Banks are creating detours in their core systems to be able to calculate the Secured Overnight Financing Rate (Sofr) rather than trying to replace the entire system before the end of 2021.
However, these complementary systems can come with additional risks, say bankers. There is also uncertainty around the level of time required to implement, test, and run systems to be ready for the cessation of Libor - answers from market participants vary between 12, 18, and 24 months.
“You can put in place things like [detours] and then work through the actual systems at a later point [after 2021],” says the chief data scientist at a major US financial institution.
“When you are modifying code and there are interdependencies, you can do lots of different types of testing but unless you work through the entire process there is always some possibility that there are some interdependencies which you haven’t accounted for. I think you can get a sense of that risk with enough testing, but it can still be a risk.”
Firms in the US are currently preparing for the transition away from the dollar London Interbank Offered Rate (Libor) reference interest rate which has been used for decades to the recommended alternative - the Sofr.
According to the Alternative Reference Rate Committee (Arrc) – a group of market participants co-ordinated by the Federal Reserve - roughly $200trn financial products reference USD Libor.
There is no doubt among market participants that existing core systems which are used for Libor calculations will require modifications. But there is disagreement as to whether even temporary system modifications will be ready in time.
Issues lie around systems used for analytics, pricing methodologies, and the computer code which sets the rate in advance. To transition away from the forward-looking term rate Libor, systems will now be required to reset in arrears. But to make the necessary changes would require a substantial knowledge of the coding which is only known by those who created the systems, and who have since left the firm, or having intricate documentation on the system which may be difficult to find. To get around this issue, many firms will have to create detours in their existing coding which take the data out of the main systems, calculate Sofr, and push those numbers back into the core system, according to Marty Ross-Trevor, senior project manager global markets Libor transition, HSBC.
“People will avoid making drastic changes to core technology and then over time there will be projects going on for the next three or four years where they will actually say, ‘let’s get rid of our old systems, integrate the new requirements, and we will put in a whole new system and transition to that rather than try to hack through potentially millions of lines of code looking to where they have to make changes.’
“A lot of systems will require rates up front, and changing those so that they don’t reset in advance anymore but reset at the end with a five day look back and compound the daily rate - it’s a huge change to some of these old systems which just aren’t built to support that kind of product,” says Ross-Trevor.
Finding the potholes
The success of thdetours in existing systems will depend on how easily data can be taken out and put back into the existing core system. According to Ross-Trevor, some of the older systems may actually be easier to work around because they have less checks and balances in place to override values that have been calculated by the system itself.
The sheer volume of work needed to update and correct these systems cannot be underestimated. For large institutions simple things like updating libraries within code can take quite a long time, says the chief data scientist.
And some of these systems have rarely been updated over their lifetime. Rei Shinozuka, director of capital markets research at the Federal Home Loan Bank of New York (FHLBNY) says he is aware of banks that haven’t modified their loan systems in 10 or 20 years, meaning any wholesale change is problematic.
With Sofr, systems will have to able to handle multiple different kinds of the overnight reference rate – compound average or simple average.
On February 10, Arrc held a webinar to discuss a survey it sent out to technology vendors to understand the current state of preparations in the market. On the webinar, David Bowman, board of governors of the Federal Reserve said that a forward looking Sofr term rate would be created, hopefully sometime next year which would likely be based off the Sofr futures markets and perhaps Sofr IOS markets.
In 2013 the International Organisation of Securities Commissions (Iosco) published principles for financial benchmarks. The principles focus on governance, the quality of the benchmark and its methodology, accountability.
“The Arrc does expect and hope that a robust Iosco compliant forward-looking term rate can be produced in 2021. The exact timing of that is uncertain and the publication of that rate cannot be guaranteed because it really depends on continued growth of liquidity in Sofr derivatives markets,” said Bowman.
But the system detours can only be a temporary fix to the problem as a lot of legacy systems are computationally intensive, and inefficient in collection data and feeding it into various models, says the chief data scientist. Greater flexibility in the architecture is needed, and that can only come from more modern technology.
There are also concerns that creating detours in the coding will create a split of focus with firms having to monitor the calculation outputs from both the existing system and additional systems.
“I think there is going to be a lot more work on places like operations, middle office and risk management to make sure the raw numbers they are looking at reflect the numbers they expect,” says Ross-Trevor.
But while the chief scientist says these system workarounds are likely to occur in commercial and regional banks in the US which historically have not put technology at the forefront of their capital expenditures, HSBC’s Ross-Trevor says it is will happen in all banks regardless of size.
“There aren’t many [banks] that run on the latest up to date vendor products only. I think that almost every financial institution will have whole systems that will require some kind of external module to calculate the numbers for them,” says Ross-Trevor.
Adding to the problem, testing of these re-routes will take longer than many in the market have allowed time for, says Ross-Trevor, as any changes will have downstream impacts on everything from the front office through to the back including systems for accounting, regulatory reporting etc. Firms need to have intricate testing plans prepared, he says.
Making a list, checking it twice
Stalling preparations is the uncertainty of a post-Libor world.
On the Arrc webinar, Bowman warned: “We have essentially 23 months of guaranteed stability and really, past that point, nobody should be certain about what will happen to Libor, and so it is prudent for financial institutions, all market participants to move away from Libor before that time.”
Despite regular statements from regulators, industry associations, and the market in general that all firms which are implicated in the transition away from the reference rate must be ready by the end of 2021, there is a reluctance to transition.
For example, Sofr’s volatility and its impact on pricing will be unclear until the entire market has moved to it.
“What is the spread that we should be using between Libor and Sofr? How do we see that spread? How do we point at it? Whether it is a borrower and a real estate loan or anything else. Without having this ability, it is very hard to ask people for economic acceptance,” says John Coleman, senior vice president and managing director of the fixed income group at RJ O'Brien & Associates
“The banks will tell you they make a three basis point wide market on term Sofr. I’m not so sure those term rates are even right, right now.”
There are still concerns in the industry as to the impact of the lack of credit sensitivity in the recommended Sofr reference rate.
“The absence of a credit component leaves most of these banks saying, ‘wait a second, if we have a systemic shock, I could be lending money to my borrowers at a negative rate.’ And for that reason, they are looking for the Fed to come up with a way to reconstitute something that shows both the risk-free rate sensitivity, and the credit sensitivity,” says Coleman.
“Nobody is talking about a plan B for what is going to be our default in the event that Sofr or whatever other index may show up doesn’t work. There can be only one answer - to go to what’s liquid, and it’s treasury. It still doesn’t take care of the credit problem.”
The Intercontinental Exchange Benchmark Administration (IBA) sought to solve this issue last year after it published the methodology for a forward-looking and credit-sensitive benchmark – the US Dollar Ice Benchmark Yield Index. Though the benchmark remains at testing stage, the IBA has said with satisfactory testing and adequate funding from banks, it hope to begin producing the benchmark by the second half of 2020.
In preparation for the cessation of US dollar Libor, focus by regulators and associations has turned to technology systems and solutions.
On December 23, the superintendent of the New York State Department of Financial Services, Linda Lacewell, sent a letter to ask the chief executives of those institutions which come under its supervision to provide plans on how they are addressing the transition risks, operational risks, process of communications with customers, and governance framework. The deadline for submissions was February 7, but this has been extended to March 23.
A survey of 177 firms from across financial services and corporate industries published by Accenture on September 16 found that one in five respondents were operationally ready for the Libor transition broadly. Additionally, 17 percent of firms said they plan to apportion funds over the next three years to risk models, and 14 percent said they would allocate funds to technology.
“If only one out of seven are investing in technology as of September 2019, and knowing how budget cycles work, that number ought to be a lot higher considering that technology is one of the potential stumbling blocks for making this successful,” says the FHLBNY’s Shinozuka.
On January 31, Arrc published the vendor survey and a buy-side checklist to assist the market. The deadline for submissions to the survey is March 16, with an aggregated anonymised report being published soon afterwards.
The checklist is pushing the buy-side to establish program governance – understanding what systems are used in transactions, and the downstream impacts of changes to Libor, says Shinozuka.
But with so many unanswered questions, Ross-Trevor says vendors are waiting for clarity before making new releases.
“A lot of companies are still waiting on the vendors to do their releases or actually publicise what they are doing. Everyone knows that the vendors are working on [Libor], but exactly what they are working on, how they are working on it, and what products and methods they are going to support is a little more difficult to nail down in some cases,” he says.
For smaller banks who will rely solely on tech vendors to provide solutions for the transition, it is a nerve-wrecking time, according to Ross-Trevor.
“[Smaller banks] are in a situation where they can’t now just develop their own systems to do what needs to be done. They have to rely on the vendor that forms the core of trading or accounting for a particular product.
“A lot of the thinking is, there are 2,000 companies that use this product. It’s got to be right in the end.”