IFRS quantitative finance series
Lifetime PD Term Structures for IFRS 9 Banking Portfolios
Lifetime PD is a sequence of conditional credit-risk estimates over the remaining life of an exposure. Treating it as a flat annual rate can materially distort the timing and amount of expected losses, especially for amortising loans and portfolios with seasoning effects.
From one-year PD to a curve
Banks may use survival models, transition matrices, vintage curves or hazard-rate approaches. Whatever the method, marginal PDs must sum consistently to cumulative default risk and survival cannot become negative. Extrapolation beyond observed data requires explicit assumptions.
Conditioning on the economy
Point-in-time term structures should respond to current and forecast conditions. The mapping from macro variables to hazard rates must avoid implausible oscillations and should preserve rank ordering where that is part of model design.
Validation over a long horizon
Short histories make lifetime validation difficult. Useful tests include calibration by horizon, survival-curve comparison, transition stability, benchmark curves and sensitivity to extrapolation. Governance should distinguish model evidence from management judgement.
Quantitative expression
Implementation controls
- Check cumulative PD is monotonic and bounded by one.
- Separate observed calibration from extrapolation.
- Validate by vintage and remaining maturity.
- Monitor economic sensitivity and rank stability.