The Quantum Paradox: Why Big Banks Can’t Do What Small Banks Already Should

The collapse of Silicon Valley Bank (SVB) was a ghost from the past.

It’s 2008, and Donald R. van Deventer, founder and former chief executive officer of Kamakura Corporation (now a subsidiary of SAS), watches SVB’s collapse unfold, three decades after he predicted it would happen. Not SVB specifically, but exactly this type of failure — the same asset-liability mismatch that decimated 1,507 savings and loan associations in the 1980s.

“Silicon Valley Bank had 10-year treasuries, but they ignored the market value sensitivity,” van Deventer explains. “They made the same mistake the banks made 40 years ago.”

The irony goes deep. While quantum computing promises to revolutionize financial risk management, most major banks struggle to implement even basic Monte Carlo simulations — a technology that has been available for decades. Van Deventer, now managing director of the Center for Applied Quantitative Finance at SAS, reveals that the banking industry is trapped by its own success, where legacy systems worth billions become computational prisons.

The Monte Carlo mirage

Traditional wisdom suggests that large banks should lead technological adoption. Van Deventer's experience tells a different story. “Big North American banks are using legacy systems,” he notes, “and the vendors would say, we have Monte Carlo, but when the banks try to run that system, it gets slower and slower and slower until each time statement finally stops.”

Monte Carlo simulations enable an ALM team to model thousands of potential future scenarios, including shifts in interest rates, changes in customer behavior, or credit defaults. They help to stress-test a bank’s balance sheet. Today’s legacy systems might only run a few thousand scenarios over a quarterly time step. A modern system, however, could run hundreds of thousands of scenarios daily at the transaction level.

The regulatory landscape compounds this dysfunction. Van Deventer recounts a conversation with a senior person at the Federal Reserve, who explained why Monte Carlo wasn’t required in regulations: “We thought the big banks could do it, but the little banks couldn’t.” Van Deventer noted that it showed that large institutions and even regulators may be “out of touch with the reality of it.”

The reality really goes against conventional assumptions. Big banks have “spent so much money linking their application systems to one specific vendor’s database format, that the switch costs to a more modern system are really high.” Small banks face lower switching costs and “have the easiest time doing Monte Carlo.” While the SVB collapse highlights the banking sector’s vulnerability, the same legacy paralysis affects a wide range of financial institutions, from insurers to pension funds, which also struggle with long-term asset-liability mismatches.

Quantum leap or quantum prison?

This legacy system paradox extends into the quantum era. Van Deventer envisions a future where quantum computing enables daily Monte Carlo simulations at the transaction level — the gold standard for risk management. “If I’m a bank CEO…I’d be doing Monte Carlo every day,” he declares.

But quantum computing introduces its own complications. Current systems require hybrid approaches, where “there are those assets and liabilities where Monte Carlo is not necessary” alongside “really serious, complex securities, like collateralized debt obligations, where if you own one of the liabilities associated with a pool of assets, you’ve got to value every asset and all its cash flows.” This hybrid model would involve using classical computers to handle the vast, structured data sets and less complex calculations, while offloading specific, highly complex tasks to a quantum machine.

The computational demands will be staggering. Van Deventer describes simulating “the whole term structure, both of interest rates and default probabilities, for every scenario, for every time step, for every counterpart.” Traditional systems collapse under this load, where "you either have to cut down, summarize the data, or cut down a number of scenarios."

The ESG complexity

Adding complexity, banks now face mounting pressure around ESG and climate factors. These are also areas where van Deventer sees fundamental issues emerging. His focus at SAS spans quantitative finance, credit risk, asset and liability management, and portfolio management, giving him a unique perspective on these multi-dimensional challenges.

Meanwhile, climate modeling faces its own computational limitations. Van Deventer references a book analyzing U.N. climate reports: “There are about 60 climate change studies that were done, but there were only three software systems used to do the work.”

In “Unsettled: What Climate Science Tells Us, What It Doesn’t, and Why It Matters” by Steven E. Koonin, the statistical conclusion highlights the challenge: “There was so much uncertainty among the 60 studies...that he said, there isn’t the kind of proof you would normally want if you're going to ask people to spend trillions of dollars.” These computational challenges are a core theme of the book. As a former Under Secretary for Science in the Obama administration, Koonin argues that “the science” as presented to the public is often a misrepresentation of the more nuanced and uncertain data found in official scientific reports.

Beyond legacy thinking

Van Deventer’s work at SAS represents an attempt to solve these systemic problems. “What we loved about SAS as a partner,” he explains, “is with a Viya platform and the quantum research that the R&D group is working on, there’s just no limit to the computer science that we can bring to bear for those clients who want everything.”

The solution involves a scalable architecture that can “drop individual aspects of the calculation into that risk engine platform. So when the client finally says, Okay, I’m ready to do everything, they don’t need a new platform.”

This approach addresses the fundamental contradiction in modern banking: institutions are desperate for better risk management but trapped by sunk costs and regulatory inertia. “Most ALM teams are desperately trying to improve what they have, and it will be quite a while until they ask the CEO for a budget for a quantum computer.”

The SVB collapse crystallizes these tensions. Despite sophisticated risk models and regulatory oversight, the bank failed because of basic asset-liability mismatches. These were problems that a proper Monte Carlo simulation could have flagged years earlier.

The failure wasn’t technological but organizational. “Both internal oversight failed, and the regulatory oversight…had a problem with MOU understanding that they needed to make progress, but there was no penalty for failure to make the progress,” says van Deventer.

Key takeaways

Overall, van Deventer’s insights reveal a banking industry that is at a significant crossroad:

• Legacy paralysis: Big banks are trapped by their own success, with multi-billion-dollar systems creating a competitive disadvantage against smaller, more agile firms.
• Scale inversion: Contrary to regulatory assumptions, smaller banks can more easily adopt advanced risk technologies than their larger counterparts.
• Regulatory gaps: Current regulations don’t require Monte Carlo simulation, despite its superiority in risk assessment, while new ESG requirements may lack scientific rigor.
• Systemic risk amplified: When major institutions use identical but inadequate models, their collective failure could create a domino effect, leading to a broader financial crisis.

Quantum computing could deliver unprecedented risk management capabilities. But institutions need to escape the gravitational pull of their legacy systems and be savvy about hybrid classical-quantum approaches.

Will today’s financial leaders have the organizational courage to leave their computational prisons before the next crisis forces them? That’s the one thing that Monte Carlo simulations can hardly predict. 

Image credit: iStockphoto/Andrii Yalanskyi