ICE Data Services believes that in certain markets (e.g. Fixed Income) one-way markets are typically as valid as two-sided markets and thus should be allowed to be considered a verifiable price for the purpose of demonstrating modellability.
We have aggregated real price observation data for the month of May 2019 within the corporate credit markets in order to support this recommendation. Firstly, as was highlighted in the proposal, there is significantly more volume of one-way quotations as compared to two-sided quotations - to a magnitude of 5x in this market. Please note that this is not true of all markets, where for example, quotations are generally only aggregated when two-sided in certain markets (e.g. CDS) where over 98% of our available data is two-sided in that market. One additional note is that one-way quotations were significantly more likely to be quoted throughout the day than two-sided markets and thus after filtering for one per day, and additional filters described below, we had more two-sided quotations than one-sided.
Our analysis started by looking at only one quotation per day. If both a two-sided market and one-way quotations were both available on a particular day, we considered that a two-sided market for the sake of this analysis. We then looked at how many of those quotations had ultimately resulted in a transaction on the same calendar day, 19% of total two-sided quotations consummated with a transaction on that day whereas only 3% of the one-way only quotations resulted in a transaction.
However, when the aggregated results are compared to any trade in the month, as opposed to same day, the results are starkly different. There were nearly 22,000 unique ISINs with two-sided quotes in the month where 61.2% ultimately resulted in a transaction in the month. There were also over 19,000 unique ISINS with one-way quotes only (without two-sided markets available on the same day) where 26.2% ultimately resulted in a transaction in the month.
ICE Data Services agrees that two-sided quotations are statistically more likely than one-way quotations to lead to a timely transaction. However, there are still significant volumes of one-way only quotations that ultimately do lead to transaction activity. We believe EBA should not ignore all of these quotes for demonstrating modellability. We believe that the price validation approach discussed below concerning our response to Question 4 should offer comfort against the possibility of “manufactured” off-market one-way quotes solely for improving modellability metrics.
Fixed income instruments by nature trade very differently than either equity or derivative instruments. For example, certain corporate bonds trade differently than other instruments, such as an equity security or a 5-year point on a CDS curve, where the latter is more likely to have a tighter bid-offer spread. Institutional investors that trade the bond market typically have longer-term investment strategies, hence only a one-way market is required to transact, indicating the direction the investors intends to trade. Various Federal Reserve and industry publications resonate that corporate bond inventories at banks are low despite the increase level of bonds issued. One-sided corporate bond quotes are just as valid as two-side corporate bond quotes, as these tend to be long term investments and supported by the Volcker rule and certain Basel III mandates.
The Volcker rule and Basel III have influenced and discouraged dealers from holding a significant amount of bonds according to Euromoney . On June 20, 2014, Bloomberg mentioned the low-interest-rate environment and the increased popularity of exchange-traded funds have led bonds to not actively trade and more are held longer, sometimes until maturity . Furthermore, in Q2 2017 the Federal Reserve has also stated, “dealers’ inventories of corporate and foreign bonds increased a bit in Q1 2017, but remained near their lowest levels over the past decade”. Dealers are less likely to put out bids, given dealers hold less inventory because of regulation and bonds investors hold onto bonds longer. One-sided quotes are one-sided in response to the market interest and constraints on buying bonds.
Although different banks may have unique definitions of what a “non-negligible volume” is for a particular trade or asset type, ICE Data Services believes that allowing each institution the flexibility to set their own volume threshold would result in unintended consequences. For example, a smaller entity that has a lower negligible volume threshold could feasibly have certain modellable risk factors that a large institution would not be able to demonstrate as modellable if they only look at transactions and committed quotes of a larger size. For example, a retail driven market such as US municipal debt, or certain corporate issuers mainly have small average sizes that could be considered negligible for a large bank. Another unintended consequence is the increased complexity of examining FRTB compliance programmes with this additional degree of freedom, as each desk may have its own definition of negligible volumes, even for different assets within a desk.
As an alternative, ICE Data Services suggests creating an industry-standardised threshold as a function of existing MiFID II requirements. Among other potential options, we propose leveraging the MiFID II rules that define a retail size as any transaction below €100,000, which also could be used as the definition of a negligible size.
ICE Data Services has worked with the industry and other regulators to educate how the pricing service industry has become increasingly sophisticated and have benefitted from significant technological advancements. These enhancements have allowed ICE Data Services, as well as other pricing vendors, to generate intra-day, independent valuations that could be used to assess the validity of a price associated with a committed quote, even if one-sided only (i.e. no explicit quote-level bid-ask spread is available).
ICE Data Services believes that there are multiple tools available to the banks to verify appropriately the validity of prices associated with quotes and smaller transactions. The final text should explicitly permit the use of independent third party pricing. . This type of validation is more robust than the current proposal, and alleviates the rationale for requiring an eligible committed quote to have both a bid and offer side of the transaction, as we discussed above in Question 1.
ICE Data Services reviewed corporate bond quotes that matched to trades with at least one million notional from March 2019 to May 2019. Corporate bond trade prices, executed within an hour after the quote (with both a bid and ask), were compared to quotes and ICE Data Services’ Continuous Evaluated Pricing (CEP). The histogram below buckets the trades into the basis point ranges of the difference between a CEP or quote and a trade price. The data shows that over the 90-day period, trade prices were closer to the CEP levels. Figure 2 is a histogram of how close the trade price was to the CEP verses the quote for May 2019. Trade prices for May 2019 in Figure 2 are more in-line with CEP than quote prices. Over each of the three-month period, consistently the first 5 bins account for between 66%-70% of the trades when compared to CEP and between 58%-60% of the trades when compared to quotes. The data supports that using an independent third party pricing tool, such as CEP, provides a smaller differential and therefore more accurate price validation.
The table below in Figure 3 further demonstrates that CEP, with typically tighter bid-ask spreads than those of two-sided quotations, would be an appropriate benchmark to validate the pricing quality of potential real price observations.
ICE Data Services derivatives pricing data is available at any point in time throughout the trading day for all broad OTC asset classes. The ICE Data Services Derivatives pricing service allows for request and retrieval of a valuation at an execution timestamp for any OTC derivatives. Figure 4 below shows curve data by currency and tenor with associated quoted prices, liquidity data, dispersion value of quotes via min-max data points, total notional traded, and the average traded notional. The data in Figure 4 allows the client to compare bid-offer spreads to market data to verify if the spreads are in-line with normal market conditions.
“Legally Obliged” Language
ICE Data Services believes that the term “legally obliged” is very difficult to demonstrate and suggests the final legal text, under Article 2, section 1(c) not have it. We firmly believe that in conjunction with the robust price validation requirements that this language should be softened or removed. Many quotes, including even those submitted on certain electronic execution platforms, allow for a right-to-refusal, or refresh capabilities to the provider of the quote in the event the market moves from the time of the initial quote. This workflow and trading protocols do not invalidate the intent to transact by the quote provider. The workflow acknowledges that markets move and tradeable prices move over time, as should the associated quotes.
ICE Data Services acknowledges the goal of EBA that “transactions shall not be conducted and quotes shall not be committed with the sole purpose of identifying a sufficient number of verifiable prices that allows to meet the criteria specified in Article 1.” (Page 21 on the CP). On the quote side, we believe that robust price verification (as described above in our response to Question 4) combined with other restrictions inherently minimises the likelihood of these types of “manufactured” quotes without the term “legally obliged” included in the final text. Moreover, we believe it is exceptionally burdensome for banks and their vendors to demonstrate appropriately any sort of legal obligation associated with any individual quote with minimal to no benefit in the validity of the quote population used to demonstrate modellability.
ICE Data Services does not see any issues with requiring external data providers to obtain an acceptable independent audit in order to be utilised by a reporting entity to assist with demonstrating modellability of their risk factors. We are currently preparing preliminary work and research to arrange for this independent third-party audit.