Common mistakes about credit and spreads

The common belief: A record level of corporate debt is, by itself, a warning that a credit crisis is near.

What the data shows: Nominal debt almost always sets records because the economy grows; the level alone carries little information. US nonfinancial corporate debt reached roughly $14.2tn at end-2025 (FRED), a nominal record — yet relative to the market value of corporate equities it fell from about 26% in 2022 to roughly 16% in 2025. What matters is the change, the maturity profile, and interest coverage, not the headline figure.

→ Full breakdown: Is corporate debt at record levels a problem?

The common belief: High yield is “junk” to avoid, while investment grade is safe.

What the data shows: The line is not binary — it is a probability gradient. Since 1983, Moody’s speculative-grade default rate has averaged roughly 4–5% per year, ranging from about 1.6% to above 10% across the cycle, while investment-grade issuers have defaulted at well under 1% annually and in some years not at all. High yield compensates for that higher default frequency with a wider spread; the relevant question is whether the spread pays for the realized loss, not whether the label reads “junk.”

→ Fuller explanation: What is the difference between investment grade and high yield debt?

The common belief: A credit rating is an objective, reliable measure of default risk.

What the data shows: Ratings are through-the-cycle opinions, they lag market pricing, and the issuer pays for them. Their most visible failure came in 2008: according to the FCIC, over 90% of the AAA-rated subprime mortgage securities from 2006–2007 were downgraded to junk within roughly two years. Ratings remain useful as a coarse ordinal ranking, but treating a letter grade as a precise, real-time risk gauge repeats the error that preceded the crisis.

→ Extended explanation: How do credit rating agencies work and why do they matter?

The common belief: Credit spreads widen because the recession has arrived — a lagging consequence of the downturn.

What the data shows: The causality runs the other way. Spreads price expected default ahead of activity data, so they typically widen before a recession is visible and before equities fully react. The ICE BofA US High Yield OAS rose from about 360 bps at end-2019 to roughly 1,090 bps by late March 2020, ahead of the official contraction; it peaked near 2,150 bps in 2008–09 and also exceeded 1,000 bps in 1990–91 and 2001 (CFA Institute, ICE BofA data). Contrary to the textbook framing, the credit market is an anticipation system, not a rear-view mirror.

→ The complete explanation: Why do credit spreads widen in recessions?

The common belief: The credit cycle simply follows the business cycle.

What the data shows: It more often leads it. Lending standards tighten and spreads widen before output and employment turn, which is why the Fed’s Senior Loan Officer survey and high-yield spreads sit among the more reliable leading indicators. The mechanism is the one Hyman Minsky described: long expansions encourage looser underwriting, which seeds the fragility that the next contraction exposes.

→ Full account: How does a credit cycle differ from a business cycle?

The common belief: Credit default swaps are speculative instruments that caused the 2008 crisis.

What the data shows: A CDS is, at its core, an insurance-like hedge and a real-time price for credit risk. What went wrong in 2008 was concentration and opacity, not the instrument: AIG wrote protection on senior structured tranches assuming a near-zero default probability, with little collateral and no central clearing. Post-crisis clearing and margining addressed much of that plumbing; conflating the tool with its 2008 misuse misreads the history.

→ Complete breakdown: What is a CDS and how does it measure credit risk?

The common belief: A fallen angel — a bond downgraded from investment grade to high yield — is something to flee.

What the data shows: The forced selling the downgrade triggers is often the opportunity, not the warning. Investment-grade mandates must sell bonds that cross below the line, pushing prices below fundamental value, after which they have historically tended to recover over roughly six months. The ICE/VanEck fallen-angel index outperformed the broad high-yield market in 15 of the past 22 calendar years — a technical, mandate-driven pattern rather than a quality signal.

→ Complete explanation: What is a fallen angel and why does it matter for markets?

The common belief: Leveraged loans are safe because they are senior secured and float with rates.

What the data shows: Seniority is only as protective as the documents behind it, and those have eroded. Covenant-lite structures exceeded 90% of outstanding US leveraged loans by end-2024 (roughly $1.3tn, S&P data), stripping the maintenance covenants that once let lenders act early. Senior secured loan recoveries historically averaged around 80% (Moody’s, 1983–2011); Moody’s has estimated recent covenant-lite vintages could recover closer to 60%, though academic work disputes the magnitude. Senior secured is a starting point, not a guarantee.

→ Detailed explanation: What is the leveraged loan market and why is it fragile?

The common belief: CLOs are the subprime CDOs of 2008 repackaged — the next bomb.

What the data shows: The collateral and the structure differ. CLOs hold diversified senior secured corporate loans — typically capped near 2% per issuer and 15% per sector — not correlated subprime mortgages. No AAA-rated CLO tranche has defaulted in the market’s history; across roughly 21,000 S&P-rated tranches only about 0.3% defaulted, whereas AAA subprime CDOs recorded around $325bn in losses in 2008 (S&P, Wharton). The open question is whether a corporate-correlation shock could yet test that record, not whether CLOs are mechanically CDOs.

→ In-depth explanation: What are CLOs and how did they evolve after 2008?

The common belief: The Altman Z-score reliably predicts which companies will go bankrupt.

What the data shows: It has real predictive power but bounded scope. Edward Altman built the model in 1968 on a small sample of manufacturers, and its weightings reflect that origin: it is markedly less reliable for asset-light technology and services firms, for financials, and across different accounting regimes. It works best as one screen among several for industrial issuers, not as a standalone verdict — a calibration limit, not a universal law.

→ The full explanation: What is the Z-score and how is corporate distress measured?