Most people who read about banks read about one bank at a time. The earnings release for a regional. The stress-test result for a money-center. A failure write-up in the FT. What's harder to build is intuition for the system-level behavior. A small rise in charge-offs can compress capital, drive deposit flight, and feed back into more failures. A policy-rate move works its way through funding-cost beta and loan-repricing speed before it shows up in net interest margin and return on assets. Those are loops, not single numbers, and reading about them in prose is a slow way to feel how they actually behave.
The U.S. Banking Industry Simulator is a small attempt to close that gap. It's a free site, no login, that runs three system-dynamics models of the U.S. banking industry in the browser. You move the sliders, the time series redraw, and the second-order effects (the ones that are hard to hold in your head) become visible.
The three scenarios
Each scenario is its own model, with its own state variables, its own parameters, and a short panel of historical context next to the chart.
Bank Failure Cascade. Credit stress to capital erosion to contagion. The sliders are the initial charge-off rate, a stress-amplification factor, deposit-flight sensitivity, the starting Tier 1 capital ratio (FDIC historical range 8–18%), and the regulatory capital floor at which a bank is resolved. The model lets you see how a charge-off shock that looks manageable at the loan-level can still produce a wave of failures once capital starts to draw down faster than it can be rebuilt and depositors start to move.
Net Interest Margin Squeeze. Rate dynamics to NIM compression to profitability feedback. The sliders are a policy-rate shock (Fed Funds, plus or minus), funding-cost beta, loan-repricing speed, loan-demand elasticity, and the starting NIM (FDIC historical range 2.5–6.0%). The point of the model is to make repricing asymmetry visible: when funding costs reprice faster than loan yields, a rate move that reads neutral on day one shows up later as a NIM compression that drags ROA along with it.
Industry Consolidation. Merger dynamics over the long arc that took U.S. banking from more than 18,000 institutions in 1984 to about 4,400 today. The sliders are annual merger-activity rate, annual involuntary-closure rate, new-charter formation rate, a scale-economy pressure term that captures how strongly cost advantages push the count down, and a structural floor for the minimum viable number of institutions. The chart runs forward from the current count and lets you see what range of institutional populations is plausible over the next decade or two under different assumptions about M&A appetite, charter activity, and the strength of scale economies.
The parameters in each model are calibrated to FDIC historical ranges rather than picked to produce a pre-baked result. The starting values are reasonable for the current environment; the slider bounds are reasonable for things that have actually happened.
How to read it, and what it is not
The simulator is built for intuition, not for prediction. It is not a forecast and it is not a regulator-grade stress test. It is a small set of differential equations with a handful of legible parameters, designed to make the feedback loops in each scenario easy to see and easy to play with. The numbers it produces are illustrative of how the loops behave under the assumptions you choose; they are not point estimates of where the industry will be next quarter. Where the real banking system has dozens of additional moving parts (loss-given-default by loan type, FHLB advance dynamics, deposit-insurance behavior, off-balance-sheet exposures), the simulator deliberately does not.
What it earns its keep on is the question "what does this loop actually do." A reader who has spent ten minutes pushing the charge-off rate up and the capital buffer down in the Failure Cascade model comes away with a sharper sense of why bank failures cluster in time than a reader who has only read about them. That kind of intuition is the precondition for a useful conversation with someone (a CFO, a board member, a journalist, a student) about a specific bank in a specific quarter.
We built this for the same audience as our other free tools: executives whose treasury sits inside the banking system and want to think more clearly about what could move it, board members and policy staff who want a model they can poke at without booting up a Bloomberg terminal, researchers and students who want a clean way into the dynamics of the industry, and the curious reader who has wondered why the count of U.S. banks keeps falling year after year and wants to see the merger and failure terms separately. None of it asks for a background in banking; the parameters are explained in plain language next to each slider.
If you want the raw data behind any of the dynamics the simulator illustrates, the Banking Industry Data Explorer is the companion piece. It pulls the same FDIC call report data on every active U.S. institution and lets you read the historical record directly.
The U.S. Banking Industry Simulator is free to use at /insights/bank-simulator/. If your team has a use we haven't built for yet, or a scenario the public dataset would support that we haven't modeled, get in touch.