What to do with US banks' unrealized losses?

Unrealized losses are back at the center of the banking discussion. According to the latest data from Office of Financial Research (OFR), US banks ended the first quarter of 2025 with $413 billion in unrealized losses on their securities portfolios, a staggering figure rooted in the rapid rise in long-term rates and the resulting markdown in the value of Treasuries and agency mortgage-backed securities (RMBS). While unrealized, they *could* represent a latent vulnerability embedded deep within bank balance sheets, particularly for regional institutions whose business models and capital structures leave them more exposed than their larger peers.

Figure 1: Unrealized gains (losses) on specific investment securities (as of March 2025):

Source: The Federal Deposit Insurance Corporation (FDIC)

At the heart of the issue lies the accounting treatment of investment securities. US banks report securities either as “held-to-maturity” (HTM) or “available-for-sale” (AFS). HTM securities are carried at amortized cost, effectively shielding the balance sheet from price fluctuations so long as the securities are not sold. AFS securities, by contrast, are marked-to-market, meaning changes in their value are reflected in the bank’s book equity and, for large “advanced approach” banks, in regulatory capital via the accumulated other comprehensive income (AOCI) account. This dual system aims to balance the desire for transparency with the recognition that not all paper losses are economically meaningful, so long as the intent to hold to maturity remains credible.

But credibility is precisely what is at stake. The run on Silicon Valley Bank back in 2023 was a sharp reminder that intentions can shift under stress (also here). The episode shattered the illusion that HTM portfolios are immune to market perceptions. When depositors fear asset impairment, accounting classifications do not prevent real-world liquidity crises. Regional banks, in particular, have learned this lesson the hard way, but not hard enough, I guess. Despite holding only 40% of total equity in the banking system, they are carrying around 60% of total unrealized losses. This is partly a function of their higher exposure to long-dated securities and agency RMBS, assets that have suffered the most from rising rates (because they are longer-term) (see Figure 2 below).

Figure 2: Instruments most affected by unrealized losses (as of March 2025):

Source: Office of Financial Research

But it's also about funding models. These institutions rely disproportionately on uninsured deposits (close to 40%), which tend to flee quickly when interest rate risk rears its head. This wouldn’t be inherently problematic. What is problematic is that these regional banks (if not the entire US banking system) turn a blind eye when it comes to interest rate swaps, the very instruments that could serve as a hedge against risk. The importance of this oversight was made painfully clear in the case of Silicon Valley Bank. Today, although RMBS account for the bulk of unrealized losses, over 70% of these instruments remain unhedged, as per FDIC. Worse still, although US banks report large notional amounts of swaps (close to $93tn), their actual exposure to interest rates is near zero. This is because gains and losses from individual swap positions typically offset each other, leaving the overall portfolio “balanced” (more about this here).

What banks do or could do? A growing number of institutions have begun restructuring their securities portfolios, accepting realized losses in the near term in exchange for longer-term stability. This strategy - essentially de-risking the balance sheet - has gained traction in recent months. Associated Banc-Corp and Heritage Financial Corp. are just two examples of regional banks that have conducted multiple rounds of securities sales. Still, such moves are far from painless. A realized loss compresses earnings, weakens dividend capacity under regulatory rules, and risks triggering shareholder dilution if capital needs to be replenished through equity issuance. In several cases, banks have been forced to sell stock below tangible book value to fill the hole, an unpopular but increasingly common solution.

Some banks have started engaging in leaseback transactions, as in the case of Bogota Financial Corp. More specifically, the bank sold three branch offices, resulting in a $9 million pre-tax gain. However, it also booked a pre-tax loss of $8.9 million from selling around $66 million in amortized cost ($57.1 million in market value) of HTM and AFS securities, with a weighted average life of 5.5 years and a yield of just 1.89%. This is likely a form of earnings management. By selling profitable real estate assets, the bank generates enough gains to offset the losses from selling underwater securities. More on that here.

Why are banks doing this? Although, on paper, unrealized losses across the banking system fell by $69 billion quarter-over-quarter (from $483 billion at the end of Q4 to $414 billion by the end of Q1 2025), the very existence of these losses has fundamentally reshaped how investors perceive risk in the sector. Debt and equity holders are now likely to demand higher compensation, increasing the cost of capital for banks. That cost will inevitably be passed on to borrowers in the form of higher loan rates, tightening credit conditions. If sustained, this dynamic will drag on investment and consumption, adding another channel through which monetary tightening exerts pressure on the broader economy. Moreover, in the absence of interest rate swaps acting as a natural hedge, unrealized losses (whether net or gross) may not affect a bank’s regulatory capital, but they do impact the economic value of the bank’s equity by reducing the economic value of equity or EVE. Unrealized losses reduce a bank’s EVE because they mean that the assets the bank holds - like Treasuries and RMBS - are worth less if sold today, even if they haven’t been sold yet. EVE looks at the long-term value of a bank by comparing what it owns (assets) to what it owes (liabilities), adjusted to current market values. So when asset values fall (like when interest rates rise and bond prices drop), EVE shrinks. And a lower EVE means the bank’s overall financial position is weaker in the long run.

This, in turn, weakens banks’ ability to attract fresh capital. Faced with funding needs, banks may look to raise debt instead. But since they can’t afford to sell securities that would trigger accounting losses, investor risk perceptions rise again - driving up debt funding costs as well. This is a mouse trap.

Even more complicated are the regulatory and accounting constraints. Reclassifying securities from AFS to HTM is not a free option. Once a bank does it and sells even a small portion of its HTM securities - whether due to liquidity pressures or a shift in strategy - it may be forced to reclassify all HTM securities as AFS securities, dragging the mark-to-market pain onto the balance sheet. Such reclassifications are tightly restricted to discourage opportunistic behavior. The upshot is that banks must get it right at the moment of acquisition. There are no easy do-overs.

This rigidity in the classification system creates a powerful feedback loop. As rates rise, AFS securities drag down equity and capital buffers, forcing banks either to take the hit or to ride it out in the hope of future rate cuts. But if stress emerges - whether from deposit outflows or broader market volatility - those paper losses can quickly become real. Selling at a steep discount crystallizes the loss. Not selling risks a liquidity crisis. The SVB debacle made this dilemma explicit.

The debate over whether unrealized losses should count in regulatory capital highlights a deeper tension. On one hand, fair value accounting provides a more transparent view of a bank’s financial health, particularly in volatile markets. On the other, mark-to-market regimes can inject procyclicality into capital requirements, forcing banks to hold more equity precisely when market conditions are deteriorating. This can discourage lending during downturns, amplifying the credit cycle. The question is not whether the losses are “real” - economically, the opportunity cost is undeniable - but whether recognizing them too quickly creates new forms of fragility.

But this isn’t meant as a doomsday warning, the current situation is far from what we saw in 2023. In 2023, it became clear that many of the regional banks hit hardest by losses were closely tied to venture capital and crypto.

Still, the risks are real. Bank balance sheets remain largely unhedged against interest rate risk, especially when it comes to CMBS and RMBS exposures. That could become a problem in upcoming quarters (if commercial mortgages are not going to do well). For now, though, there’s no liquidity crisis and that’s good news. A quick tip: trouble tends to start when the Federal Home Loan Banks (FHLBs) begin issuing large amounts of debt. That’s usually a sign regional banks are facing liquidity stress. What typically follows is a spike in discount window usage. Fun fact - back in 2023, the FHLBs lent over $250 billion to support regional banks. Today, that number is something around $15 billion.

In the end, what we're witnessing is a slow-moving capital revaluation across the banking system. Rate hikes have repriced duration, and that revaluation is now colliding with regulatory structures and funding fragilities, especially at the regional bank level. The industry is responding - strategic repositioning, securities sales, and capital raises are underway, you name it - but the process is uneven. Unless long-term yields begin to fall or investor confidence in deposit stability returns, regional banks will remain under pressure. Their capacity to extend credit will be limited, not only by capital constraints but by investor expectations and the higher cost of funding that now accompanies them. We’re also likely to see a slowdown in M&A activity and a rise in bank funding costs.

These are the main challenges posed by unrealized losses - at least for now. We're not facing a liquidity crisis, nor is the US banking system insolvent. But the last thing it needs right now is additional costs and growing investor pessimism, which could further erode the trust in the US banking system going forward.