Our Perspective on the CLO Scare

As stocks continue their jagged recovery from March 23’s low, a second COVID wave isn’t the only worry on investors’ minds. For weeks, pundits have warned a second shoe waiting to drop lurks on bank balance sheets, in the form of collateralized loan obligations (CLOs)—securitized corporate debt. In the wake of some high-profile bankruptcies—and with analysts presuming many more will follow—pundits draw parallels between CLOs and their mortgage-backed cousins, collateralized debt obligations (CDOs), the securities widely seen as responsible for 2008’s financial crisis. If corporate debt goes the way of the housing market, people fear bank balance sheets will be collateral damage, triggering a 2008 repeat. But as we will discuss, there are quite a few reasons this scenario is unlikely.

For one, lost in most coverage is the simple fact that US banks’ exposure to CLOs is tiny. At 2019’s end, S&P Global’s analysis of Federal Reserve data put banks’ holdings of CLOs at just under $100 billion—a sliver of the $17.8 trillion in total bank assets at that time.^[i] Of that, around $30.2 billion is designated as “held to maturity,” while roughly $69.3 billion is marked for sale. This is crucial since accounting rules have changed since the financial crisis. Back then, banks had to value all assets at market price or “fair value,” regardless of whether they intended to sell them. This created problems for rarely traded, hard-to-value assets like CDOs. When hedge funds sold CDOs at fire-sale prices, banks had to mark down the value of all comparable assets accordingly. This rule created the vicious cycle of fire sales and writedowns that wrecked bank balance sheets and forced several major financial institutions out of business. Eventually, at the prompting of formed FDIC chief William Isaac and others, regulators recognized the error of forcing mark-to-market accounting on illiquid assets banks never intended to sell. So in mid-March 2009, they suspended the rule—and revised it months later. Now, when these securities hit rough patches, banks need only mark those designated “for sale” to the most recent sale price. For those marked “held to maturity,” banks can disclose the market value in a footnote and carry on. That further defangs the already very small risk CLOs pose.

Worries about CLO credit quality also seem overstated. In general, CLOs largely echo high-yield corporate bond markets, which usually do quite well early in a recovery—even as bankruptcies mount. That was the case in 2009, and all signs point to a repeat now. US high yield bond interest rates peaked at 11.4% on March 23, the date stocks bottomed.^[ii] Friday, they closed at 6.68%, not far removed from their pre-bear levels.^[iii] Bond markets are forward-looking, just like stocks. If credit risk were off the charts, yields probably wouldn’t have nearly halved in just three months.

Moreover, after 2008 shone a bright spotlight on securitized debt, buyers got wiser. Banks aren’t the only player in the CLO market. According to an analysis from the Financial Stability Board, US banks own only about 20% of CLOs globally. Hedge funds, specialty credit funds and international non-bank investors collectively own about 40%. This is critical because CLOs are sold in tranches, with each slice carrying varying levels of risk (largely based on credit ratings, which aren’t perfect, but this is the way it is). Banks generally shy away from the riskier tranches, leaving them for that less risk-averse 40% of the investor base. So in the event that problems do arise, banks are less likely to take a hit.

In our view, 2008’s financial crisis stemmed from the unintended consequences of a well-intended accounting rule that forced banks to transform around $200 billion of loan losses into roughly $2 trillion of writedowns. This time, the max financial hit banks could take in a completely unrealistic worst-case scenario is $69.3 billion. It is exceedingly unlikely a hit that size would be sufficient to trigger a devastating cascading failure in the financial system.