We use a panel of investment-grade bonds to investigate why credit spreads are so much larger than expected losses from default. We find that systematic factors contribute little to spreads, even if higher moments or downside effects are incorporated. Instead, two idiosyncratic risk factors,...