Why the Conditions Behind SVB’s Collapse Still Matter

When Silicon Valley Bank (SVB) collapsed three years ago this month, many were quick to dismiss it as a one-off … a niche institution with an unusually concentrated client base, brought down by a sudden loss of confidence. This explanation was comforting … and incomplete.

SVB did not fail because of a single bad decision or an unexpected shock. It failed because its long‑standing structural vulnerabilities were exposed all at once. Those vulnerabilities included excessive exposure to interest rate risks, fragile liquidity, siloed risk management and overreliance on historical patterns and trends.

These vulnerabilities were not unique to SVB; in many cases they remain embedded in today’s banking system. Thus, the most relevant question is not whether SVB was an outlier, but whether banks today have meaningfully addressed the conditions that made its collapse possible.

A decade of assumptions meets a new regime

For more than a decade, banks operated in an environment of exceptionally low interest rates. For many regional and community banks, generating acceptable returns without taking on duration (or interest rate) risk was difficult. Asset‑liability mismatches were not accidental; they were structural responses to the rate environment.

When interest rates rose sharply, those assumptions were tested. At SVB, unrealized losses tied to rate movements were likely manageable as long as depositor confidence held. Once that confidence broke, those losses quickly became destabilizing.

Crucially, that dynamic was not specific to one institution. While unrealized losses across the system have declined from their peak, they are still substantial, standing at $306 billion as of Q4 2025. Inflation has proven stickier than markets initially expected, and there are many catalysts that may lead interest rates to remain elevated for longer than many models had assumed. That means maturity mismatches still matter, even if they are less visible today.

At the same time, concentrated deposits remain common, and in a digital environment, they are increasingly mobile. The assumption that depositors will behave tomorrow like they did in the past is becoming less and less reliable.

Risks do not happen in isolation

One of the clearest lessons from SVB is how quickly different forms of risk can cascade. What began as an interest rate problem rapidly became a credit issue and then a liquidity crisis.

Yet many institutions still manage these risks separately. Credit, liquidity, interest rate and capital risks are frequently measured using different systems, reported on different timelines and governed by different teams. This fragmentation can appear manageable when things are calm, but it becomes dangerous during periods of stress.

Due to slow escalation paths and delayed reporting, a bank can be left flying blind at precisely the moment when visibility is needed most. By the time decision makers see the full picture, market confidence may already have disappeared.

This is why integrated balance sheet management is more than just creating better predictions. It is about surfacing interconnected risks, understanding the assumptions embedded in the models, and having a clear picture of how applicable those assumptions are to market conditions.

Many common balance sheet dynamics, from prepayments to deposit sensitivity, are quantified using behavioral models. When the environment shifts, models calibrated to the past can become liabilities.

To respond to market stress, banks need to reforecast more frequently and with clear attribution, use a wide spectrum of scenarios that go beyond the typical supervisory ones (base, adverse and severely adverse), and perform rigorous and repeated backtesting.

Deposit behavior has entered a new regime

SVB also demonstrated how modern bank runs differ from those of previous decades. Depositors no longer queue outside branches. Confidence can evaporate digitally, amplified by social media and instant communication within tightly connected networks.

This creates a difficult trade‑off. Transparency and open discussion of risk are critical for good governance. At the same time, public scrutiny of interest rate or liquidity exposure can accelerate the erosion of confidence. Traditional deposit models, which rely heavily on historical behavior, are poorly suited to capturing these dynamics.

The underlying reality has not changed: banking is based on trust. Capital and liquidity buffers matter, but they are not always sufficient once confidence is lost. This places greater emphasis on governance, clarity of risk disclosures and the ability to quickly assess the situation and act on it.

Regulation helps, but can be a double-edged sword

Regulatory reforms introduced after SVB’s collapse have strengthened oversight in some areas. These measures are important, but regulation alone cannot ensure resilience.

There is an active debate over whether bank capital and leverage constraints are limiting dealer intermediation capacity and thereby weighing on U.S. Treasury market liquidity. At the same time, Treasury market resilience is a legitimate policy concern. Official reviews of the March 2023 episode found that market functioning deteriorated at exceptional speed. And subsequent analysis has shown that Treasury market liquidity can again weaken sharply when volatility rises or dealer balance sheet capacity is strained.

The challenge lies in how those concerns are addressed. Some studies suggest that simple leverage has historically been more predictive of systemic risk than risk‑weighted capital measures, which can create incentives to optimize for regulatory models rather than underlying exposures. Shifting the balance too far toward risk‑based capital can have unintended consequences.

At the same time, regulatory complexity carries its own risks. In some jurisdictions, the volume of reporting requirements is so large that it strains both banks and regulators. When institutions devote disproportionate resources to producing data and reports rather than analyzing them, risk management suffers. Furthermore, these costs ultimately flow through to customers and the broader economy.

A system increasingly defined by divergence

Perhaps the most underappreciated trend in banking today is divergence. Large global systemically important banks (G-SIBs) generally have the scale, diversification and earnings capacity to invest heavily in technology, controls and operating infrastructure, including AI and automation. There is some evidence that these investments are improving resilience at least for some of the largest firms, as seen in sustained high technology spending, bank-specific control and transformation programs, and continued strong stress-test performance across the largest U.S. banks.

Smaller regional and community banks face a different reality. They are subject to many of the same regulatory requirements but lack the scale to absorb the cost. For some, even managing interest rate risk at a granular level can be challenging. Scenario analysis may be performed quarterly or annually rather than continuously, despite greater vulnerability to funding shocks.

This divergence mirrors broader economic patterns. Aggregate indicators suggest strength, but averages often obscure stress in specific segments. For example, most wealth creation in the U.S. over the past decade has been driven by asset ownership (i.e., stocks, homes) rather than wage growth. And since most Americans do not own stock, that wealth creation has been very unevenly distributed. The same principle applies to banking: system‑level stability does not guarantee institution‑level strength and resilience.

Integration and solid data foundation is the path to resilience

Avoiding the next SVB-like crisis is not about predicting the precise trigger. It is about building the ability to see across the balance sheet clearly, consistently and quickly.

That requires integrating data, models and governance so that risk signals are coherent and actionable. It requires breaking down organizational silos that delay recognition of emerging threats. And it requires acknowledging that advanced analytics amplify whatever foundations they are built on. Without strong data quality, transparency and governance, automation can accelerate failure just as easily as it can improve decision making.

Silicon Valley Bank should not be remembered as a historical anomaly. It is a reminder of how quickly interconnected risks can overwhelm institutions that underestimate complexity and speed. The conditions that contributed to its collapse have not disappeared. They have evolved and banks that recognize that reality and act on it will be better positioned for what comes next.