The Bank

Between the saver, who wants a balance close at hand, and the borrower, who needs funds committed for years, stands the bank. Its familiar spread is the difference between the interest it receives and the interest it pays. But the two sides of that difference buy different promises. The depositor has a claim against the bank; the bank has a claim against the borrower. The depositor need not wait for that particular borrower to repay before making a withdrawal, and the borrower need not replace the loan whenever that particular depositor leaves. This separation is the beginning of the service. It is also the beginning of the problem. The bank must judge the borrower, collect what is owed, arrange the payments its own creditors demand, and absorb losses somewhere in its balance sheet. The question is therefore more demanding than why the saver and borrower do not meet directly. It is what would have to replace the bank's promises, information and resources if they did. The institution earns a place between them by doing work that the two quoted rates leave largely out of sight.

Even the measured spread requires care. A loan rate of six percent and a deposit rate of two percent are four percentage points apart. The bank's net interest margin need not be four percent. Suppose, for one unchanged year, it holds one hundred of interest-earning assets yielding six percent, financed by ninety of deposits costing two percent and ten of equity. Interest income is six, interest expense is one point eight, and net interest income is four point two. Dividing by the hundred of average earning assets gives a margin of 4.2 percent. The quantities on the two sides matter as well as their rates.¹ This margin is still before the operating expenses, credit-loss provisions and other items that separate it from the bank's final profit. A larger margin can accompany more costly or riskier lending; a smaller one can accompany lower-yielding assets, more expensive funding, or stronger competition. The number records an accounting relation. To call the whole of it payment for useful work, or the whole of it a toll on access, would already assume the answer the volume is trying to find.

One part of the work is the watching that lending requires. A borrower may know more than the lender about the proceeds of a project and whether the promised conditions have been met. Obtaining that information costs time and expertise. If a firm borrows from a hundred small lenders and each pays to investigate the same facts, the work is duplicated; if each relies on the others to undertake it, too little may be done. Douglas Diamond's account of delegated monitoring starts from this difficulty. A bank can investigate once on behalf of the funds it has raised, enforce the loan's terms, and use information that the dispersed lenders would find expensive to acquire separately.² The service is more than a first introduction. Lending establishes a relationship that continues after the money has been advanced. A covenant may have to be interpreted, a shortfall examined, a contract renegotiated. The advantage of a specialist lies partly in having someone responsible for that continuing work. It does not require every saver to be incapable of lending directly; it requires circumstances in which the combined arrangement is less costly than the available alternatives.

The objection is immediate: who watches the watcher? The depositor who could not cheaply verify a business's receipts cannot simply assume that a bank will report and distribute its own receipts faithfully. Delegation creates another incentive problem. Diamond's answer depends on the bank's portfolio and the contract it offers its creditors. Combining many loans whose returns are independent, or independent conditional on observable information, makes the bank's aggregate receipts more predictable. In his model, the cost of giving the delegated monitor the appropriate incentives falls as the portfolio grows; under the stated assumptions it approaches zero in the limit.³ This is a reason diversification can make the institution economical, rather than a certificate that a large bank has become safe. Loans exposed to the same shock do not become independent because their borrowers have different names. Nor does a predictable asset portfolio by itself settle the timing of the bank's obligations. The theory's achievement is to explain how the extra link in the chain can reduce the total cost of trust. The conditions that allow it to do so remain part of the explanation.

Timing gives the bank a second service and a distinct vulnerability. Some depositors will need to spend early; others can wait. A pool can meet ordinary early demands while leaving part of its assets committed to longer projects. Diamond and Dybvig showed how demand deposits can provide insurance against uncertain consumption needs, improving on each person holding and liquidating an individual investment. Their model also admits a run. If enough depositors expect the others to withdraw, waiting becomes unattractive; early liquidation then destroys the return that patience would have produced.⁴ An asset can be sound at maturity and costly to realize today. The promise of access is valuable because the bank does not ordinarily have to turn every asset into cash at once. Its fragility appears when that ordinarily becomes the disputed fact. Credible deposit insurance can remove the run incentive in the model, but its force comes from the backing of the promise; the authors also distinguish that result from the further problem of managers choosing risky assets. Neither diversification nor a positive interest margin makes these questions disappear. The spread is earned across time, and the institution must remain able to make its payments along the way.

The balance sheet also prevents a misleading picture of money passing unchanged from a named saver to a named borrower. In its 2014 explanation of money creation, the Bank of England describes a loan being made by crediting a deposit account: the bank acquires a loan asset and creates a matching deposit liability. It does not first have to find an existing deposit of precisely that amount and hand it onward. Yet creating the account entry does not abolish the cost of funding. When the borrower pays someone at another bank, the lending bank may have to transfer reserves to settle the payment. It must manage those outflows, attract or retain funding, and meet the capital and liquidity demands associated with its business.⁵ The power to create a deposit is a power to incur a promise, not a power to make the promise costless. This account of the transaction fits the larger account of intermediation: the bank issues one kind of claim while holding another, and must keep the arrangement viable. Deposits, other borrowing and equity perform different roles in that arrangement. The interest margin reflects some of their prices without describing the whole mechanism.

The bank's defense is therefore substantial, and its fee remains open to examination. Monitoring, payments, liquidity and the bearing of losses can make an intermediary useful; the absence of alternatives can also affect the terms it obtains. To distinguish them, one would need to compare the service and risks as well as the quoted loan and deposit rates. A high margin alone cannot show how much came from costly work, how much from exposure, and how much from power over customers with few other routes. Nor can removing the bank's name remove the functions its balance sheet performed. Someone would have to undertake the watching, provide the claims and bear the consequences when their timing diverged. The next plate returns to The Maker, the dealer who supplies immediacy by remaining ready to trade. The bank supplies a different accommodation: it lets one party make a longer commitment than another is prepared to make. What it keeps between the two promises must be judged against the work, the risk and the position that make their coexistence possible.