Why Credit Spreads Do Not Directly Measure Default Probability

Summary

The document explains why the yield spread between a corporate bond and a risk-free bond cannot be read as a default probability on its own. The spread reflects expected default losses, including both the chance of default and the loss given default; collateral and recovery prospects can therefore affect yields even when default risk is high.

Liquidity also contributes: corporate bonds may be harder to sell than government securities, especially under pressure. The answer distinguishes realized or physical default probabilities from risk-neutral probabilities inferred from market prices under modeling assumptions, and notes that substantial empirical literature studies their relationship. It points to a long historical study of corporate bond risk but supplies no calibration statistics or detailed empirical findings. The discussion is conceptual, so it does not provide a formula for converting spreads into default forecasts or assess predictive performance for particular bonds or periods.

Key ideas

• A corporate bond spread reflects more than the probability of default.
• Loss given default and collateral influence the spread alongside default likelihood.
• Corporate bond liquidity can add a yield premium relative to government bonds.
• Realized default probabilities differ from risk-neutral probabilities inferred from prices.
• The document identifies empirical research as relevant but gives no calibration results.

Tags

# Do default risk implied by bond yield actually predict it?


# Do default risk implied by bond yield actually predict it?












AFAIU in perfect market difference between risk-free rate and corporate bond yield should be only influenced by default risks.

How well do they actually predict defaults? Are there any calibration stats or research about this?

## Answer by Dimitri Vulis (score 4)

https://quant.stackexchange.com/a/81656

The amount of additional yield that investors demand for holding credit risky bonds is driven not only by the probability of default (PD), but also by loss given default (LGD) (more important as PD increases - a highly collateralized secured bond may trade at relatively low spread even on the verge of default), liquidity (it may be harder to sell a corporate bond than a treasury bond in secondary markets, especially in a hurry), and other moving parts. There are physical / actual / realized / observed probabilities (historically, what percentage of similar issuers defaulted in real life; the likelihood of future defaults perceived by various people) and risk-neutral probabilities (derived from observables using various assumptions). There's a huge body of literature on empirical analysis of their relationships. For a historical perspective, I highly recommend this paper: Giesecke, Longstaff, Schaefer, Strebulaev. Corporate Bond Default Risk: A 150-Year Perspective.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.