Systemic financial crises marked by the collapse of many financial institutions are enormously costly for society. When banks fail together, credit dries up, and aggregate investment collapses. The 2008 global financial crisis showed just how deep and persistent these effects can be, with widespread job losses, housing foreclosures, and a prolonged, anemic economic recovery.

Given the scale of these costs, it seems obvious that efforts by banks to actively manage their risks are a good thing. And, in many ways, they are. Modern financial institutions have access to an array of sophisticated risk-management tools. They lend across industries and geographies. They use complex financial derivatives. They hold securitized assets that bundle risks in creative ways. In short, they do not put all their eggs in one basket.

So, are they now more resilient to shocks?

At first glance, the answer seems straightforward: of course. If each bank holds a better-diversified portfolio, each is less vulnerable to a shock in any one area. But here is the paradox: these superior risk-management technologies can actually make the system more fragile, not less.

Why? First, when banks feel safer, they are willing to lend more: risk-sharing allows for more risk-taking. The result is often a credit boom—a period of rapid expansion that sows the seeds of future instability.

Second, as all banks diversify, they tend to do so in similar ways. They use the same models, target the same “safe” sectors, and rely on the same hedging instruments. Over time, their portfolios converge. In trying to protect themselves, they become similarly exposed to underlying risk factors.

This dynamic—individually safe, systemically fragile—is not just a theoretical curiosity. It is a pattern we have seen in real events, like the run-up to the 2007–2008 crisis, where multiple banks were exposed to the same opaque risks.

So, how exactly does the financial system move from being safe and sound to becoming fragile and susceptible to a sudden systemic crash?

The journey is gradual. It starts during good times.

Suppose the economy is enjoying a streak of strong productivity growth. New technologies emerge, firms are optimistic, and investment opportunities are plentiful. Banks respond as expected: they expand lending by borrowing more, thus growing their balance sheets. At this stage, risk is low and profits are high. Because their core lending opportunities are performing well, banks feel little pressure to diversify beyond what they know best. As a result, portfolio overlap across banks remains low—and so does the risk of systemic collapse. In Figure 1, this stage occurs between periods 0 and 10.

But credit booms do not last forever. As the economy becomes saturated with investment, and the best opportunities are used up, returns begin to decline. To protect themselves from growing risk, they begin to diversify more aggressively.

Here is where things take a turn.

As each bank seeks to hedge its own risks, it shifts more of its portfolio toward projects outside its area of expertise—often toward the same set of remaining, decent-looking opportunities. The result is that banks’ portfolios start to overlap. On paper, everyone is diversified. In practice, everyone is exposed to the same underlying risks. In Figure 1, this stage happens between periods 10 and 16.

At this point, the system is vulnerable. Even a modest negative shock to a subset of investment projects—for example, a downturn in a particular sector or region—can ripple through all the banks holding similar exposures. What would have been a manageable loss in isolation becomes a systemic issue. This is what happens in period 17 in Figure 1.

This mechanism explains why systemic crises do not hit the economy at random times but instead tend to happen at the end of credit booms. That is when banks have lent the most, diversified the most, and unknowingly synchronized their risks the most. Even a moderate negative shock to a relatively isolated part of the economy is now enough to bring down a large part of the financial system.

Figure 1: Credit boom and portfolio overlap in the run-up to a systemic financial crisis

Does this mean that diversification is all bad? Certainly not.

True, banning or severely restricting banks from diversifying would virtually eliminate systemic crises in the model. But the cure would be worse than the disease. Without diversification, banks would become far more vulnerable to the risks in their core investments. They would lend less, pull back from marginal borrowers, and face a much higher probability of failing in isolation. Instead of rare, system-wide crises, the economy would suffer frequent small-scale bank failures. These individual defaults may not be as painful as full-blown crises, but their constant presence would choke credit, increase financing costs, and hurt long-term growth. All in all, a full diversification ban is equivalent to a loss of more than 6% of annual consumption.

So, limiting diversification severely is not a viable option. Instead, portfolio overlap can be used as a powerful early warning signal. When banks’ portfolios start to look increasingly alike, it is a sign that financial fragility is high and a systemic financial crisis is likely. This is the moment for regulators to pay close attention.

Specifically, rising portfolio commonality should trigger a stricter supervisory stance. Banks should be required to hold larger liquidity buffers and reduce reliance on short-term borrowing. Stress tests should place greater emphasis on the potential for correlated defaults. In short, regulators should aim to slow the pace of credit expansion when portfolio overlap is high and productivity growth is declining.

If done right, such targeted regulation can make a real difference. A quantitative assessment suggests it can reduce the frequency of systemic crises by more than half. Crucially, this is achieved without a large shrinking of credit supply or causing an uptick in isolated bank defaults. Therefore, the key is not to restrict banks from diversifying, but to prevent the economy from overheating.