Table of Contents
1. Introduction
This paper addresses a critical gap in credit risk modeling: the explicit incorporation of exchange rate (FX) risk into the assessment of a borrower's Probability of Default (PD) and the asset correlations between borrowers. Intuitively, a borrower whose assets and liabilities are denominated in different currencies faces additional volatility, increasing their default risk. This increase manifests not only in a higher individual PD but also in stronger default dependence (higher asset correlation) among similarly exposed borrowers. The author combines established models—Merton's (1974) structural default model, the Garman-Kohlhagen (1983) currency option model, and Vasicek's (2002) asymptotic single risk factor model—to derive parsimonious formulas linking PDs and correlations with and without FX risk.
2. Background of the Model
The model's foundation lies in representing key economic variables as stochastic processes.
2.1 Asset Value Process
The borrower's asset value $A(t)$ follows a geometric Brownian motion (GBM):
$dA(t) = \mu A(t)dt + \sigma A(t)dW(t)$
Equivalently, $A(t) = A_0 \exp\left((\mu - \sigma^2/2)t + \sigma W(t)\right)$, where $\mu$ is the drift, $\sigma$ is the asset volatility, and $W(t)$ is a standard Brownian motion.
2.2 Exchange Rate Process
The exchange rate $F(t)$ (units of debt currency per unit of asset currency) is also modeled as a GBM:
$dF(t) = \nu F(t)dt + \tau F(t)dV(t)$
Equivalently, $F(t) = F_0 \exp\left((\nu - \tau^2/2)t + \tau V(t)\right)$, where $\nu$ is the drift, $\tau$ is the FX volatility, and $V(t)$ is another standard Brownian motion. The two Brownian motions are correlated with parameter $r$: $\text{corr}[V(t)-V(s), W(t)-W(s)] = r$.
2.3 Default Condition with FX Risk
Default occurs at time $t=1$ if the asset value converted into the debt currency falls below the debt level $D$:
$F(1)A(1) \leq D$.
This can be conveniently normalized by today's exchange rate $F_0$ to express the debt in the asset's local currency: $F^*(1)A(1) \leq D^*$, where $F^*(t)=F(t)/F_0$ and $D^*=D/F_0$.
3. Derivation of Key Results
Under the model assumptions, the author derives closed-form expressions for PD and asset correlation under FX risk.
3.1 Adjusted Probability of Default (PD)
The PD under FX risk, $p^*$, is given by the probability that the combined log-asset process falls below the log-debt threshold. Assuming independence between the asset and FX processes ($r=0$) and a zero drift for the FX rate ($\nu = 0$), the adjusted PD is:
$p^* = \Phi\left( \frac{\ln(A_0/D^*) - (\mu - \sigma^2/2)}{\sqrt{\sigma^2 + \tau^2}} \right)$
Compared to the one-currency PD $p = \Phi\left( \frac{\ln(A_0/D^*) - (\mu - \sigma^2/2)}{\sigma} \right)$, the denominator increases from $\sigma$ to $\sqrt{\sigma^2 + \tau^2}$, leading to a higher PD ($p^* > p$) for the same distance to default, as the total volatility increases.
3.2 Adjusted Asset Correlation
The asset correlation $\varrho^*$ between two borrowers under FX risk also increases. If both borrowers are exposed to the same FX risk factor, their asset values become more correlated because they share an additional common shock from the exchange rate movement.
3.3 The Core Consistency Condition
The most powerful result is a parameter-free consistency condition linking the changes in PD and asset correlation. For two borrowers with identical risk profiles, it simplifies to:
$\frac{1-\varrho^*}{1-\varrho} = \frac{[\Phi^{-1}(p^*)]^2}{[\Phi^{-1}(p)]^2}$
This equation (Equation (1) in the paper) implies that one cannot arbitrarily adjust PDs and asset correlations for FX risk independently; they are intrinsically linked. An increase in PD ($p^* > p$) must be accompanied by an increase in asset correlation ($\varrho^* > \varrho$).
4. Key Insights & Analyst's Perspective
Core Insight: Tasche's work isn't just a mathematical exercise; it's a formal indictment of the common, siloed approach to market and credit risk. The paper proves that FX volatility doesn't merely add a flat premium to credit spreads—it fundamentally alters the joint failure dynamics of obligors. The derived consistency condition is a powerful sanity check: if your FX-adjusted PDs go up but your correlations remain static, your model is internally inconsistent and likely underestimating portfolio tail risk.
Logical Flow: The argument is elegantly simple. 1) Model assets and FX rates as correlated GBMs. 2) Define default via the converted asset value. 3) Observe that the effective volatility driving default is $\sqrt{\sigma^2 + \tau^2}$. 4) This higher volatility increases both the marginal default probability (PD) and the co-movement (correlation) between firms exposed to the same FX factor. The final consistency condition emerges naturally from this geometry.
Strengths & Flaws: The major strength is tractability. By making standard (if strong) assumptions—GBM, independence, zero FX drift—the model yields a clean, usable formula. This is far more actionable for risk managers than complex, computationally heavy simulations. The flaw, however, is in those very assumptions. The Garman-Kohlhagen model, while foundational, is known to struggle with capturing FX volatility smiles and jumps, as noted in more recent literature (e.g., Bakshi, Cao, and Chen, 1997). Assuming independence between a firm's asset value and the FX rate is also a significant limitation, especially for export-oriented firms whose fortunes are directly tied to currency moves. The model, as presented, is a first-order approximation.
Actionable Insights: For practitioners, this paper mandates a procedural change. First, validate your correlations. Use the consistency condition to back-test whether historically estimated PD-correlation pairs for internationally active firms align with the model's predictions during periods of high FX volatility. Second, stress-test your portfolio. Apply the formula to shock PDs and correlations simultaneously under a severe FX shock scenario, rather than in isolation. This will reveal concentrated vulnerabilities that standard models miss. Finally, this work underscores the need for integrated risk platforms. As the regulatory landscape evolves towards principles like Basel III's interest rate risk in the banking book (IRRBB), which acknowledges currency risk, models like Tasche's provide a foundational quantitative argument for breaking down the silos between market and credit risk departments.
5. Technical Details & Mathematical Framework
The core mathematical derivation involves characterizing the log of the normalized asset value $X = \ln(F^*(1)A(1)/A_0)$. Under the model assumptions:
$X \sim N\left(\mu - \frac{\sigma^2 + \tau^2}{2}, \sigma^2 + \tau^2 + 2r\sigma\tau\right)$
The default condition $F^*(1)A(1) \leq D^*$ becomes $X \leq \ln(D^*/A_0)$. The PD is therefore $p^* = \Phi\left( \frac{\ln(D^*/A_0) - (\mu - (\sigma^2+\tau^2)/2)}{\sqrt{\sigma^2 + \tau^2 + 2r\sigma\tau}} \right)$. The consistency condition is derived by considering the asset values of two firms and applying the Vasicek (2002) asymptotic single risk factor model, which links default thresholds to asset correlations.
6. Analytical Framework: A Practical Case Example
Scenario: A European bank has a loan portfolio containing two manufacturing firms, Firm A (German, assets in EUR, debt in USD) and Firm B (Japanese, assets in JPY, debt in USD). The bank has estimated their one-currency PDs as $p_A = p_B = 1\%$ and an asset correlation of $\varrho = 15\%$, ignoring FX risk.
Analysis: The bank now wants to incorporate USD/EUR and USD/JPY risk. Using internal models, they estimate that the additional FX volatility increases each firm's PD to $p^*_A = p^*_B = 1.5\%$.
Application of Consistency Condition: The bank must now adjust the asset correlation. Using the formula:
$\frac{1-\varrho^*}{1-0.15} = \frac{[\Phi^{-1}(0.015)]^2}{[\Phi^{-1}(0.01)]^2} = \frac{(-2.17)^2}{(-2.33)^2} \approx 0.87$
Solving gives $\varrho^* \approx 1 - 0.87*(0.85) \approx 26\%$.
Interpretation: The introduction of a common FX risk factor (USD strength) not only raises individual default risk by 50% (from 1% to 1.5%) but also increases the default dependence between the two firms significantly, from 15% to 26%. A portfolio model that only adjusts the PDs would substantially underestimate the risk of multiple defaults occurring simultaneously during a USD appreciation event.
7. Application Outlook & Future Directions
The implications of this research extend beyond traditional corporate lending.
- Climate Risk & Just Transition: The framework can be adapted to model how physical climate risks (e.g., floods) or transition risks (carbon taxes) act as a new, systematic "factor" increasing both PDs and correlations for exposed sectors, similar to the FX factor.
- Cryptocurrency & DeFi Lending: In decentralized finance, where loans are often collateralized in volatile cryptocurrencies, the model's logic is directly applicable. The volatility of the collateral asset ($\tau$) drastically increases counterparty risk and correlation in lending pools.
- Regulatory Capital (Basel IV): The model provides a theoretical basis for arguing that the Foundation Internal Ratings-Based (F-IRB) approach's fixed asset correlation assumptions may be inadequate for portfolios with significant FX mismatch, potentially justifying the use of advanced approaches.
- Future Research: Key extensions include relaxing the independence assumption to model firms with natural hedges or export dependencies, incorporating stochastic volatility for both assets and FX rates (e.g., Heston model), and empirical validation of the consistency condition across different economic cycles and currency regimes.
8. References
- Merton, R. C. (1974). On the pricing of corporate debt: The risk structure of interest rates. The Journal of Finance, 29(2), 449-470.
- Garman, M. B., & Kohlhagen, S. W. (1983). Foreign currency option values. Journal of International Money and Finance, 2(3), 231-237.
- Vasicek, O. (2002). The distribution of loan portfolio value. Risk, 15(12), 160-162.
- Bakshi, G., Cao, C., & Chen, Z. (1997). Empirical performance of alternative option pricing models. The Journal of Finance, 52(5), 2003-2049.
- Basel Committee on Banking Supervision. (2016). Standards: Interest rate risk in the banking book. Bank for International Settlements.