks

Industrial capitalism is not a system where people push around a fixed amount of “money.” Instead, capitalist economies melt upward (until the process reverses, usually temporarily). The act of lending grows liabilities and assets, and those assets end up being traded away.

The role of banks is to understand the liquidity flows created by asset allocation trends, and to tap into them to fund their own balance sheet. 

Email subscription: Go to 

(c) Brian Romanchuk 2024

No comments:

Post a Comment

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.