Deposits a Bank Bought Rather Than Earned
A bank short of funding can buy deposits through intermediaries rather than gathering them from local customers. Regulators treat that money as less reliable, and the definition of what counts has been fought over for decades.
Two Ways to Fund a Bank
A bank makes loans and funds them with deposits. The traditional source is a core deposit: money from a customer with a relationship, typically a chequing account holder in the bank market area, who keeps the balance for reasons unrelated to the interest rate.
Core deposits are the most valuable funding in banking because they are stable, cheap, and insensitive to rates. That stability is why a bank branch network has value beyond the transactions conducted in it.
The alternative is to buy deposits. A brokered deposit is placed by a third party intermediary, who gathers funds from investors seeking yield and directs them to whichever bank offers the best rate, frequently in amounts sized to stay within deposit insurance limits.
Why Regulators Treat Them Differently
The concern is behavioural rather than legal. A depositor placed by a broker chasing yield has no relationship with the bank and will move at maturity to whoever pays more.
That produces two problems. The funding is rate sensitive, so its cost rises immediately when rates rise, compressing margin exactly when a bank may be under pressure. And it is flighty, meaning it does not roll over if the bank credit deteriorates or a competitor bids higher.
| Core Deposits | Brokered Deposits | |
|---|---|---|
| Source | Customer relationship | Intermediary seeking yield |
| Rate sensitivity | Low | High |
| Behaviour under stress | Sticky | Leaves at maturity |
| Speed of raising | Slow, requires distribution | Very fast |
| Cost | Low | Above market |
The dangerous property is the combination. Brokered deposits can be raised very quickly, which means a bank with a bad lending strategy can fund it faster than supervisors can react, and the funding disappears exactly when the strategy fails.
The Historical Pattern
The concern is grounded in a documented pattern rather than in theory. Institutions in the savings and loan crisis used rapidly gathered brokered funding to grow balance sheets far faster than any organic deposit base could support, deploying it into aggressive lending.
The mechanism recurs. A bank that has decided to grow quickly finds that gathering deposits from customers takes years and buying them takes days. The constraint on growth becomes the willingness to pay for funding rather than the ability to attract it, which removes the natural brake that a deposit franchise imposes.
The Restrictions
Federal law addresses this by tying access to capital condition. A bank that is well capitalised may accept brokered deposits freely. One that is adequately capitalised requires a waiver. One that falls below that may not accept them at all.
The design is deliberate and severe. As a bank weakens, the funding source it can raise fastest is removed, which forces it to shrink rather than to grow its way out of trouble.
It is also, from the bank perspective, a cliff. A capital downgrade triggers a funding constraint that can precipitate exactly the liquidity problem the rule was meant to prevent, which is a recognised criticism of the design.
The Definition Fight
The rules were written before deposits could be gathered through technology platforms, and the resulting definitional question has been contested for years.
The core issue is the deposit broker definition, which turns on whether a party is engaged in the business of placing deposits at insured institutions on behalf of third parties.
Modern arrangements complicate it enormously. A fintech company holding customer balances at a partner bank, a wealth management platform sweeping client cash across multiple banks, or a reciprocal network exchanging deposits among institutions all involve a third party directing where money sits.
Rules finalised in 2020 narrowed the definition, introducing a primary purpose exception for arrangements where placing deposits is not the main objective of the relationship, and treating reciprocal deposits more favourably where the bank is well capitalised.
Subsequent proposals moved in the opposite direction, seeking to broaden the definition again, particularly in light of bank failures where deposits gathered through technology platforms proved highly concentrated and rapidly mobile. The question remains genuinely unsettled.
What Recent Failures Showed
The bank failures of 2023 complicated the traditional framing in a useful way.
The deposits that fled fastest were not brokered in the technical sense. They were large uninsured balances from concentrated customer bases, connected socially and professionally, who could move money instantly through digital channels.
Those deposits were classified as core under the existing definitions and behaved far worse than most brokered funding would have. The episode suggested that the characteristics regulators care about, concentration, insurance status, and speed of withdrawal, are only loosely correlated with whether a broker was involved.
What to Look At
For anyone analysing a bank, several disclosures matter more than the brokered label. The share of deposits that are uninsured. Concentration by industry or customer type. The mix of time deposits versus transaction accounts. Deposit cost against peers, since paying materially more indicates buying funding regardless of classification. And growth in deposits far exceeding branch or customer growth, which indicates the money is arriving through a channel rather than a relationship.
The Bottom Line
Brokered deposits are funding a bank purchased rather than earned, and the restrictions on them exist because rapidly available money has repeatedly financed rapidly deteriorating lending. The regulatory definition has struggled to keep pace with how deposits are actually gathered now, and the 2023 failures demonstrated that uninsured, concentrated, digitally mobile deposits can be less stable than brokered ones while carrying none of the same restrictions. The label is a proxy, and the properties it was meant to capture are worth measuring directly.