Self-Funding And Deposit Hoarding

Example Recap

We again assume that the banking system has five banks — Bank A to Bank E. We assume that Bank A extends $100,000 million in residential mortgages in one day, while the other five banks do nothing (and there is no net loan repayment). We assume that all banks have 20% market share, so cash flows between non-banks are evenly distributed.

We assume that the mortgages all clear in the same day, and the proceeds immediately sent to the sellers of the dwellings. This immediately results in $20 million in new deposits at the five banks (because of the 20% market share assumption). This implies that there is an $80 million outflow from Bank A to the other four banks, which means its draws down its liquidity portfolio by $80 million. As previously emphasised in the example discussion, banks have to operate with excess liquidity buffers, and this outflow posed no particular risk.

We then assume that Bank A issues $100 million in mortgage-backed securities, and those securities are bought by non-banks. Under the 20% market share assumption, that means that all five banks lose $20 million in deposits — which implies a net inflow of $80 million to Bank A, which returns its liquidity portfolio back to where it previously was.

If we look at the formal banking system (Banks A to E), there is no change to their balance sheets — other than the fact that Bank A will have pocketed the usual mortgage fees, generating some immediate income. The non-bank sector expanded its balance sheet by $100 million — home buyers will have plopped an additional $100 million of mortgage debt onto the household sector balance sheet, and various entities will have $100 million in new mortgage-backed assets on their balance sheets. If those mortgage-backed securities are held directly by households or pass-through entities with household owners, this $100 million of new assets is either directly or non-directly owned by the household sector without being clouded by firm’s balance sheets.

(This is an example of how the non-bank financial system can grow its balance sheet even if all the non-bank entities are pass-through entities that superficially appear to obey the laws of loanable funds. If we are working with monthly data, the transit through the balance sheet of Bank A would not be visible. This is why it is a bad idea to pretend that the non-bank financial system is wildly different than the formal banking system in terms of credit creation — the whole system acts like a bank. The non-bank sector exists to get assets off the balance sheets of traditional banks.)

Since the formal banking system balance sheet is unchanged at the end of this two step process, we can justify the plausibility of the ability to issue the mortgage-backed securities immediately — it looks like a return to the status quo ante. There is a change to overall portfolio allocations, but that is the topic of the rest of the article.

Non-Bank Balance Sheets

The previous example text just looked at what was happening on the balance sheets of the five traditional banks. The reasoning being is that I wanted to describe the logic as seen by the treasury desk of a bank. If we want to see why would should expect funding to be available from outside the banking system, we need to spend some time looking at the other balance sheets in the private sector.

Although there might be some advantages to having a model of the entire economy — like a stock-flow consistent (SFC) model — I think that this would not be the best approach to get an understanding (and deal with the inevitable objections). A full model of the economy has a lot of moving parts — connections between all the sectors need to be formalised — as well as requiring behavioural rules. The behavioural rules of a tractable model will be too simplistic, and discussions would get derailed if those are questioned. I would rather describe balance sheet effects that apply to any set of behavioural rules (as will be done below).

We do not need to look at every balance sheet entry for the household sector, non-financial firms, and non-bank financial firms, just what matters for the example. Mortgage debt on the liability side, and deposit balances, mortgage-backed security holdings, as well as the total of all financial assets on the asset side.

After Bank A issues the $100 mortgage-backed security, there are two changes to the balance sheet: household mortgage debt increased by $100 million, and mortgage-backed securities by $100 million.

The example did not specify the amount outstanding of various asset classes other than (demand) deposits — there was $50 billion outstanding across all banks. The transactions represent 0.2% of demand deposits, and probably even a smaller percentage of the market value of all other financial assets (since equity and bond holdings tend to be much larger than demand deposits outstanding).

The new transaction does imply a small change in the portfolio weightings of the other sectors. Deposits did not grow, but mortgage-backed security holdings did. We can decompose this portfolio shift at two levels.

1. The deposit to other financial assets ratio dropped. To the extent that is unwanted, the expected return for aggregate financial assets would need to tick up a small amount if believe that there is a money demand function that depends upon the level of expected returns (which is standard).

2. Within the financial asset aggregate, the weighting of mortgage-backed securities would rise. To the extent that we believe portfolio allocation models, mortgage-backed securities might need to cheapen versus other financial assets (the spread versus the benchmark curve would widen).

In practice, it is extremely hard to detect such relative value shifts. Although new issues might be priced cheap to help underwriting, it is hard to see a persistent effect once we take into account the relatively wide bid-offer spreads in the secondary market. Even if a $100 million deal is too large for my example economy, it would just need to be split into issues at the size that the market can bear. Markets are not going to choke permanently on issuance that is 1% of the size of demand deposits outstanding.

Note that we do not need exact matching between the people who sold houses and the buyers of mortgage-backed securities. If they buy other securities (or send the money to fund managers), this will just trigger rebalancing across financial market participants. For example, if they buy equities, the equity seller might give the proceeds to a bond manager, who then buys mortgage-backed securities. What matters is the total size of the flows for the aggregate.

Why Deposit Hoarding is a Bad Faith Response

We can now address a highly predictable response to this example: why do I assume that Bank A is able to issue the $100 million in new mortgage-backed securities? Why will the recipients of money courtesy of selling their homes not hoard the deposits?

This response appears plausible on the surface because it deliberately obscures everything except the deposit transfer. If the new deposits are hoarded, then that implies that the weighting of deposits in portfolios rises. This implies a change in portfolio preferences, which raises the obvious response: why would they change? And even if they did, that is just adding a new step to the scenario.

(In the case of mortgages, it is extremely unlikely that the sellers of homes will keep the proceeds as deposits. They would likely have balances that are beyond deposit insurance limits, and since the sellers are likely older, they will be allocating the proceeds to their retirement portfolio.)

Portfolio Allocations Change

It is entirely possible that non-banks change their target portfolio allocation, and prefer deposits over bank debt issues (including securitisations). This is something that can happen at any time, and in conventional theories, it is met by relative price changes. Most conventional models have “money demand functions,” but instead of just the interest rate on the only other financial asset in the model (“bonds”) we need to look at expected returns across all asset classes. Since the lending transactions created the $100 million of deposits that can be used for funding, the deal can get done, just the price needs to be determined.

That is, under normal circumstances, funding for (solvent) banks is always available — at some price. The key to long life for a bank is knowing where to get their hands on that funding (hopefully cheaply). In practice, banks are only cut off from funding just ahead of a restructuring, or all the banks are cut off and they have to run to the central bank to be bailed out.

The entertaining part of this is that the most likely people to suggest that deposits might be hoarded are fans of free capitalism and believe that price changes are supposed to regulate market behaviour.

Reserve Hoarding

Mystical beliefs about money leads to far too much time being spent discussing what is happening with reserve balances. Since the central bank has to supply the minimum the system needs, only the allocation of excess reserves is a point of discussion.

Excess reserves are just another financial asset in bank liquidity portfolios (although it might have some regulatory advantages). The reason why Bank A is not going to face the risk of liquidity event is straightforward: it had plenty of assets in its liquidity portfolio in excess of the minimum. Unless the other banks form a cartel (which is legally and practically dubious), all that should be required is the relative price shift in money market assets to allow Bank A to rebalance its liquidity portfolio to generate the required reserve inflows. (And since Bank A would just go the discount window if the other banks tried locking it out, forming a cartel would not accomplish anything other than angering regulators.)

If the phrase “relative price shift” sounds scary, I just want to point out that money market participants tend to get greatly excited about spread changes of 10 basis points (0.1%). As soon as that happens, “experts” will be flooding financial media with stories about “the central bank is losing control of the money markets!” From the perspective of term funding markets, nothing interesting happens in the money markets until the money market participants do something really stupid. This does happen — many modern financial crises are the result of credit risk being concentrated in the hands of the people least able to evaluate credit risk — money market portfolio managers.

Once again: price changes are the mother’s milk of free market capitalism.

Concluding Remarks

Lending creates the deposits that can be drawn upon to fund debt issuance by banks. The issue is not whether the funding exists, but rather: at what price? In practice, nobody is going to find a measurable effect of issuance on a bank’s debt pricing — partly larger issues have a lower secondary market illiquidity premium. What raises banks’ lending cost is a weakening credit profile — which is driven by credit losses (or interest rate losses if bank management is dumber than a sack of hammers).

We will next give an overview of the behaviour of non-bank credit markets, since they are where banks draw funding from. What we will see is that relative pricing — credit spreads — do not move the way crude supply and demand theories suggest.

Technical Appendix: Mortgage Lending

Given the importance of housing bubbles in recent decades, it is perhaps useful to look at some of the empirical regularities about residential finance. My comments here are mainly based on the “Anglo” countries — the U.S., Canada, Australia, New Zealand, and the United Kingdom (which I covered as a macro analyst at the beginning of my career).

Other than for the very wealthy (or corrupt), residences are too expensive to buy without mortgage finance (or the proceeds of a home sale). This implies that a home purchase quite often requires the simultaneous sale of the purchaser’s home. This creates a chain of transactions that typically works as follows (for existing homes).

1. A new buyer enters the housing market (possibly making parents very happy), buying House #1. We will assume that the new house costs $200,000, and the new buyer has a 10% down payment — requiring a new mortgage of $180,000. This injects $180,000 of new mortgages into the system.

2. The seller of House #1 had an existing mortgage of $100,000, and decided to buy House #2 that costs $300,000. (This example ignores all the fees in the transaction.) Once the sale is finalised, they need a new mortgage of $200,000 — they received $200,000 for House #1, used $100,000 to pay off the old mortgage, and used the proceeds as a down payment on House #2. What we see is that the net issuance of mortgages is $100,000 — which matches the difference in price between House #1 and House #2. That is, the owner of House #1 was not exposed to the total price of House #2, rather the difference between its price and their already-owned house.

3. The seller of House #2 may buy another house, repeating the process.

4. Finally, the seller of House N will pocket the proceeds of the home sale, paying off whatever mortgage was left on the house.

There are other possibilities than such a chain — one seller could decide to hold on to two houses for a period of time, the initial buyer may buy a new home directly from a developer, or there is the theoretical possibility that the “last” person in the chain could buy from the “first” (forming a loop).

What we see is that there is a frenzy of new mortgage issuance, but there is also a matching repayment of mortgages in the middle of the chain. The net issuance is capped by the highest house price within the chain (but will tend to be less since existing homeowners will likely have existing home equity). The reality that new mortgage issuance can fire off a chain of transactions means that lending by one bank (Bank A in our example) will trigger mortgage transactions at other banks.

The people entering the housing market tend to be young, and the owners of homes that exit the housing market (end of he chain) tend to be old — or dead. Not everyone ends up renting spaces in senior homes — many people die while owning a home. Unless there is a surviving partner, the house becomes the property of the estate — and typically it is eventually sold by the heirs.

The people exiting the housing market will generally be receiving large cash inflows when they are older. They will not want to keep the cash as a demand deposit, as it would overshoot deposit insurance limits, and is not a useful asset in a retirement portfolio. So the proceeds will likely end up in the financial markets (or term financing of a bank, which is the retail version of a bond).

The effect of house price changes is that higher prices will create a need for larger net mortgage issuance — since the households entering the housing market will tend to have constraints on their ability to increase down payments. Households that already own homes are only exposed to the difference in price, and their existing home equity will be improved by the price change. As such, they do not face great strains buying the new home. This creates a mechanical link between mortgage debt and house prices: they move together, and it is not useful to ask which “causes” the other. People can either be more willing to bid up house prices (house prices “cause” more mortgage debt), or banks may be more willing to extend loans (more mortgage debt “causes” higher house prices).

In summary, the answer to the question “Who were the savers that funded the various housing bubbles?” is that the savers were largely the people exiting the housing market. And they were only able to exit the housing market because they buyer got a mortgage in the first place. Which tells us that “savers” are not appearing first — at best, the savings are created simultaneously with the draw on savings.