FCA takes next step towards model-driven, machine-readable regulation

Chris Hamblin

6 November 2017

Around 100 developers from 30 organisations took part in November's event, which generated ideas that spanned the whole process of regulatory reporting, starting with the conversion of the FCA's rulebook into machine-readable text as a step towards automated advice or "personalised filtering that uncover which parts of the handbook would apply to a particular type of firm." Other teams thought about changes to regulatory reporting that included a move from firms 'pushing' reports into the FCA, with the regulator storing all this data, to the idea of the FCA 'pulling' the agreed data instantaneously from firms as required.

The FCA is looking at four fundamental ways in which to change regulatory reporting. It has bestowed a title on each one.

Efficiency and collaboration - changing the use of the rulebook by using the semantics of business vocabulary and business rules (SBVR) and natural language processing (NLP) to create machine-readable versions of the rulebook; exploring the use of semantics and 'triples' (statements in subject/predicate/object form) in order to map and translate many "internal and external data ontologies" (ontology is branch of metaphysics that studies the nature of existence or being as such; the FCA evidently thinks that it can exist in the plural) from different domains into a universal format; and building on all this to link regulation (the FCA/PRA rulebooks), compliance procedures, databases and data standards together into a universal machine readable format. Model-driven machine executable regulatory reporting is the catchphrase.
Integration, standards and understanding - here the FCA is looking at setting up: MITOC/ISDA, a standardised model for expressing data and processes with the aim of expressing transactions as collections of economic features and trade events: RegHome, a platform for intra-bank knowledge exchange on regulatory related matters; a 'Wiki-style approach' that might allow people to share best practices and knowledge in a crowd-sourced way; and ITRAC, a piece of banking 'risk and controls' software that might help banks with new IT.
'Predict, learn and simplify' - here the FCA is thinking of setting up an "intelligent regulatory assistant" that acts as a regulatory lawyer and helps clients fill in their authorisation forms; an intelligent regulatory advisor (robo-advisor) that guides an applicant for authorisation through the process with basic advice; and something that might use NLP and artificial intelligence to interpret regulations issued by the European Union in accordance with its gigantic Markets in Financial Instruments Directive II (MiFiD-II), this last with the help of CBA, ING and Pinsent Masons.
Blockchain technology for automating regulation and compliance. TechSprint partners UCL and Santander are working on SmartReg, a project to use smart contract and distributed ledger technology to allow the FCA to find out whether firms are complying. On Project Maison, the FCA is working with the enterprise software firm known as R3, RBS and another global bank to explore the possibility of using distributed ledger technology for regulatory reporting. At some stage it wants to set up a pilot project with live data. Incidentally, the Colombian financial services regulator Superfinanciera has just joined R3's global network of members.

The next TechSprint gathering will welcome people who attended last year’s event along with others who have presented ideas to the FCA this year.