Portfolio Book & Macro-Conditioned Default Probabilities ($P_{j, t}$)
CreditPortfolioView Engine| Obligor & Sector | Rating | Base PD (%) | Cycle Factor ($r_{j, t}$) | Conditioned PD (%) | EAD ($M) | LGD (%) | Expected Loss ($M) | Component UL | Action |
|---|
Non-Linear Sector Default Rate Sensitivity to GDP Shocks
Notice how the econometric logit curve captures asymmetric credit cycle non-linearities: during sharp GDP contractions, default rates accelerate exponentially before tapering.
Unconditional vs Stressed Default Probabilities across Rating Grades (AAA to CCC)
CreditPortfolioView dynamically stretches and skews the transition matrix. In severe recessions, default probabilities multiply across all rating grades.
Simulated Credit Loss Distribution under Macroeconomic Stress ($10,000$ Scenarios)
Monte Carlo simulated portfolio credit losses with colored quantile thresholds: Blue (Normal Loss Range), Amber (95% VaR Tail), Red (99% Extreme Tail).
Sensitivity: GDP Growth vs Unemployment on Default Rate (%)
Sensitivity: Macro Volatility ($\sigma_{\text{macro}}$) vs EAD Scale on 99% CreditVaR ($M)
The CreditPortfolioView (Wilson / McKinsey) Framework
Published in 1997-1998 by Thomas C. Wilson (Head of Risk Management at McKinsey & Company), CreditPortfolioView completed the foundational quartet of modern quantitative portfolio credit risk models. Unlike structural models that rely on equity prices (KMV) or static historical rating averages (CreditMetrics), CreditPortfolioView is explicitly designed to model credit risk as a dynamic, multi-period econometric function of macroeconomic cycles.
1. Multi-Factor Macro Logit Index
For each country and industry sector $j$, default probability is driven by a systematic macroeconomic index $Y_{j, t}$: $$Y_{j, t} = \beta_{j, 0} + \sum_{k=1}^K \beta_{j, k} X_{k, t} + \epsilon_{j, t}$$ where $X_{k, t}$ represents observable macroeconomic variables (GDP growth, unemployment rate, interest rates, government expenditures). The logit link bounds default probability strictly in $(0, 1)$: $$P_{j, t} = \frac{1}{1 + \exp(-Y_{j, t})}$$
2. Credit Cycle Index ($r_{j, t}$)
The model compares conditional default probability $P_{j, t}$ to the long-run historical baseline $\bar{P}_j$: $$r_{j, t} = \frac{P_{j, t}}{\bar{P}_j}$$ In deep economic recessions, $r_{j, t} > 1$, scaling default probabilities upward. In expansions, $r_{j, t} < 1$, reflecting benign default environments.
3. Stressed Transition Matrices
Wilson conditions the entire multi-state credit migration matrix $M_t = [P_{ik}(t)]$ on the macroeconomic state $r_{j, t}$. Default columns are scaled ($P_{i, D}(t) = \bar{P}_{i, D} \times r_{j, t}$), downgrades are elevated, and upgrades are curtailed, providing the exact engine used in Federal Reserve CCAR stress tests.
Institutional Knowledge Check: CreditPortfolioView
Frequently Asked Questions
The Canonical "Big Four" Credit Portfolio Models
CreditMetrics (1997)
J.P. Morgan rating migration model with correlated Gaussian asset returns and forward credit curves.
CreditRisk+ (1997)
Credit Suisse actuarial Poisson mixture model with Panjer probability generating function recursion.
KMV Model (1993)
Moody's structural asset inversion, empirical default point, and Distance-to-Default ($DD$) EDF™.
CreditPortfolioView (1997)
McKinsey econometric logit model linking macroeconomic cycles directly to rating migrations.