Almost a decade on, the lawyers still remember that document review. None of them had worked on it, but the mere threat they might have to — amid rumours of trainees bussed to a warehouse, of a mandated six-day working week because of the volume of data to process — meant the project code name remained seared in the collective memory.
“Just reams and reams of this stuff,” recalls one lawyer at another firm, who declines to be named, of a similar exercise in document review. “We could be there from 9am to 11pm just reading and flagging, reading and flagging — and never with any certainty that you were actually getting it right.”
Such is the life of a junior lawyer. Electronic communications — emails, automatically recorded phone calls and online messaging systems such as Bloomberg’s IB, ubiquitous among traders — mean the amount of data stored has ballooned. That has swelled the scope of litigation and regulatory reviews to a size hard to comprehend for anyone outside compliance departments and law firms instructed on the lucrative mandates. A document review that 15 years ago might have involved 20,000 documents might now hit 2m.
Document reviews have long been a staple of early-career experience for junior lawyers, whether as part of mergers and acquisitions due diligence or litigation discovery. A more recent addition is the bank-commissioned compliance reviews that proliferated as money-laundering and rate-rigging scandals came up against an already tighter regulatory net after the financial crisis.
“[One review] involved millions of documents, months of review, with dozens of reviewers reviewing eight hours a day,” says another lawyer, who declines to be named and recently quit private practice. “It would be normal to have targets of around 700 documents to review, per reviewer, per day. Some trainees could find themselves stuck reviewing documents for months.”
The resources needed to master so much data — as well as the strain that such reviews put on bright young lawyers, adding to retention problems — have sparked a hunt for ways to solve the problem. To start with, routine parts of deal due diligence and litigation disclosure reviews are often outsourced to specialist providers at significantly lower cost.
Technology has helped. Start-ups such as Luminance (founded by mathematicians from Cambridge university), Toronto-based Kira Systems, or eBrevia and Everlaw of the US have used artificial intelligence methods such as natural language processing to automate many contract review and discovery processes. That has cut down the hours junior lawyers spend on some time-consuming tasks and, the technology companies claim, boosted accuracy.
Efforts to cut the volume of documents and resources devoted to legal and compliance reviews have faced significant challenges in other areas, however. Not least is the overhaul of the European regulatory landscape over the past two years, first with the introduction of the EU’s Market Abuse Regulation, which came into force in mid-2016, then the advent of Mifid II (the Markets in Financial Instruments Directive) this year.
The new market abuse rules forced banks and other financial institutions to increase the scope of their surveillance and monitoring systems dramatically and, along with a proliferation in trading activity, led to a similar rise in the data they must sift to identify suspicious transactions and orders and notify regulators. For example, Nasdaq stock exchange’s market surveillance systems flagged up some 75,000 alerts in its Nordic region during the first quarter of the year, says Andreas Gustafsson, European general counsel.
Number of documents involved in a review that 15 years ago might have involved 20,000
The volume of alerts has prompted attempts at automated solutions. This has involved a shift in focus to behavioural patterns to help train computer systems to spot where trading patterns deviate from the norm for an individual or their peers. “Organisations are much more interested in behavioural changes so they can stop things before they get too far down the track,” says Rukshan Permal, a partner at professional services firm PwC and a specialist in market abuse and surveillance. That could cut the need for lawyers doing after-the-fact probes.
At Nasdaq, which has invested heavily in machine-learning technology to supplement its surveillance system, the in-house legal team has helped train the system to mimic the decisions human analysts reviewing those alerts typically take.
That has helped discount some of the more straightforward false positives and cut the number that must be manually reviewed. The number of reports made to regulators by the company in the Nordic region increased from 51 in the first half of 2017, before the machine-learning technology was rolled out, to 81 in the first half of this year.
Stumbling blocks remain, however. One is smoothing out the interaction between human and machine, and not merely at a technical level. “It’s not only that the machine is learning, the machine is teaching [the human] too — and sometimes it’s hard to be taught by a machine,” says Mr Gustafsson.
Another pitfall is with the data. While there has been a huge increase in technology providers, it is early days for deploying the technology at scale. Banks are still grappling with their data internally, Mr Permal says, trying to ensure information on orders and trades is held at the same level of detail across many types of system. Because of that, banks tend to hire many “regtech” (regulatory compliance technology) vendors.
Then there are the traders themselves. Better technology, monitoring and surveillance can help bring huge data sets under control, identifying likely cases of front-running or insider trading. That can limit the need for postmortems by lawyers into what went wrong at a bank, although ultimately there will still be those who are determined to evade the systems and the limits imposed on them by the bank — rogue traders such as Kweku Adoboli at UBS, who was convicted in 2012 of fraud.
“There will still be the big investigations after-the-fact,” says Mr Permal. But if they are less frequent, few junior lawyers will complain.
Business of Law
- Data, Knowledge and Intelligence • In-house
- Managing and Developing Talent • In-house
- New Business and Service Delivery Models • In-house
- New Products and Services • In-house
- Strategy and Changing Behaviours • In-house
- Technology • In-house
- Accessing New Markets and Capital • In-house
- Dispute Resolution
- Enabling Business Growth • In-house
- Managing Complexity and Scale • In-house
- Creating a New Standard • In-house
Get alerts on Law when a new story is published