Counterparty Risk and XVA

If you work on credit limit management, CVA management, or trade pricing, you may see different figures for the same counterparty. One reason the figures differ is that credit limit management uses the tail of the distribution and CVA management uses the average.

Why one counterparty is viewed in two ways

The credit risk of a derivatives counterparty is handled in two kinds of work, credit limit management and pricing. Credit limit management puts a cap on what the firm can afford to lose if a single counterparty defaults, and it is the management of counterparty credit risk. The measure it uses is potential future exposure (PFE), a high percentile of the distribution of future exposure. Pricing puts the expected cost of the loss into the trade's price, and it finds CVA from expected exposure (EE), the average of that distribution.

Limits use the tail of the distribution because they express how large a risk you are willing to accept from one counterparty. The average is considered appropriate when you add up across counterparties or use it inside CVA¹.

The two views are used together, not chosen between. The 2011 US guidance calls it sound practice to set limits on CVA or on the VaR of CVA, but says this does not remove the need to limit exposure to a counterparty by PFE¹.

The guideline that the Basel Committee finalized in December 2024 lists the uses of exposure measures as capital rules (SA-CCR and internal models), pricing (XVA), and risk monitoring (PFE and stress tests), among others. It asks firms to understand the strengths and limits of each measure and to use them in combination². The definitions of the measures are covered on the Exposure Measures: CE, PFE, EE/EPE and EAD page.

Example: a three-year uncollateralized trade with Company M

Consider a hypothetical example in which a bank has a three-year trade with Company M, with no collateral agreement. This is the only trade with Company M. The table summarizes the exposure (the positive part of the market value, or 0 if the value is negative) at 1, 2, and 3 years as four simulation outcomes. The four outcomes have probabilities of 50%, 30%, 15%, and 5% at every date.

EE is the sum of each outcome times its probability. At 2 years, it is 0 × 0.5 + 200 × 0.3 + 500 × 0.15 + 1,000 × 0.05 = 185 million yen. PFE (90%) is the first outcome at which the cumulative probability, adding from the smallest outcome, reaches 90% or more. At 2 years the running totals are 50%, 80%, and 95%, so it is the third outcome, 500 million yen.

From the credit limit view, the usage is the largest PFE across dates, which is 500 million yen (at 2 years). If the limit for this counterparty is 600 million yen, utilization is 500 ÷ 600 ≈ 83%.

From the pricing view, you multiply the sum of the EEs by the counterparty's default probability in each year and by the loss given default. Suppose Company M has a default probability of 2% in each year and a loss given default of 60%, and ignore discounting. CVA is 0.6 × 0.02 × (105 + 185 + 80) million yen = 4.44 million yen.

The same trade gives two different figures, 500 million yen and 4.44 million yen. The 500 million yen answers one question: how large can the exposure grow, on some date, in a near-worst case. The 4.44 million yen answers another: what is the expected loss, summed over time. PFE is set by one point in the tail of the distribution, while CVA is set by the sum of the averages over all dates. For that reason, the two figures do not move by the same proportion when you change the terms of the trade or the collateral.

Roles of the risk unit and the CVA desk

Credit limit management and CVA management differ in who does the work and what tools they use. The table summarizes the difference.

A unit independent of the trading business sets each counterparty's limit by looking at the counterparty's creditworthiness and financial condition. This independence is required by the Basel Committee's 2024 guideline and the 2011 US interagency guidance. The guideline requires that limits be set by a function independent of the first line (the trading business)³. The guidance requires the risk management function to be independent of the trading units that take counterparty credit risk, in both activities and reporting⁴. The Basel Committee's guideline says that cooperation between the market risk and credit risk functions is important, because counterparty credit risk has both market risk and credit risk components⁵.

On the CVA side, according to a 2018 article by a former Financial Services Agency official and a Bank of Japan staff member, advanced financial institutions manage CVA centrally in a CVA desk that is independent of the other trading desks. The CVA desk handles CVA hedging, CVA risk management, the pricing function, and the review of collateral agreements⁶. In the US guidance, a unit trading in connection with CVA management is also required to have an independent risk management function overseeing its activities⁷. A note in a 2018 paper from the Bank of Japan's Institute for Monetary and Economic Studies says that some CVA desks act as "xVA desks" that manage a wider scope, including funding, collateral management, and reduction of capital charges, in addition to counterparty credit risk management⁸.

As of 2010, the use of CVA was not necessarily established at financial institutions in Japan⁹.

How mitigants affect credit limits and CVA

Netting, collateral, and clearing are ways to reduce counterparty credit risk. Of these, netting and collateral affect both the credit limit usage (PFE) and CVA (EE), but not in the same way. Clearing (for CVA and bilateral exposure) and compression (for notional amount) are described separately in the second half of this section.

Netting works because claims and obligations under the same netting agreement can be offset if the counterparty defaults. CVA is also found per netting set, not per trade¹⁰. When netting and collateral effects are taken into account, PFE and current exposure (CE) are likewise found for the portfolio of each counterparty¹¹.

When you receive collateral, exposure falls and CVA becomes smaller¹². The CVA calculation reflects the threshold and the minimum transfer amount in the collateral agreement¹³. For a collateralized counterparty, PFE is calculated assuming a realistic close-out period¹⁴. The threshold is the upper limit on the exposure allowed without collateral, so it works as a cap on exposure¹⁵.

Now add to the trade with Company M a collateral agreement with a threshold of 200 million yen. To simplify, assume that collateral keeps up with market value immediately. Then exposure never exceeds 200 million yen, and any outcome above 200 million yen becomes 200 million yen.

The maximum PFE falls from 500 million yen to 200 million yen, and utilization of the 600 million yen limit falls from 83% to 200 ÷ 600 ≈ 33%. CVA is 0.6 × 0.02 × (70 + 100 + 70) million yen = 2.88 million yen, about 35% lower than 4.44 million yen. PFE is 60% lower and CVA is 35% lower. The reason is that PFE looks directly at the large outcomes in the distribution, so cutting them at the threshold has a strong effect. If the shape of the distribution changes, the proportions change too.

This calculation assumes that collateral transfers never lag behind market value, so exposure stops at 200 million yen.

Clearing replaces the counterparty with a central counterparty (CCP). A 2018 paper from the Bank of Japan's Institute for Monetary and Economic Studies explains two points. In a trade with a CCP, initial margin is posted so the firm avoids the gain from its own default (DVA). With a CCP, counterparty credit risk is in principle zero, so the effect of CVA can be avoided¹⁶. Under the capital rules too, trades through a qualifying central counterparty are excluded from the CVA risk capital charge¹⁷. On the other hand, counterparty credit risk is concentrated in the CCP. The US guidance says that using a CCP can reduce bilateral exposure but concentrates risk in one entity, and that a financial institution should review the CCPs it uses at least once a year¹⁸.

Portfolio compression is a mechanism in which several participants share information about each other's trades and terminate many trades together, cutting the outstanding notional amount while leaving the overall size of risk roughly unchanged. According to Bank of Japan material, the growing use of compression is linked to financial institutions preparing for the leverage ratio rule. To reduce the add-on that is tied to notional amount, the notional amount itself has to be reduced¹⁹.

Pricing CVA into a trade

This is the flow for putting the cost of CVA into a price. In step 3, the cost is allocated to the business that brought in the trade. Step 5 is handled by the CVA desk in the table in the previous section, at advanced financial institutions.

Figure description

The columns stand for views and the party that bears the cost, not for units. The steps run from the assessment of limit usage before the trade, to checking the change in CVA, allocating the cost, reflecting it in the price, and hedging and managing the risk after the trade.

1. Before the trade, the US guidance lists having procedures and systems that assess the trade's impact on limit usage²⁰.
2. In calculation terms, the difference between CVA for the netting set before and after adding the new trade is the amount the trade adds, because CVA is found per netting set¹⁰.
3. When the trade is executed, the cost of CVA is allocated to the business that brought in the trade. In the US guidance, reflecting CVA risk in the risk-adjusted return of the business is considered sound practice. Allocation is also said to encourage the people who take the risk to mitigate it, for example by asking for stronger collateral terms²¹.
4. Whether to reflect the cost of CVA in the price to the customer has a competitive side. A 2010 Bank of Japan workshop presentation explains that if the counterparty calculates CVA, a bank that does not take CVA into account wins more trades with counterparties of lower credit quality than its own, so counterparty credit risk may build up. Counterparties of high credit quality will not accept a price that ignores CVA²².
5. After the trade, at advanced financial institutions, the CVA desk handles CVA hedging and risk management. In the US guidance, firms are required to check, at least quarterly, whether changes in the fair value of CVA are adequately explained by sensitivities²³.

A 2010 Bank of Japan workshop presentation lists, as a possible reason why financial institutions do not introduce CVA, a business practice in which, for most trades, the counterparty cannot be asked to pay a markup for CVA, which bears on how the price is set in step 4²⁴. Which adjustments to put into a price also has no market consensus in places, as of the 2017 paper from the Bank of Japan's Institute for Monetary and Economic Studies. How far KVA is included in the price differs by financial institution, and including both CVA and KVA may double count the cost of hedging counterparty credit risk²⁵.

Check the difference in step 2 with numbers. Consider a hypothetical example in which Bank D has a netting set with a counterparty whose EE at 1, 2, and 3 years is 100, 100, and 50 million yen. The counterparty's default probability is 2% in each year, the loss given default is 60%, and discounting is ignored. This CVA is 0.6 × 0.02 × (100 + 100 + 50) = 3 million yen.

Suppose a new trade is added and the EE of the set becomes 160, 180, and 100 million yen with trade A. A is a trade whose exposure builds up in the same direction as the existing trades. CVA becomes 0.6 × 0.02 × (160 + 180 + 100) = 5.28 million yen, and the difference before and after is +2.28 million yen. With trade B, suppose the EE falls to 70, 60, and 30 million yen. B is a trade whose direction offsets the exposure of the existing trades. CVA becomes 0.6 × 0.02 × (70 + 60 + 30) = 1.92 million yen, and the difference is −1.08 million yen. B's EE is lower than the existing set's EE by 30, 40, and 20 million yen at 1, 2, and 3 years. Even with the same counterparty, depending on the relationship with the existing set, the CVA of a trade you add can rise or fall. The EE values are assumptions chosen to keep the result simple.

How changes in CVA affect profit and capital

CVA moves when the counterparty's creditworthiness changes, and also when market moves change the exposure. In the Company M example, if the default probability in each year rises from 2% to 4%, CVA increases to 0.6 × 0.04 × 370 million yen = 8.88 million yen. The increase of 4.44 million yen is a valuation loss for the bank. Conversely, if the probability falls to 1%, CVA falls to 2.22 million yen, and the decrease of 2.22 million yen is a valuation gain. Profit and loss moves on the change in creditworthiness alone, even if Company M never defaults.

The US guidance says that CVA has become more important because accounting changes now require the effect of CVA changes on earnings to be recognized²⁶. It also says that in the 2007–09 financial crisis, a large part of the losses related to counterparty credit risk came from CVA losses, not from actual defaults²⁶. A 2010 Bank of Japan workshop presentation explains that two thirds of these losses were declines in the market value of derivatives caused by changes in exposure or creditworthiness²⁷.

Changes in CVA come from two sources, changes in the counterparty's credit spread and changes in exposure driven by market factors such as interest rates and exchange rates. It is usual to hedge the former with the counterparty's CDS and the latter with futures, swaps, and options²⁸. The hedge is not perfect. In the US guidance, firms that hedge CVA are asked to measure gaps such as basis risk and check the effectiveness of the hedge, in particular capturing nonlinear parts such as the correlation between market risk and credit risk²⁹. What the hedge cannot remove stays.

On capital, Basel III in 2010 asked banks to hold capital against the risk of changes in the market value of CVA, separate from the capital held against counterparty default³⁰. The final framework agreed in December 2017 provides a standardized approach (SA-CVA), a basic approach (BA-CVA), and a simplified method for banks with small derivatives volumes. To use SA-CVA, a bank must meet requirements that include having a CVA desk (or a similar function) responsible for managing and hedging CVA risk, and eligible hedges can be reflected in the required capital³¹.

Data and systems used for management

To find credit limit usage and CVA for the same counterparty, you need the same trade data, netting agreements, and collateral terms. The US guidance lists calculating CE and potential exposure for each counterparty every day, based on the previous day's positions and collateral transfers³².

Assumptions and limitations

Consider a hypothetical example in which there are only four outcomes. The default probability is constant each year, discounting is ignored, and collateral transfers never lag behind market value, so you can trace the difference between the two views by hand. A real CVA, when computed in detail, calculates future exposure in each scenario by simulation and, in general, uses a credit curve built from market credit spreads. If there is no CDS that references the counterparty directly, a bond spread of the counterparty or the CDS of a similar company is used instead³⁵.

The way probability is measured can also differ between the two views. Market prices such as CVA are valued under risk-neutral probability. PFE for setting limits, on the other hand, is often calculated under real-world probability, which has the advantage that, if fitted to historical data, limits do not change often and stay stable. The probability used for capital calculation can also differ from the one used for pricing³⁶.

The CVA formula assumes that the counterparty's default and the exposure are unrelated. If exposure grows when the counterparty's creditworthiness falls, this assumption fails. The US guidance asks firms to identify the correlation between the counterparty's creditworthiness and exposure and try to build it into the CVA calculation³⁷. The treatment of this correlation is covered in Wrong-Way Risk.

What this article does not cover

How trade compression affects credit limit usage (PFE) and CVA is not covered on this page.

Notes

1. Office of the Comptroller of the Currency and others, "Interagency Supervisory Guidance on Counterparty Credit Risk Management" (June 29, 2011), Section V.1 "Counterparty Limits" (page 13). See also Appendix A "Peak Exposure" (pages 20–21). ↩ ↩²

2. Financial Services Agency and Bank of Japan, "バーゼル銀行監督委員会「カウンターパーティ信用リスク管理に関するガイドライン」について" (On the Basel Committee's "Guidelines for counterparty credit risk management") (April 2025), "⑤Exposure measurement (1)" (page 12) (in Japanese). The guideline was finalized in December 2024 (page 4). ↩

3. The same document, "Limit governance and exception management" (page 20) (in Japanese). ↩

4. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.3 "Risk Management Function and Internal Audit" (page 4). ↩

5. Financial Services Agency and Bank of Japan, "バーゼル銀行監督委員会「カウンターパーティ信用リスク管理に関するガイドライン」について," "People and risk culture" (page 17) (in Japanese). ↩

6. Soichi Maeda and Tatsuya Hasegawa, "ＣＶＡリスク計測の枠組みの見直し" (Review of the framework for measuring CVA risk), Kinzai Institute for Financial Affairs, Kinyu Zaisei Jijo (February 12, 2018 issue), "実務におけるＣＶＡ管理" (CVA management in practice) (page 33) (in Japanese). ↩

7. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.5 "Management of CVA" (page 9). ↩

8. Takashi Kawakami, "ヘッジ取引におけるデリバティブ信用評価調整の影響についての考察" (A study of the effects of credit valuation adjustments of derivatives on hedging transactions), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2018), Section 5, footnote 26 (pages 147–148) (in Japanese). The same place also mentions a move by large European and US financial institutions to set up CVA desks. ↩

9. Yoshihiko Uchida, "カウンターパーティー・リスクの管理とCVAの活用" (Counterparty risk management and the use of CVA), Bank of Japan Financial System and Bank Examination Department, Center for Advanced Financial Technology, workshop presentation of June 14, 2010, "本日のワークショップの目的" (Purpose of today's workshop) (page 2) (in Japanese). The presentation states that it reflects the author's personal views and not the official view of the Bank of Japan. ↩

10. Maeda and Hasegawa, "ＣＶＡリスク計測の枠組みの見直し," "信用評価調整（ＣＶＡ）とは" (What is credit valuation adjustment (CVA)) (page 32) (in Japanese). ↩ ↩²

11. Uchida, "カウンターパーティー・リスクの管理とCVAの活用," "CCR量の評価" (Evaluation of CCR amounts) (page 17) (in Japanese). ↩

12. Maeda and Hasegawa, "ＣＶＡリスク計測の枠組みの見直し," "実務におけるＣＶＡ管理" (CVA management in practice) (page 33) (in Japanese). ↩

13. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.5 (page 9). ↩

14. The same document, Appendix A "Peak Exposure," "Purpose" (page 21). ↩

15. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA considering the combined effects of financial regulation), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), Section 3 (1) (a) "信用極度額" (credit limit amount) (page 117) (in Japanese). The paper's term for the threshold is "credit limit amount" (信用極度額); this page uses "threshold" to match the glossary. ↩

16. Kawakami, "ヘッジ取引におけるデリバティブ信用評価調整の影響についての考察," Section 5 (page 147) (in Japanese). The same place says that, for trades with corporates, some trades do not assume collateral exchange because of the burden of collateral management. Footnote 25 (page 147) says that recognizing FVA and MVA can get in the way of hedge effectiveness. ↩

17. Maeda and Hasegawa, "ＣＶＡリスク計測の枠組みの見直し," final paragraph of "適格ヘッジの範囲拡大" (Expansion of the range of eligible hedges) (page 35) (in Japanese). ↩

18. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section VI "Managing Central Counterparty Exposures" (pages 16–17). ↩

19. Bank of Japan, "（日銀レビュー）最近のOTCデリバティブ市場の動向 ： OTCデリバティブ統計からみた特徴点" (BOJ Review: Recent trends in the OTC derivatives market, features seen in OTC derivatives statistics) (2017-J-8, June 2017), "コンプレッションの普及" (The spread of compression) (pages 5–6) (in Japanese). ↩

20. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section V.1 (pages 13–14), "trade procedures and systems that assess a trade's impact on limit utilization prior to execution." ↩

21. The same document, Section III.5 "Management of CVA" (pages 9–10). ↩

22. Uchida, "カウンターパーティー・リスクの管理とCVAの活用," "CVAと逆選択" (CVA and adverse selection) (pages 23–24) and "CVA導入のメリット・デメリット" (Advantages and disadvantages of introducing CVA) (page 25) (in Japanese). ↩

23. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.5 (page 10). ↩

24. Uchida, "カウンターパーティー・リスクの管理とCVAの活用," "CVA導入のメリット・デメリット" (page 25) (in Japanese). The presentation lists the reasons a financial institution that has not introduced CVA might give, in the form of questions. ↩

25. Saito, "金融規制の複合的影響を考慮したXVA," Section 2 (1) (page 113) and Section 2 (2) (page 114) (in Japanese). ↩

26. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.5 "Credit Valuation Adjustments (CVA)" (page 8) and footnotes 12 and 13. ↩ ↩²

27. Uchida, "カウンターパーティー・リスクの管理とCVAの活用," "顕在化の歴史と教訓（リーマン）" (History of realized losses and lessons (Lehman)) (page 12, which says "it is said that") and "現行の自己資本比率規制上の取り扱いと今次危機を踏まえた問題点" (Treatment under the current capital adequacy rules and problems in light of the recent crisis) (page 27) (in Japanese). ↩

28. Maeda and Hasegawa, "ＣＶＡリスク計測の枠組みの見直し," "実務におけるＣＶＡ管理" (page 33) (in Japanese). ↩

29. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.5 (page 10). ↩

30. Maeda and Hasegawa, "ＣＶＡリスク計測の枠組みの見直し," "現行のＣＶＡリスク計測の枠組みと課題" (The current CVA risk measurement framework and its issues) (page 33) (in Japanese). ↩

31. The same article, "新たなＣＶＡリスク計測の枠組みと特徴" (The new CVA risk measurement framework and its features) (pages 34–35) and "標準的方式（ＳＡ－ＣＶＡ）" (Standardized approach (SA-CVA)) (page 36) (in Japanese). ↩

32. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.2 "Aggregation of Exposures" (page 5). ↩

33. Financial Services Agency and Bank of Japan, "バーゼル銀行監督委員会「カウンターパーティ信用リスク管理に関するガイドライン」について," "⑦Infrastructure, data and risk systems" (page 21) (in Japanese). ↩

34. Uchida, "カウンターパーティー・リスクの管理とCVAの活用," "CVA導入のメリット・デメリット" (page 25) (in Japanese). ↩

35. Maeda and Hasegawa, "ＣＶＡリスク計測の枠組みの見直し," "実務におけるＣＶＡ管理" (page 33) (in Japanese). Kawakami, "ヘッジ取引におけるデリバティブ信用評価調整の影響についての考察," Section 3 (page 126) (in Japanese), also says that building a credit curve from credit spreads observed in the market is the general method. ↩

36. Saito, "金融規制の複合的影響を考慮したXVA," Section 3 (4) "資本計算の確率測度" (Probability measure for capital calculation) (pages 130–131) (in Japanese). ↩

37. "Interagency Supervisory Guidance on Counterparty Credit Risk Management," Section III.5 (page 9). ↩