Keeping Pace with Technology

• Client Suitability-This is another area of focus by regulators after the global financial crisis caused by mis-selling of structured products. Machine learning can facilitate banks to establish clients’ risk profile, not just relying on the questions and answers in the investment questionnaire, but also to assess the historical investment behaviour of customers. In addition, a more standardised process can be applied on product risk rating among banks. The building and maintenance of customer risk profile as well as matching with products by using Regtech are more efficient for banks to assess and prevent mis-selling risks, and provide more tailored products and services to customers.

• Market/Employee Conduct-Machine learning and NLP can be used to analyse unstructured data such as communication style of traders and trading pattern which facilitate banks to interpret those data and identify any abnormal behaviour or market misconduct such as collusion or manipulation. Further to this, AI is also being deployed to monitor internal culture and behaviours.

• Regulatory Interpretation and Reporting–NLP may be used to interpret new regulations and translate them into rules for risk assessment and reporting so as to help banks to comply with the regulations.

The above examples can show that, with the use of AI, the quality of information available for banks to manage regulatory risks is largely improved and decision can be made in a more effective and efficient way.

Are banks ready?

There are overwhelming ideas of AI solutions available in the banking industry and the expectation is high. However, it may be too early to say they have been successfully implemented. In reality, it involves significant initial investment cost, it takes time and it will subject to a new set of risks:

• Data privacy risk-This goes to another risk for how customer data can be well protected after they are on the shared platform, in particular the regulations on data protection vary across countries.

•Integration of legacy system-Banks with large scale and global coverage can develop their own AI solutions. However, can their legacy systems integrate with the new AI solutions?

• Data quality–Accurate and up-to-date data for AI to process and analyse is another pain spot by banks as it will take time and efforts to clean up the legacy data and categorise them properly.

• Outsourcing risks-It's common to adopt AI developed by vendor instead of in-house. But how can banks ensure the customer data in the vendors’ platform is safeguarded and able to understand the parameters and controls? As stipulated by the regulators, the onus of the outsourced activities is always with the banks. Can the individual obligations be properly discharged in particular the regulators such as, the HKMA and SFC are imposing more individual obligations on senior management.

• Cross border risk-The outcome of better understanding customers’ data and digitalisation of documents involve sharing of customer data across countries. Some country regulatory framework however does not permit outflow of data outside the country. Banks nowadays are no longer only subject to competition among themselves, but also non-traditional market players, which have direct impact on their earning sustainability if banks are not able to streamline their processes, broaden the variety of products and services, and operate at low cost. AI such as big data analytics or machine learning has transformed the way of compliance doing their jobs. It has saved a lot of time by reducing the manual and administrative process, and prioritizes the compliance resources on the risk assessment and strategic decision making.