Deposit Concentration Risk¶

Judge a bank's funding fragility by the correlation-adjusted effective depositor count rather than the headline number — coupled depositors collapse toward one bet, voiding the law-of-large-numbers smoothing a large base seems to guarantee.

Core Idea¶

Deposit concentration risk is the funding fragility that arises when a bank's deposit base is dominated by a small number of large depositors, or by many depositors whose withdrawal decisions are highly correlated — because they share the same industry, the same venture-capital advisor network, the same geographic exposure, or the same information event. The institution appears diversified by depositor count but is not diversified by effective exposure: when the correlation structure is activated, the nominally many depositors behave as one, and funding can leave faster than assets can be liquidated at par. The result is an idiosyncratic liquidity crisis that can threaten an otherwise solvent institution.

The mechanism is the failure of the statistical independence assumption that makes large deposit counts genuinely protective. A deposit base of ten thousand fully independent depositors is essentially immune to simultaneous large-scale withdrawal — the law of large numbers smooths individual decisions into a predictable, slow-draining aggregate. Correlation breaks this: ten thousand depositors who share a single point of information or a single relationship manager or a single industry crisis have an effective depositor count that may be closer to one, and a sufficiently negative signal causes a cluster withdrawal on a timescale the institution's liquidity buffer and asset-liquidation capacity cannot absorb. The 2023 failure of Silicon Valley Bank is the instructive canonical case: the nominal depositor count was large, insurance coverage appeared normal, but the dollar concentration in a single correlated population of venture-backed startups — coordinated additionally through a small number of venture capital advisors who collectively recommended withdrawal — made the funding base respond as a single bet, draining tens of billions in hours.

Structural Signature¶

Sig role-phrases:

the depositor base — the population of accountholders providing the bank's funding
the nominal diversification — the headline depositor count and the distribution of deposit sizes, the safety the count appears to provide
the correlation structure — a shared cause (industry, geography, advisor network, relationship manager, information channel, insurance status) coupling depositors' withdrawal decisions
the effective depositor count — the correlation-adjusted number, collapsing toward one as coupling rises, the load-bearing quantity to track against the nominal count
the broken independence assumption — the law-of-large-numbers smoothing that requires summed decisions to be independent, voided once a shared cause correlates them
the insured/uninsured coupling split — insured balances weakly coupled (cannot lose principal, slow-draining) versus uninsured balances mutually informative (each informed exit strong evidence to the next, self-accelerating)
the trigger and synchronized run — an information event reaching the coupled population at once so the nominally many behave as one bet, funding leaving faster than the buffer absorbs
the asset-side mismatch — assets that cannot be liquidated at par on the correlated-withdrawal horizon, the complement that turns a cluster outflow into a liquidity crisis in an otherwise solvent bank
the diversify-against-the-cause remedy — the lever: lower the effective count by diversifying the correlation-generating cause (and lengthen funding tenor / hold more buffer), not by adding accounts from the same coupled population

What It Is Not¶

Not made safe by a large depositor count. Diversification by count (many accountholders) is not diversification by exposure (statistically independent withdrawal decisions). The protective smoothing a large base seems to guarantee is an artifact of the law of large numbers, which requires independence; once a shared cause couples depositors, ten thousand of them can have an effective count near one. A bank that looks safe by headcount can be acutely fragile by correlation.
Not a solvency problem. Deposit concentration risk is a funding-liquidity fragility: it can threaten an otherwise solvent institution when correlated withdrawals drain funding faster than assets liquidate at par. The bank may be fully solvent on a mark-to-hold basis and still fail; reading the failure as insolvency misses that the crisis is a timing-and-liquidity mismatch, not a shortfall of assets.
Not cured by adding more accounts. Adding depositors drawn from the same correlated population raises the nominal count while leaving the effective count untouched — no real protection. The remedy is to diversify against the correlation-generating cause (industry, geography, advisor network, information channel), lengthen funding tenor, or hold more liquidity buffer; naïve count-diversification is precisely the failure mode.
Not a base where insured and uninsured behave alike. Insured depositors cannot lose principal, so their incentives to flee are only weakly coupled and they drain slowly; uninsured balances above the limit are tightly coupled, because each informed exit is strong evidence to the next. That mutual informativeness is the self-accelerating branch that converts a quiet outflow into a run, so treating all balances as equivalent misjudges the run mode.
Not the general concentration-risk pattern itself. Stripped of banking vocabulary, the structure is correlated exposure masking notional diversification — the parent concentration_risk / correlated_exposure (with run_dynamics/contagion for the cascade and liquidity_under_stress for the asset-side mismatch) — recurring in B2B customer concentration, supplier concentration, and single-counterparty exposure, where the same effective-count diagnostic and diversify-against-the-cause remedy apply. The insurance-limit coupling, run dynamics on a funding base, and liquidation-at-par horizon are bank-specific cargo; the deposit-specific framing is one worked instance, not the portable pattern.

Scope of Application¶

Deposit concentration risk lives within banking, finance, and prudential regulation, across the deposit-taking and funding settings that share its substrate — a funding base whose nominal diversification can mask a correlated withdrawal exposure against an asset side that cannot be liquidated at par on the run horizon; its reach is bounded there. (The deeper "correlated exposure masking notional diversification" pattern — with its effective-count diagnostic and diversify-against-the-cause remedy — is the parent concentration_risk / correlated_exposure, which carries to customer, supplier, and counterparty concentration as genuine co-instances; the insurance-limit and run-on-a-funding-base flavor does not.)

Commercial banking — the home turf; concentration by single large depositor (institutional cash sweeps), by industry vertical, or by a correlated population, the 2023 SVB failure being the instructive case (venture-backed startups coupled through a few VC advisors).
Prudential supervision — Basel III LCR and NSFR penalise concentrated, flighty funding; supervisors track top-20 depositor lists and large-uninsured-deposit ratios, with contingency funding plans now central post-SVB.
Money-market funds — institutional and retail MMFs are regulated differently because institutional flows are highly correlated and concentrated.
Non-bank deposit-takers — credit unions, building societies, and fintech accounts via partner banks carry the same fragility, sometimes worse where the base is even more narrowly selected.
Stablecoin reserves — concentration in the bank holding the reserve becomes concentration in the coin itself, the modern parallel.
DeFi liquidity pools — pools dominated by a few large liquidity providers whose simultaneous exit can break the pool exhibit the same structural pattern.

Clarity¶

Naming deposit concentration risk makes legible a distinction that a headline depositor count conceals: the difference between diversification by count — many accountholders — and diversification by exposure — accountholders whose withdrawal decisions are statistically independent. A bank with ten thousand depositors looks safe by the first measure and may be acutely fragile by the second, and the label is what forces a liquidity officer to see that the two can come apart. The clarifying insight is that the protective smoothing a large deposit base seems to guarantee is an artifact of the law of large numbers, which requires the summed decisions to be independent; once a shared cause — one industry, one advisor network, one information event — correlates them, the nominally many depositors collapse toward a single bet, and the funding can leave faster than assets liquidate at par. The practitioner's question sharpens from "how many depositors, and how large?" to "what is the correlation structure across them, and what is the effective depositor count once it is applied?"

This reframing turns funding-stability analysis into a concrete diagnostic and explains features of bank behavior that the nominal view leaves opaque. The object to manage is no longer the headline count but the effective count under correlation, and the right remedy is to diversify against the correlation-generating cause — industry, geography, relationship manager, information channel — rather than merely to add more accounts drawn from the same correlated population. The framing also makes legible why the deposit-insurance limit matters in the precise way it does: insured depositors, unable to lose principal, have only weakly correlated incentives to flee, whereas uninsured balances above the limit are tightly coupled, because each informed depositor's exit is strong evidence to the next, and that mutual informativeness is exactly what converts a quiet outflow into a run. The clarity is in relocating fragility from the size of the deposit base to the hidden dependence among its members, where a stress that the count says is impossible becomes the expected outcome.

Manages Complexity¶

The sprawl this concept tames is the seemingly unbounded variety of ways a deposit base can be coupled — by shared industry, shared region, a common venture-capital advisor network, a common relationship manager, a single information channel, a common insurance status — each of which, examined on its own, suggests a different and ad hoc story about how funding might suddenly leave. Deposit concentration risk collapses that variety onto a single quantity the liquidity officer must track: the effective depositor count once the correlation structure is applied, as against the nominal headline count. Whatever the particular coupling, its consequence for run risk enters through that one number — many depositors who share a cause have an effective count collapsing toward one, and the smaller that effective count, the faster funding can drain relative to the bank's capacity to liquidate assets at par. The analyst therefore stops cataloguing scenarios and instead reads fragility off the gap between nominal and effective count, with the qualitative outcome — quiet slow outflow versus cluster run — settled by how far correlation has shrunk the effective number below the protective threshold the law of large numbers would otherwise guarantee. The branch structure the framing supplies is keyed to that same correlation: insured balances, which cannot lose principal, stay weakly coupled and drain slowly; uninsured balances above the limit are tightly coupled, because each informed exit is strong evidence to the next, and that mutual informativeness is exactly the branch that tips a trickle into a run. The remedy is determined too — diversify against the correlation-generating cause rather than adding more accounts from the same coupled population, since the latter raises the nominal count while leaving the effective count untouched. The high-dimensional problem "how fragile is this funding base, given everything that might make depositors leave together" reduces to "what is the effective depositor count under the operative correlation, and how much of the base is uninsured and thus mutually informative" — one correlation-adjusted count and one coupling test in place of an open list of withdrawal scenarios.

Abstract Reasoning¶

Deposit concentration risk licenses a set of bank-funding inferences, all keyed to one constructed quantity — the effective depositor count once the correlation structure is applied — read against the nominal headline count.

Diagnostic (compute the effective count behind the nominal one). The signature move is to look past the depositor count to the correlation structure and infer the effective number. The analyst reasons FROM "this base has ten thousand depositors, but they share one industry / one advisor network / one information channel / one insurance status" TO "their withdrawal decisions are coupled, so the effective count collapses toward one," and reads fragility off the gap between nominal and effective count. The load-bearing inference is that the protective smoothing a large base seems to guarantee is an artifact of the law of large numbers, which requires the summed decisions to be independent — so a shared cause is diagnosed as the thing that voids the smoothing, and a bank that looks safe by count is judged acutely fragile by exposure.

Predictive (the effective count and the insurance split fix the run mode). From the effective count and the insured/uninsured composition, the framework predicts whether an outflow stays a trickle or tips into a run. The analyst reasons FROM "the effective count is small and much of the base is uninsured" TO "a sufficiently negative signal causes a cluster withdrawal faster than assets liquidate at par"; FROM "balances are insured, cannot lose principal" TO "incentives stay weakly coupled and the base drains slowly." The decisive predictive branch is the mutual-informativeness of uninsured depositors: reasoning runs FROM "each informed depositor's exit is strong evidence to the next" TO "the outflow is self-accelerating among uninsured balances," which is exactly what converts a quiet drain into a run — so the analyst predicts the run mode from coupling and insurance status, not from size.

Diagnostic / order-of-events (read a trigger forward through the correlated population). Given a correlation-generating cause, the framework predicts the path of a stress: a single information event reaches the coupled population at once, the nominally many depositors respond as one bet, and funding leaves on a timescale the liquidity buffer and asset-liquidation capacity cannot absorb. The analyst reasons FROM "a negative signal hits an industry / an advisor network coordinates withdrawal" TO "a synchronized cluster outflow," and FROM "the asset side cannot be liquidated at par on that horizon" TO "an idiosyncratic liquidity crisis in an otherwise solvent institution" — the order-of-events reading that explains how a base the count says is safe drains in hours.

Interventionist (diversify against the cause, not the count). Treating the funding structure as the design handle, the framework predicts the sign of a remedy by whether it lowers the effective count or merely the nominal one. The analyst reasons FROM "add more accounts drawn from the same correlated population" TO "the nominal count rises while the effective count is untouched — no real protection"; FROM "diversify against the correlation-generating cause (industry, geography, relationship manager, information channel), lengthen funding tenor, or hold more liquidity buffer" TO "the effective count rises or the asset side can absorb the correlated load." The lever is the correlation structure, and the predicted failure mode of naïve diversification is precisely adding accounts that leave the coupling intact.

Boundary-drawing (the independence assumption, and the substrate edge). The protective reasoning of a large base applies only where withdrawal decisions are statistically independent; the analyst reasons FROM "a shared cause couples them" TO "the law-of-large-numbers smoothing no longer holds, and the count is the wrong safety measure." The same logic marks the concept's edge: stripped of banking vocabulary the structure is correlated exposure masking notional diversification — the same skeleton as customer concentration in a B2B firm or counterparty concentration in a derivatives book, all calling for the same correlation-adjusted effective-count diagnostic — so the cross-substrate inference travels under the broader concentration/correlated-exposure pattern, while the deposit-specific framing (insurance limits, run dynamics on a funding base, liquidation at par) is bound to the bank-liquidity substrate.

Knowledge Transfer¶

Within the home domain — banking, finance, and prudential regulation — deposit concentration risk transfers as full mechanism. The effective-versus-nominal depositor-count construction, the law-of-large-numbers-requires-independence diagnostic, the insured/uninsured run-mode branch (uninsured balances mutually informative and self-accelerating; insured balances weakly coupled and slow-draining), the order-of-events reading of a trigger through a correlated population, and the diversify-against-the-cause-not-the-count remedy all port intact across the deposit-taking and funding settings the concept governs: commercial banking (industry-vertical and single-large-depositor concentration, institutional cash sweeps), money-market funds (institutional flows highly correlated and concentrated), non-bank deposit-takers (credit unions, building societies, fintech accounts via partner banks), stablecoin reserves (concentration in the reserve bank becoming concentration in the coin), DeFi liquidity pools dominated by a few large LPs, and post-SVB supervisory practice (LCR/NSFR penalties on flighty funding, top-20 depositor lists, large-uninsured-deposit ratios, contingency funding plans). The same diagnostic reads each because the substrate is shared — a funding base whose nominal diversification can mask a correlated withdrawal exposure against an asset side that cannot be liquidated at par on the run horizon. The transfer is mechanistic because the load-bearing content (the correlation-adjusted effective count, the insurance-limit coupling, the run dynamics, liquidation at par) travels with the vocabulary; the 2023 SVB failure is the instructive case in every one of these settings.

Beyond bank funding the honest report is a strong shared abstract mechanism case — strong enough that the deposit-specific concept is best understood as a canonical worked instance of a more general fragility prime rather than a thing that transfers in its own right. Stripped of banking vocabulary, the structure is correlated exposure masking notional diversification: a population of nominally many independent units whose decisions are coupled by a shared cause, so the protective smoothing that count seems to guarantee is voided and the effective count collapses toward one. That pattern is precisely the catalogue prime concentration_risk / correlated_exposure, and it genuinely recurs across radically different substrates as co-instances, not resemblances: customer concentration in a B2B firm, supplier concentration in a manufacturer's supply chain, geographic concentration in an insurer's policy book, single-counterparty exposure in a derivatives book, single-point-of-failure in cloud-vendor dependencies. In every one, the same diagnostic applies literally — compute the correlation-adjusted effective count, not the nominal count — and the same remedy applies — diversify against the correlation-generating cause, not by adding more units from the same coupled population. So the genuinely portable object, including the diagnostic and the intervention, is the broader prime: "compute correlation-adjusted effective count" transfers across all those substrates, while "compute deposit concentration risk" does not, because none of them would call the pattern "deposit" anything.

The cargo that stays home is the deposit-specific flavor: the insurance-limit coupling, the run dynamics on a funding base, the liquidation-at-par horizon, the funding-liquidity framing. And the concept decomposes cleanly into already-covered primes — correlation (the underlying coupling of nominally independent units), concentration (a few units dominating the aggregate), run_dynamics / contagion / cascade (the self-accelerating exit mechanism), and liquidity_under_stress / funding_liquidity (the asset-side complement, the liquidity mismatch). So the correct cross-domain lesson carries concentration_risk / correlated_exposure (with run_dynamics/contagion for the cascade and liquidity_under_stress for the asset-side mismatch) — not "deposit concentration risk," which is the bank-funding instantiation of those primes and one of the most operationally important worked instances of them in modern prudential practice. Within banking the mechanism transfers in full across every deposit-taking and funding structure; one level up the correlated-exposure / concentration-risk pattern carries the cross-domain lesson — diagnostic and remedy intact — as genuine co-instances; the deposit-specific framing does not travel past the bank-liquidity substrate (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

The construction at the concept's core is the effective-count adjustment for correlated units. For N depositors whose withdrawal decisions share a common (equicorrelated) pairwise correlation ρ, the effective number of independent depositors is N_eff = N / (1 + (N − 1)ρ), which for large N approaches 1/ρ. Take a bank with N = 10,000 depositors — a base that looks amply diversified. If those depositors are fully independent (ρ = 0), N_eff = 10,000 and the law of large numbers smooths withdrawals into a slow, predictable drain. But if a shared cause couples them at ρ = 0.2, then N_eff = 10,000 / (1 + 9,999 × 0.2) ≈ 10,000 / 2,000.8 ≈ 5. Ten thousand accountholders behave, for run-risk purposes, like five independent bets. Raise the coupling to ρ = 1 and N_eff collapses to 1 — the whole base is a single decision.

Mapped back: The headline 10,000 is the nominal diversification; ρ encodes the correlation structure of the shared cause. Computing N_eff ≈ 5 is reading the effective depositor count against the nominal one, and the fact that the 1/√N smoothing evaporates once ρ > 0 is the broken independence assumption the whole concept turns on.

Applied / In Practice¶

The March 2023 collapse of Silicon Valley Bank is the worked case that made the risk vivid to regulators. SVB's deposit base was dominated by a single correlated population — venture-backed technology startups, many advised by the same handful of venture-capital firms, and about 94% of its deposits sat above the $250,000 FDIC insurance limit, hence uninsured. When SVB announced securities losses and a capital raise, that coupled network moved as one: prominent VCs told portfolio companies to pull funds, and customers requested roughly $42 billion in withdrawals in a single day on 9 March. The asset side could not absorb it — the bank held long-dated bonds deeply underwater from rate rises, unsellable at par — and regulators seized the bank the next day. A base that looked large and diversified by count drained in hours.

Mapped back: The shared VC-advised startup population is the correlation structure collapsing the effective depositor count toward one. The 94%-uninsured share is the insured/uninsured coupling split at its most flammable — mutually informative balances. The one-day $42 billion request is the trigger and synchronized run, and the underwater bonds are the asset-side mismatch that turned it into failure.

Structural Tensions¶

T1: Effective-count precision versus the unobservable, endogenous correlation it rests on (the formula is clean, ρ is not). The concept's analytic core, N_eff = N / (1 + (N − 1)ρ), gives a crisp correlation-adjusted count that exposes the fragility the headline number hides. But all the difficulty is loaded into ρ, which is neither observable nor stable: in normal times depositors act nearly independently (ρ near 0, so N_eff looks large and reassuring), and ρ jumps toward 1 precisely when a shared trigger fires — the correlation is endogenous to the crisis it predicts. So the metric that is supposed to reveal fragility is calibrated on a parameter that reads benign right up until the moment it does not, and estimating ρ from quiet-period behavior systematically understates it. The precision of the effective-count construction and the treachery of its key input are the same fact. Diagnostic: Is the ρ behind this effective-count estimate drawn from calm-period behavior (where it looks near zero), or stress-conditioned to the regime in which a shared trigger couples the base?

T2: Deposit insurance as stabilizer versus as concentrator (the limit manufactures the flammable tail). Deposit insurance genuinely stabilizes: insured depositors cannot lose principal, so their incentives to flee are weakly coupled and their balances drain slowly, which is exactly why insured-heavy banks are less run-prone. The same limit, however, sharply concentrates flight risk in the uninsured balances above it — each informed uninsured exit is strong evidence to the next, making that tail mutually informative and self-accelerating — so the insurance boundary itself carves the base into a placid insured majority and a combustible uninsured minority. SVB's 94%-uninsured base is the extreme. Insurance thus does not merely reduce run risk; it relocates and concentrates it into the segment above the limit, and a bank can be stabilized in aggregate while its uninsured tail is more flammable than an uninsured base would be. Diagnostic: Is the funding base's stability being read off the insured majority, while the uninsured tail above the limit is left as a tightly-coupled, self-accelerating population that carries the actual run risk?

T3: Concentration as fragility versus concentration as franchise (the niche that funds the run is the business). The remedy is to diversify against the correlation-generating cause — industry, geography, advisor network. But that same concentration is frequently the bank's reason to exist: SVB's dominance of the venture ecosystem, a community bank's regional depth, a specialist's industry relationships are the source of relationship value, pricing power, and deposit franchise. Diversifying against the correlated population means diluting the niche expertise and network that made the deposits sticky and cheap in the first place. The coupling that creates the funding fragility and the specialization that creates the competitive advantage are often the identical fact, so the prudential fix and the business model pull against each other. Diagnostic: Would diversifying against this correlation-generating cause reduce the effective-count fragility without dissolving the specialization that is the bank's franchise — or are the risk and the value the same concentration?

T4: Solvent-but-illiquid versus the run that manufactures insolvency (the liquidity crisis makes itself real). The concept frames the danger as a funding-liquidity fragility that can fell an otherwise solvent bank — a timing mismatch, not an asset shortfall. But the separation of liquidity from solvency, clean in principle, collapses under the run: to meet a synchronized cluster withdrawal the bank must liquidate assets that cannot be sold at par (SVB's underwater long bonds), and crystallizing those losses can convert a hold-to-maturity-solvent balance sheet into a mark-to-market-insolvent one. So the run does not merely exploit a pre-existing insolvency; it can produce the insolvency by forcing fire-sale realization. The reassuring "otherwise solvent" framing understates that the liquidity event and the solvency event are causally linked, not independent. Diagnostic: Is the bank solvent independent of the run, or would meeting the correlated withdrawal force par-below liquidations that manufacture the insolvency the "liquidity-only" reading assumes away?

T5: Rational individual exit versus collective run (mutual informativeness makes fleeing first optimal and fatal). Among uninsured depositors, each informed exit is strong evidence to the next that withdrawal is warranted, and because funding leaves faster than assets liquidate at par, the depositor who withdraws first is repaid while latecomers may not be. That makes running individually rational — even for a depositor who believes the bank is fundamentally sound, since the payoff depends on others' behavior, not fundamentals. The coordination structure that makes each exit informative is exactly what makes the collectively-ruinous run the individually-optimal move, so a base of rational actors can drain a solvent bank precisely because each is reasoning correctly about the others. Suppressing the run therefore requires changing the coordination payoff (a credible backstop, insurance), not persuading depositors the bank is sound. Diagnostic: Is each depositor's exit driven by a fundamental judgment about the bank, or by the mutually-informative first-mover advantage that makes running optimal regardless of fundamentals — and does the remedy address the coordination payoff or only the fundamentals?

T6: Autonomy versus reduction (a bank-funding fragility or a domain instance of concentration risk). Deposit concentration risk carries bank-specific cargo — the insurance-limit coupling, run dynamics on a funding base, the liquidation-at-par horizon — and within banking it transfers as full mechanism across commercial banks, money funds, non-bank deposit-takers, stablecoin reserves, and DeFi pools, with SVB the instructive case throughout. But stripped of banking vocabulary the structure is exactly concentration_risk/correlated_exposure: nominally many independent units coupled by a shared cause, so the count's protective smoothing is void and the effective count collapses toward one. That parent recurs literally — customer, supplier, geographic, counterparty, and cloud-vendor concentration — and carries both the diagnostic (compute the correlation-adjusted effective count) and the remedy (diversify against the cause). The deposit concept is one operationally important worked instance of the parent, decomposing into correlation, concentration, run_dynamics/contagion, and liquidity_under_stress. The tension is between a prudentially central bank concept and the recognition that its portable content is the concentration-risk parent. Diagnostic: Resolve toward the parent (concentration risk / correlated exposure, with run-dynamics and liquidity-under-stress) when the units are customers, suppliers, or counterparties; toward deposit concentration risk when insured/uninsured coupling and a run on a funding base against par-illiquid assets are actually present.

Structural–Framed Character¶

Deposit concentration risk sits at mixed on the structural–framed spectrum — a genuine statistical fragility (correlated exposure voiding law-of-large-numbers smoothing) at its core, but one constituted by, and named as a hazard within, human financial institutions. On evaluative_weight it leans framed: "risk"/"fragility" is not an evaluatively silent mechanism-name but a hazard concept — the entry is throughout an account of how a funding base can fail, so the term carries a normative charge (this configuration is dangerous) heavier than "correlation" or "concentration" alone. On human_practice_bound it is framed: the phenomenon exists only where there are banks, deposits, and depositors who withdraw — remove the human deposit-taking practice and there is no funding base to be concentrated; even the load-bearing insured/uninsured split presupposes a deposit-insurance institution. Institutional_origin points the same way — the deposit-insurance limit, the funding-liquidity framing, the liquidation-at-par horizon, and the prudential apparatus (LCR/NSFR, top-20 depositor lists) are artifacts of banking and its regulation, not facts a nature observer would find. Vocab_travels fails at the deposit-specific layer: "deposit concentration risk" does not travel because, as the entry says, none of the co-instance domains "would call the pattern 'deposit' anything." Import_vs_recognize is bimodal but structural at one level up: within banking the mechanism is recognized intact across commercial banks, money funds, stablecoin reserves, and DeFi pools, and — unusually strong — the parent pattern recurs literally (not by analogy) as genuine co-instances in customer, supplier, geographic, and counterparty concentration, all running the same effective-count diagnostic and diversify-against-the-cause remedy.

The portable structural skeleton is correlated exposure masking notional diversification — nominally many independent units coupled by a shared cause so the count's protective smoothing is void and the effective count collapses toward one — and this is precisely what deposit concentration risk instantiates from its umbrella prime concentration_risk / correlated_exposure (with run_dynamics/contagion for the cascade and liquidity_under_stress for the asset-side mismatch, decomposing further into correlation and concentration). That parent is what carries cross-domain, diagnostic and remedy intact; the insurance-limit coupling, the run on a funding base, and the liquidation-at-par horizon are the domain accent that stays home and keeps the entry domain-specific — indeed the entry calls deposit concentration risk "one operationally important worked instance" of that parent. The cross-domain reach belongs to the concentration-risk parent, not to "deposit concentration risk." Its character: a genuinely structural correlated-exposure fragility dressed in banking-institution vocabulary and framed as a prudential hazard, structural in the effective-count/masked-diversification skeleton it borrows from concentration_risk but held at mixed by the deposit-insurance-and-run practice that constitutes it.

Structural Core vs. Domain Accent¶

This section decides why deposit concentration risk is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — there is no separate section for that.

What is skeletal (could lift toward a cross-domain prime). Strip the banking and a thin relational structure survives: a population of nominally many independent units is coupled by a shared cause, so the law-of-large-numbers smoothing that a large count seems to guarantee is void and the effective count collapses toward one — making the aggregate behave as a single bet the moment the coupling is activated. The portable pieces are abstract — a large notional count, a hidden correlation structure, an effective (correlation-adjusted) count read against the nominal one, and a broken independence assumption. That skeleton is genuinely substrate-portable, which is exactly why deposit concentration risk instantiates the catalog prime concentration_risk / correlated_exposure (decomposing further into correlation, the coupling of nominally independent units, and concentration, a few units dominating the aggregate), with run_dynamics/contagion supplying the self-accelerating cascade and liquidity_under_stress/funding_liquidity the asset-side complement. Unusually, the shared core carries not just the diagnosis but the diagnostic and the remedy intact — compute the correlation-adjusted effective count; diversify against the cause, not by adding more units from the coupled population — which is why the entry is best read as a worked instance of that parent. This is the core it shares, not what makes it distinctive.

What is domain-bound. Almost all the content is bank-funding furniture and none of it survives extraction intact: the depositor base and its funding role; the deposit-insurance limit and the insured/uninsured coupling split (insured balances weakly coupled and slow-draining, uninsured balances above the limit mutually informative and self-accelerating); the run on a funding base; the liquidation-at-par horizon on the asset side; the funding-liquidity-versus-solvency framing; and the prudential apparatus (Basel III LCR/NSFR, top-20 depositor lists, large-uninsured-deposit ratios, contingency funding plans). These are the worked vocabulary, the instruments, and the canonical case (SVB's venture-backed startups coupled through a few VC advisors, 94% uninsured, $42bn requested in a day against underwater long bonds) the discipline actually studies — all specific to deposit-taking against a par-illiquid asset side. The decisive test: remove the deposits, the insurance limit, and the funding-liquidity mismatch — none of the co-instance domains would call the pattern "deposit" anything — and what remains is the bare correlated-exposure structure, a looser and more general thing than this named risk.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. Deposit concentration risk's transfer is bimodal. Within banking, finance, and prudential regulation it travels intact as full mechanism: the effective-versus-nominal count construction, the law-of-large-numbers-requires-independence diagnostic, the insured/uninsured run-mode branch, the trigger-through-a-correlated-population reading, and the diversify-against-the-cause remedy all port without retuning across commercial banking, money-market funds, non-bank deposit-takers, stablecoin reserves, DeFi liquidity pools, and post-SVB supervision, because all share the funding-base-against-par-illiquid-assets substrate. Beyond bank funding the named concept does not travel — but here the boundary is unusual: what travels is not a mere analogy but the parent prime, which recurs literally as genuine co-instances in customer concentration, supplier concentration, geographic concentration, single-counterparty exposure, and cloud-vendor single-point-of-failure, each running the identical effective-count diagnostic and diversify-against-the-cause remedy. And when that bare structural lesson is needed cross-domain — compute the correlation-adjusted effective count, and diversify against the correlation-generating cause — it is already carried, in more general form, by the prime deposit concentration risk instantiates: the masked-diversification structure is concentration_risk / correlated_exposure, the cascade is run_dynamics/contagion, and the asset-side mismatch is liquidity_under_stress. The cross-domain reach belongs to that parent; "deposit concentration risk," as named, carries the insurance-limit coupling, the run on a funding base, and the liquidation-at-par horizon that stay home and should.

Relationships to Other Abstractions¶

Current abstraction Deposit Concentration Risk Domain-specific

Parents (3) — more general patterns this builds on

Deposit Concentration Risk is a kind of Funding Fragility Domain-specific

Deposit concentration risk is funding fragility caused by concentrated or correlated withdrawal-capable funding claims.
Deposit Concentration Risk is a decomposition of, conditional Apparent Variety Masks Shared Driver Prime

In the nominally broad but correlated-depositor branch, stripping banking reveals surface multiplicity coupled by a hidden driver and an effective count that collapses under stress.
Deposit Concentration Risk is a decomposition of Dependency Distribution Concentration Prime

Removing deposits, insurance, run dynamics, and asset-liquidity framing leaves dependency weight concentrated across funding providers, with fragility governed by effective rather than nominal provider count.

Hierarchy paths (12) — routes to 9 parentless roots

Deposit Concentration Risk → Funding Fragility → Coordination Problem and Equilibrium Selection → Coordination → Concurrency

Show alternative paths (11)

Not to Be Confused With¶

Credit concentration risk. The classic banking concentration hazard on the asset side: too much lending exposure to one borrower, sector, or geography, so a single default event impairs the loan book. Deposit concentration risk is the mirror on the liability/funding side: correlated withdrawal by a coupled depositor population. Both are concentration_risk instances, but on opposite sides of the balance sheet. Tell: is the coupled, correlated exposure in the loans the bank holds (credit concentration) or in the deposits that fund it (deposit concentration)?
Interest-rate / duration risk. The exposure of asset values to rate moves — SVB's long-dated bonds fell deeply underwater as rates rose. This is the asset-side mismatch the entry names as a complement, not the concept itself: deposit concentration risk is the funding-side coupling that triggers a synchronized run. Rate risk made SVB's assets unsellable at par; deposit concentration made the withdrawal arrive all at once. Tell: is the problem that assets lost value to rate moves (duration risk), or that a correlated depositor base pulled funding faster than those assets could be liquidated (deposit concentration)?
Bank run (general / Diamond–Dybvig). The self-fulfilling coordination failure in which depositors rush to withdraw because they expect others to. A classic run can strike even a dispersed, independent deposit base purely from a coordination shift. Deposit concentration risk is the specific structural amplifier — a correlated base whose effective count is near one, so a run is faster and more easily triggered. The run is the event; concentration is the coupling that primes it. Tell: is the referent the withdrawal cascade itself (bank run), or the correlation structure that collapses the effective depositor count and makes such a cascade acute (deposit concentration)?
Funding-liquidity risk (general). The broad category — the risk that a bank cannot meet obligations as funding drains faster than it can raise cash. Deposit concentration risk is one driver within it, keyed specifically to correlated depositor behavior; funding-liquidity risk also covers wholesale-funding rollover, maturity mismatch, and market-wide freezes with no concentration at all. Tell: is the frame the whole capacity-to-fund-obligations problem (funding-liquidity risk), or the specific correlated-depositor mechanism that produces a synchronized outflow (deposit concentration)?
The parent prime it instances (concentration_risk / correlated_exposure). The substrate-neutral structure — nominally many independent units coupled by a shared cause, voiding the law-of-large-numbers smoothing so the effective count collapses toward one. This carries the effective-count diagnostic and diversify-against-the-cause remedy literally to customer, supplier, geographic, and counterparty concentration as genuine co-instances. Deposit concentration risk is the bank-funding worked instance with insurance-limit coupling and run-on-a-funding-base cargo. Tell: strip the deposits, the insurance limit, and the par-illiquid asset side and what remains — correlated exposure masking notional diversification — is this parent, not deposit concentration risk. (Treated more fully in earlier sections.)

Neighborhood in Abstraction Space¶

Deposit Concentration Risk sits in a sparse region of the domain-specific corpus (74^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Monetary Mechanics & Macro Trilemmas (7 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12