One week ago, Silicon Valley Bank was the sixteenth-largest bank in the United States. Today it no longer exists as an independent functioning bank, American authorities have guaranteed all of its deposits, financial markets are scrutinising regional banks across the United States, and Switzerland has just engineered an emergency takeover of Credit Suisse by UBS.
The speed matters because banking crises are fundamentally crises of time.
A normal company can lose customers gradually. Revenue falls, management restructures, investors reassess the business and bankruptcy may eventually occur. Banks operate differently because much of their funding can disappear almost immediately. Depositors expect to withdraw money whenever they want, while banks invest much of that money in assets that cannot necessarily be converted into cash at the same speed without losses.
That maturity mismatch is one of the foundations of modern banking.
It is also one of its fundamental vulnerabilities.
SVB’s particular vulnerability came from the interaction between its depositor base and rising interest rates. The bank served large numbers of technology companies, venture-capital firms and start-ups. During the technology boom, deposits grew rapidly, and SVB invested substantial amounts in long-duration US government securities and mortgage-backed securities.
Those assets were not equivalent to reckless subprime mortgages. US Treasury securities are generally considered among the safest assets in the world in terms of default risk.
But safe from default does not mean safe from price movements.
When interest rates rise, the market value of existing fixed-rate bonds falls. A bond paying 1.5% becomes less attractive when newly issued securities offer 4%. Investors therefore demand a discount before purchasing the old bond.
For a bank intending to hold the asset until maturity, the price decline can remain largely an accounting issue.
The problem begins when the bank needs cash now.
SVB’s technology-sector customers began withdrawing deposits as venture-capital funding became more difficult and companies burned through existing cash. The bank therefore needed liquidity and sold securities whose market values had fallen substantially.
That converted unrealised losses into realised ones.
SVB then attempted to raise new capital.
Instead of reassuring depositors, the announcement revealed how vulnerable the balance sheet had become.
Depositors rushed to withdraw money.
The bank collapsed.
This is what makes bank runs so strange economically: fear can make the feared outcome happen.
Suppose I believe my bank is perfectly healthy. I leave my deposit there.
Suppose I believe everyone else is about to withdraw. Even if I personally think the bank’s underlying assets are sound, my rational response may be to withdraw first.
If enough people make the same calculation, the bank has to liquidate assets rapidly.
Prices fall.
Losses materialise.
A liquidity problem can become a solvency problem.
Expectations therefore become causal.
Banking is partly an exercise in collective confidence.
This is why governments created deposit insurance.
If ordinary depositors know their money is guaranteed up to a certain threshold, they have much less reason to rush towards the bank whenever rumours appear.
The United States normally guarantees deposits up to $250,000 per depositor, per insured bank, for each ownership category.
SVB created a problem because an unusually large share of its deposits exceeded that limit.
Its customers were not primarily ordinary households holding salaries and savings.
They included companies holding millions of dollars needed for payroll and operations.
When SVB failed, those uninsured depositors theoretically faced substantial losses.
The government then made an extraordinary decision.
Federal regulators announced that all depositors would have access to all of their money, including balances above the normal insurance ceiling.
Shareholders were not protected.
Management was removed.
The bank itself failed.
Yet depositors were made whole.
Is that a bailout?
The answer depends on what we mean by the word.
If a bailout means protecting the bank’s owners from the consequences of failure, this is importantly different from many interventions during 2008. SVB shareholders can lose their investment.
If a bailout means government intervention preventing private actors from bearing losses they otherwise would have suffered, then uninsured depositors have clearly received protection.
The distinction matters because different forms of rescue create different incentives.
Protecting shareholders creates enormous moral hazard. Investors receive profits when risks succeed and taxpayers absorb losses when risks fail.
Protecting depositors creates a subtler version.
A company holding $20 million in one bank theoretically has an incentive to investigate whether that bank is safe or diversify its deposits.
If the government demonstrates that sufficiently large bank failures will result in every depositor being protected regardless of the formal insurance limit, that incentive weakens.
Why monitor the bank if the government will ultimately protect you?
That is moral hazard: insurance changes behaviour because people take less care when they do not bear the full cost of failure.
The concept appears everywhere.
Drivers with comprehensive insurance may behave differently from drivers bearing every repair cost themselves.
Governments protected by international lenders may borrow more aggressively.
Banks expecting central-bank rescues may take more risk.
Insurance solves one problem while creating another.
The solution is rarely abolishing insurance.
It is designing insurance so the protected actor still bears enough consequences to behave carefully.
Banking makes that extraordinarily difficult because the actors we want to discipline are connected to actors we want to protect.
Suppose regulators had refused to protect SVB’s uninsured depositors.
A venture capitalist with $10 million in the bank might deserve the loss if they knowingly accepted concentration risk.
But a start-up with $3 million needed for payroll may suddenly become unable to pay employees despite operating a perfectly viable business.
Workers who had no role in choosing the bank lose salaries.
Suppliers go unpaid.
Companies fail.
The cost spreads beyond the people responsible for the original risk.
That is the externality that makes banking regulation different from ordinary bankruptcy.
If a badly managed restaurant fails, neighbouring restaurants do not normally collapse because customers suddenly conclude restaurants as a category are unsafe places to store money.
Banks are interconnected through confidence.
One failure changes beliefs about another.
Depositors begin asking whether their own bank holds similar securities.
They withdraw money.
Banks sell assets.
Losses spread.
The government therefore faces a collective-action problem.
Punishing one bank’s depositors may improve long-term market discipline while creating short-term contagion capable of destabilising institutions that were previously healthy.
That is why promises never to intervene are difficult to make credible.
Before a crisis, policymakers want everyone to believe:
If you take excessive risk, you will bear the loss.
During the crisis, policymakers confront another question:
What if making you bear the loss causes everyone else to panic?
The optimal policy changes once systemic risk appears.
Markets understand this.
That knowledge changes behaviour before the crisis.
This is the time-consistency problem of financial regulation.
Governments want to promise one thing ex ante and rationally choose another ex post.
Investors anticipate the reversal.
The original promise loses credibility.
This is why regulation has to operate before failure rather than relying entirely on market discipline afterwards.
If governments know they may be forced to protect depositors during crises, they need to reduce the probability that banks reach the crisis.
Capital requirements.
Liquidity requirements.
Stress tests.
Interest-rate risk management.
Supervision.
Diversification.
These can look like bureaucratic costs during normal periods.
They are the price of reducing the probability of socialised losses later.
Again, prevention looks inefficient until catastrophe occurs.
SVB’s failure also demonstrates how monetary policy produces effects beyond the variable central banks are targeting.
The Federal Reserve has raised interest rates rapidly because American inflation became extremely high.
Higher rates reduce demand and should eventually help reduce inflation.
But interest rates are also prices connecting almost every part of the financial system.
When rates rise, mortgage costs change.
Corporate borrowing changes.
Asset valuations change.
Government debt-servicing costs change.
And the market value of long-duration bonds falls.
That means monetary tightening can expose vulnerabilities accumulated during the previous era of extremely low interest rates.
This does not mean the Federal Reserve caused SVB’s failure in the sense that it should have kept rates low to protect badly positioned banks.
Price stability remains part of the Fed’s mandate.
Banks are supposed to manage interest-rate risk.
The broader lesson is that policy transmission is nonlinear.
Raise rates gradually and the system adjusts.
Raise them rapidly after years near zero and previously hidden balance-sheet risks can become visible suddenly.
The financial system built itself around one price of money.
The price changed.
Some business models were more dependent on the old price than anyone realised.
Credit Suisse demonstrates why the story is already larger than one Californian bank.
Its problems are not identical to SVB’s. Credit Suisse has suffered years of scandals, losses, management changes and declining investor confidence. The bank entered this week already weak.
Then the global banking environment changed.
Investors became more sensitive to fragility.
Credit Suisse shares fell dramatically after its largest shareholder said it could not provide additional capital. The Swiss National Bank offered substantial liquidity support, but confidence continued deteriorating. Swiss authorities have now brokered an emergency acquisition by UBS.
Again, confidence operates as a transmission mechanism.
SVB did not create Credit Suisse’s underlying problems.
It changed how markets priced them.
This is how crises spread without every institution sharing the same cause.
A forest can contain several dry trees for different reasons.
One spark changes the risk facing all of them.
Financial contagion works similarly.
The question therefore becomes whether regulators should rescue institutions that are systemically important even when those institutions behaved badly.
This is where the phrase “too big to fail” becomes politically toxic.
If a bank becomes sufficiently large that government cannot allow it to collapse, the bank receives an implicit subsidy.
Creditors lend more cheaply because they expect rescue.
Management can take greater risks.
The bank becomes even larger.
The implicit guarantee strengthens.
Eventually private profit operates on top of public insurance.
That is a terrible equilibrium.
One solution is preventing banks from becoming systemically important.
Another is making failure possible without systemic collapse through resolution regimes that impose losses on shareholders and creditors while maintaining critical functions.
The post-2008 regulatory architecture attempted precisely this.
Today’s events are a test of how much progress was made.
The principle should be straightforward even if implementation is not:
protect the financial system, not necessarily the financial institution.
Those are different objectives.
A government may need deposits to remain accessible, payments to continue and credit markets to function.
That does not mean executives, shareholders and every creditor should be protected.
The more precisely intervention can preserve systemic functions while imposing losses on risk-takers, the smaller the moral hazard.
This is why the details of rescue matter more than the headline “government intervenes”.
Financial crises are distributional events.
Somebody absorbs the loss.
Shareholders.
Bondholders.
Depositors.
Other banks.
Central banks.
Governments.
Ultimately taxpayers.
The loss cannot be made to disappear simply through financial engineering.
Policy determines where it lands.
That makes banking crises political.
During 2008, many citizens concluded that governments protected financial institutions while ordinary households bore unemployment, foreclosures and austerity afterwards.
That perception damaged political trust across several democracies and contributed to populist movements on both left and right.
Policymakers therefore enter the current episode with a legitimacy problem inherited from the last crisis.
They need to stabilise banks without appearing to recreate a system in which wealthy financial actors keep gains while society absorbs losses.
Protecting SVB depositors while allowing shareholders to lose everything is partly an attempt to draw that distinction.
Whether citizens find it convincing will depend on who ultimately bears the costs.
There is another issue: concentration risk.
SVB’s customer base was heavily concentrated in the technology and venture-capital ecosystem. Depositors therefore behaved more similarly than a diversified retail customer base would.
They knew one another.
They communicated through the same networks.
They responded to the same venture-capital investors.
Digital banking allowed enormous transfers to occur almost instantly.
This may have changed the speed at which bank runs can happen.
A nineteenth-century bank run required people physically queuing outside a branch.
A twenty-first-century company can transfer millions of dollars from a laptop.
Information spreads through group chats and social media before regulators can organise a press conference.
Financial stability institutions designed around slower runs may therefore need to update.
Technology has reduced transaction costs.
It has also reduced the friction that once slowed panic.
Friction is usually treated as economically inefficient.
Sometimes friction provides stability.
This is another example of optimisation producing unexpected systemic consequences.
Making banking faster improves ordinary transactions.
It also makes withdrawals faster during crises.
The policy response should not be deliberately making banking inconvenient.
It should recognise that liquidity requirements and emergency communication need to operate at digital speed too.
Regulation has to evolve with the system it regulates.
The politics of deposit insurance may also change. If governments repeatedly protect uninsured deposits during systemic failures, the formal $250,000 limit becomes less credible.
Perhaps the limit should be redesigned.
One option would be broader insurance financed through higher risk-based premiums paid by banks.
Another would preserve limits for investment balances while providing stronger protection for business transaction accounts used for payroll.
Every design creates trade-offs.
Unlimited deposit insurance virtually eliminates classic bank runs but substantially weakens depositor discipline and increases the government’s contingent liability.
Limited insurance preserves some market discipline but creates incentives for runs above the threshold.
There is no free solution.
The correct design depends on deciding which actor is best positioned to monitor bank risk.
Ordinary depositors probably are not.
Regulators and professional investors may be.
If that is true, relying on depositors to discipline banks may be inefficient in the first place.
A small business should perhaps spend its time developing products rather than reading a bank’s balance sheet.
This strengthens the case for strong supervision.
But supervision creates its own moral hazard if banks assume regulators will identify every risk.
Again, redundancy matters.
Capital.
Supervision.
Market discipline.
Deposit insurance.
Resolution mechanisms.
Central-bank liquidity.
Financial stability emerges from several imperfect safeguards overlapping.
This is the same systems principle from the Greek railway disaster last week.
A safe system should not depend on one actor never making a mistake.
Banks will mismanage risk.
Regulators will miss things.
Depositors will panic.
Markets will misprice assets.
The institutional question is whether enough independent layers exist that one failure does not become catastrophe.
Financial regulation is essentially safety engineering with money.
The Credit Suisse takeover tonight is another emergency layer being activated.
Swiss authorities appear to have concluded that allowing uncertainty to continue into another trading week created unacceptable systemic risk.
UBS will absorb its historic rival under extraordinary conditions.
That may stabilise markets.
It also creates an even larger Swiss banking institution.
The immediate solution to “too big to fail” can therefore produce something even bigger.
That is the recurring irony of crisis management.
Emergency policy optimises for Monday morning.
Structural policy has to worry about the next decade.
The two objectives can conflict.
During panic, governments prioritise stopping contagion.
After panic, regulators need to address the concentration and incentives created by whatever intervention stopped it.
Failure to complete the second stage means every crisis leaves the financial system more dependent on rescue.
This is why crisis response should always contain an exit strategy.
Temporary guarantees.
Resolution plans.
Regulatory review.
Accountability for management.
Otherwise emergency intervention becomes permanent architecture accidentally.
There is also an international dimension. Banking is global while regulation remains substantially national.
Credit Suisse operates across jurisdictions.
American banks interact with global markets.
A confidence shock in California can affect a Swiss bank whose underlying problems developed independently.
Financial globalisation therefore creates another mismatch between economic integration and political authority.
No world government regulates the entire system.
National central banks and regulators have to coordinate.
This resembles climate change and pandemics in institutional structure.
The externality crosses borders.
The authority remains national.
International cooperation becomes necessary because one country’s regulatory failure can impose costs elsewhere.
This is why institutions such as the Financial Stability Board, Basel Committee and Bank for International Settlements matter even though almost nobody outside finance pays attention to them.
Boring institutions become visible when they fail.
Again, governance is often most important where politics is least visible.
As of tonight, there is no reason to conclude that the world is entering another 2008-style financial crisis.
Banks are generally better capitalised than before the global financial crisis, and the underlying asset problem is different.
It would be equally irresponsible to declare the danger over.
Confidence is dynamic.
The relevant question is whether regulators have convinced depositors and investors that vulnerabilities are contained without convincing banks that risk no longer matters because governments will always intervene.
That is the impossible balance.
Too little reassurance produces panic.
Too much reassurance produces moral hazard.
Financial stability requires people to believe two things simultaneously:
your money is safe enough that you do not need to run, but your investment is risky enough that you still need to care where you put it.
Policy has to decide which actors receive which message.
Depositors may need protection.
Shareholders need exposure to loss.
Executives need accountability.
Bondholders need credible risk.
Regulators need authority.
Taxpayers need protection.
No single guarantee can accomplish all of that.
Perhaps this is why banking crises return despite centuries of experience.
The underlying institution performs two socially valuable functions that naturally conflict. We want banks to transform short-term deposits into long-term investment because that finances economic activity.
We also want deposits to remain available immediately.
The maturity mismatch creates value.
It creates fragility.
Government insurance reduces the fragility.
It weakens discipline.
Regulation restores discipline.
Regulation creates compliance costs.
Financial innovation finds ways around regulation.
Then another vulnerability appears.
There is no final equilibrium.
Only institutions adapting faster than risk migrates.
SVB failed because a particular balance sheet became vulnerable to rising rates and concentrated withdrawals.
Credit Suisse arrived at crisis through a different history.
The common factor is confidence.
Governments have spent this week trying to preserve it.
That may be necessary.
But every rescue writes part of the incentive structure for the next crisis.
The question is not simply whether policymakers saved the financial system this weekend.
It is what financial institutions learn from being saved.
If the lesson is that depositors are protected while owners and
