An Australian Perspective on ASTIN’s Global IFRS 17 Actuarial Best Practice Survey

In July 2025, the “Actuarial Studies In Non-Life insurance” (ASTIN) Working Party released its final report documenting the results of a global survey on actuarial best practices for IFRS 17. The report is the culmination of over one and a half years of work, spanning 34 jurisdictions and covering both general insurers and reinsurance companies.

The chief objective of the survey is to understand the different actuarial approaches taken to implement IFRS 17, including nuances around the measurement models, the risk adjustment for non-financial risk, onerous contracts and discounting, thereby allowing actuaries around the world to contrast their work and explore opportunities for further improvements.

The support the Working Party received from Australia, both from the Actuaries Institute and the wider general insurance profession, was greatly appreciated. Australia is well-represented in the survey, receiving responses from a range of insurers and covering all major classes of insurance.

In this article, the Australian representatives in the Working Party provide their perspective on the results of the global survey with a focus on how Australia compares to the rest of the world. The full report is available on the International Actuarial Association’s (IAA) website for member access under ASTIN (Non-Life Insurance). The results were also presented by lead author Joana Raposo at the IAA’s May 2025 joint colloquium and the presentation is available online .

Measurement models

The overwhelming majority of Australian respondents apply the premium allocation approach (PAA), with the general measurement model (GMM) used for more complex, multi-year contracts. The results are comparable to responses from the rest of the world, where PAA is applied in 95% of cases and GMM is applied in 4%. A small percentage also applied the variable fee approach, but this was limited to general insurers that also sold life products.

Consistent with the fact that many general insurance contracts in Australia have a coverage period of one year or less, the application of the PAA was automatic for many Australian respondents (under paragraph 53(b) of IFRS 17). Where testing was required to show the PAA results were not materially different to the GMM (often the case for respondents from the rest of the world), the percentage of premiums (written or earned basis) was often used as the materiality threshold, with the adopted percentage typically between 0.1% to 5%. These results are broadly similar to our own experience in Australia, where we have observed premiums being favoured when establishing accounting policies around PAA eligibility (due to premiums providing more stable results compared to using claims liability balances). 

One opportunity noted in the survey relates to the lack of a GMM contingency plan should PAA eligibility testing fail and a GMM implementation is required. This conclusion applied equally to Australian participants. Whilst we understand that the Australian industry has progressed in strengthening internal controls around IFRS 17, including building out GMM contingency plans since the launching of the survey in early 2024, the survey indicates this is a fruitful area for actuarial involvement to future-proof the valuation process against more complex multi-year insurance and reinsurance contracts. From a financial reporting perspective, more granular modelling required in the GMM can provide further insights into profitability, especially in relation to revenue recognition.

Risk adjustment

The survey notes that close to 90% of respondents applied a confidence level approach (whether that be using value at risk or margins for adverse deviation), with the remaining ~10% using a cost of capital approach. The results for Australia are similar, with the majority of respondents applying a confidence level over a cost of capital approach.

We conducted further analysis by analysing publicly available financial reports in Australia to better understand why an insurer might favour one approach over another. We have found that the larger ASX-listed insurers in Australia have typically favoured a cost of capital approach, whereas a confidence level approach was more common for smaller insurers, with the confidence level selected being 75%.

One possible interpretation is that, for the smaller insurers, the compensation required is largely the amount required to satisfy and meet solvency regulations in the jurisdiction of operation. As the regulator in Australia (APRA) requires the risk margin to be calibrated at a 75% probability of sufficiency (or one-half standard deviation), the risk adjustment has also been calibrated to this amount. For larger listed insurers operating complex businesses and navigating through a range of stakeholder expectations, a range of internal performance indicators as well as the availability of market information, has allowed compensation to be more easily articulated through the organisation’s weighted average cost of capital, therefore being amenable to cost of capital techniques.

Across the entire survey, the average gross risk adjustment as a percentage of the discounted cash flows is 7.9%, although the difference between insurers could be sizeable with a range between 0% to 35%. In determining the reinsurance risk adjustment, the majority of respondents (75%) applied a proportionate approach to the gross business, with commentary suggesting that the amount of non-proportional reinsurance may have been a contributing factor to those that applied an alternative approach. Similar observations on the range of the gross risk adjustment and approach to the reinsurance risk adjustment apply to the Australian survey responses.

Onerous contracts

The survey suggested that both in Australia and the rest of the world, the general approach used in onerous contract testing is to assess the insurer’s combined ratio (claims cost plus operating expense as a percentage of premiums). However, differences exist in the exact methodology, including the basis used (e.g., relying on historical information, budget/business plans, premium liability assumptions, pricing or market benchmarking) and level of aggregation (e.g., per contract, set of contracts or insurance class level). The more popular techniques incorporated more forward-looking information to be relevant for the group of contracts under consideration rather than simply relying on historical information without any adjustments.

With 50% of respondents noting that actuaries are only involved in the classification of insurance contracts during valuation, as opposed to earlier in the value chain (e.g. designing, pricing and selling), one risk to insurers is actuaries being involved too late in the process. As the classification of a contract would impact the portfolio and level of aggregation, which would in turn impact the measurement of the contract, including onerous contract testing, there is an opportunity for actuaries to be engaged earlier in the process.

From our own experience, we have found actuarial involvement particularly useful when a new insurance product is released or a portfolio has undergone substantial changes (e.g., remediation, changes in business mix, passing through large rate changes) so any impact to onerous contracts/groups of contracts can be identified early, and the financial implications understood and communicated across the business well before the valuation process.

Discounting

The majority of participants (67%), both in Australia and the rest of the world, indicated using the bottom-up approach in establishing the IFRS 17 discount rate. The risk-free yield curve is typically derived based on government bond yields in the local currency. An illiquidity premium is typically, but not always, included in the IFRS 17 discount rate for both respondents from Australia and the rest of the world.

An assessment of publicly available financial statements in Australia and New Zealand show cases where insurers have adopted a nil illiquidity premium, with commentary in the reports noting its immateriality on the overall results. We also reflect in the context of general insurance reserving in Australia that the illiquidity premium concept is relatively new, with limited published research describing a quantitative technique to estimate the illiquidity premium for an archetypal insurance contract without resorting to more sophisticated methods from quantitative finance.

At any rate, as financial statements enter the public domain, we expect this would be an important source of information in setting the illiquidity premium, as it will open up an opportunity for these assumptions to be tested through benchmarking with insurers that offer similar insurance products and with similar liquidity characteristics.