Inflation in Capital Modelling – what does the crystal ball say?

It seems that every time an inflation forecast is released by a financial institution, the short-term estimate climbs higher and higher. 

Insurers have largely had the luxury of not having to worry about inflation for a while and consequently may lack the experience and tools to deal with the levels of inflation that we are now experiencing. This makes it somewhat tricky to work out how best to deal with rising inflation, particularly within the capital modelling space. We lay out below some considerations that we would expect to see this year.

Be clear about what we mean

The definition of inflation has become more nuanced in the last few years. In the pre-Covid years, inflation was split between economic inflation and social inflation. At the start of the year, the term "excess inflation" entered common lexicon, which encapsulates social inflation and more. Precisely how these terms are defined is quite important when it comes to working out how we untangle the different causes of inflation – and not all (re)insurers or regulators have a common definition.

For example, do we consider the inflation arising from political upheaval to be part of economic inflation or excess inflation? Or do we want to define economic inflation to be consistent with a published index such as the Consumer Price Index (CPI), noting the volatility in what the CPI forecasts have been recently? How we define inflation becomes key to how we will attempt to measure it.

Mean vs volatility

Understanding whether sufficient inflation is captured within the mean estimates is important, as volatility measures may well leverage off the mean inflation volatility, impacting the amount of capital held. Whilst the temptation is therefore to say that volatility is already higher due to higher mean inflation, the reality is that capital modellers should also consider if the volatility factors themselves need to be even higher given how uncertain future inflation is.

To that end, it is important for capital modellers to assess if the output presented in Economic Scenario Generators (ESGs) are sufficient to capture the forward-looking view of inflation volatility. Scenario testing becomes important here, with extra effort potentially required to help subject matter experts expand their horizons, given very few of us would have seen inflation like this in our careers. 

Within the Lloyd’s market, the consensus is forming that the unadjusted outputs of most ESG models would be insufficient, given the lag in ESG calibrations, whilst the situation evolves daily. Where an ESG is not used, careful consideration must be given to how much inflation (both mean and volatility) is implicit within existing assumptions and whether it should be increased to ensure that an adequate level of capital is maintained.

Test, test, test

Capital modellers should revisit their tests suite to ensure that they have a sufficient number of tests to help understand the impact that inflation uncertainty can bring. We would generally expect more scenario and sensitivity testing to be employed this year to help management and Boards understand how bad things could get. 

Separately, it is important to also work out if any additional loadings for inflation would be double counted within capital models. This year, more so than in previous years, attempts to quantify the levels of inflation (and the split between economic and excess inflation) and to estimate the duration of this high inflation period have become important. This would serve as another data point to help management understand if any proposed loadings are sufficient.

Also, bear in mind that the past is no longer a good predictor of the short-term future. Therefore, capital modellers and validators should be more critical when performing backwards looking tests, such as back-testing, as it could provide false comfort.

For most insurers, there is insufficient time to develop an explicit inflation model and so working out how one can use the existing modelling and parameter framework to bake in future inflation would be more practical. That should be coupled with additional testing to prove how inflation within the model has changed and why. There will also likely be a number of different views of the results needed to justify and/or validate the internal model outputs and so the ability to visualise these outputs may help management and the Board absorb the information and draw conclusions more quickly.

Do not jump to conclusions

Just because inflation levels are expected to rise in the short term does not automatically mean that capital will increase. Inflation will have different impacts on different risk profiles. For longer tailed lines, the current environment may be transitory and so a more measured response may be appropriate. Also, as much of economic inflation is driven by energy price increases, the level of inflation for lines of business not associated with energy may well be different, as an example. Inflation may also be curtailed by other management action such as rate increases or policy limits as well as differentiated by geography.

Capital modellers should try and understand all the different tools and levers that management are using to combat inflation and adjust the modelling methodology to reflect that reality. We have also seen where some insurers have relied upon their Events Not In Data (ENID) processes to provide some mitigation when it comes to inflation, although these arguments must also be considered in relation to the evolving economic environment and insurers’ risk profile.

Contact the general insurance team to discuss how they can help you.