Credit Spreads and Risk Appetite

What a spread is made of

Start with the actuarial part: the probability of default over the period, multiplied by the loss given default, which is one minus the recovery rate. That is the expected loss, and it is the only component that can be estimated from credit fundamentals. Everything above it is a risk premium, and the risk premium is not a fudge: a lender has to be paid for the fact that defaults cluster in recessions, that the loss is uncertain in a way that a single expected value conceals, and that the market for the bond may not exist when the lender wants to sell. The clustering is the part that matters most and the part least visible in a table of expected losses. A portfolio of high-yield bonds that loses 1.9% a year on average does not lose 1.9% every year: it loses a fraction of a percent in good years and many times that in a recession, and the losses arrive at the same moment as falling equity prices and rising unemployment. Diversification across issuers does not help with that correlated component, which is why the premium has to be large — and why the spread can be simultaneously a fair price for the risk and a bad investment, depending on when it was bought. The third component is liquidity and uncertainty, and it behaves differently from the other two. It is small in calm markets and enormous in a crisis, which is why the widening of a spread during a risk-off episode is partly a repricing of probability and partly a repricing of the ability to sell at all. That distinction matters for the investor because the liquidity component mean-reverts while the default component does not: a spread that widened purely on liquidity comes back without any improvement in credit, and a spread that widened because defaults are genuinely expected does not. Where a spread goes — Investment grade: 110bp spread: expected loss 6bp, excess 104bp · High yield: 390bp spread: expected loss 192bp, excess 198bp ← · High yield ÷ investment grade spreads: about 3.55× — for roughly thirty times the expected loss · And the excess spread pays for: clustering in recessions, uncertainty, and illiquidity Expected loss is estimated from the present composition of the index, and the index changes. Downgrades move issuers from investment grade into high yield, so the high-yield index deteriorates as the cycle turns — which is why realised default rates rise faster than any forecast made from today’s ratings.

The level is a price; the direction is the information

A spread is quoted as a level and read as a signal, and the two uses require different questions. As a level, the question is what the spread pays relative to the loss it is likely to incur, which requires an estimate of the default cycle. As a signal, the question is where the spread sits relative to its own history and what has changed: 390bp when the long-run median for that index is 450bp is offering less cushion than the same 390bp after a rally from 700bp, and the same number carries opposite implications in those two states. The information content runs in both directions. Spreads tighten when growth is expected, when credit availability improves and when investors reach for yield, and they widen when default expectations rise or when liquidity is scarce. Credit spreads have a documented record as a leading indicator of activity, and specifically as a signal about the availability of credit rather than about its price — the excess bond premium, the part of the spread that is not explained by expected defaults, is the component most associated with future output. That is the same distinction the last lesson drew in the transmission chain: the part of a credit move that is about default is an earnings signal, and the part that is about risk appetite is a conditions signal. For a portfolio, three rules follow. Read the spread against its own percentile rather than against a level you remember. Ask what would have to happen to defaults for the spread to be adequate, and check whether you believe that default rate. And treat the spread as a statement about the whole credit cycle rather than about one issuer — because the largest risk in a corporate bond portfolio is not any company failing, it is every company being repriced at once, which is the same lesson the 2008 balance sheets taught on the bank side. • Spread = expected loss + a premium for clustering and uncertainty + a liquidity component. • Defaults cluster: a 1.9% average annual loss is not 1.9% every year. • The liquidity component mean-reverts; the default component does not. • The excess bond premium — the part not explained by defaults — is the best activity signal. The same decomposition governs equity credit analysis: a company whose bond spread has widened 200bp while its equity is flat is a company where the debt market has already repriced the default that the equity market has not.

Where the stress shows up first

A credit spread is a summary of a market, and markets tell you less about what is fragile than the **funding structure** does. The question to ask of any borrower or any asset class is the same one the failure cases ask: when does the money have to come back? A borrower with debt maturing in five years is insulated from a repricing today; a borrower with debt maturing next quarter is at the mercy of whoever is willing to lend then. This is why the credit cycle moves through **maturity walls** rather than through spreads alone — spreads can fall while a wall approaches, and the refinancing itself is the event that matters. Reading the maturity profile is the analysis that a spread chart cannot replace. The second structural feature is where the assets are held and how they are marked. Credit that trades in a market has a price that moves daily and therefore a spread that responds immediately. Credit held by lenders or funds that report valuations periodically, or that hold to maturity, has a price that moves slowly and sometimes not at all — the marks are an estimate rather than a transaction. That opacity is not fraud and it is not necessarily wrong; it does mean that **a smooth return series in a less-liquid credit strategy is not the same evidence as a smooth return series in a traded one**, and that the adjustment, when it comes, arrives in a step rather than as a drift. The 2022–2023 period showed both directions of this: traded credit repriced visibly while some privately held credit reported far less movement. The third feature is the **divergence between credit and equity**, which is a signal worth watching because the two markets price overlapping cash flows. Equities are a claim on residual cash flows and are usually more sensitive to growth; credit is a claim on a fixed payment and is more sensitive to default and liquidity. When credit is deteriorating while equities are making highs, the two markets disagree about something — typically about whether the growth is durable — and the disagreement resolves. It is not a timing tool, and it does not tell you which market is right. It tells you that one of them is going to be paid for being wrong. Three things a spread chart does not show you — The maturity profile: When the money has to come back — the real constraint · How the assets are marked: Traded prices move continuously; periodic valuations move in steps · The equity-credit divergence: Two markets pricing the same cash flows and disagreeing ←

The instruments: cash bonds, index derivatives and the basis between them

The spread in this lesson lives on several instruments that are related but not interchangeable, and knowing which one is being quoted prevents a large category of errors. In the **cash market** you buy a bond from an issuer and receive a coupon; the spread over the government curve is a credit spread and the position is funded by owning the bond. In the **derivative market** you buy protection through a credit default swap on a single name, or on an index of names: the main families are the North American investment-grade and high-yield indices and their European equivalents, each with a series that rolls periodically and a set of standard maturities. The spread on a CDS is the price of insurance rather than a yield, and it responds immediately to the same news as the cash bond with far less settlement friction, which is why the derivative is usually where the market’s opinion is expressed first. The difference between the two prices is the **basis**: cash bond spread minus the equivalent CDS spread, and it is positive or negative for reasons that are mostly technical — balance-sheet costs of holding the bond, the deliverability of the cheapest security into the contract, and the supply of the cash bond relative to demand for protection. The index versions add a second layer of machinery that matters for reading the level. An index of credit default swaps is a portfolio of insurance contracts on a fixed set of issuers, weighted, with a fixed maturity and a periodic roll from one series to the next; buying the index is not buying bonds and does not pay a coupon in the ordinary sense, so its “spread” is a price that must be compared with the cash market carefully. The practical advantage is liquidity: a position in an index can be taken and unwound in a size that would move a single-name cash bond, which is why the index spread is the number quoted in a headline and why a widening index spread with stable single-name spreads is usually a signal about liquidity or about the composition of the index rather than about any issuer’s credit. The trap is the same one the previous read named from the other direction: the index changes composition at each roll and through downgrades, so a level that has moved can be measuring who is in the index rather than how the market feels about credit. The last piece is the **equity-credit divergence**, which the previous read introduced and which here gets its instrument detail. Because credit is a claim on a fixed payment and equity is a residual claim, the two markets disagree most usefully when a company’s debt is trading as though default were plausible while its equity is priced as though growth were intact. The signal is stronger when the driver is idiosyncratic — a single issuer whose spread has widened sharply against a stable sector index — than when the whole index has widened with equities, which is usually a common discount-rate move rather than a disagreement about default. And the direction of the trade is not implied: credit can be early and equities can be late, or credit can be reacting to a refinancing wall that the equity market has correctly decided is solvable. What the divergence tells you with confidence is that the two markets cannot both be right, which makes it a prompt to read the filing rather than a reason to take either side. The same credit, four ways to own it — Cash bond: A coupon and a credit spread — funded, and priced by the issuer’s own supply and demand · Single-name CDS: The price of insurance; reacts fastest, settles least · Index CDS: Liquid and headline-quoted, with a rolling series and a changing composition ← · The basis: Cash spread minus CDS spread — balance-sheet, deliverability and supply effects ← The comparison that keeps an analysis honest: a widening index spread with stable single names is a liquidity or composition story; a widening single name against a stable index is a credit story. Both are real information, and they lead to different actions.

Sources

• Credit spreads, default rates and recoveryStandard fixed-income analytics; rating-agency default and recovery studies
• Credit spreads as a predictor of activityGilchrist & Zakrajšek (2012), “Credit Spreads and Business Cycle Fluctuations”
• Liquidity and uncertainty in the credit risk premiumStandard literature on the credit risk premium and the excess bond premium