Nitzan Solomon, VP, Head of Surveillance and Financial Crime Technology, Nomura EMEA
The increasingly complex regulatory landscape, together with the exponential growth of data, present a significant challenge to compliance leaders in the financial services industry. There is a growing number of highly effective and innovative RegTech solutions on the market that can play a key role in addressing this challenge but choosing the right solution is a challenge on its own.
The waves of change
“Why can’t we just use excel” said a senior stakeholder in a leading financial institution (FI) many years ago in a discussion about their transaction monitoring strategy. But they soon realised that it is not a sustainable solution and a few years later almost every leading FI deployed automated rule-based surveillance/monitoring tools.
Fast-forward to today, the next wave of change is here. While Compliance costs continue to grow, the levels of effectiveness and efficiency remain low and once again the wind of change is blowing. In a few years from now, relying solely on rule-based solutions is likely to be laughed off as outdated and unsustainable, the same way the use of excel is being laughed off today.
You can’t stop the waves but you can learn how to surf
Innovative RegTech solutions can significantly reduce cost and increase both efficiency and effectiveness. They have the potential to be truly transformative but there are many risks involved and many hurdles to overcome. Deploying the wrong products, or the right products in the wrong way, can result in moving backward and leave FIs with an even weaker defence and higher risk exposure, but getting it right may prove extremely rewarding.
The best RegTech solution
“The best” is quite subjective and would depend on many factors such as line of business, characteristics, use-case, and jurisdiction. FIs should pick a solution that is flexible and customisable, one that can support their current and future requirements. When evaluating products, FIs should also consider the vendor related risks. For smaller vendors, the focus would be onmaturity, scalability, longevity, and financial stability, whereas for the more established vendors it would be on flexibility, agility, innovativeness and willingness to tailor to custom requirements. A good solution would not only increase true positive alerts and reduce false positive ones, but it will also reduce the manual effort required by an analyst across a wide range of tasks including alert review, escalation and reporting.
AI in RegTech
Artifical Intelligence (AI) based solutions have an enormous potential.
They can analyse vast amounts of data and can (to some extent) imitate human intelligence and reasoning, and allow systems to learn, predict, and make recommendations. Most compliance teams today still have a fairly labour-intensive and inefficient operation and there are countless ways in which AI-based solutions can deliver value including hyper automation of existing processes, generating new insights, and reducing cost.
Notable examples include the use of Natural Language Processing (NLP) in surveillance tools, enabling FIs to screen an immense number of conversations to identify potential market abuse or misconduct. A recent study has demonstrated how major scandals such as the LIBOR and FX scandals could have been uncovered had the technology been incorporated into the surveillance function back then. NLP also supports the screening of media and open web, enabling FIs to identify bad actors. Another highly practical use-case is alerts review and disposition. Various RegTech solutions use Machine Learning (ML) to perform the level one alert review instead of, or in addition to, a compliance analyst using past alerts as an input. Another powerful example is solutions that leverage deep learning or other types of unsupervised learning to detect potential money laundering by identifying anomalous trading patterns and suspicious behaviour. The most important factor when it comes to compliance systems is the completeness and quality of the input data and there are various solutions that can assist with identifying and remediating data issues.
The AI Hype: reality check
“Hitler was Right”. Was a comment made by a Microsoft experimental AI persona named Tay. AI-based technology is indeed powerful but sometimes the results fail to meet expectations.
One might think, especially when reading through marketing material, that they simply need to deploy the solution and the machine will do the rest.
Unfortunately, it doesn’t work this way. FIs must understand the capabilities, limitations and risks of the tools they are looking to deploy. For solutions that rely on ML, there is sometimes a proof-of-concept-to-production gap which means that while some solutions may perform well on test data, they may not necessarily do so in production. It is critical to ensure that sufficient training has taken place over a large enough dataset, and that the data used for training is appropriate. In some cases, a hybrid approach between the traditional rule-based solutions and the more advanced ones can prove to be the optimal solution.
Over the coming years, as RegTech products continue to improve and mature, and as success stories continue to emerge, the adoption rate is likely to increase, further accelerating the growth of the RegTech industry.