Digital regulatory reporting: automated compliance still a work in progress

The FCA is running a project to see if machines could read and execute their rules. The results of the first phase of testing are now out. The findings show that the project has promise but the Report also concludes that more work needs to be done.

The Report is the latest milestone in the Digital Regulatory Reporting project. For more background to the project, see our previous posts on what digital regulatory reporting is and a previous update on the pilot.

Codifying rules proves tricky

Unsurprisingly, the Report explains that there is no obvious way to convert natural language regulations into code. Challenges include not only the content of the regulation – and how it should be interpreted – but also how the regulation is presented. The fact that rules are subject to change was also highlighted as a sticking point.

Even so, the Report concludes that the problem of translating unambiguously between natural language rules and machine-readable code is not insurmountable. There are alternative technologies that have not yet been fully tested. And, even if the perfect solution is not found, other aspects of the project e.g. data standardisation would still benefit some use cases.

“The process of translating, agreeing and verifying regulatory instructions expressed in general purpose programming languages is difficult.” – FCA’s Digital Regulatory Reporting Pilot Phase 1 Report
Standardising data is also a challenge

The Report suggests that data quality is another important challenge. The pilot participants found that data must be provided in a standardised format. This required definitions to be precise and well understood by both the regulators and the participating firms. The Report acknowledges that any future roll-out of the framework must have in place agreed technical, data and security standards to ensure high quality of data.

Live reporting via DLT

Despite these challenges, the pilot has successfully built a prototype system that demonstrates the possibility for real time regulatory reporting. The system uses distributed ledger technology. Machine executable versions of the FCA rules are created as smart contracts and loaded onto the network. This code is then executed against synthetic data supplied by the participating firms.

The path to implementation

The Report also sets out the pilot’s findings on how digital regulatory reporting could be governed, operated and ultimately implemented. These acknowledge the opportunities offered by a digitised regulatory reporting system but also the risks that industry-wide adoption would present.

“It is prudent… to recognise that the path to [digital regulatory reporting] becoming a reality is uncertain.” – FCA’s Digital Regulatory Reporting Pilot Phase 1 Report
Next steps

Phase 2 of the pilot has already begun. This is intended to build on the first phase by focusing on:

  1. Testing new ways to develop machine-executable code from regulation
  2. Understanding if the data model is scalable
  3. Undertaking a cost-benefit analysis of implementing the project in the industry