What Is Probability of Default (PD)?
Probability of default estimates the chance that a borrower experiences a credit event within a defined window, usually the next twelve months. Default itself can take several forms — a missed interest payment, a covenant breach that accelerates the debt, a bankruptcy filing, or a distressed exchange in which creditors accept less than they were promised. PD captures the likelihood of any of these occurring, without saying anything about how severe the resulting loss would be.
Credit ratings are essentially a shorthand for PD. Historical rating-agency studies show that investment-grade issuers default within a year roughly 0.1% of the time or less, single-B issuers default around 3% to 4% of the time, and issuers rated CCC and below default at rates that can exceed 25%. Cumulative default probabilities rise steadily as the horizon extends beyond one year.
How PD Is Estimated
Practitioners estimate PD from several angles. Historical approaches map a borrower to a rating or internal score, then read off the observed default frequency for that cohort. Structural models in the Merton tradition treat a company's equity as a call option on its assets and infer default risk from the market value and volatility of the firm relative to its debt load. Market-implied approaches back PD out of bond spreads or credit default swap premiums, given an assumed recovery rate.
PD then plugs into the standard credit-loss formula: expected loss = PD × LGD × EAD, where LGD is the loss given default and EAD is the exposure at default. A further distinction matters in practice — point-in-time PDs move with the economic cycle, while through-the-cycle PDs smooth those swings, which is why a bank's internal estimates and an agency rating can disagree at any given moment.
Why PD Matters
PD sits at the center of how credit gets priced and regulated. Lenders set spreads so that expected losses are covered with a margin for risk, banks using the Basel internal-ratings-based approach feed their own PD estimates into required capital calculations, and accounting standards such as CECL and IFRS 9 force lenders to provision for expected losses using PD-based models from the day a loan is made.
For investors, comparing a market-implied PD against an independently modeled one is a classic way to find mispriced credit — if the bond market is pricing a 10% one-year default probability but fundamentals suggest 4%, the bonds may be cheap. Candidates targeting credit research, leveraged finance, or distressed roles should be comfortable explaining the expected-loss formula and how PD differs from loss severity.
