Speech

Imagining Tomorrow's Financial Stability Challenges

June 6, 2024

Thank you, Tobias, and good morning, everyone—you are all most welcome!

On behalf of the IMF and our esteemed co‑organizers at the U.S. Federal Reserve and the World Bank, I welcome you to our 23rd annual conference on policy challenges for the financial sector.

For 23 years, this conference has promoted deep thinking on banking supervision, financial sector regulation, bank resolution, and financial stability more broadly. Above all, it has promoted cooperation.

Before I dive into my remarks, let me note up-front that I see you—the international regulatory and supervisory community—as a model of cooperation. You talk to each other. You work together. You meet regularly in conferences such as this one.

Three words from me: keep-it-up!

Here at the Fund, we of course recognize that effective banking supervision is vital to financial stability, which in turn is vital to sustainable economic growth. We are talking about productivity, prosperity, human progress—these are not small things! And yet, without you—the guardians of financial stability—we know how easily economic growth can fall into the canyon of economic despair. You are the ones who hold the line!

Yours is a serious and consequential duty. You can never let down your guard. The world around you changes. Then it changes again. You need to stay ahead of it—and you do.

***

This takes me to my subject today: imagining the financial stability challenges of tomorrow.

Why do I choose this topic?

Because the changes all around us are unfolding at a dramatic pace—much faster than in the past, and probably much faster than we fully realize.

Changes are being driven by many factors—from economic developments to geopolitics to climate—and by the force of ever-accelerating technological advancement.

In finance, blockchain, crypto, tokenization, and central bank digital currencies are healthy challengers to traditional demand-deposit banking.

And, of course, changes are driven by artificial intelligence: “generative AI”—not totally new, yet something that is now poised for breakthrough. Something with true transformative potential. These forces demand our attention.

They pose multi-faceted challenges. On the one hand, they will take our 21st century financial system to a new frontier full of opportunity. On the other hand, we can be sure they will bring new risks.

At a time when we face the grim prospect of the lowest global growth in decades, new technologies—especially AI—have the potential to lift productivity in ways we have never seen before.

I look with hope to the coming revolution in economic productivity. Because it means faster wealth-creation. More people climbing out of poverty. Real gains for real people.

But, to get there, we face the tough job of making sure technological progress creates not just a more-efficient financial system that makes the best use of people’s savings, with high returns and high impacts, but also benefits that stretch far beyond the financial sector.

Access to finance, for everything from housing to green transportation to entrepreneurship. A more-inclusive financial system—cheaper and faster payments, especially across borders, benefiting vulnerable populations that rely on remittances from hard-working relatives abroad. Vital basis points shaved off corporate financing costs.

***

So, how will digitalization, fintech, and AI change the supervisory world?

Basically, I see us heading to a financial system with more market liquidity. More assets tradable, more assets traded, and—yes—more of the trading automated. Trillions of dollars of assets in a constant search for the tiniest high-frequency arbitrage opportunity. Investment offerings micro-targeted to individual customers’ risk‒return preferences after analyzing their social media postings and shopping habits. Bank risk models that sweep constantly for data anomalies to fine-tune capitalization and provisioning levels.

Yes, this could be a world of financial efficiency as never before. Costly middlemen replaced by algorithms. Better price discovery. Shock-absorbing buffers that strike a proper balance between risk mitigation and low-cost lending, adjusting in real time to help dampen financial cycles. The possibilities are huge—as are the ethical challenges, a subject for another speech, not today.

Looking at AI specifically, it is hard not to notice its vast potential. The benefits are already visible in all manner of information processing, from credit decisions to fraud detection, regulatory compliance, and enhanced supervision.

***

But you will never find the IMF getting carried away by positivity! As we look ahead, we must ask: what could go wrong? Is there a risk that technology will raise financial sector profits but not really improve savings allocation? Will it create major new risks for the real economy?

We must be alert and take history as a guide.

I remember, in the early 2000s, a financial sector that for the first time fully embraced risk-management models. It was state-of-the-art, we were told. It all seemed logical and sophisticated. Until it all went wrong.

In 2007 and 2008, how quickly we discovered that not all of the models had been properly calibrated and properly understood—that in some cases they were misused to game capital requirements. That the famous originate-to-distribute paradigm had not truly offloaded credit risk because of the backstops banks promised to their off-balance-sheet conduits. That the magic of collateralized debt obligations—the famous CDOs—was often just old wine in a new bottle, misleading investors.

Remember how quickly wholesale funding evaporated when everyone panicked all at once—panicking that somewhere deep within the fancy pools of private collateral were buried some very bad lemons?!

It is a cautionary tale—of how damaging financial innovation can be. A global collapse of finance, including trade finance. One investment bank bringing down the world economy. The deepest recession since the 1930s. Real human hardship.

***

Now, let us pause for a moment to consider what could go wrong with the financial innovations underway today.

Obviously, I can start with the ever-present risk of data loss, whether by systems breakdown or cyber-attack. “Cloud computing” does not really sit in a cloud—ultimately it sits in physical data warehouses somewhere! We are all aware of the growing ability of hackers to pursue ever-more-sophisticated data theft and fraud. I need not tell you that, when the data goes, instability arrives.

Indeed, as our latest Global Financial Stability Report tells us, almost one-fifth of the reported cyber incidents in the past two decades have affected the financial sector.

Potentially, a cyber incident at a financial institution or critical infrastructural node could generate systemic financial instability in ratcheting steps:

First, there might be limited substitutes for the services lost—think of a big payment system going offline.

Second, a loss of confidence could follow, triggering a funding run.

Third, interconnectedness could lead to broader contagion.

Consider the ripples if one cloud provider or settlement system that serves many banks is compromised. Bad news!

Turning back to AI specifically, we must recognize that, along with its vast promise, it will also bring new sources of risk. New transmission channels too. If we were to be hit by a “tail event,” AI could quickly amplify and spread the shock throughout the financial system. Three concerns:

One, AI may lead to new third-party dependencies and the emergence of new systemically important players and service providers outside the regulatory perimeter.

Two, the adoption of machine learning further increases cybersecurity risks like the ability to manipulate data at great speed and scale, for fraud.

Three, AI may also be vulnerable to mechanical groupthink. It is not hard to see how the widespread use of algorithmic trading, directed by robots trained on historical data, could result in much greater homogeneity in risk assessments, credit decisions, and trading strategies—creating a risk of abrupt reversals and contagious fire sales when unprecedented events occur.

***

Bottom line: we may find ourselves facing many new types of financial stability risks. How should we manage them?

One new priority for oversight, certainly, will be to ensure that outsourcing arrangements are properly managed. We will need to mitigate systemic operational and cybersecurity risks. We will need to prevent too much concentration, ensuring layered data back-ups and redundancy. We will need to look beyond the traditional supervisory perimeter.

Another part of the solution, no doubt, will rest with technology itself—using machines to police machines. Supervisors could employ self-educating algorithms to monitor the flow of transactions, constantly scanning for evidence of mechanical or human groupthink, as a trigger for supervisory action.

There will be many new priorities. A broader perimeter for regulation. More transparency on new AI-driven leverage. New ways to ensure originators of new financial instruments have real “skin in the game.” New forms of circuit breakers. And other guardrails against procyclicality and herding.

Recognizing that financial innovation knows no borders, and that international cooperation is therefore vital, we at the IMF favor a three-pronged approach to help avoid, prepare for, and respond to the financial crises of the future:

First prong: improved diagnostics—which means better data as well as top-quality stress testing. Data-gathering: easier said than done, especially when we speak of data on nonbank financial institutions. Nonbanks almost by definition are subject to lighter reporting requirements than banks—we can ask for more reporting, but there are limits. As for stress testing, we need to think ahead creatively to future risks, including cyber threats.

Second prong: strengthened regulation and supervision—including full and faithful implementation of Basel III and lessons learned from the banking turmoil last year, here in the United States and in Switzerland. Selected upgrades are needed in the nonbank space too. How best to handle nonbanks? Not always straightforward. We know they have grown massively and come with many different business models and risks. For pension funds, investment funds, insurers, other institutional investors managing retail savings—strict rules, strict limits on leverage. For hedge funds, private equity, and high-net-worth individuals—a lighter touch, letting sophisticated investors take losses when they get things wrong, allowing market discipline to work.

Third and last prong: robust crisis management arrangements—close cooperation between supervisors and resolution authorities, at home and across borders. Flexible resolution regimes and resolution plans for banks and holding companies. Robust deposit insurance systems. Well-functioning lender-of-last resort facilities—including liquidity backstops for institutions coming out of resolution. Again, many complex questions. Standing liquidity facilities for selected categories of nonbanks? Perhaps, but striking a very careful balance between financial stability and moral hazard. Vigilance. Above all, readiness to act fast when trouble strikes!

Many difficult questions. Many leading experts engaged in serious debate. And in all these areas, the IMF and World Bank helping, including through our deep dives under the Financial Sector Assessment Program.

***

To conclude: we are entering a new era in global finance, one that holds huge promise—but let us not assume all will be benign. Through our actions, our vigilance, we must ensure that it will be so. Our financial destiny is ours to shape—and we must start now.

Before I step off this stage, I want to respectfully pause for one moment to note that today is the 80th anniversary of D‑Day—a day that has become synonymous with action. Today and every day, as you, the supervisors, work together to protect our hard-won stability and progress, let action be your motto.

Thank you—may your discussions here be fruitful!