Changing modelers’ view of U.S. earthquake risk

fallen utility pole and road destroyed.

What do we know and what can we expect? Those questions are ones that modelers tackle daily.

Let’s start with what we know: The 2014 National Seismic Hazard Maps (NSHM) and the 2015 Uniform California Earthquake Rupture Forecast (UCERF3) describe the prevailing scientific view of seismic hazard in the United States. These studies were each the result of a massive amount of scientific research and debate that occurred over six years. That academic work formed the foundation of the catastrophe model updates that will be introduced by vendor modeling firms in 2017.

These updates will ultimately have a significant impact on the models’ view — and therefore the industry’s view — of risk to Property and Workers’ Compensation portfolios.

Implications of changes

Willis Re reviewed the results of these studies and published a whitepaper on the changes to USGS National Seismic Hazard Maps (NSHM) back in October 2015. Our goal was to draw preliminary conclusions on what they would mean for the insurance industry and what impact they could have on catastrophe model loss estimates.

Three key takeaways from our review include:

  1. The greatest magnitude changes in the modeled seismic risk can be expected in California, with significant but lesser-degree changes in the Pacific Northwest.
  2. Overall, key return period loss estimates for portfolios concentrated in Southern California are expected to decrease, and in some parts of Northern California to increase.
  3. Tail Value-at-Risk (TVAR) measures of loss estimates are expected to change for almost all regions.

Of course, the actual change specific to an insurance portfolio is highly dependent on where the exposure is located and the concentration of exposure within each region.

Vendor model changes differ from the USGS NSHM

The new version of USGS NSHM and UCERF3 are primary drivers of vendor U.S. earthquake catastrophe risk model updates.

However, the modeling companies cannot directly take the end-product of the NSHM and UCERF and simply implement it into their models. These NSHM and UCERF studies must first be translated into an event-based catastrophe model that the insurance industry uses for probabilistic loss estimation.

AIR and RMS are currently finalizing the upgrade of their U.S. earthquake models

Due to this, revised vendor probabilistic models are typically available to the industry 12 to 18 months after the release of the NSHM.

Changes to vendor models are broader in scope and include factors such as

  • site-specific soil amplification
  • basin effects
  • fire following
  • loss amplification
  • time dependency
  • refinements to damage functions

Specifically, modeling companies planned to incorporate lessons learned from the 2010 Tohoku, Japan, 2011 Christchurch, New Zealand and 2012 Maule, Chile earthquakes.

In addition, at the same time, vendor models refine and revise the impact of earthquake-induced perils such as landslide and liquefaction, tsunami and fire-following on property insurance portfolios.

AIR and RMS are currently finalizing the upgrade of their U.S. earthquake models. RMS’s updated earthquake model is expected to be released to the industry in April 2017 via RiskLink v17. AIR’s updated earthquake model is expected to be released to the industry in June 2017 via Touchstone v5.0.

The landscape of vendor catastrophe risk models is always changing. At the Exceedance 2017 conference, RMS informed us that model changes on portfolio loss estimates were likely to be decreasing.  Our catastrophe analytics team will also be attending Envision 2017 and will let you know what we learn there.

What these model updates mean for catastrophe risk managers

Model changes will be significant for many portfolios, and the patterns of change will be multifaceted. These changes are expected to impact underwriting guidelines, capital requirements and portfolio management strategies. In addition, these changes will affect the Workers’ Compensation and fire following earthquake (FFEQ) model loss estimates as well as shake loss estimates.

In almost all regions, changes to portfolio loss estimates will be highly influenced by the updates to earthquake fault source parameters, earthquake catalogs, earthquake rates and ground motion models (a.k.a. attenuation equations).

What does this mean? Three key applications catastrophe risk managers need to review where the vendor model results are used include:

  • Risk relativities between various regions and internal rating territories may change significantly (e.g., Southern California to Northern California)
  • Business rules that are based on the distance to a fault, such as underwriting guidelines or insurance rates, exposure aggregate thresholds will be moderately affected
  • Tail view of risk to change significantly

So, what do we know and what can we expect? We know that the changes we anticipated in the models have come to fruition and that, as always, there is more to be expected.

About Prasad Gunturi

Prasad Gunturi is Senior Vice President of Willis Re Analytics, where he leads the North American catastrophe model…
Categories: Natural Catastrophe, Reinsurance | Tags: , ,

Leave a Reply

Your email address will not be published. Required fields are marked *