Pre-COVID-19, the consensus was that pandemic is a shock-loss scenario that the life sector needed to worry about more than the non-life sector. Is it fair to say the reverse turned out to be true?
Life insurers do anticipate an event like this in terms of regulatory capital and risk management. So COVID-19 isn’t a extreme mortality event. The additional mortality on a life portfolio is 5-10% more than normal at worst.
But there is an impact through the financial markets and interest rates being further depressed. It’s put more pressure on the balance sheet of life insurers. COVID-19 is more of an economic event than a mortality event for life insurers in most countries.
Also, life companies usually write longevity business, and when a lot of older people die prematurely it releases reserves. Well diversified insurers and reinsurers have been writing longevity for 10-15 years as a diversifier against mortality. So there are releases on the longevity side and not too serious losses on the mortality side.
The wake-up call for insurers is that you need to be diversified. It is a market risk event and you need to compensate for that by having enough longevity, mortality and disability risk in the mix.
Do life reinsurers need to review their approach, in light of their experience and that of their cedants?
The big life reinsurers are composites and they had big losses on the non-life side. However, counter-intuitively, their outsize non-life losses are making their boards more conservative on the life side of the business.
Life reinsurers have actually been good at modelling pandemics and understanding their exposure. They understand COVID-19 versus any other virus; they’ve run thousands of scenarios and can articulate them well. But they only use their analytics to run their own book. They don’t really deploy that understanding to support their clients.
At Aon we have a book of pandemic cover with around $1bn of exposure, but reinsurers don’t want to write it; they only want to write the full quota share on that portfolio. They’re happy to take ‘at-the-money’ mortality risk but they’re not happy to offer the ‘tail’ pandemic protection.
Conversely, insurers are saying that they are happy to take the mortality risk, but they do want to protect themselves against a pandemic. This represents a disconnect in the market whereby reinsurers have a good understanding of pandemic risk but are unwilling to provide capacity that’s wanted by insurers.
Reinsurers have been content to provide capacity for what they want to write – but it isn’t what the client needs. Leaving an unsatisfied demand like this creates an opportunity for new players to come in. If reinsurers covered life insurers’ pandemic risk they would have a new source of profit – and deliver more value to their clients.
Did the models used by the industry stand up?
The models have stood up well so far but they are rapidly becoming insufficient. Prior to this year the industry has used stochastic models, and that worked with COVID-19: it’s a 1-in-10 or 1-in-20 year loss. But the picture changes as the pandemic progresses to the next stages, and the future development is unknown. There is no precedent for knowing whether a vaccine will work, for example.
It means that we all now have to take a scenario-based approach and make some assumptions around say, the four best and worst case outcomes and decide which you want to protect against. In other words, the stochastic modelling has worked well to a point, but it won’t work for the next two years.
On the P&C side, pandemic wasn’t a concern before and now it is. But sickness and deaths are not relevant for non-life portfolios. Models will have to be developed that extend from pandemic per se to economic impact, bringing in detail about industrial activity, public socialising, school attendance and seasonal factors.
COVID-19 will stimulate a lot of discussion about modelling styles and which model types are most appropriate in different circumstances and how users view uncertainty. For example, how do you express the uncertainty around the effectiveness of a vaccine?
How will COVID-19 affect the life ILS sector?
Models will be under the spotlight and the assumptions that need to be made. It will be hard to convince investors that a given model is “correct”, where there are several different scenarios. Basis risk is another issue. A parametric risk trigger, possibly based on a WHO declaration, is problematic for Solvency II-based regimes in terms of the basis risk that exists.
There is a way ahead: an industry consensus is needed on establishing standard scenarios. In South Africa amid the AIDS epidemic the market established 10 different scenarios on how HIV could play-out in the population. A similar approach could be taken with a COVID-type pandemic, whereby a set of base tables establish five or six valid scenarios that the whole industry understands.
Instead of talking variously about R numbers, infection rates or excess mortality, the industry needs a common terminology to use when looking ahead. Such a taxonomy will smooth the way for insurance-linked securities structures to be used effectively in pandemic risk transfer.