International Association of Insurance Supervisors issues report titled ‘IAIS Report on FinTech Developments in Insurance Sector’
Using Application Programming Interfaces and Open Data … 5
* Introduction … 5
* Definition of “open insurance” and use cases … 6
* Possible risks and challenges for developing open insurance … 7
* Adequacy of current framework … 8
* Barriers, incentives and coercion … 9
* Sequencing Open Insurance Adoption … 9
* Conclusions and next steps … 9
Distributed Ledger Technologies and Blockchain … 10
* Introduction … 10
* DLT in insurance … 10
* Potential benefits for the insurance industry and consumers… 11
* Risks for regulatory and supervisory assessments … 11
* Conclusion and next step … 14
Artificial Intelligence and Machine Learning … 14
* Introduction … 14
* Model for risk management and governance … 14
* Data use and administration … 16
* Ethics, bias and discrimination … 16
* Conclusions and next steps … 18
* * *
Introduction
Given the rapidly increasing ease of accessibility and digital innovation in financial technology (“FinTech”) and its far-reaching effects on the insurance sector,
* Using application programming interfaces (APIs) and open data;
* Distributed ledger technologies (DLT) and blockchain; and
* Safe, fair and ethical use of artificial intelligence (AI) and machine learning (ML) and the use and management of data.
The assessment activities included input gathered through member surveys and interviews with market players and experts. The purpose of the deep dive assessments was to better understand today’s digital transformation landscape, identify issues and trends in specific areas and assess their potential implications for insurance oversight.
This report presents the high-level findings of these assessments for informational purposes. It is not intended to present a definitive assessment of the risks and opportunities of these trends; nor does it aim to state a preference as the IAIS takes a technology neutral approach. The IAIS will continue to monitor these trends and their impact on insurers, consumers and supervisory objectives.
* * *
Conclusions and next steps
The use of AI/ML and big data by insurance companies may lead to supervisory concerns. Across the IAIS membership, several advisors have issued high-level guidance around various thematic areas, including the use of alternative data, model risk management, third-party providers and/or fair use of data. The role of the IAIS will be to take stock of these initiatives and identify good practice as well as potential gaps. 4.5.1 Analyze existing guidance/provide new guidance
* How current regulations could/could not apply to AI/ML;
* Whether further clarification may be helpful;
* How policy can best support and further safeguard the use of AI/ML by taking into account and utilizing broader and general regulations such as data/information protection and general consumer ethics/behaviour outside insurance regulations; and
* Necessary guidelines or standards for AI/ML are needed to explicitly supplement conventional policy frameworks and affect the sustainability of expected future insurance business models, if any.
More concrete, practical guidance on the use of AI/ML that may not only include principles, but also use cases as well as a discussion on how existing regulation is enforced, should be considered.
Additionally, any guidance proposed by the IAIS should ideally suggest the type of skills and resources (especially technical) needed to put such systems in place, particularly with regard to AI auditing and bias analysis capability. Specific areas of interest for possible further exploration include actuarial questions such as the line between risk differentiation and unfair discrimination, and more technical ones (e.g. what methodology is most adequate to validate AI-based insurance engines on pricing/underwriting/investment/claims/valuation reservation and solvency).
4.5.2 Monitor developments in ethics and justice
Despite the importance attached to the topic of discriminatory biases by most members, concrete deliveries from
4.5.3 Alternative data
Indeed, the use of alternative data is relevant both for model risk management and for IoT (an important source of such data), which are two key study topics identified for
* * *
The report is posted on:
TARGETED NEWS SERVICE (founded in 2004) features non-partisan “edited journalism” newsletters and information for news organizations, public policy groups and individuals; as well as “collective” public policy information, including news releases, reports, speeches. For more information contact