March 06, 2020

 Michael Bommarito

contracts ai machine learning ai and data science libor contraxuite governance risk compliance

The market has changed dramatically in the six months since our original white paper discussing the LIBOR transition. Organizations are making the transition to other reference rates, such as SOFR and SONIA, but some regions are farther along than others.

We know that LIBOR will play a diminished role in the global financial system going forward, and that it is likely to disappear altogether by end of next year. The pros and cons of various alternative rates have been extensively analyzed. At this point, the major concern with any new rate–whether it’s SOFR, SONIA, Euribor, or one of a dozen-plus of others–is whether the risks can be foreseen and ameliorated before the end of LIBOR’s hegemony.

The questions that law firms, corporate legal departments, ALSPs, banks (and everyone in between), have are:

• Do our customers have “LIBOR-impacted” contract paper?
• How much money is at stake?
• What is the best way to scope, plan, and facilitate a LIBOR transition project?

Recommendations

We have dealt extensively in LIBOR transition work since early last year, helping customers design and implement unique service- and tech-driven solutions to meet their LIBOR transition goals. The first step is to identify all of a customer’s LIBOR-impacted contracts. Then, teams of lawyers and legal professionals work together to create and execute a LIBOR transition project.

We have three recommendations for the most common scenarios we’ve encountered.

Situation 1: A bank and a law firm agree to work together on a LIBOR transition project. The bank is multinational. The law firm has a software solution they can provide the bank, but this tech solution does not meet all their needs.

Recommendation 1: Often, a software solution is cloud-based. Banks and larger multinationals often prefer on-premise hosting so that their software solution comports with internal security protocols. Software must be able to be deployed either in the cloud or on-premise, with security architecture that a large bank, financial institution, or other large asset manager can personally take charge of.

Situation 2: Another common scenario involves a bank and a law firm that agree to work together. The bank and firm originally choose a cloud-based software solution, establishing a secure instance on the cloud to get to work. The bank then realizes that many of the results the software returns are cumbersome or inaccurate. They need a tool that is extensible and customizable, with data science support.

Recommendation 2: For effective and efficient contract analysis and remediation, software tools need to have customizable data fields, as well as the ability to perform conceptual search (search using regular expressions, word embedding models, and other techniques, discussed below). In addition, a law firm, legal department, bank, or other institution benefits greatly from having an internal data science team to support ongoing technical administration and refinement of contract analysis protocols.

Situation 3: A bank and a law firm agree to work together. Contract data extraction and analysis initially proceeds with their chosen software solution, but the results require substantial analysis and interpretation beyond the bandwidth of available resources.

Recommendation 3: We have found that an internal data science team is useful here, but it is also useful to have teams of reviewers that can be called in for extra support at certain stages of a project. A LIBOR transition project frequently requires this kind of flexible resourcing, even when the extra resources provided by a third party are only a temporary need.

LIBOR Transition Projects

Any effective contract analysis regime will have machine learning (ML) at its heart. A well-trained ML platform will find the most relevant data points in a large set of documents, and can be adapted and customized with additional ML techniques to meet the unique challenges of a LIBOR transition project. As discussed above, we have found over the past year that technology and people working together creates the most effective solutions. Designing and building the right data fields and review processes requires strategic planning and communication between various types of subject matter experts, both on the customer side and on the provider side. A software development team can build ML algorithms in a multitude of different ways, but it takes experts in law, tech, and finance to fully implement a comprehensive LIBOR transition project.

Machine Learning is not a conveyor belt where data goes in and perfect results come out. It takes time and iteration. Thus, ML is a natural companion discipline for legal: both disciplines require taking in imperfect data, then developing creative and effective solutions with that data.

Software solutions powered by machine learning play a critical role in solving the unique computational and scaling problems in projects like LIBOR transition projects. There are dozens of software solutions available, but software is only part of the picture: Without knowledgeable people using and administering a software solution, a LIBOR transition project may falter.

The best LIBOR transition projects are those conducted by teams of various specialists who understand and work with resources in both software development and service delivery. Our most successful collaborations have involved Elevate teams working nimbly with customer teams to integrate strategy and process in ways that significantly augment the customer’s work. With the Clifford Chance Data Science Lab, for example, we have defined dozens of the most critical data points in LIBOR contracts, and extracted them with ML-based techniques into structured data objects.

What We’ve Learned

Since the announcement of LIBOR’s sunset, major regulators and advisory bodies have issued orders and published guidance for organizations to follow as they transition. There are numerous approaches, including the adoption of alternative reference rates for all new contract paper (and all remediations), but there are as many ways to implement new processes as there are organizations looking to make prompt changes. Every customer is different, but there are some commonalities that characterize all of our LIBOR transition projects.

1. AI makes process improvement faster and more flexible

A customizable AI platform allows for building solutions that meet the exact needs of the customer. The best platform is one with pre-built data fields, and the flexibility for users to create custom data fields that capture data specific to syndicated commercial credit agreements that reference LIBOR and other rates.

2. Collaboration is highly effective amongst multiple subject matter experts

LIBOR transition projects provide fertile ground for collaboration. Teams of lawyers, document reviewers, and software developers bring knowledge and insight to transition strategies, while financial experts and data scientists bring experience in diagnosing and remediating customers’ LIBOR transitions.

The Clifford Chance Data Science Lab, as well as other similar operations at large law firms and other organizations, has collaborated with Elevate to implement our AI platform ContraxSuite, as well as other software tools, in ongoing LIBOR transition projects. The goal of collaboration is to deliver customers successful solutions, while providing education and knowledge transfer between organizations about the most efficient processes, and the best technology solutions, for sophisticated remediation.

3. Effective innovation comes in the form of “Humans + Machines”

Humans have been building new tools (and improving old ones) forever. When it comes to new technologies like AI, process improvement initiatives within organizations, and even the new kinds of collaboration that law companies bring to the market, we see all kinds of innovative techniques coming together to solve the global, time-sensitive issue of LIBOR transition and remediation.

Law firms like Clifford Chance have collaborated with Elevate to develop flexible, custom solutions to legal problems. Every organization is different. Every matter is different. But with an ultimate goal of “Humans + Machines”, great outcomes are possible. Combining people, process, and technology enhances outcomes for all types of customers, with needs that go beyond just the LIBOR problem and similar short-term projects. Eventually, the LIBOR problem will be behind us, but scalable solutions that combine expertise with technology will continue to be necessary as the volume of contracts grows at a global scale.