Credit Risk Modeling Approaches › Portfolio Credit Risk Models

Individual default probabilities tell only part of the story of portfolio risk. When managing a portfolio of loans or bonds, the key risk driver is default correlation: the tendency for multiple obligors to default together. A portfolio of 1,000 independent loans with 1% default probability each has predictable losses around the mean, with the law of large numbers ensuring relatively modest deviations. But if defaults are correlated, as during economic downturns, the portfolio can experience catastrophic losses far exceeding expectations. Understanding and modeling this correlation is the central challenge of portfolio credit risk management.

Consider two obligors with individual default probabilities $p_1$ and $p_2$. If defaults were independent, the joint probability would simply be the product $p_1 \times p_2$. But in reality, defaults are positively correlated, economic conditions that cause one firm to default often affect others as well. Under the Gaussian copula framework, we model each obligor's creditworthiness through a latent variable that follows a standard normal distribution. This latent variable can be thought of as an abstract measure of the firm's financial health, with lower values indicating worse conditions. Obligor $i$ defaults when this latent variable falls below the threshold $\Phi^{-1}(p_i)$.

When the two latent variables have correlation $\rho$, we can calculate the joint default probability using the bivariate normal distribution. The probability that both obligors default is the probability that both latent variables simultaneously fall below their respective thresholds:

$$P(\text{both default}) = \Phi_2\left(\Phi^{-1}(p_1), \Phi^{-1}(p_2), \rho\right)$$

where:

• $P(\text{both default})$: probability that both obligors 1 and 2 default
• $p_1, p_2$: individual default probabilities for obligors 1 and 2
• $\rho$: correlation coefficient between the latent variables driving defaults (ranges from -1 to 1)
• $\Phi^{-1}(\cdot)$: inverse of the cumulative standard normal distribution function
• $\Phi^{-1}(p_i)$: standard normal threshold corresponding to default probability $p_i$ (the critical value below which default occurs)
• $\Phi_2(z_1, z_2, \rho)$: bivariate standard normal cumulative distribution function, giving $P(Z_1 \leq z_1, Z_2 \leq z_2)$ when $Z_1$ and $Z_2$ are correlated standard normals with correlation $\rho$
• This Gaussian copula approach elegantly separates marginal behavior (individual default probabilities) from dependence structure (correlation)

The copula approach gains its flexibility from this separation of concerns. The marginal default probabilities can be estimated or calibrated independently for each obligor using any method (structural models, credit scores, historical data). The copula then combines these marginals with a dependence structure specified by the correlation parameter. This modularity makes the framework highly flexible and practical for large portfolios with heterogeneous obligors.

The figure illustrates how joint default probability increases with correlation across different credit quality levels. For investment-grade obligors with 1% PD, the joint default probability under independence is just 1 basis point (0.01%), but rises to approximately 30 bps at 70% correlation. For speculative-grade obligors with 5% PD, the increase is even more dramatic: from 25 bps under independence to over 200 bps at high correlation. The dotted horizontal lines show the independent joint probability for reference, which shows how correlation can multiply joint default risk by factors of 10x or more. This visualization shows why correlation is central for portfolio credit risk.

The Gaussian copula framework, widely used before the 2008 crisis for CDO pricing (as discussed in Part II Chapter 14), models defaults through a latent factor structure. The key insight is that defaults across a portfolio can be driven by both common factors affecting all firms, and idiosyncratic factors unique to each firm. Each obligor's creditworthiness depends on both systematic (economy-wide) and idiosyncratic (firm-specific) factors:

$$A_i = \sqrt{\rho_i} M + \sqrt{1 - \rho_i} \varepsilon_i$$

where:

• $A_i$: latent "asset return" variable driving obligor $i$'s default (higher values indicate better creditworthiness)
• $i$: index for the obligor
• $M$: common systematic factor representing economy-wide conditions (e.g., GDP growth, market returns), with $M \sim N(0,1)$
• $\varepsilon_i$: obligor $i$'s idiosyncratic factor representing firm-specific risk independent of other obligors, with $\varepsilon_i \sim N(0,1)$
• $\rho_i$: correlation coefficient between obligor $i$'s asset return and the systematic factor (determines systematic risk exposure, ranges from 0 to 1)
• $\sqrt{\rho_i}$: weight on the systematic factor
• $\sqrt{1 - \rho_i}$: weight on the idiosyncratic factor
• The weights ensure $A_i$ has unit variance when $M$ and $\varepsilon_i$ are independent standard normal random variables
• Since $M$ and $\varepsilon_i$ are independent standard normals, $A_i$ is also standard normal: $A_i \sim N(0,1)$

Obligor $i$ defaults if $A_i < \Phi^{-1}(p_i)$, where $p_i$ is their individual default probability. The threshold is chosen so that the unconditional default probability matches the calibrated value for each obligor.

The factor structure elegantly captures how default correlation emerges from exposure to common economic conditions. When the systematic factor $M$ is low (representing an economic downturn), all obligors' $A_i$ values decrease simultaneously by $\sqrt{\rho_i} M$, increasing the probability that multiple firms cross their default thresholds together. The correlation parameter $\rho_i$ controls this sensitivity: obligors with high ρ_i are strongly affected by systematic conditions and thus highly correlated with each other, while obligors with low $\rho_i$ are driven primarily by idiosyncratic factors and have weak correlation with others.

This single-factor structure explains why defaults cluster during recessions even among apparently unrelated firms. A retailer and a manufacturer in different industries may seem independent, yet both are affected by consumer confidence, credit availability, and overall economic growth. The systematic factor captures these common influences. During normal times, the idiosyncratic factors dominate and defaults appear somewhat random. But during severe recessions when $M$ takes extreme negative values, the systematic component pushes many firms toward their default thresholds simultaneously, creating the default clustering that devastates credit portfolios.

The loss distribution exhibits the characteristic positive skew of credit portfolios. Most simulated scenarios produce losses near the $2.72 million expected value, clustering around the peak of the distribution. But the fat right tail extends materially beyond, which reflects scenarios where many defaults coincide. The$16.14 million 99% VaR is about 5.9 times expected loss, showing that tail risk materially exceeds typical outcomes. This asymmetry arises because losses are bounded below at zero (you cannot have negative credit losses) but can become much larger when multiple defaults coincide. Extreme losses, while rare, are the primary concern for risk managers because they represent scenarios that can threaten portfolio solvency.

Default correlation is the key driver of portfolio tail risk. Let's examine how correlation affects the loss distribution to develop intuition for this critical relationship: