By Akber Datoo
Partner, D2 Legal Technology LLP
Decades of neglect are coming home to roost. In an increasingly regulated and competitive world, it is incumbent on financial institutions to fully understand the complex contracts in place with their clients. Failure to do so risks catastrophe as witnessed by the fates of Lehman Brothers and AIG. If the financial crisis taught us anything, it was that there was an extremely poor understanding of the swathes of contracts entered into. Firms were caught in a whirlwind of panic, unable to find documents, let alone identify what crucial information they contained.
Regulators globally were shocked by this state of affairs. They imposed multiple requirements on financial firms to have better management of their legal contract data. For example, on 6 March 2015, the European Banking Authority (EBA) published a consultation paper containing a draft regulatory standard on the minimum set of information on financial contracts that should be contained in detailed records maintained by financial firms, tying this in to requirements under the Bank Recovery and Resolution Directive (colloquially known as the “living wills” requirement for systemically important financial institutions).
Certainly, these regulatory demands are not unreasonable – the legal contracts are an embodiment of how financial instruments manifest themselves, through the various terms, conditions, and contractual obligations undertaken by each of the parties to the financial contracts. The stability and smooth running of the financial system relies on the satisfactory management of these contractual terms. However, the reality over the last three decades has been that many legal and documentation teams have failed to keep up with the rapidly growing business lines that they support. Ironically, this is in stark contrast to these business lines that have wholeheartedly embraced technological innovation to scale, while the legal and documentation teams have not taken the opportunity to empower themselves in a similar manner with tools, systems, and utilities available to help them manage growth.
It is a sad reflection of the current state of the financial industry, that the stick of regulation is very much trumping the significant carrot of business optimization. The stick, however, is particularly large and forceful at the moment, manifesting itself in very heavy fines and sanctions imposed on institutions, many of which stem from a failure to track and manage contractual obligations, and as a result of an arrogant “too big to fail” mind-set. Nonetheless, the stick has proved effective in forcing the use of FinTech to assist the rectification of this state of affairs by the creation of systems and processes to store and manage legal documentation and data. This is a task that would have been far less complex and time-consuming had the correct processes been in place from the start, i.e. if the documentation teams had filed and stored the contracts, and kept track of key terms negotiated in their contracts. Fixing it after a number of decades of neglect has, however, entailed a different approach due to the sheer volume of legacy contracts of relevance. This is exactly where today’s technology comes into its own.
The big data excitement is all about seeing and understanding the relations within unstructured and structured data that until recently, we have struggled to grasp. Big data operates through comprehensive datasets and messiness, rather than small data and accuracy. One could argue that the challenge is not even big enough to really be a big data one, but it certainly is big enough to force legal teams to become more comfortable with disorder and uncertainty, taking on the role of a data scientist to find associations in the data, and acting on that, rather than a precise trawl through each and every word of every contract.
The initial challenge lies in simply ensuring that technology can be used to assist. Hard copy documents bound and kept safe in a vault are always going to require manual effort to obtain and read through. The legal contracts are therefore now being digitized and stored in electronic form by financial firms, using optical character recognition (OCR) and other related technologies, such as ICR (intelligent character recognition) and IWR (intelligent word recognition), to deal with some of the issues with handwritten rather than typed text. This is even being applied to areas such as the oral contracts between traders, with the use of speech recognition software to identify key terms of the verbally agreed contract and ensuring these match the trade confirmations generated by financial firms thereafter in order to avoid trade disputes.
The digitization phase is not without its challenges, such as the use of poorly defined tables in the documents. It is, however, greatly simplified through the legal documentation being drafted with reliance on precedent documents and tightly defined contract vocabulary.
Even the application of some basic classification algorithms to the electronic documents helps tremendously. The documents are now, post-digitization, searchable, and the legal department is already empowered through the ability to find the documents required, a task that was difficult for no reason other than the poor classification and filing of the documents at the time of their execution. Furthermore, through some basic management of identifiers and mapping to other data sources such as risk and client static data systems, the size of the problem can be sensibly reduced to a far more manageable one, by linking the documents to the underlying trade exposures applicable to them, and focusing on where the risks truly lie.
Finding the contracts is only the beginning of the battle. Whilst the lawyer may still be sceptical of the idea of a Google-style search delivering value on par with years of training and expertise, there is perhaps a parallel to be drawn with moments such as the success of Deep Blue, the chess-playing computer developed by IBM – the first piece of artificial intelligence to win both a chess game and a chess match against a reigning world champion under regular time controls, and more recently, IBM Watson, when it competed with and beat at Jeopardy! former winners Brad Rutter and Ken Jennings.
The automation of tasks and processes in a very procedural “if A, then B” statement is the old model of computing, in which each and every step and scenario is determined in advance by the programmer, offering little help in automating the task of reading the legal contract, understanding its nuances, and deriving key data and intelligence from it.
The FinTech approach is the application of the new model of computing, applying interpretative capabilities that learn from the data and human guidance, adapting over time as new knowledge is gained.
A number of techniques are used to excellent effect:
Correlation: at its core, this quantifies statistical relationships between data values. This allows a key legal contract data requirement to be met, not by understanding the intricate workings of it, but simply by identifying a useful proxy for it, including ones that are both multi-dimensional and nonlinear. There is no need to identify the exact nuances of the client asset and money provisions, as they are applied in the contract, through the client terms and conditions, a complicated documentation stack, and general application of a plethora of overriding rules and regulation. Of course, even strong correlations are never perfect, and one may simply be fooled by randomness. However, that is the increasingly important guiding role of the lawyer – to avoid such issues. What the correlations do is drive the user to the high priority, high-risk areas within the legacy documentation portfolio.
Validation: data values cannot be viewed in isolation. A legal contract and underlying law and regulation impose natural constraints on the data. By mapping these out and applying them, not only is it possible to refine the search and data extraction process, but this can also highlight some of the areas of very high concern. These are ones that a human review is far less likely to detect due to the laborious and repetitive nature of the check, such as ensuring that each party has an appropriate threshold defined in the document (rather than, as seen in a number of documents, the threshold being repeated twice to the same party in error through a simple copy and paste of the language by the drafter, forgetting to then amend the important contractual party reference).
Positive feedback: increasingly, through the use of text analytics and data extraction techniques, more granular and detailed structured legal contract data can be made available to downstream consumers. There is therefore greater visibility of the nuances of the contractual obligations contained within the legal contracts. Through use of this data, in areas such as collateral optimization and credit/funding valuation adjustments (CVA/FVA), any inaccuracies in the data are immediately noticed and can be corrected through the use of workflow.
The positive feedback to cleanse the accuracy of the legal data also forces the legal team to consider the legaleze through the lens of data. Ultimately, barring a dispute with a counterparty, this is the impact of the legal documents the financial firms have entered into. Of itself, this is transforming the real understanding of the financial instruments.
The infrastructure also allows not only better understanding of the current position and risk exposure of an institution through understanding the specific contractual terms applicable across one’s document portfolio, but it also allows continual optimization and improvement.
The application of FinTech need not end with the analysis and control over the legacy document portfolio. Rather, institutions are now looking at the current process of putting the documentation in place, using document generation and assembly tools to drive standardization in cases where bespoke and varied language does nothing other than to complicate the regulatory and compliance requirements applicable. It also allows, through workflow and what-if analysis of terms at the point of negotiations with clients, to really understand the risk and financial impact, enabling true optimization along the way.
The FinTech journey is set to transform the role of the lawyer, empowering and placing them in a position of truly advising and being part of the business, removing the more administrative parts of the role, and replacing them with genuine advisory decision-making.