Risk Reduction through Netting and Collateral

In credit risk management work that measures exposure by counterparty, the exposure numbers change with how far you can count on the effects of close-out netting and collateral.

How close-out netting reduces exposure

With close-out netting, the exposure (the amount you could lose) when a counterparty fails is not the sum of each trade's positive market value. It is the positive part of the total market value of all the trades, because trades with a negative market value offset those with a positive one. Japan's close-out netting act covers trades for which a financial institution or similar firm has agreed on close-out netting with its counterparty in a master agreement. When a petition such as one for bankruptcy proceedings is filed against the counterparty, the net amount from adding up the values of all trades under that agreement becomes a single claim or obligation¹. The big picture of counterparty credit risk is covered in What Is Counterparty Credit Risk?.

The group of trades covered by close-out netting is called a netting set. Japan's capital adequacy notice for banks treats the trades under a legally effective bilateral netting agreement as one netting set, and treats any other trade as a netting set on its own². Even trades with the same counterparty fall into separate netting sets, one per trade, if there is no agreement on close-out netting.

Take an example in which Company B and Bank N have two interest rate swaps. From Bank N's side, trade 1 has a market value of 500 million yen and trade 2 has a market value of −300 million yen. Without close-out netting, Bank N's exposure is the 500 million yen of trade 1, which has a positive value, and the negative value of trade 2 cannot be offset. With close-out netting, the exposure is the total of the two values, 200 million yen.

In the second row, the total is −100 million yen, so Bank N owes Company B 100 million yen and Bank N's exposure is zero. When every market value is positive, as in the third row, there is nothing to offset, and close-out netting does not reduce exposure. How much close-out netting saves depends on how much positive and negative value sits in the same netting set.

Formulas for exposure from market value and collateral

The exposure of a netting set is the total market value of its trades minus the collateral received, set to zero if the result is negative³. If $V_i$ is the market value of trade $i$, the market value $V$ of the netting set is as follows.

$$V = \sum_i V_i$$

If $C$ is the amount of collateral received (negative when you have posted collateral), exposure $E$ is given by the following formula.

$$E = \max(V - C, 0)$$

Without collateral, $C = 0$ and exposure is $\max(V, 0)$. Without netting, exposure is $\sum_i \max(V_i, 0)$, which adds up the market values of the trades, counting negative values as zero. Because $\max(V, 0)$ never exceeds this amount, close-out netting never increases exposure.

Collateral exchanged as the market value changes is called variation margin. The terms for variation margin are set in a credit support annex (CSA) attached to the master agreement for OTC derivatives. The main terms that affect exposure include the threshold, the minimum transfer amount, and how often collateral is requested. The request frequency is also called the frequency of margin calls⁴.

The threshold is the amount up to which you do not ask the counterparty to post collateral. You receive collateral only for the part of the market value above the threshold, so exposure up to the threshold is not covered by collateral. The minimum transfer amount is the smallest amount of collateral that is transferred. If the difference between the required collateral and the current collateral is below it, no collateral is transferred. This term reduces the work of transfers, but exposure equal to that difference remains.

Even under an agreement that requests collateral every day, not every change in market value is covered by collateral. When a counterparty fails, it takes days to get from the point where collateral stops being exchanged to closing out the trades and hedging the market risk again. This period is called the margin period of risk (MPOR), and changes in market value during it are not covered by collateral. The MPOR is not a term set in the agreement but the time needed to deal with the default. For agreements that exchange collateral daily, it is often assumed to be about 10 business days⁵. If $\delta$ is the MPOR and $TH$ is the threshold, the collateral held at time $t$ is set by the market value one MPOR earlier³.

$$C(t) = \max\bigl(V(t - \delta) - TH, 0\bigr) - \max\bigl(-V(t - \delta) - TH, 0\bigr)$$

The first term on the right is the collateral Bank N receives, and the second is the collateral Bank N posts. This formula leaves out the minimum transfer amount.

To include the effect of collateral when measuring exposure, you need not only the market values of the trades but also how the netting sets are divided and the terms of the CSA. The capital adequacy notice also lists these terms. When a bank reflects the effect of collateral under the internal model method (IMM; the notice calls it the expected exposure method), it takes into account the form of the agreement, the MPOR, how often collateral is requested, the threshold, and the minimum transfer amount⁶.

The regulatory formula for replacement cost and the number of MPOR days used in capital adequacy calculations are covered in Overview of the Regulatory Treatment of Counterparty Risk.

Worked example: two trades and collateral terms

Using the example of Company B and Bank N (a market value of 500 million yen for trade 1 and −300 million yen for trade 2), this section works out Bank N's exposure under different collateral terms. Suppose collateral from Company B stopped being exchanged one MPOR (10 business days) before the trades are closed out, and the netting set's market value was 140 million yen at that time. Over the next 10 business days, the market value rises to 200 million yen, and the trades are closed out at that point.

Figure description

The figure shows market value and collateral during the MPOR, the period from when collateral stops being exchanged until the trades are closed out. Read it from left to right: the point where exchange stops, the MPOR, and the point where the trades are closed out. Because collateral does not increase during the MPOR, the rise in market value remains as exposure. The spacing between points is not proportional to the length of time.

Collateral is set by the market value in step 6, and exposure is the market value in step 4 minus collateral. The table below shows the results for each set of collateral terms. The last row adds initial margin, which the next section covers.

With a threshold of 0 and a minimum transfer amount of 0, Bank N holds the 140 million yen of collateral from step 6. The remaining exposure is the 60 million yen (200 − 140) by which the market value moved during the MPOR. With a threshold of 100 million yen, collateral is only 140 − 100 = 40 million yen, and exposure is 160 million yen. In the row with a minimum transfer amount of 50 million yen, the gap between the required collateral of 140 million yen and the previous collateral of 100 million yen is 40 million yen, which is below the minimum transfer amount. So collateral stays at 100 million yen, and exposure is 100 million yen.

The same calculation holds from Company B's side. If the market value moves the other way and Bank N's side becomes negative, the collateral Bank N has posted and the move during the MPOR become Company B's exposure.

Covering moves during the MPOR with initial margin

Initial margin is collateral received separately from variation margin to cover the exposure that remains from changes in market value during the MPOR. Variation margin covers changes in market value that have already happened (current exposure). Initial margin covers changes in market value that could happen between the counterparty's default and the close-out of the trades⁸. MPOR exposure remains even under agreements that exchange variation margin daily. For that reason, margin rules for OTC derivatives that are not cleared through a central counterparty require variation margin that matches the market value. They also require financial institutions above a certain size to exchange initial margin⁹.

In the last row of the worked example, Bank N receives 80 million yen of initial margin in addition to 140 million yen of variation margin. The 60 million yen rise in market value during the MPOR stays within the initial margin. Because 200 − 140 − 80 is negative, exposure is zero. If instead the market value had risen to 250 million yen during the MPOR, 250 − 140 − 80 = 30 million yen would remain. Initial margin does not cover moves in market value beyond the range it was estimated for.

Netting and variation margin compared

If you have offsetting swaps with the same counterparty, close-out netting reduces exposure substantially. The simulator below shows an example in which Bank N has two swaps with Company B: a five-year pay-fixed swap (trade A) and a three-year receive-fixed swap (trade B). Each has a notional of 10 billion yen. Expected positive exposure (EPE), the average of the expected exposure (EE) at each date up to maturity, is 69 million yen in the initial setting with close-out netting and no collateral. That is 53 million yen less than the 122 million yen without netting or collateral. The maximum potential future exposure (PFE) is 444 million yen at 3.75 years. When trade B matures at three years, only trade A remains, and from then on EE and PFE are the same with or without netting.

If you set "Collateral" to "Variation margin" and move the threshold, the minimum transfer amount, and the MPOR, EPE changes as follows. With "Netting" on and the threshold and minimum transfer amount left at 0, raising the MPOR from 10 to 20 business days increases EPE from 10 million yen to 13 million yen, and setting the MPOR to 0 brings EPE to 0. With an MPOR of 0, collateral always keeps up with the market value.

With a threshold of 100 million yen, EPE is 37 million yen and the maximum PFE is 137 million yen. With a minimum transfer amount of 100 million yen, EPE is 21 million yen. If you turn "Netting" off and exchange variation margin, EPE is 21 million yen, about twice the 10 million yen with close-out netting. The simulator assumes that variation margin is exchanged in cash, and it does not handle initial margin or haircuts (discounts applied to the value of collateral).

Payments during the MPOR and spikes

If you make a payment during the MPOR, the market value rises by the amount paid, but collateral does not increase, so exposure jumps. Collateral is set by the market value one MPOR earlier, and that value includes the amount you are about to pay as a liability. Once you pay and the liability disappears, the market value exceeds collateral by that amount. If collateral was keeping up with the market value before the payment, exposure rises by exactly the amount paid.

For example, suppose Bank N is due to pay Company B a net 30 million yen during the MPOR, and the netting set's market value was zero (with zero collateral) when collateral stopped being exchanged. Even if no other market prices move, once Bank N pays the 30 million yen, the market value rises to 30 million yen while collateral stays at zero, so exposure is 30 million yen. If Company B is the one paying Bank N, the amount due was already in the market value before collateral stopped being exchanged, and Bank N had received collateral for it, so no spike occurs.

The market value of a swap jumps around each interest payment date, so even an uncollateralized swap has a stepped exposure profile¹⁰. Even with variation margin, if an interest exchange in which Bank N is the payer falls within the MPOR, collateral does not catch up, because it is set from a market value that is one MPOR old, and the jump remains as a spike.

In the simulator, if you set "Collateral" to "Variation margin" (threshold 0, minimum transfer amount 0, MPOR 10 business days), the maximum PFE is 205 million yen at four years, and PFE at other dates is 50 million yen or less. At four years, only trade A exchanges interest, and on paths where Bank N pays on a net basis, collateral does not keep up. Up to three years, the two swaps have the same payment dates. Because the initial curve is flat, both fixed rates equal the same par rate. The payments between Bank N and Company B therefore net to zero, and no spike appears. If you turn "Netting" off, the payments are no longer offset, and PFE spikes to 162 million yen at two years and 180 million yen at three years.

Risks that remain even with collateral and netting

Collateral and close-out netting both have limits. Close-out netting works only for trades under a legally effective agreement.

Collateral can lose value after it is transferred. Collateral other than cash changes in price and can take time to sell, so a CSA sets a haircut for each type of collateral¹¹. During the financial crisis of 2007–2008, even highly rated bonds fell sharply in price, and by around 2017, cash collateral, whose value does not change with prices, was becoming common, mainly in Europe and the United States¹¹.

The exchange of collateral can be delayed or disputed. The notice covers any netting set that, in the most recent two quarters, had three or more disputes over the amount of collateral that lasted longer than the MPOR. For the next two quarters, the bank must at least double the MPOR for that set¹².

Even after the counterparty fails, it takes time to liquidate collateral and close out the trades, and losses can grow in the meantime. An explanatory document from the Financial Services Agency and the Bank of Japan on the Basel Committee on Banking Supervision's guidelines cites the 2021 Archegos case to show that this liquidation can take a long time. The same document says firms should also expect that holding trades or collateral that are hard to sell can lead to large falls in asset prices when they are liquidated. For counterparties with high leverage, it points to risks that even variation margin and initial margin may not fully cover when markets move sharply¹³.

The value of collateral can also move in the same direction as the counterparty's default. The relationship in which the counterparty is more likely to fail when the value of its collateral is likely to fall is called wrong-way risk between the collateral and the counterparty¹¹. The explanatory document also calls for care that collateral and margin do not carry risk in the same direction as the counterparty and are not overly concentrated¹⁴.

Notes

1. "金融機関等が行う特定金融取引の一括清算に関する法律" (Act on Close-out Netting of Specified Financial Transactions Conducted by Financial Institutions, etc., in Japanese), Article 2, paragraph 2 (financial institutions, etc.), paragraph 4 (close-out netting events), paragraph 5 (master agreement), and paragraph 6 (close-out netting), and Article 3. ↩

2. Financial Services Agency, "銀行法第十四条の二の規定に基づき、銀行がその保有する資産等に照らし自己資本の充実の状況が適当であるかどうかを判断するための基準" (Standards for determining whether a bank's capital adequacy is appropriate in light of the assets it holds, under Article 14-2 of the Banking Act, our translation; FSA Notice No. 19 of 2006, consolidated version updated on August 24, 2026), Article 79, paragraph 7 (pages 189–190) (in Japanese). ↩

3. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA taking into account the combined effects of financial regulations), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), equation (1) in Section 2 (page 115), and equations (6) and (7) in Section 3 (2) (page 119) (in Japanese). The paper writes collateral as $X$ and the MPOR as $\delta$, and it separates the thresholds ("信用極度額") of the firm and the counterparty into $T_B$ and $T_C$. Here, collateral is written as $C$, and the same threshold $TH$ is used for both parties. ↩ ↩²

4. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA taking into account the combined effects of financial regulations), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), the opening of Section 3 (1) and "イ. 信用極度額" (a. threshold), "ロ. 最低受渡担保額" (b. minimum transfer amount), and "ハ. 評価頻度" (c. valuation frequency), page 117 (in Japanese). The paper calls the threshold "信用極度額," the minimum transfer amount "最低受渡担保額," and how often collateral is requested "評価頻度" (valuation frequency). ↩

5. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA taking into account the combined effects of financial regulations), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), "ハ. 評価頻度" (c. valuation frequency) in Section 3 (1), pages 117–118 (in Japanese). The paper calls the MPOR "リスクのマージン期間（MPR）" and says that many examples set it at about 10 business days for daily exchange of collateral. ↩

6. the same Notice, Article 79-3, paragraph 7 (pages 209–210). ↩

7. the same Notice, Article 79-3, paragraph 3, item 2 (page 207). ↩

8. Committee on Payment and Settlement Systems of the BIS and Technical Committee of the International Organization of Securities Commissions, "金融市場インフラのための原則" (Principles for financial market infrastructures, provisional Japanese translation; April 2012), explanation of Principle 6, paragraph 3.6.1 (page 69) (in Japanese). The English original is CPSS-IOSCO, "Principles for financial market infrastructures" (April 2012), which uses the same paragraph numbers. The explanation is about margin at central counterparties. ↩

9. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA taking into account the combined effects of financial regulations), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), the end of Section 3 (2) (page 121) and Section 4 (1) (pages 121–122) (in Japanese). ↩

10. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA taking into account the combined effects of financial regulations), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), the explanation of Figure 2 in Section 3 (2) (pages 120–121) (in Japanese). ↩

11. Yuichi Saito, "金融規制の複合的影響を考慮したXVA" (XVA taking into account the combined effects of financial regulations), Kinyu Kenkyu, Institute for Monetary and Economic Studies, Bank of Japan (April 2017), "ニ. 適格担保" (d. eligible collateral) in Section 3 (1), and footnotes 9 and 10, page 118 (in Japanese). ↩ ↩² ↩³

12. the same Notice, Article 79-2, paragraph 5 (page 192). ↩

13. 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 on Banking Supervision's Guidelines for counterparty credit risk management) (April 2025), "Potential future exposure (PFE)" under "⑤Exposure measurement / エクスポージャー計測" (Exposure measurement), page 13 (items #53–#58 of the guidelines) (in Japanese). The year of the Archegos case is on page 6. ↩ ↩²

14. 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 on Banking Supervision's Guidelines for counterparty credit risk management) (April 2025), the continuation of "④Credit risk mitigation / 信用リスク削減手法" (Credit risk mitigation), page 11 (item #32 of the guidelines) (in Japanese). ↩

15. 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 on Banking Supervision's Guidelines for counterparty credit risk management) (April 2025), "④Credit risk mitigation / 信用リスク削減手法（１）" (Credit risk mitigation (1)), page 10 (items #23 and #26 of the guidelines) (in Japanese). ↩