Modeling the pandemic risk
How to model the COVID-19 pandemic and forge a path to better risk insurance.
The insurance industry continues to navigate the COVID-19 pandemic. This article outlines the decisions made by insurers and reinsurers in response to the pandemic, how these decisions have worked out and what risks remain.
We discuss our observations on whether these decisions were beneficial to the companies or not, either from a financial or reputational perspective.
Effective executives and board members are continuing to challenge the thinking and decisions of themselves and their teams as they manage the ongoing impact of the pandemic.
During the early stages of the pandemic, insurers were asking the following:
During the first six months of the pandemic, we saw insurers begin to take deep dives into the current profitability of their products, particular credit life, income protection and funeral. In some cases, previously profitable products were suddenly loss-making. For others, poorly performing products were tipped into severe losses.
Either way, the pandemic triggered a more granular product-level and benefit-level experience investigation. Level 5 lockdowns triggered an enormous spike in retrenchments, partly as marginal businesses quickly saw the writing on the wall and retrenched staff. It took a few months for these claims to be reported, due to retrenchment processes and further delays caused by the lockdown itself.
Some insurers considered these reported claims and, taking a view on future adverse retrenchment experience, decided to close certain lines of credit life and income protection products to new business. (Most compulsory credit life products continued open as they remained a condition of loans being granted, although credit extension itself slowed.)
Some insurers had reinsurance agreements in place. For those without reinsurance, a key factor was the cost of obtaining this cover during the pandemic.
Acquiring reinsurance during the pandemic was difficult as reinsurers were going through a similar process of assessing their risk exposure and risk appetite. Reinsurers’ risk appetite for retrenchment risk was limited before the pandemic. The pandemic reinforced this position and slowed down decision making, making it even harder for insurers to access reinsurance.
Insurers that were able to find new reinsurance cover had to decide whether to accept the reduced margins given the increasing cost of reinsurance.
Others that were in stronger solvency positions could afford to keep their products open to new business, despite not being able to acquire reinsurance on terms acceptable to them.
Some insurers sought legal opinion on the terms and conditions of their products when it came to premium reviews and claim repudiations. These legal views did not always adequately consider policyholders’ reasonable expectations, the mood of the country and the developing perspective of the Ombud.
We saw divergent practices among insurers, including on premium increases, claim repudiations through tight application of policy conditions requiring total loss of income and questions around how to treat TERS (Temporary Employer/Employee Relief Scheme) payments to offset loss of income. Like the challenges experienced by underwriters of business interruption cover, the policy wording was likely not written with these scenarios in mind and therefore interpretation was necessary without the benefit of precedent.
Most compulsory credit life products are priced at the regulated premium caps, and therefore premium increases were not an option.
Some insurers increased premiums on existing funeral policies significantly, in some cases more than 100%. It should be no surprise that this has attracted the attention of the Ombud and Financial Sector Conduct Authority (FSCA) and has resulted in draft proposals to require reporting of premium increases above 20% to the FSCA.
Insurers are also reconsidering the maximum age at entry for funeral policies, and the extent to which further differentiation in premiums by age is required.
In Q3 2020, when insurers first had to report figures with a definite expectation of adverse experience from COVID-19, many considered the additional mortality over a 12-month period from a single wave.
As the pandemic continues and more data has become available, it is clear that death claims have been higher, and the period of higher claims has continued longer, than initially modelled.
In August 2021, the Association for Savings and Investment SA (ASISA) announced that claims paid from 1 April 2020 to 31 March 2021 were 64% higher in the last 12 months compared to the prior period.
Mortality experience on products with limited underwriting and limited waiting periods was initially severe. The limited underwriting on funeral and the short or zero waiting periods on credit life has meant these products have been particularly hit, although credit life experience has varied by type of credit, target market and provider.
Microinsurance products with the required maximum waiting period of three months meant some microinsurers’ experienced a period of very heavy mortality, particularly for extended family members. The FSCA is now reconsidering the three-month maximum waiting period.
The average age of group funeral and group life is sometimes lower than individual funeral, especially when adult dependents are considered. Group funeral and group life were therefore not generally as badly hit from claims. Still, across group and individual funeral, claim experience was up between 30% and 60%.
The second and third waves saw a shift of claims from non-underwritten products to higher sums assured, underwritten products.
The number of individual life claims was up 57% over the year to 31 March 2021, where the Rand value of claims paid was up 70%. The reasons for the difference relate at least in part to higher mortality rates and sums assured at higher ages.
Retrenchment experience turned out better than most had initially feared. After an initial spike, retrenchment rates have returned to more moderate levels. The outlook for the economy remains bleak, so there remains a risk of an extended period of adverse retrenchment, albeit not to the level experienced in the second half of 2020. For credit life and income protection, which are relatively more exposed to retrenchment and disability, the relatively good news on retrenchment experience has at least partially offset the bad news on mortality experience.
In responding to the pandemic, some insurers took a narrow, short-term view of profitability, while others balanced current profitability, ongoing volumes and business purpose. Those with adequate capital to weather the storm were better positioned to take this longer-term perspective. Some did not have this option.
Generally, insurers that kept their credit life and income protection products open to new business, with or without reinsurance, were better placed to continue business as usual when retrenchment experience turned out to be better than initially feared.
They benefited from more stable business volumes and spreading of fixed expenses, as well as better reputational outcomes and appreciation from customers for continuing to provide insurance when it was most needed.
The key remaining issue is what the long-term impact of COVID-19 will be on claim experience and business volumes.
Insurers are considering the use of vaccination status as a rating factor. The actuarial justification for vaccine status as a risk factor seems clear—and the positive externalities of encouraging more citizens to become vaccinated are also valuable. Given the “other pandemic” of vaccine hesitancy, there has been some vocal pushback, although it’s difficult to assess whether this is a small minority or a somewhat bigger group.
What remains to be seen is whether vaccine status is a further proxy for other lifestyle and health choices around conscientiousness and respect for preventive medical care, which could be a stronger predictor of mortality and morbidity than just the direct COVID-19 health impacts.
Given the reports of “long COVID” and protracted cognitive decline in those having survived SARS-CoV-2 infections, some underwriters may ask whether “every SARS-CoV-2 infected” status should be an underwriting factor, at least for permanent health insurance and long-term care.
To our knowledge, and as of writing this article, no insurers have announced plans along these lines. This is for good reasons.
Some insurers benefited from their existing operational efficiencies and robust policies. They were able to analyse the data and results, and convey the insights to the necessary decision-making forums timeously.
For other insurers, the pandemic brought existing governance, risk and compliance weaknesses to the fore.
The circumstances under which an insurer’s out-of-cycle Own Risk and Solvency Assessment (ORSA) would be triggered should be defined in its ORSA policy. For many insurers, the change in risk profile due to the pandemic was sufficient to trigger an out-of-cycle ORSA.
Some other insurers did not experience a material change in risk profile and therefore did not perform an out-of-cycle ORSA.
A few insurers, despite having their risk profiles impacted dramatically by the pandemic, chose not to perform an out-of-cycle ORSA.
To reach this conclusion, some relied on narrow definitions of triggers from ORSA policies to demonstrate that they did not need to perform an out-of-cycle ORSA. This is an unusual decision—we have yet to see an ORSA policy that prohibits an out-of-cycle ORSA if management and/or the board request one.
For example, the following changes affected almost all insurers:
While we recognise that some insurers were not materially affected by COVID-19 and the lockdowns, we saw some insurers that were affected by the issues mentioned above, but relied on narrow measures such as current Solvency Capital Requirement (SCR) cover to not perform an out-of-cycle ORSA.
Some insurers viewed an out-of-cycle ORSA as the entirety of processes required to completely redo their annual ORSA reports. As a result, some of them were produced more than six months after the start of the pandemic, where their usefulness to inform and help make decisions was modest.
However, a complete ORSA process and full report is not necessarily required.
The ORSA is a set of processes to understand the insurer’s own risk and solvency, which typically culminates in a report once a year.
An out of-cycle ORSA could then include a subset of:
An out-of-cycle ORSA is often more useful when produced quickly, focussing on the key elements that have changed, rather than complete but late.
Some insurers invested too little time in understanding the medium-term impacts of the change in risk profile and their decisions. Some invested too much time in producing heavy ORSA reports that were too late to be useful and distracted risk teams in supporting other, more urgent, business decisions.
We found that insurers that performed out-of-cycle ORSAs or simplified scenario testing, whether or not this was triggered by an ORSA policy, were better placed to inform and manage their responses to the pandemic.
Reinsurers experienced a tightening of central control functions. Decisions relating to claim practices and pricing reviews were taken by parent companies to ensure consistent action was taken across geographies.
This introduced layers of bureaucracy, which slowed down claim assessment processes within the reinsurer and therefore sometimes the insurer too.
Our economy would be in a significantly worse position if we experienced the pandemic without the internet. However, the extensive addition of new technologies and ways of working increased cyber risk exposures for insurers as employees moved to working remotely.
Although most IT teams have now improved their controls, the risk is still higher than it was pre-lockdown when everyone was in office.
Insurers have been increasingly relying on digital channels for distribution and servicing. While this was driven by a need to maintain sales volumes during the pandemic, it increases the number of “attack surfaces” from which insurers are exposed to cyberattacks.
Insurers that have had to deal with large claim volumes due to the pandemic have seen their risk controls placed under pressure. Controls are more likely to be overlooked in such an environment, reducing the effectiveness of these controls. Insurers’ own administration staffs are acutely aware of the weakened controls and, in some cases, actively exploited this to engage in noncyber fraud.
With the increasing number of cyberattacks, there has been an increased demand for cyber insurance. Premiums for such cover are likely to increase, which will have a knock-on impact on insurers’ total expenses.
Insurers that were dependent on face-to-face distribution channels experienced significant reductions in new business sales and policy renewals due to the lockdown.
The reduced sales volumes resulted in higher expense allocations at a policy level, reducing the solvency position of insurers.
For policies where the policyholder could afford to maintain their policies, lapse experience has been surprisingly good, even better than pre-pandemic as policyholders recognised the increases in risk and value of their policies. It will be interesting to see how long this greater appreciation of the value of life insurance continues.
Most insurers initially set up additional provisions for the anticipated adverse experience related to COVID-19. They included expectations of worse mortality, worse morbidity, higher lapses and higher retrenchment or other loss of income claims. At the time, only early emerging experience from China, some cruise ships, Italy and the US was available. Estimating likely future infections, case fatality rates and infected fatality rates was extremely uncertain.
There were still some outlier groups grossly underestimating the impact of the pandemic. Even for those applying their best efforts, many points were extremely uncertain, including the extent of infections and deaths, the number of future waves and the secondary impact on the economy and therefore lockdown.
Initially many insurers only considered a single wave or weren’t even thinking in terms of “waves” and looked at possible experience over the coming 12 months.
We now know that mortality rates were understated due to second and subsequent waves, and more infectious and lethal variants. Despite the unprecedented speed at which safe and reliable vaccines were developed, the rollout of the vaccine in South Africa has been disappointingly slow, first due to government procurement and logistical challenges, and now due to troubling “vaccine hesitancy” fuelled by social media misinformation.
Initial COVID-19 provisions held against future excess mortality claims were insufficient for many insurers and needed to be increased at subsequent valuations.
On the other hand, COVID-19 provisions held for retrenchment claims generally turned out to be higher than necessary as retrenchment experience in 2021 improved compared to 2020.
Insurers that were primarily exposed to mortality experienced more volatile earnings and more bad news where the provisions turned out to be insufficient.
Bancassurers have mortality exposure, but also significant retrenchment risk through their large credit life books. The under-provision on mortality was offset by overprovision on retrenchment (more or less depending on the product portfolio and assumptions used in setting the provisions). This resulted in a more modest impact on earnings during 2021.
Many insurers have expressed a growing concern that COVID-19 may become endemic, resulting in permanently higher mortality, amplifying comorbidities and denting longevity improvements.
When setting demographic assumptions, insurers need to decide what proportion of the higher mortality, morbidity and retrenchment claims may persist for years, or even indefinitely.
The pandemic has highlighted that the retrenchment capital requirement in the standard formula is likely inadequate. Most materially affected insurers are now allowing for more severe retrenchment stress in their ORSA scenarios.
The Actuarial Society of South Africa has been exploring whether to recommend a different structure or parameterisation of the standard formula, even if only for insurers’ internal use.
A related consideration is that of the mortality catastrophe calibration, which may also need to be updated.
Reinsurers have felt the impact of COVID-19 particularly heavily, with typically lower profit margins and product portfolios without the benefit of diversification from investment products and others with lower mortality and morbidity exposure.
Some reinsurers experienced a modest offset through reduced payments of profit commission to their cedants.
On the other hand, for the fair number of insurers that did not correctly allow for profit commission in their solvency calculations, the benefit of reinsurance was lower than modelled and expected.
Some direct writers are considering the loss-amplifying impact of profit commission with new unease, at the same time as reinsurers have been reminded of the loss-absorbing benefit for themselves.
Many insurers paused dividends, initially out of an abundance of caution. With hindsight, insurers generally found this to be a good decision, if not because experience turned out worse than expected, then because of the additional flexibility it gave to management in terms of product, reinsurance and capital management decisions. Some insurers have since started declaring dividends, while others are carefully monitoring their solvency positions with thoughts of declaring dividends still in the future.
The many moving parts as described above highlight the need for insurers to improve their financial projections and scenario analysis capabilities to be able to identify potential solvency concerns as experience continues to unfold.
COVID-19 will have a lasting impact on individuals, countries and insurance markets. Management teams and boards of insurers have had to make decisions without the benefit of precedent or the luxury of time.
For some, this was the opportunity to reap the rewards of prior investment in risk management processes and data analytics.
For others, it highlighted shortcomings in several areas.
The insurers that will succeed from now into the next crisis are those that have learnt from this experience and are investing for the future across underwriting, customer expectations and product design, digital distribution and market conduct.
If you would like to discuss any of the above, or anything else, with us, then please contact us. Milliman can provide a range of services including: