Welcome to Portfolio Credit Risk!
Hi there! If you’ve made it to FRM Part II, you already know that credit risk is the possibility that a borrower won’t pay back a loan. But in the real world, banks don't just hold one loan; they hold thousands! In this chapter, we transition from looking at individual loans to looking at the portfolio as a whole.
Why is this important? Because in a portfolio, the "magic" of diversification can reduce risk, but the "danger" of correlation can magnify it. Understanding how these loans interact is the difference between a stable bank and a financial crisis. Don't worry if this seems a bit math-heavy at first—we’ll break it down piece by piece!
1. The Foundation: Expected Loss (EL) vs. Unexpected Loss (UL)
To manage a portfolio, we first need to distinguish between what we expect to lose and what surprises us.
Expected Loss (EL): This is the average loss a bank anticipates over a period. It’s a "cost of doing business." Banks cover EL using provisions (like a rainy-day fund).
For a portfolio, the Total EL is simply the sum of individual ELs:
\( EL_p = \sum_{i=1}^{n} EL_i \)
\( EL_i = EAD_i \times PD_i \times LGD_i \)
Unexpected Loss (UL): This is the volatility or uncertainty around the EL. This is the risk that keeps risk managers awake at night! We use Economic Capital to protect against UL.
Important: Unlike EL, you cannot simply add up individual ULs to get the portfolio UL. Why? Because of correlation.
Quick Analogy: Imagine you are a florist. You expect 5% of your roses to wilt every week (EL). That’s a known cost. However, a sudden heatwave might cause 40% of them to wilt (UL). The EL is your cost; the UL is your risk.
Key Takeaway: EL is additive; UL is NOT additive because of the diversification effect.
2. The Role of Default Correlation
Default correlation measures the tendency of two borrowers to default at the same time. This is the "secret sauce" (or the poison) of portfolio credit risk.
Why do defaults correlate?
• Systemic Factors: A recession hits everyone. If the economy crashes, many businesses fail together.
• Contagion: If a major car manufacturer fails, all its small parts suppliers might fail too.
Did you know? During the 2008 financial crisis, many models failed because they assumed correlations were low. In reality, when things get bad, correlations tend to "spike" to 1. Everything crashes at once!
The Math of Portfolio UL:
For a two-asset portfolio, the UL is:
\( UL_p = \sqrt{UL_1^2 + UL_2^2 + 2 \rho_{12} UL_1 UL_2} \)
Where \( \rho_{12} \) is the correlation between the two assets. If correlation is 1, there is no diversification. If correlation is less than 1, the portfolio UL is less than the sum of the parts!
3. Credit Value-at-Risk (Credit VaR)
Credit VaR is the maximum loss a portfolio is expected to suffer over a given time horizon at a specific confidence level (e.g., 99.9%).
Credit VaR vs. Market VaR:
Market VaR is usually calculated over 1 or 10 days. Credit VaR is usually calculated over 1 year because credit events happen slowly. Also, credit loss distributions are not "Normal"—they are skewed (most of the time you lose nothing, but occasionally you lose a lot).
Economic Capital formula:
\( Economic Capital = Credit VaR - EL \)
Banks hold capital to cover the "unexpected" part of the tail risk, not the average loss.
Quick Review Box:
• EL: Budgeted as an expense.
• UL: Requires capital to buffer against.
• Credit VaR: The worst-case loss at a specific confidence level.
4. The Vasicek Model (Large Homogeneous Portfolios)
The curriculum often highlights the Vasicek Model (also known as the One-Factor Model). This is used to calculate the Credit VaR of a very large portfolio of similar loans (like credit cards or mortgages).
The Big Idea: We assume there is one common factor (the "State of the Economy") and many idiosyncratic factors (individual bad luck).
As the number of loans in the portfolio grows to infinity, the individual (idiosyncratic) risks cancel out, and only the systemic risk remains.
Common Mistake: Students often forget that the Vasicek model assumes the portfolio is infinitely granular. This means no single loan is big enough to impact the whole portfolio.
5. Risk Contributions
If you have a portfolio, you need to know which loan is adding the most risk. We use two main measures:
Marginal Risk Contribution (MRC): This tells us how the total portfolio risk changes if we add a small amount of a new loan. It helps in deciding if a new loan is "worth" the risk it adds.
Component VaR: This breaks down the total portfolio VaR into parts, showing how much of the total risk is "owned" by each asset.
Memory Trick: Think of a soup. The EL is the cost of the ingredients. The Portfolio Risk is how spicy it is. The Marginal Contribution is how much one extra drop of hot sauce would increase the overall spiciness.
6. Copulas: Linking Everything Together
Copulas are mathematical tools used to model the dependence structure between variables without worrying about their individual distributions.
The Gaussian Copula: This was the famous "formula that downed Wall Street." It assumes that the joint default relationship follows a Normal distribution.
The Problem: The Gaussian Copula often underestimates "Tail Dependence." In simple terms, it fails to realize that when one borrower defaults in a crisis, others are much more likely to follow than a Normal distribution would suggest.
Key Takeaway: Copulas allow us to combine individual probabilities of default into a joint probability for the whole portfolio.
7. Summary and Final Tips
Keep these points in mind for the exam:
• Diversification reduces UL, but never EL.
• Default Correlations are the biggest drivers of portfolio risk; as correlation increases, the "tail" of the loss distribution becomes fatter.
• Economic Capital is designed to cover Credit VaR minus EL.
• Granularity is your friend—the more small, independent loans you have, the more individual risk cancels out.
Don't worry if the formulas for Copulas or the Vasicek model look intimidating. On the FRM exam, focus on the relationships: If correlation goes up, what happens to VaR? (It goes up!) If the portfolio becomes more granular, what happens to idiosyncratic risk? (It goes down!)
You've got this! Portfolio Credit Risk is all about seeing the "Big Picture" beyond individual loans.