Latest / Monetary Policy Explained with Fexingo: Central Banks, Money Supply, and Interest Rates / How Central Banks Stress Test Their Own Balance Sheets
Transcript
- Lucas: When a commercial bank fails a stress test, the headline writes itself. But here's a question that gets far less attention: who stress-tests the central bank? Luna: You mean the Fed stress-testing itself? Lucas: Exactly. Central banks hold trillions of dollars in assets—Treasuries, mortgage-backed securities, foreign reserves, and now, with quantitative easing, longer-duration bonds that can lose value fast when rates rise. They also have liabilities that can run. And if a central bank's balance sheet takes a hit, its ability to conduct policy gets impaired. Luna: Right, and we saw a glimpse of that in 2023 when the Fed's own deferred asset position—the thing that happens when its income drops below its expenses—blew up to something like $100 billion. Lucas: Yes, the deferred asset. That's essentially an accounting loss that the Fed records when its interest income from its bond portfolio isn't enough to cover the interest it pays on reserves and reverse repos. It doesn't affect the Fed's ability to set policy, but it does erode its remittances to the Treasury. And it's a signal of balance sheet strain. Luna: So what exactly does a central bank stress test look like? Is it like the CCAR that the Fed runs on big banks? Lucas: Similar in spirit, but the scenarios are different. The Fed's internal stress test for its own balance sheet, which it started publishing more detail on around 2024, models what happens under severe adverse scenarios: a sudden spike in short-term rates, a rapid increase in long-term yields, a liquidity crisis where the reverse repo facility is drained overnight, even a scenario where the dollar weakens sharply and foreign reserve holdings take a hit. Luna: And what did they find? Was there a real vulnerability? Lucas: The 2024 exercise highlighted something interesting. Under the severe adverse scenario, the Fed's net income turned deeply negative—we're talking a loss of over $150 billion over two years. And the deferred asset grew to more than $200 billion. But the bigger issue was liquidity. In a scenario where the overnight reverse repo facility drained from around $500 billion to zero within weeks, the Fed would have to rely more heavily on its standing repo facility to manage reserves. That creates operational risk. Luna: And the standing repo facility is still relatively untested at scale, right? Lucas: Exactly. It was introduced in 2021 as a backstop, but it's never been used during a real liquidity scramble. So the stress test essentially said: we have a tool, but we don't fully know how it behaves under duress. That's a vulnerability. Luna: The ECB does something similar, don't they? I remember reading about their own balance sheet stress test in 2023 after the rate hikes started biting. Lucas: Yes, the ECB runs what they call a 'balance sheet projection exercise' under adverse scenarios. Their biggest risk is different. The ECB holds a lot of government bonds from weaker eurozone members—Italy, Spain, Greece. In a stress scenario where spreads blow out, the market value of those bonds drops. If the ECB had to sell them to manage liquidity, it would crystallise losses. And that could trigger political questions about burden-sharing. Luna: So it's not just about financial loss—it's about credibility and independence. Lucas: Exactly. A central bank that takes a big accounting loss can still function. But if the loss is so large that it needs a recapitalisation from the government, that opens the door to political influence. The Bank of Japan knows this better than anyone. Their balance sheet is enormous—over 130% of GDP—and they hold huge amounts of Japanese government bonds and ETFs. In 2024, when the Bank of Japan started to normalise policy, the mark to market losses on their bond portfolio were staggering. Luna: Right, I think at one point the unrealised losses on their bond holdings exceeded ¥10 trillion, about $70 billion. And they had to keep buying bonds to defend their yield curve control policy. Lucas: Exactly. And the Bank of Japan's stress tests—which they publish in their financial stability reports—show that under a severe adverse scenario where JGB yields spike to 2% or higher, the Bank's capital could be wiped out. It wouldn't go insolvent in a traditional sense, but it would be technically in negative equity. That's why they've been so cautious about exiting yield curve control. Luna: So the stress test itself becomes a constraint on policy. If the test says 'you can't raise rates too fast without blowing a hole in your own balance sheet,' that limits what the central bank can actually do. Lucas: That's the real insight. Central banks aren't just setting policy based on inflation and employment—they're also modelling the feedback loop from their own balance sheet. And that loop can be binding. The Fed's 2025 stress test actually incorporated a scenario where quantitative tightening itself triggers a liquidity crunch, forcing the Fed to stop QT earlier than planned. And that's essentially what happened in 2019 and again in 2023. Luna: It's almost like the central bank is its own patient. It has to diagnose its own vulnerabilities before they become acute. Lucas: And the diagnosis isn't always comfortable. Let me give you a specific number. In the Fed's 2024 stress test, under the severely adverse scenario, the Fed's capital ratio—its capital relative to its assets—fell from about 0.5% to negative 3%. That's a small capital base. Most commercial banks target capital ratios above 10%. So the Fed is operating with very thin equity. It's designed that way—central banks don't need much capital to function—but it means there's almost no buffer. Luna: Which is fine as long as the losses are paper. But if they become realised, you have a problem. Lucas: Right. And that's why the stress tests also look at 'realised loss' scenarios—what happens if the central bank is forced to sell assets. For example, if a foreign central bank suddenly withdraws its dollar swap line and the Fed needs to liquidate some of its Treasury holdings to provide dollars, that could crystallise losses. The stress tests model that. Luna: And there's an international dimension too. The Bank for International Settlements coordinates some of these stress tests across central banks, right? Lucas: Yes, the BIS runs a regular 'central bank balance sheet stress test' exercise among member central banks, looking at cross-border contagion. For instance, what happens if a large emerging-market central bank that holds a lot of US Treasuries suddenly needs to sell them? That could depress US bond prices and hurt the Fed's portfolio. It's a network effect. Luna: So it's not just about each central bank in isolation—it's about the system. Lucas: Exactly. And that's where the stress tests get really interesting. They're not just internal risk management exercises anymore. They're becoming a public communication tool. The Fed now publishes summary results of its own balance sheet stress tests in its Monetary Policy Report. The ECB includes a dedicated section. The Bank of Japan does too. It's a way of saying to markets: 'We know our risks, and we're prepared.' Luna: But there's a tension, right? If you publish the results, you might spook markets. If you don't, you're opaque. Where's the line? Lucas: That's the central bank communication dilemma. The Fed, for example, doesn't publish the exact scenarios in full—it aggregates them. They'll say 'under a severe adverse scenario, net income falls by X' but they won't reveal the exact trajectory of assumptions. Too much transparency could trigger a self-fulfilling panic. Too little, and no one trusts the results. Luna: And the real test hasn't happened yet. No major central bank has actually been forced to recapitalise. But the risk is real. Lucas: It's worth remembering that in 2023, the Swiss National Bank reported a loss of over 130 billion Swiss francs—one of the largest losses relative to GDP for any central bank—and it didn't require a bailout. But that was because its losses were mostly unrealised and it had a strong capital position going in. Not every central bank is that fortunate. Luna: If walking through the economy with us has made something click, you can support the ad-free choice at buy me a coffee dot com slash fexingo. We keep this show free of sponsors because we want the analysis to be independent, and listener support is what makes that possible. Lucas: Yeah, and it's a small way to keep this kind of detailed policy conversation going without the pressure to sell anything. Now, back to the stress tests—one thing I find fascinating is how central banks are now stress-testing their own digital currencies. The People's Bank of China, for instance, runs scenarios on what happens to its balance sheet if the digital yuan is adopted rapidly and bank deposits shift massively. Luna: Right, that's a whole new dimension. If people move deposits from commercial banks to the central bank's digital wallet, the central bank's liabilities grow, and it might have to buy more assets. It changes the balance sheet structure entirely. Lucas: Exactly. And the PBOC's stress tests show that under a rapid adoption scenario, their balance sheet could expand by 20% within three years. That would require them to hold more foreign reserves or government bonds, which changes their risk profile. So even the digital currency rollout is being stress-tested before it happens. Luna: It's almost like central banking has become an exercise in simulated self-destruction. You try to break yourself on paper so you don't break yourself in real life. Lucas: That's the whole point of stress testing. And the irony is that the more credible the stress test, the less likely the scenario is to occur. Because if markets know the central bank has a plan for a liquidity crunch, they're less likely to run. The stress test itself becomes a stabilising force. Luna: So maybe the most important output of these stress tests isn't the number—it's the confidence that the central bank has done the homework. Lucas: Exactly. And that's the real asset on the balance sheet. It doesn't show up in the accounting, but it's probably the most valuable thing a central bank owns.