The previous article in the series was by Danielle Ling and Olivia Ai, presenting the results of a recent survey: Insights From the 2024 Survey of General Insurance Reserving Practices in Australia.

In this article, I have summarised the key findings and recommendations from a 2006 paper released by the UK’s General Insurance Reserving Issues Taskforce (GRIT): A Change Agenda for Reserving: Report of the General Insurance Reserving Issues Taskforce. Many of the recommendations have relevance for actuaries in the modern Australian context.

Purpose and scope of the paper

The overarching goal of the paper (and the GRIT in general ) was to identify opportunities for actuaries to improve general insurance reserving – with respect to both the accuracy of actuarial estimates and the perception of actuaries in the financial industry more broadly.

This was in response to increasing scrutiny of the actuarial profession in the UK, noting in particular:

• the downfall of Equitable Life (the world’s oldest mutual life insurance company) in 2000 after the UK’s Supreme Court ruled that it had breached its contracts with policyholders;
• the public inquiry (led by Lord Penrose) into the closure of Equitable Life, with a report published in 2004 criticising both the management of the company and the actuarial profession as a whole (for inadequate professional standards and insufficient technical guidance, amongst other issues); and
• the Morris Review of the UK Actuarial Foundation, which was set up by the government in response to the findings of the Penrose Report and which released its Final Report in 2005.

While the GRIT paper was not written to directly respond to the Morris Review (in fact, GRIT developed its work programme before Morris published its findings), the authors did review the findings of the Morris Review as they were released, and noted some (but not complete) overlap in the scopes of the two pieces of work.

The authors of the paper initially released a set of thoughts and recommendations in a Consultation Paper, and then sought feedback from actuaries and the users of actuarial reserving services (such as CEOs, regulators and rating agencies). The March 2006 paper then served as a final report to encompass all of the research, discussions and feedback. It includes recommendations across a number of key themes, which I have condensed into the following components of the valuation process:

• improving base knowledge of the business;
• enhancing the consistency and quality of actuarial methodologies;
• appropriately disclosing uncertainty that surrounds our work; and
• clearly communicating results to a variety of stakeholders.

Recommendations

Recommendation 1: Improving base knowledge

The authors of the paper determined that a better understanding of the business that makes up reserving classes would lead to “considerable improvements”.

Specific suggestions included:

• Improving the homogeneity of classes – that is, adding additional granularity and/or cutting the data in such a way that the sub-groups of policies used in the reserving process would have more similar development characteristics. One such proposition suggested by the authors was splitting claims into new notifications and previously notified claims, regardless of the accident year to which the claim belongs.
• Identifying atypical policies for separate consideration – that is, determining outlier policies (or claims) in a database that may be inappropriate for aggregation within a sub-class and analysing these policies separately (to improve the accuracy of estimates for both that specific policy and for the remainder of the sub-class).
• Understanding the claims handling and case reserving processes – that is, being aware of behaviours (or changes to behaviours) amongst the claims team which may affect the pattern of claims development or the accuracy of underlying data.
• Tracking changes in policy wording – that is, being explicitly aware of any terms and conditions which may cause future experience to differ from that observed historically, which would then allow the actuary to make the necessary adjustments to models and assumptions.
• Developing diagnostics which are specific and appropriate to the setting in which they are used – that is, ensuring that the diagnostic measures that are used are simple and understandable, as well as being tailored to the specific models and classes of businesses being assessed.

Recommendation 2: Consistency and quality of actuarial methodologies

The GRIT also concluded that there should be greater consistency in the application of existing actuarial methods, and that the methodology used by actuaries should be interrogated to understand whether it is sufficiently accurate.

My key takeaways from these recommendations include:

• The authors did not believe that “more sophisticated mathematical and statistical methodology” was an area requiring substantial focus – more on that later. That said, the paper does recommend considering which methods perform well in which circumstances (i.e., for certain valuation classes or for recent/older accident periods) as well as greater sophistication around the estimation of reinsurance recoveries (as opposed to the use of simple gross to net ratios).
• Testing of data quality and reasonableness was a key focus.
• Hindsight analysis and testing of prior results were also highlighted, specifically in determining the ‘quality of fit’ of selections and whether or not the selected reserving models were appropriate.
• The authors emphasised the importance of using actuarial judgement as opposed to blindly applying averages – for example, by encouraging a greater amount of discretion and justification with the selection of development factors (particularly in the tail) and IELR assumptions for the BF method. It is noted that overconfidence is greatest for tasks with low predictability and a lack of fast, clear feedback, which is very true of reserving (especially for long tail classes). At the same time, the authors noted how anchoring bias can limit the responsiveness of actuaries in changing environments, and suggested that actuaries should be more aware of this phenomenon.
• With respect to significant judgements and the selection of assumptions, the GRIT paper notes that actuaries should engage with key individuals (for example, underwriters and claims handlers), and should look for additional sources of information where judgements are particularly significant.
• The paper also makes an interesting observation with respect to the underwriting cycle. It confirms the conclusion of the Reserve Cycle Working Party, which suggested that “the claim development profile lengthens in the soft part of the underwriting cycle” and that “most actuarial methods… do not allow for this”. In doing so, the paper highlights the need for further research to be done to understand the drivers of the reserving cycle and how actuarial methods could be adapted to account for impacts on claims performance.

The GRIT suggested that a large number of these recommendations could be addressed through the actuarial training syllabus (in addition to being considered by qualified actuaries) in order to improve the robustness of reserving processes moving forwards.

Such changes to the training syllabus might include learning about practical issues related to the traditional reserving methods, a wider range of insurance types and the effect of changing terms and conditions on liabilities.

In addition, the GRIT highlighted better documentation of reserving practices as a means of pursuing greater consistency and accuracy between valuations.

Recommendation 3: Disclosing uncertainty

While the recommendations above could be used to improve the estimation of a best estimate in the reserving process, the paper also specifically identifies the need for actuaries to do a better job of acknowledging and quantifying the uncertainty of our work.

Specifically, the paper recommends that:

• A numerical measure of uncertainty should accompany any point estimate of reserves (although the paper is not specific on what that measure of uncertainty should look like).
• The key drivers of uncertainty should be disclosed and discussed so that actuaries and non-actuaries alike are able to understand the reason for potential differences between the best estimates and the actual experience.
• The profession should “urgently” agree on a common vocabulary to be used when discussing uncertainty – again, so that there is transparency and consistency in actuarial communication that will help to set expectations and allow for comparisons to be made between different actuarial documentation.

Recommendation 4: Communicating results

Following on from the conclusions on uncertainty (and the need to discuss it transparently and clearly), the paper also recommended that all actuarial communication could be done more effectively.

The authors noted that there was a disconnect between the perception of actuaries held by actuaries and that held by other stakeholders and the press – specifically, that actuaries were generally more satisfied than other stakeholders that reserving performance had been at least adequate.

The paper therefore concluded that there needs to be a clearer framework (in terms of how we communicate, including specifically addressing how we engage with non-actuaries) which enables non-actuaries to assess the key assumptions and the level of reserves. It is not enough for our reserving techniques to be solid if we cannot justify the selected estimates with effective communication.

To be continued…

We aren’t done yet! The GRIT paper has a range of recommendations which remain pertinent to this day – and in my next article, I will outline my views on where things have changed since 2006. Stay tuned!

CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.