Wrong-Way Risk in Counterparty Exposures

Why the Relationship Between Exposure and Default Matters

Counterparty credit risk depends not only on how likely a counterparty is to default, but also on how large the amount owed to us would be if default occurred. If exposure tends to rise at the same time that the counterparty becomes less able to pay, the two risks reinforce one another. This relationship is called wrong-way risk.

For example, a financial institution might be owed money on a contract whose value increases when a broad market factor moves against a counterparty. Or the institution might hold collateral that loses value when the counterparty itself deteriorates. In either case, assuming exposure and default risk are independent can give an expected loss that is too low.

Definition: Wrong-way risk is the adverse dependence between exposure to a counterparty and the counterparty’s credit quality: exposure tends to be higher when the counterparty is more likely to default or less able to meet its obligations. General wrong-way risk arises from a broad market or economic factor that affects exposure and the counterparty’s credit quality. Specific wrong-way risk arises from a direct link between the exposure and the particular counterparty, such as collateral issued by that counterparty.

Wrong-way risk is about how variables move together in adverse conditions. A high exposure by itself is not enough to establish it, nor is a high probability of default by itself. The concern is their joint occurrence. The converse pattern, in which exposure tends to be low when default risk is high, is sometimes called right-way risk.

Expected Loss When Exposure and Default Are Dependent

Consider a single counterparty over a fixed risk horizon. Let $EAD$ be the nonnegative exposure at default, let $D$ denote the event that the counterparty defaults during the horizon, and let $LGD$ be the fraction of exposure lost if default occurs. For the calculations below, assume $LGD$ is constant and the exposure is measured consistently at the relevant default time.

L=EAD \times LGD \times \mathbf{1}_D

Here $\mathbf{1}_D$ equals 1 if default occurs and 0 otherwise. This makes the loss zero in a nondefault state and equal to exposure times loss given default in a default state. Taking expectations gives the dependence-aware expected loss:

E[L]=LGD \times E[EAD\mathbf{1}_D] =LGD \times P(D) \times E[EAD\mid D]

The second equality follows from the definition of conditional expectation: $E[EAD\mathbf{1}_D]=P(D)E[EAD\mid D]$. If exposure and default are independent, then $E[EAD\mid D]=E[EAD]$, and the familiar product of average exposure, probability of default, and $LGD$ applies. With wrong-way risk, that independence condition does not hold.

Important condition: The product $E[EAD]\times P(D)\times LGD$ is valid when $LGD$ is constant and $EAD$ is independent of default. If $LGD$ also varies with the state, the expected loss must retain that dependence: $E[EAD \times LGD \times \mathbf{1}_D]$. Averages should not be multiplied as though the variables were independent unless that assumption is justified.

General Wrong-Way Risk: A Shared Market Driver

General wrong-way risk occurs when a broad market or economic condition affects both the value of an exposure and the counterparty’s ability to pay. The factor need not be unique to that counterparty. A commodity-price decline, for instance, could increase an institution’s claim under a hypothetical contract while also weakening a counterparty whose business depends on that commodity.

Worked Example: Market Stress Raises Exposure and Default Risk

A hypothetical institution has a contract with a commodity producer. At a one-year horizon, a market review assigns an 80% probability to a stable market state and a 20% probability to a severe price decline. Exposure at default is estimated at $2.00 million in the stable state and $10.00 million in the decline state. The conditional probabilities of default are 1% and 20%, respectively. Assume a constant $LGD$ of 60%. Calculate expected loss and compare it with an independence approximation.

Method. Calculate the default-weighted loss within each market state, then weight each state’s result by its probability. All scenario values are hypothetical.

In the stable state, the conditional expected loss is $2.00 \times 0.01 \times 0.60=0.012$ million dollars. Weighting by the state probability gives $0.80 \times 0.012=0.0096$ million dollars.

In the price-decline state, the conditional expected loss is $10.00 \times 0.20 \times 0.60=1.20$ million dollars. Weighting by the state probability gives $0.20 \times 1.20=0.2400$ million dollars.

Adding the state contributions gives $0.0096+0.2400=0.2496$ million dollars, or $249,600.

For comparison, the marginal probability of default is $0.80 \times 0.01+0.20 \times 0.20=0.048$, or 4.8%. Mean exposure is $0.80 \times 2.00+0.20 \times 10.00=3.60$ million dollars. The independence approximation is $3.60 \times 0.048 \times 0.60=0.10368$ million dollars, or $103,680.

Check. The dependence-aware expected loss exceeds the independence approximation by $0.24960-0.10368=0.14592$ million dollars, or $145,920. The calculation avoids the common mistake of multiplying mean exposure by marginal default probability when both are elevated in the same adverse state.

This example illustrates the mechanism, not a universal size of adjustment. The difference depends on the exposure and default patterns across states. If high exposure occurs in states where default is unlikely, the independence approximation could instead overstate expected loss.

Specific Wrong-Way Risk: A Direct Link to the Counterparty

Specific wrong-way risk is more directly tied to the named counterparty. One important example is collateral issued by that same counterparty. If the counterparty’s financial condition worsens and the collateral loses value at the same time, the unsecured amount can rise just as the counterparty becomes more likely to default.

For a simple illustration, suppose the institution is owed a fixed amount and measures exposure as the amount owed less the value of available collateral, floored at zero. This calculation assumes the stated collateral values are available at the measurement point and does not apply any additional recovery to collateral beyond those values.

Worked Example: Collateral Issued by the Counterparty

A hypothetical counterparty owes an institution $8.00 million. The institution holds shares issued by that same counterparty as collateral. In a healthy state, the shares are worth $6.00 million; in a distressed state, they are worth $2.00 million. The probabilities of the two states are 90% and 10%. Conditional default probabilities are 1% and 25%, and $LGD$ on the calculated exposure is 50%. Find expected loss and compare it with the independence approximation.

Method. First calculate exposure after collateral in each state. Then use the state-specific default probabilities and state probabilities to calculate expected loss. All inputs are hypothetical.

In the healthy state, exposure is $8.00-6.00=2.00$ million dollars. Its conditional expected loss is $2.00 \times 0.01 \times 0.50=0.010$ million dollars, contributing $0.90 \times 0.010=0.009$ million dollars.

In the distressed state, exposure is $8.00-2.00=6.00$ million dollars. Its conditional expected loss is $6.00 \times 0.25 \times 0.50=0.750$ million dollars, contributing $0.10 \times 0.750=0.075$ million dollars.

Therefore, expected loss is $0.009+0.075=0.084$ million dollars, or $84,000.

Mean exposure is $0.90 \times 2.00+0.10 \times 6.00=2.40$ million dollars. The marginal probability of default is $0.90 \times 0.01+0.10 \times 0.25=0.034$, or 3.4%. Under independence, the estimate would be $2.40 \times 0.034 \times 0.50=0.0408$ million dollars, or $40,800.

Check. The state-based result exceeds the independence estimate by $43,200. The collateral value falls from $6.00 million to $2.00 million in the same state where default probability rises. Treating the average exposure and average default probability as independent misses this specific link.

Reading the Dependence Through Conditional Exposure

A useful way to make the connection explicit is to compare average exposure conditional on default with unconditional average exposure. For constant $LGD$, expected loss can be written as:

E[L]=LGD \times P(D) \times E[EAD\mid D]

When $E[EAD\mid D]$ is materially larger than $E[EAD]$, exposure tends to be elevated in default states, consistent with wrong-way risk. When it is smaller, the relationship may reduce expected loss relative to an independence calculation. This comparison is useful, but it does not describe every feature of the joint distribution: a few extreme exposure outcomes can matter even if a single average conceals them.

Worked Example: Using Exposure Conditional on Default

For a hypothetical counterparty over a one-year horizon, the default probability is 5%. An exposure model estimates mean exposure at default of $12.00 million conditional on default and $3.00 million conditional on survival. Assume $LGD$ is a constant 40%. Calculate expected loss and compare it with a calculation that incorrectly treats exposure as independent of default.

Method. Use conditional exposure directly in the dependence-aware expected-loss identity. All amounts and probabilities are hypothetical.

The dependence-aware expected loss is $0.05 \times 12.00 \times 0.40=0.240$ million dollars, or $240,000.

To construct the unconditional mean exposure, weight each conditional exposure by the probability of its state. Survival probability is $1-0.05=0.95$, so mean exposure is $0.05 \times 12.00+0.95 \times 3.00=0.60+2.85=3.45$ million dollars.

The independence estimate is $3.45 \times 0.05 \times 0.40=0.069$ million dollars, or $69,000. The difference is $0.240-0.069=0.171$ million dollars, or $171,000.

Check. The calculation uses the $12.00 million exposure conditional on default for the default-loss calculation; substituting the $3.45 million unconditional mean would incorrectly erase the specified dependence.

Common Traps and Practical Interpretation

Assuming independence without checking. The product of mean exposure and marginal default probability is not generally equal to expected exposure in default states.
Confusing higher default probability with wrong-way risk. A counterparty can have high default risk while exposure remains unrelated or even falls. Wrong-way risk concerns the adverse joint behavior.
Looking only at broad market factors. A direct link to one counterparty, such as its own securities being used as collateral, is a specific form of wrong-way risk.
Using average exposure as if it were exposure at default. The relevant quantity for expected loss is $E[EAD\mid D]$ when exposure and default are dependent.
Assuming a single correlation measure settles the question. The important issue is how exposure behaves in adverse default states. A summary correlation can obscure that tail behavior or nonlinear links may matter.
Ignoring horizon and measurement consistency. Exposure, default probability, and $LGD$ must refer to a compatible horizon and loss definition. A calculation combining mismatched measures may not represent the intended credit loss.

Wrong-way risk matters because it can make losses more concentrated precisely when a counterparty is least able to pay. Identifying the common driver or direct counterparty link helps analysts assess whether an independence assumption is credible and where stress analysis is warranted. This conceptual assessment is aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus.

Key takeaway: Calculate credit loss from the joint behavior of exposure and default whenever they may be dependent. General wrong-way risk links both through a broad market factor; specific wrong-way risk links exposure directly to the counterparty.

Check Your Understanding

For each problem, identify the relevant dependence and calculate the requested quantity. Monetary amounts are hypothetical.

A hypothetical market has a 70% probability of a calm state and a 30% probability of a stressed state. Exposure at default is $1.50 million in calm conditions and $7.00 million in stress. Conditional default probabilities are 2% and 15%, respectively, and $LGD$ is 40%. Calculate the state-based expected loss and the independence estimate. (Answer: state-based expected loss $134,400; independence estimate $74,340.)
A counterparty’s own shares secure an obligation. The shares lose value when that counterparty’s financial condition deteriorates, and the unsecured exposure rises as its default probability rises. Is this general or specific wrong-way risk, and why? (Answer: Specific wrong-way risk, because the exposure is directly linked to the credit condition of the same counterparty.)
A hypothetical institution is owed $5.00 million and holds the counterparty’s own securities as collateral. Healthy and distressed states have probabilities 80% and 20%; collateral values are $4.00 million and $1.00 million; conditional default probabilities are 1% and 30%. Assume $LGD=50\%$. Calculate expected loss using the state-specific information. (Answer: $124,000.)
A one-year default probability is 4%. Mean exposure conditional on default is $9.00 million, and constant $LGD$ is 45%. Calculate expected loss using conditional exposure. (Answer: $162,000.)
Exposure is $2.00 million in the low-exposure state and $8.00 million in the high-exposure state, with each state having probability 50%. Conditional default probabilities are 1% and 9%, respectively, and $LGD$ is 40%. Calculate the state-based expected loss and the independence estimate. (Answer: state-based expected loss $148,000; independence estimate $100,000.)

What You'll Learn