TORF

Contracts for yen loans and bonds that reference TORF may need to set out, in their interest calculation terms, the publication time and how a late or revised rate is handled. The process that loads published values is likewise built on the assumption of delays and revisions.

Why a term rate set in advance was needed

TORF is a benchmark rate created to provide a rate that, like 3-month and 6-month yen LIBOR (London Interbank Offered Rate), is set at the start of the interest period. TORF contains almost no bank credit risk.

TONA (Tokyo Overnight Average Rate), which was identified as the yen risk-free rate (RFR), is an overnight rate. Existing contracts, however, referenced term rates such as 3-month and 6-month yen LIBOR. The Cross-Industry Committee on Japanese Yen Interest Rate Benchmarks therefore concluded that moving away from yen LIBOR required building new term rates based on the RFR¹. How TONA is calculated and published is covered in TONA.

There are two ways to build a term rate from an RFR. One compounds the actual TONA fixings over the period. The other derives the rate from derivatives trades that reflect where the market expects TONA to go. TORF takes the second approach: it is built from Japanese yen OIS market data. This corresponds to "Option (3)" in the committee's 2019 public consultation². As a result of that consultation, three candidates to replace yen LIBOR emerged: a term rate built from OIS (later TORF), TONA compounding in arrears, and TIBOR (Tokyo Interbank Offered Rate)³.

QUICK Benchmarks began publishing TORF production rates on April 26, 2021⁴.

The difference between setting a rate in advance and setting it in arrears shows up in when the borrower learns the interest amount. Consider a hypothetical example in which a loan of 1 billion yen runs for 182 days, the 6-month TORF is published as 0.50% at the start of the period, and the spread is 0.30%. Interest is calculated on an actual/365 basis. For a loan at TONA compounded in arrears, the spread is the same 0.30%, and the compounded rate fixed at the end of the period is set to three different values for comparison.

A borrower at TORF pays 1 billion yen × 0.80% × 182 ÷ 365, or 3,989,041 yen, and this amount is known at the start of the period. A borrower at the compounded rate pays more if TONA runs higher than expected during the period (4,487,671 yen in the table) and less if it runs lower (3,490,411 yen). The TORF borrower knows the interest amount at the start of the period, but in exchange the interest does not fall if TONA declines during the period. How the compounded rate is calculated is covered in Compounding in Arrears.

The yen OIS trades behind the calculation

TORF is calculated from the fixed rate of Japanese yen OIS. An OIS is an interest rate swap that exchanges a fixed rate for a floating rate calculated by compounding TONA over a set period. Its fixed rate reflects the market's expectation of TONA over that period⁵.

Yen OIS trades are reported only if they meet all of the following requirements⁶.

• Spot start (the swap starts two business days after the trade date)
• A standalone trade that is not combined with other trades (an outright trade)
• A term (tenor) of 1 month, 3 months, or 6 months
• An actual/365 basis
• Traded on a Tokyo business day
• Cleared at a central counterparty, either Japan Securities Clearing Corporation (JSCC) or LCH
• Executed between 3:00:01 p.m. on the previous business day and 3:00:00 p.m. on the current day, or, for orders, live during that window

Only centrally cleared trades are used so that counterparty credit risk does not enter the calculated rate. In OIS that are not cleared through a central counterparty, the traded rate includes the credit risk of the counterparty⁷.

The data are reported by reporting brokers selected by QUICK Benchmarks. They report two types of data: executed trades (transaction data) and orders placed with the intention to trade (quote data). Indicative quotes, which are not meant for trading, are not used⁸. As of the review of the operational framework for 2024, the reporting brokers were three voice brokers, which arrange trades by telephone and similar means⁹.

Calculation by the waterfall methodology

TORF ranks the types of data in order of priority. If data of a higher rank are available, they are used; if not, the calculation moves down to the next rank. This is called the waterfall methodology, and it lets the rate be calculated mechanically, without expert judgment, on days with many trades and on days with few¹². The rank is determined for each tenor using the data from all reporting brokers combined¹³.

For Rank 1, the transaction data are sorted by traded rate, and trades outside the 25th and 75th percentiles of cumulative notional amount are removed as outliers. The traded rates $r_i$ of the remaining $n$ trades are then averaged, weighted by their notional amounts $N_i$¹⁴.

$$\text{TORF} = \frac{\sum_{i=1}^{n} r_i N_i}{\sum_{i=1}^{n} N_i}$$

The quality-weighted average used for Ranks 3 and 5 gives more weight to order pairs with a narrower gap between bid and offer (the spread). Let the bid of order pair $i$ be $b_i$ and the offer be $o_i$. The mid rate $m_i = (b_i + o_i) \div 2$ is averaged, weighted by the inverse of the spread, $1 / (o_i - b_i)$¹⁵.

$$\text{TORF} = \frac{\sum_{i} m_i \,/\, (o_i - b_i)}{\sum_{i} 1 \,/\, (o_i - b_i)}$$

Consider a hypothetical example in which the market has the two order pairs in the table below.

The weighted average is (0.50 × 25 + 0.51 × 50) ÷ (25 + 50) = 38 ÷ 75, or 0.506666...%, which rounds to 0.50667% at the sixth decimal place¹⁶. The result is closer to 0.51%, the mid rate of pair B with its narrower spread, than the simple average of the mid rates, 0.505%. QUICK Benchmarks explains that it chose this average to prevent an order with an extremely wide spread from becoming the best quote and moving the calculated rate¹⁷.

Deep diveHow order pairs with the same bid and offer are handled

In yen OIS, the two sides sometimes agree on the rate but not on the notional amount or the central counterparty (JSCC or LCH), so the bid and the offer end up at the same value just before a trade. With a spread of zero, the quality-weighted average cannot be calculated. Ranks 3 and 5 are therefore split into A, which gives priority to pairs with the same bid and offer and takes their simple average, and B, which takes the quality-weighted average of pairs whose bid is below the offer¹⁸.

Which rank is actually used differs by tenor. The 2024 review by QUICK Benchmarks gave the following shares¹⁹.

For the 1-month tenor, the previous day's rate was used on more than 70% of days. QUICK Benchmarks noted that the share of days calculated from same-day data rose from the previous year for all tenors, but that for the 1-month tenor it was still only in the 20% range, and said it would continue to monitor the situation²⁰.

Publication time and what happens when the rate is not published

TORF is calculated from data as of 3:00 p.m. and published around 5:00 p.m. on business days in Japan. The rate is expressed in percent to five decimal places, rounded at the sixth decimal place²¹. It is published through information vendors: QUICK Benchmarks distributes it to QUICK, and QUICK distributes it to other information vendors (Refinitiv Japan, Bloomberg, Nikkei Inc., and others). The QUICK Benchmarks website does not normally show TORF²².

The calculation requires reports from at least two reporting brokers. If two or more have reported by 4:15 p.m., the rate is published at 5:00 p.m. as usual²³. If fewer than two have reported, publication may be delayed until 6:30 p.m. at the latest²⁴. If an error is found after publication, the published rate may be revised. This applies, as a guideline, to calculation errors that are found by 6:00 p.m. that day and affect the published rate by more than 2bp (0.02%). The board of directors of QUICK Benchmarks decides whether to revise, and any revision is published by 7:00 p.m. that day²⁵.

The figure below puts these times in order, from the start of the period covered by the calculation to the publication of a revision.

Figure description

The figure lists the times of one TORF business day in order. The calculation covers orders that were executed or live from 3:00:01 p.m. on the previous business day to 3:00:00 p.m. on the current day. If two or more reporting brokers have reported by 4:15 p.m., the rate is published around 5:00 p.m. If fewer than two have reported, publication may be delayed until 6:30 p.m. at the latest. A calculation error found by 6:00 p.m. that day may be revised, as a guideline when it affects the published rate by more than 2bp (0.02%). The board of directors of QUICK Benchmarks decides whether to revise, and a revision is published by 7:00 p.m.

If publication is suspended temporarily because of a major disaster or a similar event, the previous business day's published rate is used as that day's TORF²⁶. Before stopping publication permanently, QUICK Benchmarks announces the timing and reasons at least six months in advance, and before changing the definition or calculation methodology, it announces the changes and the effective date at least three months in advance²⁷.

The administrator and regulation under the Financial Instruments and Exchange Act

TORF is administered by QUICK Benchmarks Inc., a wholly owned subsidiary of QUICK. It handles collecting the reported data, calculating and publishing the rate, governance, and compliance on its own²⁹.

Under the Financial Instruments and Exchange Act, the Prime Minister may designate a person who calculates a financial indicator as a specified financial index calculation agent. This applies when ensuring the proper conduct of the calculation business is necessary for the public interest or the protection of investors³⁰. TORF was designated as a specified financial indicator after production rates began to be published in April 2021, and in October 2021 QUICK Benchmarks received approval from the Financial Services Agency (FSA) for its operational rules under the Act³¹. According to the FSA's list, there are two specified financial index calculation agents: the JBA TIBOR Administration and QUICK Benchmarks (as of July 8, 2024)³². QUICK Benchmarks states that it complies with the Principles for Financial Benchmarks issued by the International Organization of Securities Commissions (IOSCO)³³.

QUICK Benchmarks reviews the definition and calculation methodology of TORF at least once a year³⁴. In its review for 2024, it concluded that the yen OIS market is an appropriate market for TORF to measure and that no change to the operational framework, including the calculation methodology, is needed at this time³⁵.

Where TORF is used and what to watch for

TORF is used as a reference rate for transactions such as loans and bond issues, and for valuing assets³⁶. Three-month TORF, along with TIBOR, is listed as a base rate for loans linked to market interest rates³⁷.

For derivatives, the committee's 2019 public consultation stated that TONA compounded in arrears was generally expected to be used. It also stated that when loans or bonds reference a term RFR, derivatives trades matching them would be carried out³⁸. The same consultation noted that the Financial Stability Board (FSB), in a 2018 statement, considered it desirable to move many derivatives to overnight RFRs³⁹. As more loans and bonds use TONA compounded in arrears, more OIS are traded to hedge them, which in turn makes TORF, which is built from OIS trades, more robust⁴⁰.

When you use TORF in a contract, keep two points in mind. First, TORF is a registered trademark of QUICK. You can view it on information vendors' terminals, but using it in financial contracts and similar uses may require a license agreement with QUICK, depending on the use⁴¹. Second, publication may be delayed, rates may be revised, the definition may be changed, and publication may stop. To prepare for these cases, QUICK Benchmarks advises that contracts that reference TORF include measures such as fallback provisions⁴².

Notes

1. Cross-Industry Committee on Japanese Yen Interest Rate Benchmarks, "日本円金利指標の適切な選択と利用等に関する市中協議" (Public Consultation on the Appropriate Choice and Usage of Japanese Yen Interest Rate Benchmarks, in Japanese), July 2019, "（ⅱ）ターム物金利構築の必要性" (The need to build term rates), page 9. An English version is available: "Public Consultation on the Appropriate Choice and Usage of Japanese Yen Interest Rate Benchmarks". ↩

2. The same document, "（２）代替金利指標の選択肢" (Options for alternative benchmarks), "（ⅰ）概要" (Overview), pages 10–11, and "選択肢（３）ターム物 RFR 金利（スワップ）" (Option (3): term RFR rate (swaps)), page 11. ↩

3. Bank of Japan Financial Markets Department and Financial System and Bank Examination Department, and Financial Services Agency Strategy Development and Management Bureau and Supervision Bureau, "円LIBOR移行対応の振り返りと今後の取り組み" (A review of the transition from yen LIBOR and future initiatives) (Bank of Japan Review 2022-J-8) (in Japanese), May 2022, the main text on page 3 and Chart 3, "検討委員会の主な取り組み" (Major initiatives by the committee). An English version is available: "Review of JPY LIBOR Transition and Future Initiatives" (Bank of Japan Review 2022-E-4, May 2022), the main text on page 3 and Chart 3 on page 4. ↩

4. QUICK Benchmarks, "TORF概要書" (TORF Methodology, in Japanese; established April 21, 2021, revised October 13, 2021), page 2. An English version is available: "TORF Methodology". ↩

5. The same document, section "2 用語" (Terms), the entry "OIS", page 6. That the floating rate is compounded is from "円LIBOR移行対応の振り返りと今後の取り組み" (Bank of Japan Review 2022-J-8, in Japanese), footnote 31, page 8. That the fixed rate reflects expectations is from the committee's public consultation, "選択肢（３）" (Option (3)), page 11. ↩

6. QUICK Benchmarks, "ＴＯＲＦ算出要綱" (TORF Calculation Guidelines, in Japanese; established April 21, 2021, revised October 13, 2021), the transaction requirements on page 5. An English version is available: "TORF Calculation Guidelines". ↩

7. QUICK Benchmarks, "TORF概要書" (in Japanese), section "2 用語" (Terms), the entry "清算集中取引" (Centrally cleared transactions), page 7. ↩

8. The same document, section "2 用語" (Terms), the entries "気配データ" (Quote data), "報告データ" (Reported data), and "レポーティング・ブローカー" (Reporting brokers), page 6. ↩

9. QUICK Benchmarks, "ＴＯＲＦの運営態勢の定期的な見直し結果について" (Result of a Periodic Review of TORF Operational Framework (2024), in Japanese only; September 30, 2025, covering January 4 to December 30, 2024), pages 1–2, "2.（1）背景" (Background). ↩

10. QUICK Benchmarks, "TORF概要書" (in Japanese), section "3.2 報告データの種類" (Types of reported data), requirements ⑧ and ⑨ for (１) transaction data, page 9. ↩

11. The same document, section "3.2 報告データの種類" (Types of reported data), (２) quote data, pages 9–10, and section "2 用語" (Terms), the entry "Bid/Offer", page 6. ↩

12. The same document, section "1.3 TORFの説明" (Description of TORF), "② ウォーターフォール手法の採用" (Adoption of the waterfall methodology), pages 4–5, and section "4.4 専門家判断の行使" (Use of expert judgment), page 18. ↩

13. QUICK Benchmarks, "ＴＯＲＦ算出要綱" (in Japanese), section "６ 順位判定と各順位の算出方法" (Rank determination and the calculation method for each rank), pages 14–19, and section "７ 第5 順位までのデータがない、または第5 順位の閾値未達の扱い" (Treatment when there are no data up to Rank 5 or Rank 5 does not meet the threshold), pages 19–20. ↩

14. The same document, page 14, the calculation method for Rank 1 and note 8. Note 8 gives the exclusion threshold as "確定値開始当初は25 パーセンタイル" (the 25th percentile at the start of production rates). In the notation, $r_i$ stands for Ratei and $N_i$ for Notionali in the document. ↩

15. The same document, pages 16–17, the calculation method for Rank 3 B. In the notation, $b_i$ stands for BidRatei, $o_i$ for OfferRatei, and $m_i$ for MidRatei in the document. ↩

16. The number of decimal places follows "小数第５位（小数第６位四捨五入）" (five decimal places, rounded at the sixth) on pages 2–3 of QUICK Benchmarks, "ＴＯＲＦ算出要綱" (in Japanese). ↩

17. QUICK Benchmarks, "TORF概要書" (in Japanese), the box "第 3 順位で想定元本を用いない理由" (Why notional amounts are not used in Rank 3), page 15. ↩

18. QUICK Benchmarks, "ＴＯＲＦ算出要綱" (in Japanese), "(３) 第３順位の判定と算出方法" (Determination and calculation method for Rank 3), page 15. ↩

19. QUICK Benchmarks, "ＴＯＲＦの運営態勢の定期的な見直し結果について" (in Japanese; September 30, 2025), Table 4, "TORF算出順位の割合" (Shares of TORF calculation ranks), page 8. Ranks 3 and 4 were 0.4% or less for every tenor and are omitted from the table. ↩

20. The same document, pages 9–10, the main text of "(4)" and the summary in "(5)". ↩

21. QUICK Benchmarks, "TORF概要書" (in Japanese), the table in section "1.3 TORFの説明" (Description of TORF), page 4. ↩

22. QUICK Benchmarks, "TORF ご利用上の留意点" (Notes on Referencing TORF, in Japanese; April 21, 2021), page 1, first paragraph. An English version is available: "Notes on Referencing TORF". ↩

23. The same document, the note on page 1. ↩

24. QUICK Benchmarks, "TORF概要書" (in Japanese), section "5.2 公表の遅延" (Delayed publication), page 19. ↩

25. The same document, section "5.3 修正方針" (Revision policy), page 19. ↩ ↩²

26. The same document, section "5.4 一時的な公表停止" (Temporary suspension of publication), page 19. ↩

27. The same document, section "5.5 継続的な公表停止" (Permanent cessation of publication), pages 19–20, and section "6.2 算出方法の変更" (Changes to the calculation methodology), page 21. ↩

28. The same document, section "5.1 公表" (Publication), page 19. ↩

29. The same document, section "1.2 指標運営" (Benchmark administration), page 4. ↩

30. "金融商品取引法" (Financial Instruments and Exchange Act, in Japanese; English translation: Japanese Law Translation), Article 156-85, paragraph 1. Approval of operational rules is in Article 156-87. ↩

31. Bank of Japan Financial Markets Department and Financial System and Bank Examination Department, and Financial Services Agency Strategy Development and Management Bureau and Supervision Bureau, "円LIBOR移行対応の振り返りと今後の取り組み" (Bank of Japan Review 2022-J-8, in Japanese; May 2022), page 6. The English version (Bank of Japan Review 2022-E-4) says the same on page 7. ↩

32. Financial Services Agency, "特定金融指標算出者一覧" (List of specified financial index calculation agents, in Japanese; as of July 8, 2024). ↩

33. QUICK Benchmarks Inc. website, home page (in Japanese; checked on September 25, 2026). ↩

34. QUICK Benchmarks, "TORF概要書" (in Japanese), section "6.1 算出方法や運営態勢等の検証" (Review of the calculation methodology, operational framework, and related matters), page 21. ↩

35. QUICK Benchmarks, "ＴＯＲＦの運営態勢の定期的な見直し結果について" (in Japanese; September 30, 2025), "1. 検証結果" (Review results), page 1. ↩

36. QUICK Corp., "TORF（東京ターム物リスク・フリー・レート）について" (About TORF (Tokyo Term Risk Free Rate), in Japanese), "何に使われるものなのか" (What it is used for) (checked on September 25, 2026). ↩

37. Bank of Japan, "金融システムレポート" (Financial System Report (April 2024), in Japanese), "BOX4 最近の金利動向" (Box 4: Recent developments in interest rates), page 92 (page 97 of the PDF). An English version is available: "Financial System Report (April 2024)". ↩

38. Cross-Industry Committee on Japanese Yen Interest Rate Benchmarks, "日本円金利指標の適切な選択と利用等に関する市中協議" (in Japanese), "（ⅲ）デリバティブ" (Derivatives), page 40, and the note to Chart 3-1, page 41. ↩

39. The same document, footnote 6, page 4. ↩

40. Bank of Japan Financial Markets Department and Financial System and Bank Examination Department, and Financial Services Agency Strategy Development and Management Bureau and Supervision Bureau, "円LIBOR移行対応の振り返りと今後の取り組み" (Bank of Japan Review 2022-J-8, in Japanese), pages 6–7. The English version (Bank of Japan Review 2022-E-4) says the same on page 7. ↩

41. QUICK Benchmarks, "TORF ご利用上の留意点" (in Japanese), page 1, second paragraph. ↩

42. The same document, page 2. ↩