Aftershock: Can ‘model change’ impact the real economy in New Zealand?

Alexander Pui discusses the interaction of catastrophe modelling, under-insurance, credit ratings, and interest rates. He warns of the potential spiralling of under-insurance that could be triggered by a higher evaluation of earthquake risk in New Zealand.

Third-party catastrophe risk models are powerful tools used by many insurers and reinsurers to manage exposures of insurers to catastrophes.

Rating agencies play an important role in providing an independent view of credit risk in financial markets. Changes in credit ratings move market prices for rated investments – could changes in third-party catastrophe models move insurance prices?

Last year, the catastrophe modelling firm RMS announced the impending release of its updated earthquake (EQ) model for New Zealand (RMS Model). Although to date model updates have not resulted in market disruption, the New Zealand (NZ) experience may differ given the significant exposure to EQ risk in NZ, the high insurance penetration and the socialised EQ insurance structure through the NZ Earthquake Commission (EQC).

The RMS Model may provide insurers and the wider NZ population a greater appreciation of the EQ risk facing NZ, and below I discuss some possible impacts of this changing view on the level of insurance risk.

 A Major Overhaul: What prompted the model update?

Catastrophe risk models are regularly refined through the validation of assumptions and outputs when new science emerges or new claims data becomes available.

A string of earthquakes from September 2010 to April 2011 in and around Christchurch (known as the Canterbury Earthquake Sequence (CES)) caused a devastating loss of life and property, with 185 casualties and insured loss estimated at NZD 33 – 38 billion to date. The CES revealed important shortcomings in existing EQ models, including previously unknown faults, progressive weakening of structures not captured by damage functions due to earthquakes occurring in quick succession as well as highly uncertain cost apportionment due to under-insurance.

The scale and complexity of the CES provided the modelling community with a plethora of new geological, building vulnerability and claims information. Preliminary indications suggest that the new RMS model will feature significant changes[1]; including increases at both short and long return periods. Naturally, the extent of changes will also largely depend on the geographical distribution of individual insurer portfolios, with implications for regulatory capital, reinsurance and pricing.

Regulatory Capital:  Is a 1 in 1,000 year catastrophe risk charge adequate?

In the aftermath of the CES, and following the failure of a number of insurers, the Reserve Bank of New Zealand (RBNZ) raised capital requirements for general insurers. The Catastrophe Risk Capital Charge progressively increased from 1 in 250 years to 1 in 1,000 years after 2016.  While it is understandable for RBNZ to adopt a conservative position following CES (ostensibly to account for model miss), the lack of guidance on how to treat uncertainty and interpret catastrophe modelling output may create a new set of model governance issues, especially since it is far more difficult to predict the size of a 1 in 1,000 year loss than a 1 in 250 year loss.

Without explicit guidance on how to arrive at a nuanced technical view of risk[2], the inherently challenging task of modelling catastrophe risk is laden with even greater subjectivity. For example, if the benchmark set by the RBNZ is higher than their internal view of risk, firms may elect to bias their estimates towards a catastrophe model that produces lower loss estimates  or blindly ‘blend’ models in an attempt to obtain a ‘holistic’ view of risk. This introduces a degree of moral hazard into a process that should instead be focused on the most accurate assessment of risk.

Affordability and Under-Insurance: A vicious economic feedback loop?

The 'double whammy' of model change and increased regulatory burden will increase reinsurance requirements for insurers and these additional reinsurance costs will likely be passed on to personal customers. While EQ insurance in NZ ought to be treated as an inelastic product, customers may no longer be able to afford further price hikes given that current premiums are already very high due to the CES. The subsequent under-insurance or protection gap will likely be filled by the NZ government (over and above existing EQC arrangements) as the insurer of last resort. 

In the absence of sufficient private insurance or public catastrophe pools, costs from major disasters have regularly been shouldered by tax payers as insurers of last resort.  For example, following the 2011 Brisbane Floods a flood levy was imposed on higher income earning Australians. Owing to the small population and size of NZ's economy, the internal transfer of the increased cost of the EQ risk may be an excessive burden on citizens.

If the NZ government does assume significantly greater insurance risk due to the higher predicted cost of EQs and resulting increases in under-insurance, and if this risk burden is significant enough to warrant a downgrade in credit ratings, then this could result in a vicious economic feedback loop (see schematic below).  Once rating agencies become aware of the heightened financial risk to the NZ government, they may downgrade the NZ government debt. This credit risk will be somewhat unique, as it is not explicitly tied to economic fundamentals or political climate, but extreme natural peril risk.  The resulting poorer credit ratings will attract higher interest rates, giving rise to financial stress for consumers with high levels of household debt.  With a greater proportion of income required to service debt, insurance may become yet more unaffordable, and under-insurance will shift even more risk to the Government.


Schematic: Potential vicious economic feedback loop perpetuated by model change and regulatory requirements

This article has explored some far reaching potential implications from a seemingly routine catastrophe model update on the NZ general insurance sector and its wider economy. The market reaction in NZ may be unprecedented due to a unique confluence of factors particularly insurance affordability issues and the large proportion of insurance risk relative to the small size of its overall economy. As the industry is already under considerable duress following the CES and recent Kaikoura earthquake, the model update could trigger a string of economic repercussions unless there is a receptive framework in place to accommodate this change.

Estimating tail risk with any degree of certainty is challenging and the RBNZ should be commended on taking active measures to impose more stringent capital requirements. However, its decision should be supplemented by detailed guidance on catastrophe modelling best practice to avoid potential model misuse and reduce ambiguity. One option could involve setting up an independent committee of expert actuaries, engineers and seismic geologists to enhance the assessment of EQ tail risk in NZ which would provide the industry with greater transparency on the science underpinning such changes.

All opinions are solely of the author and do not necessarily reflect Swiss Re's position or views.

[1] See RMS – NZ EQ (public resource) for more information. (

[2] Although it should be noted that solvency standards have been set following consultation with industry and a Quantitative Impact Survey (Source: Insurer Solvency Standards – reducing risk in a risk business, Richard Dean, RBNZ Bulletin, Vol.74, No.4, Dec 2011)

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

About the author

Alex Pui

Alex is currently a property underwriter at Swiss Re Corporate Solutions. His previous experience at Willis Re and IAG involved catastrophe risk modelling and research. Alex's industry interests include alternative risk transfer, understanding decision making under uncertainty and product innovation. He holds a Bachelor of Law (LLB) and PhD in Applied Statistics (majoring in Climate Science) from UNSW.

Comment on the article (Be kind)


No Comments

Also this month