Publication 2 in our Model Risk Management Thought Leadership Series
In our first publication we delved into Model Risk Management from a broad industry perspective, examining where model risk arises and how efficient management can help. We also discussed how to implement an effective framework and explored the common pitfalls.
In this second publication of our four part series, we turn our attention to the insurance industry and we explore the use of models within the industry and the challenges model risks can pose for insurers. We explore why Model Risk Management is gaining momentum including as a result of increased regulatory scrutiny; model demand, availability, complexity and interconnectedness; and the emerging area of climate risk modelling.
Use of models in insurance
Models have long been an integral part of insurance business operations, with financial models being widely used across the insurance industry.
Traditionally, these have been used in the calculation of regulatory reserves, pricing of new business, asset valuations, forecasting, reinsurance modelling and business planning; however, the scope of models used within insurance companies is expanding and has become more sophisticated over time.
The increased reliance on models to support business decisions and the connectivity of models used within an insurance organisation highlights the need for companies to reduce their model risk. Furthermore, as technology continues to advance rapidly, with Artificial Intelligence (AI) and Machine Learning (ML) algorithms becoming more widely incorporated, the risk that models fail to perform as expected is heightened.
Demonstrating not only the validity of individual models but also the effectiveness of the controls covering the design, development, revision, and use of models is of paramount importance. Establishing a comprehensive, robust and fully embedded Model Risk Management framework can help demonstrate this and mitigate the model risks presented.
Does model risk pose material challenges for insurers?
As explored in our first publication, there are many different types of model risk that can arise and pose challenges. Within the insurance industry, the introduction of Solvency II in 2016 raised the awareness of Model Risk Management for capital models, prompting insurers to invest more heavily to comply with the requirements. However, other fundamental models used across the business (e.g. within reserving, pricing) do not face the same level of regulatory scrutiny and there may be less incentive to invest in controlling the model risk.
There are numerous real cases that demonstrate the problems insurers have faced when model risk materialises:
- In 2011 AXA Rosenberg encountered a spreadsheet error that over-estimated client investment losses and failed to declare the mistake. This led to a $242 million fine and reputational damage.
- An error in a spreadsheet for the scheme valuation for Mouchel Pension Fund discovered in 2011 resulted in a profits downgrade of £8.6m and a fall in share price by a third.
These examples highlight the importance of robust Model Risk Management to prevent errors that can lead to financial loss and damage to reputation.
Why is Model Risk Management gaining momentum within the insurance industry?
Over recent times, the boundary of model usage is broadening as insurers are increasing the quality, pace and breadth of innovation and are embracing the technology available. While some model risks associated with Solvency II valuation are already subject to rigorous regulatory scrutiny, model risks associated with the broader use of models are gaining more attention.
In particular, the focus on Model Risk Management continues to gain increased attention as a result of regulatory scrutiny and model demand, availability, complexity and interconnectedness.
Furthermore, the risks around climate risk modelling is an emerging area for insurers, as they grapple with the complex challenges posed by modelling climate change.
1. Increasing regulatory scrutiny
- The Prudential Regulation Authority (PRA) published the policy statement SS1/23 on 17 May 2023 for banks with Internal Models for credit, market and counterparty credit risk which communicates a significant increase in expectations on Model Risk Management requirements for banks. The policy promotes model risk management as a risk-discipline in its own right and sets five key principles that must be integrated into each firm’s risk management framework including: model identification and model risk classification; governance; model development; model validation; and model risk mitigants.
In their “Dear CEO” letter to insurers published January 2023, the PRA set out its supervisory priorities for the year which included risk management as a key priority. In relation to Model Risk Management, and given the central role that models play in supporting risk assessments, the PRA set out their expectation that insurers reassure themselves of the continued validity of their models, considering the extent to which the Model Risk Management principles set out for banks could be applied and, in particular, whether current validation remains robust in the face of multiple concurrent stresses. PRA noted there will be focus on how well insurers’ capital models are operating in conditions that differ substantially from those that prevailed when much of the current modelling was developed. - In the more local context, there has been increased regulatory attention on pricing models as a result of the work performed on differential pricing by the Central Bank of Ireland (Central Bank) and the Financial Conduct Authority (FCA).
- The Bermuda Monetary Authority (BMA) released a Consultation Paper during 2023 covering enhancements to the regulatory and supervisory regime underpinning the Scenario Based Approach (SBA) and sets out updated guidance on model risk management.
The Model Risk Management enhancements proposed in this consultation paper were formally brought into regulation earlier this year, with an effective date of 31 March 2024.
2. Model demand, availability, complexity and interconnectedness
Model demand
Customer-centricity is crucial in the insurance sector and customer expectations are evolving when it comes to level of service and personalised products demanded.
This is leading to greater use of emerging technology such as chatbots and using AI to assess claims for example.
Insurers are required to embrace technological advancements, leading to further reliance and heightened demands for models with regards to capabilities, granularity and speed.
This demand for models has led to increased complexity of the typical model portfolio, the increased use of algorithms and the sophistication of underlying technologies, and the diversity of the environments in which they are used.
Model availability
Models are only set to further evolve and grow in terms of technological complexity and numbers of models used, as insurers strive to increase automation and reduce manual intervention through implementing approaches which leverage AI, ML and Big Data.
For example, the use of automation technology such as Robotic Process Automation (RPA) to automate time-consuming and manual tasks, making processes more efficient. As these approaches are integrated into the business processes, insurers can be exposed to additional and greater model risks including ethics, transparency, explainability and bias risks.
Conventional approaches may have had little exposure to these risks and the insurer might not have much expertise in handling these risks. Model Risk Management practices should evolve to address the unique risks and characteristics of AI and other advanced technology.
Model complexity and interconnectedness
As technology advances and the demand and availability of models grow, the models become more complex, interlinked, and reliant on each other.
As a result, the stakes for managing model risk are on the rise. In particular, there is greater integration of these models within key insurance business processes, such as sales and marketing, claims management, pricing and underwriting, reserving, capital modelling.
With heightened reliance placed on these models, insurers are exposed to greater model risk resulting the potential for greater operational losses, direct financial losses, reputational losses and insurers failing to meet their business strategies.
3. Climate risk modelling
- Climate risk is a complex and emerging risk and the models supporting a firms understanding of their climate risk is emerging as a specific area of insurance supervisory focus in the EU, in particular as a result of Corporate Sustainability Reporting Directive (CSRD).
- Insurers are increasingly grappling with the modelling of climate risk, which is becoming more intricate due to the evolving climate. These changes not only affect the models but also introduce heightened model risks. Model risk management becomes pivotal in this context, ensuring that the models remain robust and reliable. This applies to many types of models that used the insurance process, such as in managing capital, transferring risk, and setting prices.
- Here in Ireland
- The Risk Management Working Group of the Climate Risk and Sustainable Finance Forum recently published a report which included findings from a survey that sought the views of the financial services industry on climate risk management. Specifically in relation to risk management framework, the insurance sector identified challenges related to data availability and limitations in risk models (e.g. Climate Risk Models may be prone to data issues and model risk), alongside a consistent need for staff training.
- The Central Bank shared insights from their recent thematic review of Natural Catastrophe (NatCat) modelling practices in their Q2 2024 Insurance Newsletter. The insights shared highlight the increasing importance of modelling and the management of model risks that can arise, emphasizing the need for firms to have a thorough knowledge of their NatCat risk profiles to deal with these challenges efficiently. In particular, the Central Bank shared observations and good practices in relation to model validation and model change. The Central Bank observations included that good practice model governance includes the maintenance of a model change log that clearly sets out the details and impacts of model changes over time, and the validation work done by the firm. The Central Bank found that most entities do not keep track of model changes, and their impacts locally and they encourage firms to develop this practice.
Conclusion
Robust and reliable models are critical for a well-functioning and well-trusted insurance industry. As insurers increase the quality, pace and breadth of innovation, model risk will gain more regulatory focus.
Establishing a comprehensive, robust and fully embedded Model Risk Management framework will help insurers to understand and mitigate the model risks presented.
Furthermore, stakeholders including the board of directors, senior management, and shareholders, are demanding improved Model Risk Management, accountability, controls and documentation.
How KPMG can help
KPMG has a successful track record of providing a broad range of financial and strategic advisory services to clients across a wide array of industries related to model risk management.
We have developed KPMG’s Model Risk Management approach which can help you create a well-controlled, integrated, and comprehensive Model Risk Management programme and offers a practical framework for identifying, quantifying, and mitigating model risk by addressing the sources of risk head-on.
Depending on your specific needs, KPMG can assist with any combination of the components of a successful Model Risk Management programme including:
- Model inventory
- Model risk assessment
- Model development & implementation
- Technology solution
- Model validation
- Model policy & governance
- Model data aggregation & quality
- Internal audit assistance
Get in touch
Discover how to improve your Model Risk Management programme by talking to our in-house experts today.
