The uses with higher returns often take many years to pay off: plant fruit trees and you wait five years for the first fruit. Yet anyone storing grain may need it urgently at any time, and no one wants to lend grain away for five years. Pool the grain of many households, and in any one year only a few come to withdraw. Keep back enough to meet those withdrawals and the rest can be lent for the long term. This is the bank's maturity transformation. Its cost is the bank run: once people fear that others will withdraw first, they all come at once, and even a well-run bank can fall. Deposit insurance and the lender of last resort spare depositors the need to get there first. The bank also watches borrowers on behalf of many depositors, saving each of them the cost of checking for themselves.
Lending lets seed flow to land where it yields more: my land is fully sown, and my last few portions of seed add little to the harvest; my neighbour's land lies idle, and lent to him the same seed adds much more. That kind of loan spans a single season, but the uses with higher returns often take many years to pay off. Clearing waste land, digging an irrigation channel and planting an orchard all require years of investment before anything is harvested.
Today's counterpart. Building roads, building factories and buying homes all need money that is invested once and repaid over many years. Loans of this kind often run for ten or twenty years.
The villager who wants to plant an orchard needs to borrow for five years, but those with spare grain do not know when they will need it urgently: a family member falls ill, a house catches fire, an unexpected journey has to be made. Grain lent away for five years cannot be recovered in an emergency. For anyone storing grain, the chance of at least one emergency within five years is not small: if the chance of an emergency is one in ten each year, the chance of at least one within five years is about forty per cent. So few people are willing to lend their grain directly, for years, to someone planting an orchard.
The mismatch of terms prevents many long-term loans that could otherwise have been made.
Today's counterpart. Most households want deposits they can withdraw at any time; those willing to lock money away for many years usually ask for a higher return.
If all fifty households in the village store their spare grain in the same granary, things change. When any single household will need its grain cannot be predicted, but taken together the number of households that come to withdraw each year is quite stable, roughly a fixed share of the total. The granary need only keep enough grain to meet those withdrawals; the rest can be lent for the long term to the orchard planters. Every household can still withdraw at any time, and the orchard planters get five-year loans of grain. The granary turns grain that can be withdrawn at any time into grain lent out for the long term. Each household gets something it could not get on its own: grain it can draw on at any moment, together with a share of the orchard's harvest.
Today's counterpart. Inside the same bank, demand deposits sit alongside mortgages of twenty or thirty years. The reserves a bank is required to hold against deposits are the grain kept back to meet withdrawals.
This arrangement has a weak point. The granary keeps only part of the grain; the rest has already been planted as fruit trees, and recalling it early means selling at a discount. Suppose that one day the depositors begin to fear that others will come first, and that anyone a step behind will get nothing. Then even those with no emergency rush to withdraw. The more who come, the less the granary can hold out, and the more people rush in. Once a run begins, even a well-run granary falls, and fruit trees cut down early will never bear fruit.
For each household, as long as it believes others will not rush, it has no reason to withdraw without an emergency; once it believes others will rush, withdrawing at once is its best choice. Both beliefs are self-fulfilling. The same granary is sound while people believe it is sound, and really does fall when people fear it will. Its books are identical in the two cases; the only difference is what people expect of each other.
Today's counterpart. In March 2023 a regional bank in the United States faced withdrawal requests of about $42 billion in a single day, with about $100 billion more queued for the next day (Barr, 2023). Online banking and social media make runs far faster than in the days of queues at the counter.
Since runs come from the fear that others will withdraw first, the remedy is to remove the fear. Deposit insurance promises that even if the granary falls, each household's grain up to a set amount will be returned in full. Those with no emergency then have no need to rush, and a run is defused before it starts. The United States established federal deposit insurance in the Banking Act of 1933.
The other remedy is the lender of last resort: during a run, an institution with ample grain lends freely to the troubled granary, against good collateral and at a high rate of interest. Bagehot (1873) set out this principle in Lombard Street.
These remedies have costs too. Once deposits are protected, depositors no longer care whom the granary lends to, and the granary may take greater risks as a result. This is the moral hazard that insurance brings. That is why deposit insurance always comes together with capital requirements and supervisory inspection.
Today's counterpart. Deposit insurance has a coverage limit; banks must meet regulatory requirements on capital adequacy and liquidity; central banks provide emergency liquidity to banks in a crisis.
Will the villager who borrowed grain to plant an orchard look after the trees properly? If each of fifty depositing households had to inspect every orchard in person, the time spent would far exceed the value of the grain. If the granary inspects on everyone's behalf, each orchard needs to be inspected only once. But a new problem follows: the granary inspects for everyone, so who inspects the granary? The answer lies in the granary lending to many orchards at once. The harvests of different orchards rise and fall independently, and taken together they are roughly stable. As long as the granary does its inspecting, it is almost certain to be able to repay depositors; if it cannot repay, that shows it has failed in its duty. Depositors therefore no longer need to inspect orchards one by one; they need only ask whether the granary itself is reliable. This is another reason the granary exists: it watches the borrowers on behalf of many depositors.
The more costly monitoring is, the more valuable this delegation becomes. If monitoring becomes very cheap, depositors can see borrowers clearly for themselves, and the reasons to delegate to a bank change accordingly.
Today's counterpart. Banks examine borrowers' income and collateral and keep tracking them over the life of the loan; depositors need only ask whether the bank itself is sound.
The simulator unlocks the "Bank" rule. Fifty households in the village store grain, and orchard planters borrow five-year grain from the granary. You can switch the granary and deposit insurance on or off, adjust the share of grain the granary keeps back for withdrawals, and set how many people panic and withdraw when a rumour arrives. Then watch, over thirty years, the orchards' output, whether depositors can always withdraw, and whether a run occurs.
Without the granary, no one is willing to lend grain for five years, the orchards are never planted, and every household's stored grain wastes away year after year. With the granary, the orchards are planted; but when a rumour comes and many panic, the granary cannot pay even after cutting down trees, and it falls. With deposit insurance, those with no emergency no longer rush, and the granary comes through thirty years without trouble. The higher the share kept back for withdrawals, the steadier the granary, but the less it lends to orchards.
With banks, the mismatch of terms is resolved, but several problems remain:
Deposits, loans and wealth-management products all rest on the mechanisms of this chapter. Three implications follow. First, liquidity has a price: a deposit that can be withdrawn at any time and an investment locked up for years should earn different returns, and a product that promises withdrawal at any time should state the actual terms of what its money is invested in. Second, the stability of maturity transformation depends on confidence, and a run can happen without any bad news; the faster information and money move, the more this risk calls for defences designed in advance. Third, part of a bank's value lies in monitoring on behalf of depositors; once the cost of monitoring changes, the form of the intermediary changes too.
When many people pool their grain, the share withdrawn each year is quite stable, so only part needs to be kept back.
Each borrower is inspected once, and depositors need only care about the bank itself; the cost of monitoring sets the value of this delegation.
Fear that others will withdraw first brings everyone at once, and even a well-run granary can fall.
They spare people the need to rush; the cost is moral hazard, which calls for capital requirements and supervision.