Quick answer: a risk-based deposit-insurance assessment system charges insured banks different contributions according to both the size of the assessment base and the risk each bank poses to the insurance fund. A simple architecture is Assessment = Assessment Base × Risk-Adjusted Assessment Rate. The risk rate can depend on supervisory ratings, capital, asset quality, earnings, liquidity/funding, concentration and modeled loss severity. The deposit insurer then needs a second algorithm at the fund level: project premium income, investment income, insured-deposit growth, operating cost and failure losses to judge whether the insurance fund remains adequate across ordinary and stressed failure scenarios.

Deposit insurance changes depositor incentives; risk-based premiums try to stop that protection from making risky banking artificially cheap.

Canonical boundary: deposit insurance as a system versus premium mathematics

The wider eduKate estate already owns the general systems explanation in Civilisation | Deposit Insurance & Resolution: why insured deposits can reduce run incentives and how insurance interacts with resolution.

This Bukit Timah Tutor page owns a different question: how can a deposit insurer translate bank size, bank risk and fund adequacy into a repeatable contribution calculation without pretending every insured institution imposes the same expected cost?

1. The core premium equation

A generic risk-based assessment can be written:

Premium_i = Base_i × Rate(Risk_i).

The base measures the scale on which the bank is assessed. The rate measures the bank’s risk to the insurance system. A flat-premium system removes the second term’s variation; a risk-based system deliberately makes it non-constant.

The US FDIC provides a useful public worked example. Its assessment methodology states that the quarterly assessment is calculated by multiplying the institution’s assessment rate by its assessment base.

2. The assessment base is not the same thing as insured deposits

In the United States, since April 2011 the general FDIC assessment base is defined as average consolidated total assets minus average tangible equity, subject to detailed regulatory rules and permitted adjustments. In compact form:

Assessment Base ≈ Average Consolidated Total Assets − Average Tangible Equity.

That design matters because an assessment system can choose to price the bank’s broader liability-funded balance sheet rather than only the quantity of deposits formally insured. See 12 CFR 327.5 and the FDIC methodology page.

Other jurisdictions use different contribution bases, often involving covered deposits and risk-adjustment factors. The US formula is therefore an example, not a universal template.

3. Risk scoring converts many bank variables into one assessment rate

A deposit insurer can observe a vector of bank characteristics:

x = [capital, asset quality, earnings, liquidity, funding structure, concentrations, growth, supervisory assessment, …].

A statistical or scorecard model converts x into a risk score s. A rate schedule then maps s to an assessment rate r:

s = f(x), r = g(s).

The FDIC currently uses different risk-based approaches for small established institutions and for large/highly complex institutions. Its public Risk-Based Assessments page says the small-institution Financial Ratios Method is based on a statistical model estimating probability of failure over three years, while large-institution scorecards incorporate performance and loss-severity measures.

4. Probability of failure is useful, but expected fund loss needs severity too

Suppose Bank A and Bank B each have a 1% probability of failure over a chosen horizon. If Bank A would impose a S$10 million loss on the insurance fund when it fails and Bank B would impose S$500 million, equal failure probability does not imply equal expected insurance cost.

A simple expected-loss model is:

Expected Fund Loss = Probability of Failure × Loss to Fund Given Failure.

The FDIC’s large-bank framework explicitly incorporates a loss-severity score as well as a performance score. That is economically sensible: frequency and severity are separate dimensions.

5. A worked teaching premium

Suppose an illustrative institution has an assessment base of S$5 billion and the risk model assigns an annual rate of 8 basis points.

Eight basis points = 0.0008, so:

Annualised premium = 5,000,000,000 × 0.0008 = S$4,000,000.

If a safer peer receives 4 basis points on the same base, its annualised assessment is S$2 million. The difference is the price signal created by the risk model.

This is a teaching example. Actual assessment schedules, adjustments, invoicing and definitions depend on the applicable deposit-insurance system.

6. Why a flat premium can create cross-subsidy

If every bank pays the same rate regardless of risk, safer banks can subsidise institutions whose business models create greater expected insurance losses. That weakens market discipline because part of the downside is pooled while upside remains private.

This is the moral-hazard problem that risk-based contributions try to reduce. The European Banking Authority’s revised DGS contribution guidelines, applicable from July 3, 2024, explicitly link institution risk to contribution levels. IADI’s revised 2025 Core Principles provide the broader international framework for effective deposit insurance.

7. But risk-based premiums can become procyclical

The economically intuitive rule “riskier bank pays more” creates a timing problem. Bank risk often rises during recessions, exactly when profits and capital are already under pressure. If premiums jump sharply at the same time, the insurance system can add another drain to weak banks.

Design options include smoothing, score bands, gradual rate functions, fund buffers built in good times, or limits on quarter-to-quarter jumps. The trade-off is real: too much smoothing weakens risk sensitivity; too little can amplify the cycle.

8. Fund adequacy is a second-level stochastic problem

Individual-bank premiums feed a common insurance fund. Let F_t be the fund balance at time t. A simplified recurrence is:

F_t+1 = F_t + assessments + investment income − operating cost − failure losses ± recoveries/adjustments.

The deposit insurer therefore needs to model not only average annual loss, but clustered bank failures, insured-deposit growth, recovery timing and liquidity needs during payout or resolution.

9. Reserve ratio: the denominator can move even without a failure

For the FDIC Deposit Insurance Fund (DIF), the reserve ratio is the fund balance divided by estimated insured deposits. If insured deposits grow faster than the fund, the ratio can fall even when the fund does not suffer a large failure loss.

That happened during the extraordinary deposit growth of 2020 and led to a restoration plan. As a current concrete checkpoint, the FDIC’s May 27, 2026 Q1 2026 release reported a DIF reserve ratio of 1.43%. The FDIC’s published designated reserve ratio for 2026 is 2.00%. These are different concepts: one is the observed fund ratio; the other is the Board-designated long-run reserve ratio.

10. Fund stress testing needs correlated failures

Bank failures are not independent coin flips. A property crash, interest-rate shock or funding panic can weaken many institutions at once. If the fund model assumes independent failures, it can seriously understate tail losses.

A useful Monte Carlo or scenario model can include a common economic state Z:

P(Failure_i | Z), LossSeverity_i(Z).

When Z is severe, failure probabilities and loss severities can rise together. The model should also consider the possibility that recoveries take years while depositor access must be restored quickly.

11. Premium design changes bank behaviour

An assessment formula is not passive measurement. Banks can respond to it. If one balance-sheet item receives a particularly high premium penalty, banks may reduce that exposure or restructure funding. That may be intended—or it may create regulatory arbitrage.

A good assessment system therefore asks not only whether the score predicts failure, but also whether the score creates sensible incentives. This mirrors risk-adjusted loan pricing: price signals reshape the portfolio being measured.

12. Evidence polarity: what supports or challenges the assessment model?

Supporting evidence includes higher assessment scores preceding higher failure/distress rates, loss-severity estimates matching realised resolution losses, stable discrimination across cycles, and fund simulations that cover historical crisis losses with credible probability.

Contradictory evidence includes low-risk banks paying high premiums before remaining healthy, high-risk banks repeatedly failing below the model’s warning threshold, score components becoming easy to game, or fund adequacy depending on unrealistically independent failure assumptions.

13. Failure modes and counterexamples

Flat-premium moral hazard. Risky banks receive an implicit subsidy.
Failure-probability-only pricing. Expected severity to the fund is ignored.
Point-in-time procyclicality. premiums surge exactly when weak banks can least absorb them.
Size-only pricing. a large safe bank is treated as equivalent to a similarly sized fragile bank.
Score gaming. banks change reported ratios without changing underlying risk.
Independent-failure assumption. the fund model misses common shocks.
Coverage confusion. insured-deposit coverage limits are mixed with the assessment base used to charge banks.
Fund-ratio complacency. a healthy current reserve ratio is treated as proof against a cluster of future failures.

A useful counterexample is a deposit-insurance fund with a high reserve ratio immediately before a system-wide asset-price collapse. Fund adequacy depends on the distribution of future losses, not only today’s fund-to-deposit ratio.

14. Diagnostics and falsifier tests

Which variables drive the largest change in a bank’s assessment rate?
Does the model distinguish failure probability from loss severity?
How stable is the bank ranking across economic cycles?
What is the fund’s expected loss under a clustered-failure scenario?
How quickly could the fund become liquid enough for depositor payouts?
What happens if insured deposits grow 10% faster than projected?
Do premium increases create measurable risk reduction or only balance-sheet relabelling?
Which banks would the model have misclassified before past failures?

Falsifier: “Our premium system is risk-based because risky-looking banks pay more” is falsified if the risk score has little out-of-sample relationship with failure frequency, loss severity or fund cost. A complex formula is not automatically a useful risk price.

15. Alternatives and design choices

Deposit insurers can use flat premiums, risk bands, continuous score-to-rate functions, expected-loss models, supervisory-rating combinations or stock-based approaches linked to fund targets. There is no one universal contribution algorithm because legal mandates, banking structures and available data differ.

The design test is whether the system jointly supports depositor protection, credible fund capacity, fair risk differentiation and incentives that do not amplify instability.

16. Verification and update triggers

backtest bank risk scores against subsequent distress and failures;
compare predicted loss severity with actual resolution cost;
re-estimate models after major accounting or capital-rule changes;
stress common-factor bank failures rather than only independent events;
review premium procyclicality after recessions;
recalibrate fund targets after structural insured-deposit growth;
audit whether banks can game key score variables;
update the contribution model when the deposit-insurance mandate or coverage architecture changes.

Connections across the finance-and-banking algorithms lane

Bank resolution waterfalls — resolution cost is one tail outcome the insurance fund may face.
Bank capital models — capital strength is a major determinant of failure resilience.
Deposit-behaviour models — insured and uninsured deposit dynamics affect run risk and fund exposure.
eduKateSG Deposit Insurance & Resolution — canonical general explanation of the system’s public role.

Research anchors

The deeper lesson

Risk-based deposit-insurance assessment is the mathematics of pricing a shared safety net. The bank-level model asks who is more likely to fail and how costly that failure could be. The fund-level model asks whether pooled premiums and reserves survive correlated failures and changing insured deposits. The incentive model asks whether the premium formula changes bank behaviour in the intended direction. A strong system therefore does not optimise one number. It connects prediction, pricing, collective reserve capacity and behavioural response—and keeps each claim falsifiable with future failure and loss data.

Educational note: This article explains public deposit-insurance mathematics and institutional design. It is not legal advice, deposit advice, bank-selection advice or a contribution formula for any specific jurisdiction.

Small Group Tutorials

How Risk-Based Deposit-Insurance Assessment Algorithms Price Bank Risk: Failure Probability, Assessment Bases, Premium Schedules and Fund Adequacy