Deposits are the lifeblood of retail and community banking, yet they are fundamentally unstable. Unlike wholesale funding, which is contractual and predictable, deposits rest on the belief that your bank is safe and that alternatives (money market funds, short-term Treasuries, other banks) aren't more attractive. When either assumption breaks, deposits walk out the door—fast.
This module frames the core problem: deposits are not fixed liabilities. They carry optionality. Depositors hold call options on your balance sheet. When rates rise, when your bank faces stress, or when yield-seeking opportunities appear elsewhere, they exercise those options. Your job as an ALM practitioner is to model that behavior with enough precision to manage liquidity and earnings risk.
Why this matters: A Treasury yield that spikes 200 basis points doesn't just affect your asset repricing—it triggers deposit flight. The 2022–2023 rate cycle demonstrated this starkly. SVB collapsed not because it couldn't meet withdrawal demands at face value, but because once confidence eroded, deposits evaporated faster than management could respond. Regional banks lost 5–8% of deposits in weeks. Even strong banks with low loan losses and positive earnings faced deposit pressure because depositors rationally migrated to risk-free alternatives paying 5% overnight.
This is not market risk alone—it is liquidity risk and behavioral risk combined.
Core Concepts
Deposit optionality: Depositors can withdraw or transfer funds with minimal notice (demand deposits) or short notice (savings accounts). They do this when the opportunity cost of staying rises. That opportunity cost depends on three factors: (1) the rate they could earn elsewhere, (2) their perception of your bank's safety, and (3) the friction cost of moving deposits (search time, relationship value, switching costs).
Core vs. non-core deposits: Core deposits are those tied to relationships—payroll, direct deposit setup, linked checking and savings, credit product borrowing. They're stickier. Non-core deposits are price-sensitive rate-seekers: brokered deposits, deposits from other banks, large institutional cash. During rate normalization, core deposits deplete last; non-core deposits leave first.
Relationship to other ALM modules: Deposit behavior feeds directly into liquidity modeling (Module 29), funding strategy (Module 28), and pricing decisions (Module 25). If you don't model deposit decay correctly, your liquidity buffer is a fiction. If you don't understand deposit betas (the sensitivity of your deposit rates to market rates), your earnings hedge is incomplete.
What You'll Learn
This module introduces the behavioral and economic foundations of deposit modeling. You'll understand why simple run-off assumptions are dangerous, how to build a deposit beta curve, and why the 2022–2023 cycle exposed so many banks' blind spots. By the end, you'll see deposits not as a static liability class, but as a dynamic funding source that requires constant monitoring and modeling.
The Problem: Why Simple Assumptions Fail
Traditional ALM models often treated deposits like bonds with a constant duration and a stable rate spread. This assumption held through 2015–2021 because short rates were effectively zero, and long rates were suppressed by Fed policy. Depositors had nowhere else to earn yield, so they stayed. Deposit betas were low (maybe 20–30%), and run-off was predictable. Banks could forecast deposit balances with confidence.
Then 2022 happened.
When the Fed raised rates 400 basis points in nine months, depositors faced a genuine choice: earn 4.25% on a money market fund or 0.5% at their bank. The economic value of staying plummeted. And unlike bond investors, depositors can act on that value instantly, with no transaction costs, no capital loss, no tax consequence.
SVB's collapse in March 2023 was the punctuation mark, but the story began earlier. In mid-2022, regional banks lost deposits for the first time since 2008. Banks like Huntington, KeyCorp, and Comerica reported deposit outflows of 3–5% over a few quarters—far more volatile than historical patterns. By early 2023, even strong banks like JPMorgan (which had deposit growth) faced intra-month swings of 2–3% of total deposits as large depositors rotated into Money Market Funds (MMMFs).
The mechanical driver: every 1% rise in risk-free rates created a ~4–5% opportunity cost gap between what depositors earned and what they could earn. That gap was too large to ignore.
The Behavioral Foundation
Deposit behavior is fundamentally a rational-choice problem with stickiness parameters. A depositor stays if:
Value of staying ≥ Cost of switching + Value of bank relationship
Where:
- Value of staying = Expected return on deposits at your bank (interest rate + safety premium + service convenience)
- Cost of switching = Time to move funds, risk of mistakes, tax complexity, relationships with branch staff
- Value of bank relationship = Access to credit, convenience of linked checking/savings, employer direct deposit setup, social/psychological commitment
In 2022–2023, the switching calculus broke. For a $1M deposit earning 0.25% at a regional bank, the cost to switch (near zero) was negligible compared to the opportunity cost (4% annual difference = $40k). The relationship value had to be $40k or more to keep that deposit. For most institutions, it wasn't.
This is why core deposits held better than non-core. A small business owner with a payroll account, a commercial relationship, and embedded treasury systems faces real switching costs. A rate-seeking corporate treasurer with deposits at five banks faces none.
Modeling the Behavior: Beta, Decay, and Threshold Rates
Practitioners model deposit behavior through three lenses:
1. Deposit Beta
The elasticity of your deposit rate to the risk-free benchmark (fed funds, SOFR, T-bills). A beta of 0.40 means: if the fed funds rate rises 100 bps, your deposit cost of funds rises 40 bps (on average). Betas vary by product (savings accounts 50%, non-interest checking 10%, money market 80%) and by depositor type (retail 40%, commercial 70%, institutional 90%).
Historical betas from 2015–2021 looked like:
- Savings: 25–35%
- NOW accounts: 15–25%
- MMA: 40–60%
- Retail CD: 60–80%
- Commercial deposit: 70–90%
During the 2022–2024 cycle, betas blew higher:
- Savings: 70–85%
- NOW: 60–75%
- MMA: 95–110%
- Commercial: 95–110%
The elevated betas reflect two forces: (1) genuine demand for higher yields as alternatives appeared, (2) competition among banks trying to hold deposits by raising rates faster than historical patterns.
2. Deposit Run-Off and Decay
Not all deposits reprice; some leave. When opportunity costs rise fast (as in 2022), some portion of deposits decays into other institutions. Historical run-off was ~2–3% annually. In 2022–2023, run-off hit 5–8% in quarters when rates rose fastest.
You model run-off as a time-dependent and rate-dependent function:
Run-off Rate = Base Decay + Rate-Sensitivity Function
Base decay might be 1% per quarter (natural attrition). Rate sensitivity adds to that: for every 100 bps of opportunity cost gap, add 0.5–1.5% per quarter (varies by product and bank franchise strength).
Example: BankX has $10B in savings deposits with a 0.30 beta and a 1% base decay rate.
- Fed funds at 0.25%, BankX paying 0.10% on savings
- Risk-free alternative (MMMF): 5.25%
- Opportunity cost gap: 5.15%
- Implied rate-sensitive outflow: 5.15% × 0.005 (elasticity) = 2.6% per quarter
- Total run-off: 1% (base) + 2.6% (rate-sensitive) = 3.6% per quarter
This simple formula captured the 2022–2023 reality better than historical models.
3. Threshold Rates and Non-Linear Behavior
Deposit run-off is not linear. Below a certain opportunity-cost threshold (maybe 150 bps), outflows stay mild—relationships hold, switching costs bind. Above that threshold, behavior shifts. You see acceleration. Some banks model a discontinuous jump when spreads exceed a trigger.
Example: A bank observes that when the T-bill/deposit-rate spread exceeds 200 bps, deposit decay accelerates from 0.5% to 2% per month. This is a behavioral cliff, likely reflecting when depositors overcome inertia and actively rotate funds.
Real-World Data: The 2022–2023 Cycle
Looking at 10-K disclosures:
JPMorgan Chase (2023): Despite rising competitive pressure, JPM's consumer deposits grew 1% YoY (2023 vs 2022) because of relationship breadth and credit product linkage. However, this masked 6–7% outflows in non-interest-bearing categories and inflows in higher-rate MMAs. The bank's deposit rate paid increased from 0.68% (2021 avg) to 2.31% (2023 avg)—a 163 bps increase.
SVB Financial (2023 Failure): SVB's deposits fell from $198B (end 2022) to $91B (March 2023)—a 54% loss in three months. This wasn't a liquidity crisis in the traditional sense; it was a confidence crisis compounded by asset-liability mismatch (they held $117B in securities, many underwater). Once depositors learned about the unrealized loss and saw a 2% all-in rate environment, deposits fled. The bank failed not on day 1 of the crisis, but after a bank run that accelerated over 48 hours.
Huntington Bancshares (2023): Huntington's total deposits fell 5.4% in 2023. However, non-interest-bearing deposits fell 11.2%, while MMAs and savings rose 8.3%. This was deposit mix shift, not pure outflow. Average cost of deposits rose from 0.59% (2022) to 1.94% (2023).
Ally Financial (2023): Ally, a digital bank with no branch network, saw online savings accounts face direct competition from money market funds and Treasury direct purchases. Ally raised savings rates from 0.5% (early 2022) to 4.0% (late 2023), but still lost deposits. By year-end 2023, Ally's deposits had fallen 12% YoY. The bank's all-in deposit cost rose 380 bps in 12 months.
Implications for ALM
Liquidity Impact: Deposit decay forces you to plan for run-off. If you assume $10B in deposits stays, but 5% runs off quarterly when rates are high, your actual available funding is $9.5B—and shrinking. That gap must be filled with wholesale funding or asset sales.
Earnings Impact: If your deposit beta is higher than you model, your net interest margin compresses faster. A 50 bps error in deposit beta on a $100B deposit base = $50M earnings miss per 100 bps of rate move.
Valuation Impact: Deposit run-off reduces your funding base and forces higher-cost wholesale funding, which reduces ROA and ROE. Investors penalize this. During 2023, regional bank stock prices fell 30–50%, in part because deposit betas shocked higher and deposit levels fell.
Module Connections
This module is foundational to everything that follows. Module 22 dives deep into how to estimate and track deposit betas. Module 23 examines the structural deposit migration that occurred in 2022–2024 (flight to safety, flight to yield). Module 24 explains how deposit behavior feeds into NMD (non-maturity deposit) modeling for interest rate repricing and liquidity forecasting. Module 25 covers pricing strategy—how to price deposits competitively to slow decay without destroying margins. Module 28 examines how deposit forecasts drive term funding strategy and the management of the maturity wall.