The Credit Canary

Every day, the U.S. Treasury borrows money by selling bonds. Because the U.S. government is considered the safest borrower on Earth — it has never formally defaulted on its debt — Treasury bonds pay the lowest interest rate of any dollar-denominated security. That rate is the floor. Everything else pays more, and the amount more is the “spread.”

A credit spread, then, is a measurement of fear. When a company like Johnson & Johnson borrows money, it pays perhaps 0.80% above Treasury rates. That tiny spread reflects the market’s almost-total confidence that J&J will pay its bills. It has an AA− credit rating, $25 billion in cash, and sells products that people buy whether the economy is booming or collapsing. At the other end of the spectrum, a company like a mid-size shale oil driller with a B− credit rating and $3 billion in debt might pay 6% above Treasuries. That large spread reflects genuine uncertainty — this company might not survive if oil drops below $50 a barrel. The spread is the price of doubt.

The BofA US High-Yield Option-Adjusted Spread (FRED series BAMLH0A0HYM2) aggregates all of these risky spreads into a single number. It takes every bond in the Bank of America high-yield index — hundreds of bonds from companies rated BB+ or below by the rating agencies — and calculates the average spread over Treasuries, adjusted for embedded options like call provisions. When this number is 3%, the junk bond market is calm. When it is 5%, investors are nervous. When it exceeds 8%, something is breaking. When it crosses 10%, something has already broken.

The high-yield market is enormous and consequential. Roughly $1.5 trillion in junk bonds are outstanding in the U.S., issued by companies that range from household names like Ford Motor (whose debt was downgraded to junk in 2020) to small energy explorers, regional hospital chains, and retail leveraged buyouts. These are the companies most sensitive to economic cycles, because they operate with less margin for error. When times are good, they refinance easily, their spreads tighten, and investors chase yield. When times turn, they are the first to lose access to capital — and the spread is where that loss of access first becomes visible.

This is why bond traders call high-yield spreads the “canary in the coal mine” of financial markets. Stocks react to headlines. Credit spreads react to cash flows, balance sheets, and the actual willingness of institutional investors to extend credit to borrowers who might not pay them back. It is a harder signal to manipulate, less subject to retail sentiment or short squeezes, and it captures something fundamental: the price that real money charges real companies for real risk. When that price starts climbing, pay attention.

The Bank of America high-yield index begins at the end of 1996, giving us nearly three decades of daily data covering six distinct stress episodes. The chart below shows the annual average spread for each year — smoothing out the daily noise to reveal the tectonic shifts. The pattern it reveals is unmistakable: long periods of compression punctuated by sudden, violent blowouts.

Look at the rhythm. The late 1990s began with spreads below 3% — pure exuberance, the dotcom era in full swing — before the Russian debt crisis and LTCM collapse in 1998 spiked the average to 4.05%. Then a brief recovery, followed by the longer grind higher through 2000, 2001, and 2002 as the dotcom bubble burst, Enron imploded, and WorldCom became the largest bankruptcy in American history at that time. Annual average spreads hit 8.68% in 2002.

The compression that followed was breathtaking. From 2003 to 2007, spreads marched relentlessly lower: 6.27%, 3.91%, 3.55%, 3.26%, 3.61%. By June 1, 2007, the daily spread touched 2.41% — the all-time low. On that day, the market was saying that lending money to America’s riskiest companies was barely more dangerous than lending it to the U.S. government. Six months later, two Bear Stearns hedge funds had collapsed. Eighteen months later, the spread was at 21.82% and the global financial system was on life support.

This is the central paradox of the credit canary. The quietest readings — the moments when spreads are compressed to their tightest levels — are often the most dangerous. A spread below 3% doesn’t mean everything is fine. It means everyone thinks everything is fine, which is a very different statement. In 2007, the tightest spread in history was followed by the widest spread in history, separated by just 18 months. The canary was singing its loudest just before it dropped dead.

Over the last three decades, the high-yield spread has spiked dramatically five times. Each spike told a different story, moved at a different speed, and resolved in a different way. Understanding these patterns is the key to reading the spread in real time, because the next crisis will not look exactly like any of them — but it will rhyme with at least one.

1. LTCM and the Russian Default (1998). In August 1998, Russia defaulted on its domestic debt and devalued the ruble. This was, by any measure, a relatively small event — Russia’s GDP was smaller than the Netherlands’ at the time. But it triggered a chain reaction that nearly destroyed the global financial system, because Long-Term Capital Management (LTCM), a Connecticut hedge fund run by two Nobel laureates and a team of former Salomon Brothers traders, had built a $125 billion leveraged portfolio of bond arbitrage positions that were catastrophically exposed to exactly this kind of event. When Russia defaulted, liquidity vanished across global bond markets. LTCM lost $4.6 billion in four months. The New York Fed organized a private bailout of $3.6 billion from 14 banks to prevent a systemic collapse. The HY spread, which had been sitting at a comfortable 2.71% in March 1998, surged to 6.78% by October — a 150% increase in seven months.

2. The Dotcom Bust and Corporate Fraud (2000–2002). This was a slow-moving crisis, more like a disease than a heart attack. The NASDAQ peaked in March 2000, and the HY spread began widening almost simultaneously — from 4.78% in January 2000 to 9.16% by the end of the year. Telecom companies that had borrowed billions to lay fiber-optic cable started defaulting. WorldCom, which had $41 billion in debt, was exposed as a massive accounting fraud in June 2002. Enron had already collapsed in December 2001. The annual average spread in 2001 was 8.38%; in 2002, it was 8.68%. The daily peak hit 11.20% on October 10, 2002 — a level that wouldn’t be seen again until Lehman Brothers. What made this crisis distinctive was its duration: spreads stayed above 7% for roughly two and a half years, grinding down corporate balance sheets month after month.

3. The Global Financial Crisis (2007–2009). This was the big one. On June 1, 2007, the HY spread touched 2.41% — the tightest in history. The economy was growing, unemployment was 4.5%, housing prices were still rising in most markets, and the stock market was near all-time highs. Then, on June 22, Bear Stearns disclosed that two of its mortgage-backed securities hedge funds were in trouble. Spreads widened from 2.41% to 2.80% in a week — barely noticeable. By the end of July, after Bear Stearns told investors the funds were essentially worthless, spreads had crossed 4% for the first time since 2003. The canary was no longer singing.

What followed was an escalation so methodical it reads like a medical chart tracking a patient’s decline. August 2007: spreads averaged 4.31% after BNP Paribas froze three investment funds, marking the moment the crisis went global. November: 5.31% as Citigroup revealed $11 billion in subprime losses. January 2008: 6.80% as fears of a recession accelerated. March: 8.08% as Bear Stearns was forced into JPMorgan’s arms at $2 per share (later raised to $10 under shareholder protest). Then a deceptive calm — spreads actually fell back to 6.69% in May as markets convinced themselves the worst was over.

Then Lehman Brothers filed for bankruptcy on September 15, 2008. In the three months that followed, the HY spread went from 9.19% to 20.31%. To put that in human terms: if you were a B-rated company that could borrow at 7% above Treasuries in June 2008, by December you were looking at a 25% interest rate — if anyone would lend to you at all. The bond market wasn’t just pricing risk anymore; it was pricing extinction. Trading desks refused to bid on high-yield bonds. Mutual funds faced massive redemptions and were forced to sell into a market with no buyers. The spread peaked at 21.82% on December 15, 2008 — the worst single reading in the history of the index. It would take 25 months before spreads fell back below 5%.

4. The Oil Crash (2015–2016). This crisis was sector-specific, not systemic — but the HY spread didn’t care about that distinction. When oil prices collapsed from $107 per barrel in June 2014 to $26 in February 2016, dozens of shale oil and gas companies that had borrowed heavily to fund drilling operations found themselves unable to service their debt. Companies like Chesapeake Energy (which had $12 billion in debt), Linn Energy, and SandRidge Energy were all junk-rated, and their bonds made up a significant share of the HY index. As energy defaults surged, the spread widened from 3.35% in June 2014 to 8.87% on February 11, 2016. The crisis was real — over 200 North American oil and gas companies filed for bankruptcy between 2015 and 2017 — but it was contained. Non-energy high-yield bonds continued to function normally, and the spread recovered to below 5% by September 2016, just seven months after the peak.

5. COVID-19 (2020). The fastest crisis in the history of credit markets. On January 17, 2020, the HY spread was 3.39%. By March 23 — just 47 trading days later — it was 10.87%. That is a tripling in less than two months. The world had never seen credit spreads move this fast, because the world had never shut down its economy this fast. What made the COVID credit crisis unique was not just its speed but its resolution. On March 23, the Federal Reserve announced it would buy corporate bonds — including, for the first time in its history, high-yield ETFs. That announcement, before a single bond had actually been purchased, functioned like a circuit breaker. Spreads fell from 10.87% to 7.96% in two weeks. By August 27, just five months after the peak, the spread was back below 5%. The Fed didn’t just stop the panic; it compressed the entire crisis cycle from years to months.

Each crisis left a distinct fingerprint in the data. The table below compares all six stress episodes on the metrics that matter most: how high the spread went, how fast it got there, the ratio of high-yield to investment-grade widening (which measures panic concentration), and how long recovery took.

Several patterns emerge from this scorecard. First, the pre-crisis lows have been getting lower over time. Before the GFC, the all-time tight was 2.41%. By late 2024, spreads had compressed to 2.60%, and by early 2026 they were regularly printing below 2.80%. Each era of easy money has pushed complacency to new extremes. Second, the recovery time has shortened dramatically. The GFC took over two years to heal. The oil crisis took seven months. COVID took five. This isn’t because crises are getting milder; it’s because the Fed’s toolkit has expanded from rate cuts alone to include direct asset purchases, lending facilities, and forward guidance — each intervention compressing the cycle further.

Third, notice the HY:IG ratio. In every crisis, high-yield spreads widen roughly 3–4 times as much as investment-grade spreads. This multiplier exists because high-yield bonds face three overlapping problems during stress: higher actual default risk, forced selling as mutual funds and ETFs face redemptions, and a collapse of liquidity as market makers pull back from riskier securities. Investment-grade bonds suffer from the last two but rarely the first. The COVID crisis was the outlier: the HY:IG ratio was only 2.7×, because the Fed’s unprecedented announcement that it would buy corporate bonds (including junk-rated ETFs) short-circuited the forced-selling cascade that normally amplifies high-yield distress.

The credit market isn’t a monolith — it’s a layer cake. Investment-grade bonds (rated BBB− or above) sit at the top, where spreads are narrow and defaults are rare. BBB bonds — the lowest rung of investment grade, one downgrade away from junk status — occupy a nervous middle ground. And high-yield bonds sit at the bottom, where the spread absorbs the full force of market fear. The chart below shows all three tiers at each crisis peak, revealing how stress amplifies as you move down the credit ladder.

The visual tells the story better than any explanation. During the GFC, investment-grade spreads peaked at 6.51% — already extreme by historical standards. BBB spreads reached 8.02%. But high-yield spreads hit 21.82%, more than three times the IG level. The junk bond market wasn’t just nervous; it was pricing widespread default. And the fear was justified: Moody’s trailing 12-month high-yield default rate peaked at 14.7% in November 2009, the highest since the Great Depression.

The COVID chart tells a subtler story. All three tiers widened violently in March 2020 — HY to 10.87%, BBB to 4.88%, IG to 4.01%. But the ratio was compressed compared to the GFC. The explanation is that COVID was a liquidity crisis, not initially a solvency crisis. Companies weren’t going bankrupt because their business models were broken; they were running out of cash because the economy had been shut down by government mandate. Everyone understood this, which is why the Fed’s backstop worked so quickly: it addressed the actual problem (liquidity) rather than the symptom (spreads).

The Tariff Shock of April 2025 is instructive for how small it was. HY spreads peaked at just 4.61%, BBB at 1.48%, IG at 1.20%. The HY:IG ratio was 3.8× — actually higher than average, suggesting the market was pricing genuine credit risk in tariff-sensitive sectors (importers, retailers, auto parts companies) even though the overall stress level remained modest. Within six weeks, spreads had fully normalized. The canary chirped; it didn’t faint.

One of the most important things the credit canary reveals is how fast stress arrives. Not all crises develop at the same speed, and the pace of widening tells you something about the nature of the threat.

The GFC was the slow burn. From the all-time low of 2.41% on June 1, 2007, the spread took 160 days to double, crossing 4.91% on November 8, 2007. That is five months of steady, grinding deterioration — plenty of time for anyone watching to reduce risk, rebalance portfolios, or hedge credit exposure. Of course, most investors didn’t, because the stock market was still rising through October 2007 and the official recession didn’t begin until December. The credit canary was screaming, but investors were watching CNBC, not the OAS spread.

COVID was the opposite. From a low of 3.39% on January 17, 2020, the spread doubled in 55 days, passing 7.42% on March 12. But even that understates the velocity: more than half of the widening happened in just 15 trading days between February 24 and March 12, as the market transitioned from “this is a China problem” to “this is a global shutdown” virtually overnight. On February 21, the spread was 3.66%. On March 23, it was 10.87%. That is a tripling in 22 trading days — less than five calendar weeks. There was no time for a measured response. By the time the data confirmed what was happening, the move was already over.

The dotcom bust was the slowest of all — a multi-year grind that took the spread from the mid-4s in early 2000 to 11.20% in October 2002, with several false recoveries along the way. This kind of slow-motion crisis is actually the hardest to trade, because the signal never gets loud enough to trigger panic, yet the damage accumulates quarter after quarter. Investors who reduced risk after the first spike to 7% in late 2000 felt vindicated briefly, then watched spreads fall back to 7.38% in January 2001 before climbing all the way to 10.68% eighteen months later. The canary didn’t stop singing suddenly; it just got quieter and quieter until you realized it had been dying all along.

If the speed of the crisis tells you what kind of threat you’re facing, the speed of recovery tells you how the system healed. Below are the recovery timelines for the three most severe episodes, measured from peak spread to the first day below 5% — a level that represents “elevated but functioning.”

The GFC recovery took more than two years because the underlying damage was structural. Banks had to be recapitalized. Mortgage-backed securities had to be unwound. The housing market had to find a floor. Even with the Fed cutting rates to zero and launching QE1, the healing was slow because the problem was solvency, not just liquidity — banks and borrowers were genuinely insolvent, and no amount of cheap money could instantly make bad loans good. The HY spread stayed above 10% for all of 2009 and didn’t cross below 7% until December of that year.

The oil crisis was faster because it was contained to one sector. Energy companies defaulted, their bonds were restructured or written off, and the rest of the high-yield market continued to function. Once oil prices stabilized above $40, the surviving energy companies refinanced, and new issuance resumed. The damage was real but bounded.

The COVID recovery was the fastest because the cause of the crisis was uniquely identifiable and treatable. The economy wasn’t broken; it was paused. The Fed provided the bridge financing that companies needed to survive the pause, and once the economy reopened, cash flows resumed. The five-month recovery from 10.87% to below 5% is the most dramatic snap-back in the history of credit markets, and it set a precedent that markets still rely on: the assumption that the Fed will intervene quickly and decisively in any future credit crisis. Whether that assumption will hold in the next crisis — especially if the crisis is inflationary rather than deflationary — is one of the open questions in financial markets today.

The table below shows the annual low, high, and average HY spread for every year since 1997. Reading it chronologically reveals the repeating cycle of compression and blowout that defines credit markets. Years where the low is below 3% are years of maximum complacency. Years where the high exceeds 8% are years of genuine stress. Years where the range is narrow (like 2017 or 2021) are years when the market barely moved — the calm before the next storm.

* 2026 data through April 9.

Three patterns jump out. First, the years with the narrowest ranges — 1997 (0.72), 2006 (0.91), 2017 (0.78), 2021 (0.92), and 2026 so far (0.82) — are all years of maximum complacency, where the market barely budged. Several of these were followed within 12–24 months by significant blowouts. The 2006 calm preceded the GFC. The 2017 calm preceded the late-2018 selloff. The 2021 calm preceded the 2022 rate shock. Narrow ranges don’t predict crises, but they indicate compressed volatility that eventually has to release.

Second, the low-water marks have been falling over time. In the pre-GFC era, spreads rarely got below 2.50%. In the post-2020 era, sub-2.70% has become routine. This secular compression reflects three structural forces: the Fed’s post-2008 zero-rate policies that pushed investors into riskier assets, the growth of passive high-yield ETFs that automatically buy bonds regardless of their risk, and the market’s belief — confirmed by 2020 — that the Fed will backstop credit markets during any severe stress event. Each of these forces reduces the risk premium that investors demand, which shrinks the spread, which makes the next blowout start from a lower base.

Third, look at 2008. The range that year was 15.72 percentage points — from a low of 6.10% in early January to a high of 21.82% in December. No other year comes close. The GFC was not just a spike; it was a complete dismantling of the credit market’s pricing mechanism, where the normal rules of supply and demand were replaced by pure panic and forced liquidation. Understanding that 2008 was an outlier is important, because it means that the most useful comparable for any future crisis is probably the 8–11% range (like COVID or the oil crash), not the 20%+ range. The GFC was a once-in-a-century event whose unique combination of leverage, opacity, and systemic interconnection is unlikely to be replicated in the same form.

As of April 9, 2026, the BofA US High-Yield OAS reads 2.90%. The investment-grade spread is 0.83%. The BBB spread is 1.05%. The HY:IG ratio is 3.5× — perfectly in line with the historical average, suggesting that the relationship between credit tiers is normal. No tier is under unusual stress; no tier is sending a divergent signal.

A reading of 2.90% places the spread at the 8th percentile of all observations since 1996, meaning 92% of all historical readings have been higher. This is deeply complacent territory. The only periods with comparable calm were mid-2007 (just before the GFC), mid-2014 (before the oil crisis), late 2024 through early 2025 (before the tariff scare), and now. Each of those prior episodes of extreme compression was eventually followed by a widening — though the timing ranged from months to over a year.

Does this mean a crisis is imminent? No. Spreads can stay tight for extended periods, especially when corporate fundamentals are strong, defaults are low, and the economy is growing. The current reading reflects genuine underlying health: high-yield default rates are near record lows, corporate cash balances are high, and the vast majority of junk-rated issuers have already refinanced their near-term maturities at favorable rates. The canary is singing because, right now, there is nothing to sing about.

But the lesson of thirty years of data is that the canary is most useful precisely when it seems least necessary. The time to build a dashboard is not when the alarm is already blaring. It’s when everything is quiet, when spreads are at the 8th percentile, when the financial press has moved on to other stories, and when the most common phrase in analyst reports is “benign credit conditions.” That is when you want to know your thresholds, your historical comparables, and your action plan. The next episode covers the VIX — the market’s fear gauge — which measures stress not through credit but through the price investors pay to insure against equity losses.