The Role of Artificial Intelligence in the LIBOR Transition

Considered as the vehicle that transmitted the 2007 financial crisis far and wide, tainted by the scandal that caused lasting reputational damage to the entire financial system, the London Interbank Offered Rate (LIBOR) is expected to be phased out by 31 December 2021.

It is estimated that LIBOR affects financial contracts worth $300 trillion globally, including mortgages, loans and many other financial products. Banks are running against the clock to identify and manage LIBOR legacy contracts, how much money is at stake, and the best way to scope, plan and facilitate LIBOR transition projects.

Why LIBOR is being replaced

Two significant events triggered questions over LIBOR's credibility. The first is the manipulation scandal following the 2007 financial crisis. The second is the concerns raised by the FCA that LIBOR is not fully supported by an active market of observable transactions by market participants, which has led the regulator to conclude that LIBOR is no longer fit for purpose.

During the financial crisis of 2007, the rate behaved abruptly and out of line in contrast to other market rates and prices. I have written an article explaining the role LIBOR played in the 2007 financial crisis as well as an overview of LIBOR, which you can find here.

In the following years, financial regulators and some public authorities in the UK and internationally investigated LIBOR's alleged manipulation. Their findings exposed the weaknesses of LIBOR and challenged its credibility as a standard benchmark rate.

In June 2012, Barclays Bank was fined £59.6 million by the Financial Services Authority (FSA, now FCA) and $360 million by US authorities for tampering and false reporting of LIBOR and EURIBOR between 2005 and 2009.

Soon after the Barclays findings were announced, the Chancellor of the Exchequer commissioned the then managing director of the FSA, Martin Wheatley, to set up an independent review of the various aspects of LIBOR.

One of the recommendations suggested in the Wheatley Review was to hand over LIBOR to a new administrator. Following this recommendation, on 1 February 2014, ICE Benchmark Association took over the responsibility for compiling and distributing the rate. This has strengthened LIBOR's integrity through more transparency, robust oversight and governance framework.

Whilst there is no evidence that manipulation has occurred since 2013, the UK regulators – the FCA, the Prudential Regulation Authority (PRA) and the Bank of England – are determined to phase out the benchmark by 31 December 2021. Indeed, in 2017, the Chief Executive of the FCA told market participants that they should not rely on LIBOR being available after 2021.

Artificial Intelligence (AI) makes LIBOR transition projects faster and more cost-effective

Some organisations are employing document review software commonly used to perform discovery review in litigation departments to run their LIBOR transition projects. However, this solution does not meet the needs of banks with large amounts of LIBOR-linked contracts as this would take 1,000 lawyers more than two years to identify and switch over all its contracts.

Therefore, law firms and banks have turned to AI and other technologies, such as Natural Language Processing and Machine Learning, to identify LIBOR as referenced in various guises and languages by a well-trained computer.

In this context, AI produces value in several ways. The most evident benefit is that AI-driven document analysis can be undertaken in seconds, meaning banks can ease the switch from LIBOR. For law firms, this means a cost-effective innovation that provides greater value to increasingly demanding customers. Automating parts of the review process will reduce training time and minimise the potential costly financial risks due to human errors.

Besides, to some extent, AI can apply reason to the information in the document to interpret and produce summaries in a fraction of the time. This is because AI software can use Machine Learning to observe data, direct experience, or instructions, in order to look for patterns in data and make decisions based on the examples provided. This process enables the software to learn automatically without human intervention. It can also group together agreements with similar expected LIBOR transition handling strategies regardless of variance in language. Another value AI brings is that it can perform natural language generation to create draft documents, reports, and chatbots for interaction between internal and external stakeholders.

Final remarks

In the context of LIBOR transition projects, it is likely that only very small review jobs will be performed using manual processes. The AI LIBOR contract review is designed specifically for the complex financial services contracts the LIBOR problem affects and tailored each customer's particular needs. Moreover, AI is a highly scalable and cost-effective tool for facilitating the transition.

Key Takeaways:

• The London Interbank Offered Rate, or LIBOR, is embed into financial contracts worth $300 trillion globally.

• Financial institutions that use LIBOR are expected to transition to alternative benchmarks after 31 December 2021.

• Banks and law firms are using AI solutions to solve the time-sensitive issue of LIBOR transition and remediation.

• AI capabilities such as Natural Language Processing and Machine Learning produce value by reducing time and costs, improving processes and collaboration.