All articles

AI and the LIBOR Transition

The financial world is never calm seas, but lately the waters worldwide have been especially rough, because support for the London Interbank Offered Rate (LIBOR) benchmark is set to expire by 2021. Many institutions face the risk that millions of contracts – even hundreds of millions of contracts – will wind up invalid or exposed to risk because the interest rates they reference are formulated in terms of a benchmark no longer in play.

In fact, the scale of the LIBOR replacement problem, particularly in the developed world, is overwhelming: LIBOR has been the most widely used interest rate in the world; it has been used to price assets worth over $370 trillion. The financial instruments requiring scrutiny and possible amendment include commercial loans, securitized products, mortgages, derivatives, bonds, floating rate debt, and interest rates/swaps.

It is no exaggeration that banks are now locked, both in competition with each other and as independent players, in a race against time. Replacing LIBOR in contracts requires identifying those among them that require such a replacement, which in turn requires scrutinizing and analyzing all contracts that extend into 2021 and beyond for information. That of course requires separating out the contracts that expire before the LIBOR benchmark is phased out, to narrow down the field of inquiry accordingly. And all this taken as a whole requires as many as a few dozen careful extractions of meaning from each contract in question, with particular attention to any provisions that constitute a so-called fallback clause relevant to the replacement of LIBOR. Not to mention it’s crucial that whoever is doing the extracting isn’t allowed to drift off and make a careless mistake, or costly consequences will follow.

Mind-numbing. That is how any sane person tasked with the sorting and technical analysis of documents will describe the work if it is repetitive and mundane enough. But when it is written in merciless legalese and the number of documents exceeds a few dozen or so, to call this kind of work mind-numbing is simply not enough. Now imagine that the number of documents exceeds not several dozen but one thousand. How about ten thousand? How about one million? How about a hundred million?

For this scale of effort, human language simply lacks the words. In fact, for this kind of effort, humans lack what it takes to handle the work itself. We are too driven to search for meaning and to create it, plus we’re too likely to make the occasional error. When the documents are contracts and the numbers hit the thousands, there’s no match for AI.

Immune to boredom, thorough, methodical, and error-free, artificial intelligence with machine learning offers what may be the only way out of the LIBOR transition problem. A solution such as Applica can be trained for decision-tree-based scrutiny that sorts documents, identifying those that require further action. After an RPA bot retrieving required documents from an archive is implemented, Applica interprets and classifies the reference rates as required. Subsequently, documents are scrutinized for fallback clause content and sorted into classes requiring different sequences of steps. The system requires minimal human interaction throughout the entire process – regardless of file type, language or layout. It is only at the final stage, after Applica has performed its scrutiny and categorization, that a minimum of necessary documents is transferred to qualified human actors for work which must be performed by lawyers and paralegals.

The LIBOR transition problem is a great example of a twenty-first-century challenge that is practically impossible to solve without AI. It is also a clear example of the way Applica exists not to replace skilled professionals, but to make their work more useful to businesses and institutions in the first place.

Interested in learning more about how Applica can help solve your LIBOR transition pain points?

Download the Applica LIBOR one pager or book a LIBOR demo with an Applica expert today.