LIBOR Replacement – Determining Whether A Benchmark Is “Representative”

EU Benchmark Regulations “representative” requirement

In a speech on 7/12/2018, the head of the British financial regulator, FCA (Financial Conduct Authority), Andrew Bailey, declared that the regulator could prohibit LIBOR from continuing to be used for new businesses, if it determines that LIBOR no longer sufficiently “represents” the market. (EU Benchmark Regulations, Article 11(1)(a): “[T]he input data [of a benchmark] shall be sufficient to represent accurately and reliably the market or economic reality that the benchmark is intended to measure.”) Such a determination might not have been overly difficult, if LIBOR had remained a poll-based benchmark. As Mr. Bailey asserted in the same speech, “[LIBOR] relies on the so-called judgment of the panel banks. […] [T]o continue in the new regulated world of benchmarks, LIBOR has to be representative. I struggle to see the case for this judgment.”

LIBOR, however, has evolved.

Reformed LIBOR and permissible “waterfall” input data

ICE, the administrator of LIBOR, has reformed LIBOR to incorporate a “waterfall” of input data types, including transactions, transaction-derived data, and expert judgments. (See here.) Moreover, the EU Benchmark Regulations explicitly permit an interest rate benchmark to incorporate these hybrid types of input data. (See table below.)

Determining “representative”

Since the EUBMR explicitly permit an interest rate benchmark to incorporate such “hybrid” input data types, the Reformed LIBOR can satisfy the “representative” requirements even if it incorporates non-transactional data (e.g., judgments). The question then is: How does one determine whether the Reformed LIBOR is “representative”? According to what criteria? What are the parameters and factors to be used?

The EUBMR do not provide a clear answer to these questions. However, logically such a determination may depend on the following factors, among others:
• Total number of input data;
• Total number of transactional data;
• Total number of judgments;
• Percentage of transactional data;
• Percentage of judgments;
• Variations in the numbers & percentages;
• Etc.

Recognizing the shrinking unsecured interbank funding markets, the Reformed LIBOR broadens the sources of its transaction data to include certificates of deposit and commercial papers. Whether the expansion in data sources will be sufficient to satisfy the “representative” requirements, however, remains uncertain.

ICE Bank Yield Index

In a clear effort to further expand the sources of transactional data, ICE created the Bank Yield Index (BYI) to supplement LIBOR as a credit-sensitive interest rate benchmark. According to ICE, BYI is derived from two types of input data:

  1. Wholesale, primary market funding transactions executed by large, internationally active banks (e.g. inter-bank deposits, institutional certificates of deposit and commercial paper); and
  2. Secondary market transactions in wholesale, unsecured bonds issued by large, internationally active banking groups.

Below figure shows the average numbers and volumes of transactions underlying BYI. (See here.)

Will these expanded data satisfy the “representative” requirements? The answer remains blowing in the wind! The regulators will eventually need to step up to resolve the issue.

Libor Replacement – ICE Bank Yield Index & Benchmark Regulations

[Update 4/13/2019: Last Wed, Randal Quarles, Vice Chair for Supervision of the Federal Reserve and Chair of the Financial Stability Board (“FSB”), gave a speech at an FSB roundtable on reforming interest rate benchmarks. On the topic of choosing a replacement rate, Mr. Quarles did not foreclose the adoption of rates different from SOFR (or other risk-free rates). Instead, he established the principle that banks should “conduct at least as much due diligence on the reference rates that they use as they conduct on the creditworthiness of the borrowers”. Mr. Quarles concluded that SOFR satisfied the due diligence requirements. However, the other benchmark rates (e.g., ICE BYI) may also satisfy the requirements, and thus be a potential candidate to replace LIBOR.]

IBA, the administrator of the ICE Bank Yield Index (“BYI”) (see here & here), may have gained some ground in promoting BYI as a replacement for LIBOR. According to ARRC meeting minutes and agenda, IBA has been scheduled to make a presentation to ARRC on BYI at the ARRC April meeting. The presentation supposedly was requested by ICE in response to comments on BYI made at the February ARRC meeting. However, the fact that ARRC specifically discussed BYI in its meeting seemed significant. Moreover, there have been claims (although unconfirmed) that large banks “don’t like SOFR” (see here), but instead endorse BYI (see here).

This development increases the chances that multiple benchmarks (e.g., BYI, AMERIBOR, etc), in addition to SOFR, may co-exist post-LIBOR, as I discussed previously here & here. That is, even if LIBOR is eventually phased out, it is increasingly likely that BYI and potentially other benchmarks administered by private parties may partly replace LIBOR, for example, for the cash markets, while SOFR is used for the derivative markets.

For the regulators, however, the co-existence of private benchmarks with SOFR raises the question of whether these private benchmarks are qualified to replace LIBOR, free from the issues that plagued and doomed the latter. For example, the regulators desire a transaction-based benchmark to replace LIBOR. But, what are the minimum number or volume of transactions underlying a benchmark that can pass muster with the regulators? SOFR has a trading volume of nearly $1 trillion across thousands of transactions each day. BYI, on the other hand, may cover less than $10 billion of daily transactions across less than 200 daily transactions. Moreover, these numbers are the totals over various tenors from 5 to 500 days. The respective number for each tenor will be even less. (See Figures 3, 4, & 5 of the ICE Bank Yield Index Update, here.) Are these numbers sufficient for a benchmark designed to replace the flawed LIBOR? If not, what are the minimum thresholds?

As a result, it is likely that the US regulators will eventually need to establish a set of protocols and procedures to control and manage the quality of the private benchmarks, similar to the EU Benchmark Regulations. The Congress, however, will first have to enact legislations authorizing the regulators to establish the regulations. The legislative and regulatory processes may be time-consuming and the outcome uncertain, adding to the uncertainties already embedded in the LIBOR-replacement undertaking!


LIBOR Replacement – The “Two Benchmark” Approach

BIS paper and “two-benchmark” approach

The March 2019 issue of the BIS Quarterly Review published a paper titled “Beyond LIBOR: a primer on the new reference rates”. (See here.) As its title suggests, the paper presents a non-technical introduction to new reference rates such as SOFR intended to replace LIBOR. What’s interesting about the paper is that it discusses a “two-benchmark” approach to replacing LIBOR. The approach entails the concurrent use of a risk-free benchmark rate and a credit-sensitive benchmark rate, and presents a potential solution to the “loss of credit premium” quagmire that I discussed in my previous posting (here).

Divergences between risk-free and credit-sensitve rates

As I discussed in my previous posting, SOFR, a risk-free rate, is not intended to fully replace LIBOR, which incorporates a term credit risk premium. A credit-sensitive benchmark rate may still be desirable, particularly for cash market participants.

The BIS paper recognizes this need for a credit-sensitive benchmark. In particular, the BIS paper illustrates the potentially significant divergences between a risk-free rate (eg, SOFR) and a counterpart risk-sensitive rate (eg, LIBOR) during both normal times and times of stress. The charts below from Graph 7 of the BIS paper compare the O/N LIBOR with SOFR in 2008, and the 3m GBP (British Pound Sterling) LIBOR with futures-linked SONIA in 2018. (Note: SONIA, “Sterling Overnight Interbank Average”, is an unsecured O/N money market wholesale rate, and the British equivalent of SOFR.)

The charts show that in 2008, the O/N LIBOR surged above 6%, while the SOFR rates plunged from above 1.5% to nearly 0%. Even in normal times, the 3m GBP LIBOR rose by ~10 bps at the end of 2018, while the 3m SONIA rates implied from futures contracts remained nearly unchanged.

Reformed LIBOR, LIBOR+

As a result of the potentially substantial divergences between risk-free and risk rates, I posited in my previous post that the cash market participants are left with 3 options, none of which is satisfactory: (1) Continue to use LIBOR; (2) Adopt SOFR + a fixed spread; and (3) Use other rates having a credit component, such as PRIME, FHLB, commercial paper, CD, AMERIBOR, ICE Bank Yield Index, new auction-based rates, Investment Grade SOFR, etc. The BIS paper, on the other hand, discusses another choice as a part of the “two-benchmark” approach. – A reformed LIBOR, or LIBOR+.

LIBOR+, was proposed in a 2014 report (here) prepared by the Market Participant Groups (MPG), which was established by the Financial Stability Board (FSB) to study the reforms of benchmark interest rates. The results of the MGP report was summarized by Duffie & Stein in 2015 (here), which was in turn referred to by the BIS paper with respect to the “two-benchmark” approach. According to Duffie & Stein, in essence, LIBOR+ broadens the sources of unsecured wholesale funding transactions to those between all counterparties (specifically CPs & CDs), in addition to inter-bank transactions. Broadening the sources can increase the number of transactions underlying the benchmark, and thus reduce the risks of manipulation. Additionally, an algorithm can be used to fix the LIBOR+ rates even if submissions from panel banks are insufficient, reducing the risks to financial stability. (For example, on any day t, if a bank does not have any available transaction, the algorithm can derive the rate for the specific bank by using the transactions in the previous t-k days, giving increasingly smaller weightings to days farther away in the past.)

“Two-benchmark” = RFR & LIBOR+

According to the BIS paper, many countries have opted for the “two-benchmark” approach by choosing a RFR complemented by a local LIBOR+. (“In Japan, the reformed TIBOR will coexist with TONA; and in the euro area, there is an ongoing effort to reform EURIBOR to complement ESTER.” “In Australia, the reformed BBSW [will complement] the O/N benchmark” rate. “In Canada, the liquidity under CDOR, which is based on the bankers’ acceptance market, had actually been on the rise […], making its retention as a credit-sensitive term benchmark that much easier.”) In US, such a “two-benchmark” approach means SOFR will co-exist with LIBOR+ (or other alternatives), with the risk-free SOFR being used primarily for derivatives or for securities issued by agencies or corporations, and the credit-sensitive LIBOR+ used primarily for other cash market products.

Uncertainties

Such a “two-benchmark” approach may cause fragmentation and segmentation in the interest rate markets, reducing liquidity, and causing confusions, among other potential issues and difficulties. Furthermore, it is uncertain whether LIBOR+ can pass muster with the markets or regulators, and whether other alternative credit-sensitive benchmark rates, such as ICE Bank Yield Index, may out-compete LIBOR+. Or, it is possible that markets may reject the “two-benchmark” approach and instead adopt SOFR + a fixed spread, if the SOFR market liquidity becomes attractive and the benefit of a single rate outweighs the lack of a credit premium. Only time will tell, although the clock may be ticking!!!

LIBOR Replacement – The loss of market-based credit premium and the quagmire for cash market participants

It has been widely reported that the Federal Reserve (via a group of public & private organizations, ARRC) has identified SOFR (Secured Overnight Financing Rate) to replace LIBOR. SOFR, however, is a risk-free (or nearly risk-free) rate and does not include risk premia such as the credit risk premium. Therefore, SOFR cannot fully replace LIBOR for many stakeholders, particularly those in the cash markets.

Need for a rate with credit risk

In 2014, the FSB (Financial Stability Board) issued a report titled “Reforming Major Interest Rate Benchmarks”. (Available here.) In the report, the FSB pointed out that LIBOR, which is based on unsecured interbank markets, included two main components, a “risk-free or nearly risk-free rate” and a group of other “risk premia, including a term premium, a liquidity premium, a credit risk premium as well as potentially a premium for obtaining term funding”. The FSB report observed that it might be feasible for many derivative transactions to replace LIBOR with a risk-free rate, such as SOFR. However, the FSB report also pointed out that there would be continued need for a “reference rate with bank credit risk”, such as in the markets for bank loans, and other “bank-provided credit products”. In general, there is a greater need for such types of risk premium in the cash markets, such as bank loans, FRN bonds & securities, mortgages, structured products, capital market products, etc.

Nonexistence of a market-based rate with credit risk

Unfortunately, the need for a market-based rate with bank credit risk to replace LIBOR cannot be readily and satisfactorily met in any existing markets. In a speech presented on 7/12/2018, Andrew Baily, the head of the British financial regulator, put it bluntly: “It is difficult […] to see how the term credit premium that LIBOR seeks to measure […] can be obtained from other sources. We have not seen a compelling answer to how one-month, three-month, six-month and twelve-month term bank credit spread can be reliably measured on a dynamic and daily basis.” (See here.)

The quagmire for cash market participants

In 2014, the FSB envisioned a post-LIBOR new world, where multiple rates would provide the flexibility to meet the needs of multiple market sectors. A few years later, the reality has sunken in that the existing market liquidity can only support a risk-free rate, SOFR, but not a rate with credit risk. Therefore, for cash market participants, the new post-LIBOR reality is not a world of multiple rates to choose from, but one in which no single existing rate is satisfactory for their needs.

No satisfactory options

Given such a quagmire, cash market participants face enormous difficulties in planning for the anticipated changes in LIBOR. No options currently available paint a clear path.

  • Continue to use LIBOR: And face the liability and legal risks that LIBOR may be held inadequate or even illegal, particularly with the prospect of new bench mark regulations such as the EU Benchmark Regulations.
  • Adopt SOFR plus a fixed spread: And face the risks of mismatch between assets and liability due to large fluctuations in the credit spread (in addition to the lack of term structure and other deficiencies of SOFR).
  • Use other rates having a credit component, such as PRIME, FHLB rates (eg, 11th District Cost of Fund), commercial paper rates, CD rates, AMERIBOR (here), ICE Bank Yield Index (here), new auction-based rates (here), Investment Grade SOFR (here), etc: And face the reality that all these markets have low liquidity and other deficiencies such as data transparency & sufficiency, and may not be acceptable as a benchmark rate.

The risks

For cash market participants who are regulated, such as banks or insurance companies, the question is how the regulators will approach this quagmire. Will they establish a safe harbor or other rules? Or will they leave it to the regulated entities to establish their own safety and soundness policies & procedures? Considering the costs and uncertainties, it is most likely these regulated entities will not make any major decisions and take major actions until the regulators provide a clear guidance. The risk is high, however, that those decisions & actions may be too late.

Puzzling Definition of “Time Deposit”

IN MEMORY OF THE 6/4/1989 TIANANMAN SQUARE MASSACRE.

Reg. D Definition of “Time Deposit”

Regulation D (12 CFR 204) promulgated by the Federal Reserve Board (FRB) imposes reserve requirements on certain bank deposits. Generally, “transaction accounts” are subject to reserve requirements, but “time deposits” and “savings deposits” are not. The current definition of the term “Time Deposit”, however, can be confusing if not read in light of the historical context. Currently, Reg. D defines a “Time Deposit” as:

Time Deposit means:

(i) A deposit that the depositor does not have a right and is not permitted to make withdrawals from within six days after the date of deposit unless the deposit is subject to an early withdrawal penalty […];

(ii) A savings deposit;

(iii) […]

(iv) […]

37 CFR 204.2(c)(1).

Based on the above definition, therefore, “Savings Deposits” are a type of “Time Deposits”. But why?

If one reads the definition of “Savings Deposit” in the regulation, a “Savings Deposit” does not require a minimum term or impose an early withdrawal penalty. It is thus substantially different from the type of “Time Deposits” defined in the first sub-part of the “Time Deposit” definition.

Moreover, in practice, “Savings Deposits” are generally understood as a different type of deposit accounts from “Time Deposits”. For example, in the Federal Reserve Form, FR 2900 (Report of Transaction Accounts, Other Deposits, and Vault Cash), with which financial institutions report their deposit liabilities for reserve reporting purposes, “Time Deposits” and “Savings Deposits” are listed as two separate categories, apart from “Transaction Accounts” and “Vault Cash”. Additionally, the Reg. D section of the FRB Consumer Compliance Handbook also lists “Time Deposits” and “Savings Deposits” as two separate and distinct categories.

Under Reg. D, both “Time Deposits” and “Savings Deposits” are explicitly exempted from the reserve requirements. Therefore, it is not necessary to define “Savings Deposits” as a type of “Time Deposits” to achieve the goals of the regulation. — It is a puzzling definition, until one looks up the history of the regulation.

History

As first written in 1980, Reg. D defined a “Time Deposit” as one that “does not have a right to withdraw” for 14 days after deposit. Notably, the definition at the time did not explicitly impose an early withdrawal penalty. Such a definition, therefore, was broad enough to include certain “Savings Deposits”, for which the bank reserved rights to require an early withdrawal notice. Therefore, it was logical for the 1980 version of Reg. D to define “Time Deposits” to include those “Savings Deposits” that were not considered “Transaction Accounts”. Specifically, the definition was written as follows:

Time Deposit means:

(i) A deposit that the depositor does not have a right to withdrawals for a period of 14 days or more after the date of deposit. “Time deposit” includes funds:

(A) […];

(B) […];

(C) […];

(D) […];

(E) That constitute a “savings deposit” which is not regarded as a “transaction account.”

(ii) […].

45 Fed. Reg. 56018, 56020 (August 22, 1980).

By comparing the 1980 and current definitions, it becomes clear that after the definition of “Time Deposits” was amended, particularly by adding the early withdrawal penalty, “Savings Deposits” no longer belonged in the conventional type of “Time Deposits” and were removed from the explicit list. For unknown reasons, however, the FRB decided to keep it as a separate category under “Time Deposits”, thus creating the present confusions.

To clarify, FRB should amend Reg. D to remove the “Savings Deposit” from the definition of a “Time Deposit”.