Challenge to New Yen Benchmark and role of UTS

TORF overview and calculation process

In Japan, as a new interest rate index based on the RFR decided in advance, the construction of "TORF (Tofu / Tokyo Term Risk Free Rate)" whose reference value is calculated by QUICK Benchmarks Co., Ltd. (QBS) is being promoted. I am. Like LIBOR, this TORF is a pre-determined interest rate index and has a term structure of 1 month, 3 months, and 6 months. By using TORF, you can ensure the certainty of cash flow as well as the conventional LIBOR.

TORF is calculated based on the transaction performance and quote data of Yen OIS (Overnight Index Swap), which is traded using the unsecured call O / N rate as the underlying asset. The figure below is an image of the TORF calculation process.

Process of TORF Calculation


Contributor (Brokers)

Collect data

OIS Trades/Orders Data (1M/3M/6M)

CLOB and Voice Broking Data


RFR administrator


Calculate and publish TORF

QUICK Corp is the TORF Administrator


Comparison of TONA OIS and TORF

Difference between TORF rate fixed in advance and TONA OIS fixed in arrears

Clearing Yes(JSCC &LCH ) not yet
Clearing Tenor JSCC 40years(14,623days) / LCH 41years not decided
Cash flow frequency Interbank's current OIS is once a year, but it may diversify every six months in the future. Since TORF has a period structure of 1, 3 and 6 months, various CFs are assumed.
Floating interest rate determination and cash flow timing The floating interest rate is not fixed until the end of the calculation period, and currently payment is made on the second business day after the end of the calculation period, so the period from the settlement of the floating interest rate to payment is short and the paperwork burden is heavy. Since the floating interest rate is decided in advance, there is a grace period until payment. Therefore, the predictability of cash flow can be ensured and the administrative burden is small.
Products applied as fallback rates ISDA Derivatives Fallback Rate First-ranked Fallback Rate for loans and bonds
Market liquidity It is hard to say that TONA OIS is currently actively trading, but it has a trading history more than 10 years. Not traded at this time. It will not be traded until the middle of 2021 when the final TORF to be released .
Robustness / reliability of indicators TONA is calculated based on the transaction performance of the highly liquid call market, and is highly robust and reliable. Since TORF is calculated based on TONA OIS, robustness and reliability depend on the liquidity of TONA OIS market.
Institution that publishes indicators of underlying assets Bank of Japan Quick Benchmarks Co., Ltd.
Clearing obligation None None
ETP obligation None None

Related laws

In September 2012, IOSCO established a task force in response to investigations and legal measures against tampering with major financial indicators, and in July 2013, the task force published the "Final Report on Principles on Financial Indicators" *. it was done. The following is an excerpt of the part related to the provided data.

Principles for Financial Benchmarks (

What is required when creating and complying with Benchmarks at IOSCO
7.Data Sufficiency
The data used to construct a Benchmark determination should be sufficient to accurately and reliably represent the Interest measured by the Benchmark and should:
a) Be based on prices, rates, indices or values that have been formed by the competitive forces of supply and demand in order to provide confidence that the price discovery system is reliable; and
b) Be anchored by observable transactions entered into at arm’s length between buyers and sellers in the market for the Interest the Benchmark measures in order for it to function as a credible indicator of prices, rates, indices or values.
This Principle requires that a Benchmark be based upon (i.e., anchored in) an active market having observable Bona Fide, Arms-Length Transactions.
This does not mean that every individual Benchmark determination must be constructed solely of transaction data.
Provided that an active market exists, conditions in the market on any given day might require the Administrator to rely on different forms of data tied to observable market data as an adjunct or supplement to transactions.
Depending upon the Administrator’s Methodology, this could result in an individual Benchmark determination being based predominantly, or exclusively, on bids and offers or extrapolations from prior transactions.
This is further clarified in Principle 8.
Provided that subparagraphs (a) and (b) above are met, Principle 7 does not preclude Benchmark Administrators from using executable bids or offers as a means to construct Benchmarks where anchored in an observable market consisting of Bona Fide, Arms-Length transactions.
"This Principle also recognizes that various indices may be designed to measure or reflect the performance of a rule-based investment strategy, the volatility or behaviour of an index or market or other aspects of an active market. Principle 7 does not preclude the use of non-transactional data for such indices that are not designed to represent transactions and where the nature of the index is such that non-transactional data is used to reflect what the index is designed to measure. For example, certain volatility indices, which are designed to measure the expected volatility of an index of securities transactions, rely on non-transactional data, but the data is derived from and thus “anchored” in an actual functioning securities or options market."
8. Hierarchy of Data Inputs
An Administrator should establish and Publish or Make Available clear guidelines regarding the hierarchy of data inputs and exercise of Expert Judgement used for the determination of Benchmarks.
In general, the hierarchy of data inputs should include:
a) Where a Benchmark is dependent upon Submissions, the Submitters’ own concluded arms-length transactions in the underlying interest or related markets;
b) Reported or observed concluded Arm’s-length Transactions in the underlying interest;
c) Reported or observed concluded Arm’s-length Transactions in related markets;
d) Firm (executable) bids and offers; and
e) Other market information or Expert Judgments.
Provided that the Data Sufficiency Principle is met (i.e., an active market exists), this Principle is not intended to restrict an Administrator’s flexibility to use inputs consistent with the Administrator’s approach to ensuring the quality, integrity, continuity and reliability of its Benchmark determinations, as set out in the Administrator’s Methodology.
The Administrator should retain flexibility to use the inputs it believes are appropriate under its Methodology to ensure the quality and integrity of its Benchmark. For example, certain Administrators may decide to rely upon Expert Judgment in an active albeit low liquidity market, when transactions may not be consistently available each day.
IOSCO also recognizes that there might be circumstances (e.g., a low liquidity market) when a confirmed bid or offer might carry more meaning than an outlier transaction. Under these circumstances, non-transactional data such as bids and offers and extrapolations from prior transactions might predominate in a given Benchmark determination.
The following is an excerpt of the data quality and quantity under the EU Benchmark Regulation (BMR) enacted in June 2016. BMR was devised in response to the LIBOR tampering case that triggered the founding of TORF, and since it is a relatively new decree, it has become a global standard together with MiFID II and MIFIR. It is a regulation that is likely to be followed. The transition period for applying BMR to third-country benchmarks was the end of 2021, but it has been extended again until the end of 2023 in response to a request from related organizations to extend it until the end of 2025 (It may be extended to the end of 2025). In any case, countermeasures are urgent.

Refre to original (external site)

Article 12:Methodology
1. An administrator shall use a methodology for determining a benchmark that:
(a) is robust and reliable;
(b) has clear rules identifying how and when discretion may be exercised in the determination of that benchmark;
(c) is rigorous, continuous and capable of validation including, where appropriate, back-testing against available transaction data;
(d) is resilient and ensures that the benchmark can be calculated in the widest set of possible circumstances, without compromising its integrity;
(e) is traceable and verifiable.
2. When developing a benchmark methodology, a benchmark administrator shall:
(a) take into account factors including the size and normal liquidity of the market, the transparency of trading and the positions of market participants, market concentration, market dynamics, and the adequacy of any sample to represent the market or economic reality that the benchmark is intended to measure;
(b) determine what constitutes an active market for the purposes of that benchmark; and
(c) establish the priority given to different types of input data.
3. An administrator shall have in place clear published arrangements that identify the circumstances in which the quantity or quality of input data falls below the standards necessary for the methodology to determine the benchmark accurately and reliably, and that describe whether and how the benchmark is to be calculated in such circumstances.
In other words, although both IOSCO and BMR allow the use of non-actual transaction data, it is considered to be premised on "active markets" or "when non-actual transaction data is rather suitable". Therefore, in order for these data to be available, it is necessary to first form and maintain an "active market", and it can be said that the efforts of market participants are required, especially in the current situation where actual transaction data is insufficient.

TORF Culculation requirements

The figure below summarizes the data calculationion requirements of TORF.

Referencing Data
  • Spot Start JPY OIS outright (1M/3M/6M)
  • Transactions executed on Tokyo business day and cleared on JSCC or LCH
  • Trade rate / Notional Amount / Date and Time
  • Best Bid-Offer / Date and Time
Referencing/Publishing time Referencing : PM3:00 on Tokyo business day / Publishing: PM5:00 on same day
Time of data extraction Phase1: All day (24hours) / Phase2 : Specific period of time or all day (24hours)
Method of data extraction Phase1: Extract all data(threshold value=0) Phase2: Extract all data (threshold value=0, for the time being)
Calculating Method
  • If the total notional amount of the actual transaction data exceeds the threshold value, use only the actual tracsaction, while if the total is less than the threshold value, use only the quota data. (Waterfall method)
  • Following are considered for priorities of using data in the waterfall method
    Level1 Actual transaction data
    Level2 Tradable quotes on CLOB
    Level3 Double-sided tradable quotes (Bid and Offer) from Voice broker
    Level4 Single-sided tradable quotes (Bid or Offer) from Voice broker
    Level5 Double-sided quotes (Bid and Offer) from Voice broker
  • If the term rates cannot be calculated by the above methods, predetermine methods suck as continuously publishing the term rates of the previous business day.
Outlier tests Statistics method (Details are under consideration)
Quality weighted average Weight by reciprocal number between best Bid/Offer
Data providers Ueda Tradition Securities and other Broker institutes (Total 3 Brokers)
Data Adminstrator QUICK crop.
* Phase 1 in the above figure refers to the stage of calculating and publishing reference values ​​that are not premised on use in transactions, and Phase 2 refers to the stage of calculating and publishing fixed values ​​that are assumed to be used in actual transactions. Point to. Phase 1 started in May 2020, and Phase 2 is scheduled to start around mid-2021.

Supplemental explannation of the data waterfall structure

Level 2 CLOBs are not currently in operation in the Tokyo market.
For level 3 and 4 quotes, it is necessary to present not only the rate but also the notional amount.
Regarding level 5, we are currently referring to the quote value (Indicative rate) presented by the broker company when calculating the TORF reference value, but we will not use level 5 to calculate the final value.

The transaction data has a weighting priority, and the actual transaction data is the most important. Next is the data on electronic commerce (CLOB), and then the data on voice brokering. One important point is that in the current OIS trading practice, trading participants often do not specify the trading volume (amount) when placing Bid / Offer orders, but when calculating TORF, orders without Amount information are not specified. It means that it cannot be used as quote information.

In the current yen OIS market, there are few level 1 actual transactions, and levels 2 to 4 are almost nonexistent. Therefore, when calculating the reference value of TORF, only level 1 and level 5 are actually referred to. Since it is a policy not to use level 5 to calculate the definite value, TORF will be calculated only with a few level 1 data as it is.

In order to improve the robustness of TORF, it is important to incorporate trading practices such as clearly indicating the notional amount when presenting an order, in addition to increasing the trading volume of Yen OIS.