Reinsurance
Product Family
Claims
Services
Services
Services
Risk Consulting
Resources & Tools
Resources and Tools
Get In Touch

The upcoming 2019 Florida renewals may be an unprecedented bellwether for how reinsurance pricing in the sunshine state will respond post event. A lack of catastrophic activity (and data points) over the last decade has resulted in reinsurers and insurers taking an overly optimistic view in regards to the expected loss from a hurricane in their rate setting. Inaccurate Loss Adjustment Expenses (“LAE”) assumptions, understated litigation costs, unanticipated trends in  Assignment of Benefits (“AOB”) and unverified modeling data are just some of the examples where the (re)insurance industry fell short. Hurricanes Irma and Michael exposed weaknesses in the industry’s approach and thinking when it comes to view of risk.

It is apparent that the rate levels seen over the last few years are a good indicator that the information learned from Hurricanes Irma and Michael was not reflected in the industry’s view of risk. In addition, differentiation between companies has generally not been applied in rating and capacity deployment by the reinsurance market. AXA XL believes that we are at an inflection point where the Florida pricing dynamics will better reflect the risk being underwritten.

Due to the extensive amount of information available over the last two years from Hurricanes Irma and Michael, the (re)insurance industry is now better informed and should be able to make improved underwriting decisions.  It is our view that certain performance metrics are vital. For example, we now have a better representation of LAE, improved understanding of severity costs, and the tools to better scrutinize modeling data.  This insight should result in a more accurate view of risk in the Florida insurance marketplace. This article discusses various performance metrics and their usefulness in analyzing Florida insurance companies.

LAE Assumptions

At AXA XL we have been aware of the impact of the LAE slippage for many years. LAE slippage is caused by the 5% cap on LAE ceded to the Florida Hurricane Cat Fund (“FHCF”). For example, the FHCF only starts to pay when actual indemnity breaches the attachment point. We have observed that the average LAE in Hurricane Irma is 20%. Hence, there is a chance that the entire 20% LAE can be assumed by the layer that sits alongside the FHCF. Even once the total indemnity breaches the FHCF attachment there is a gap of 15% between what the FHCF pays and for what a reinsurer is liable. We account for this issue in our view of risk as well as a company’s actual LAE performance. Analyzing historical loss information and qualitative details assist us in applying the appropriate load on a company specific basis. It is important to note that the LAE experienced in Hurricane Irma was higher than many insurance companies anticipated.

Hurricane Irma was an eye-opening event as many catastrophe management plans and claims processes did not function as expected. One of the primary reasons for this matter was the lack of adjuster resources. Hurricane Harvey sapped a lot of the experienced adjusters from Florida. Florida Citizens compounded this issue by paying above market rate to ensure that their adjusting needs were met. This left many Florida insurers scrambling to fill voids left from unfulfilled adjuster resource contracts. The vast footprint of Hurricane Irma was another feature that added to the cost of LAE.

Many Florida counties were impacted with moderate wind speeds which led to a high proportion of claims being closed below the deductible. In other words, the damage caused from Hurricane Irma’s wind speeds was not severe in many counties that were not in the direct path of the storm. The damage in these residual counties led to a large number of insureds filing a claim and insurance companies utilizing adjuster resources - only to find out that the damage did not breach the deductible. As an example, these claims would still have incurred a similar adjustment fee as a claim which paid $10,000 to the insured. As a result, the LAE as a percentage of indemnity ended up being higher than the industry expected. In contrast, Hurricane Michael’s LAE was much lower due to the fact that this event was very powerful and concentrated. Additionally there was not a lack of adjuster resources for this event. There were many total losses in Hurricane Michael and the average indemnity was much higher than what was observed in Hurricane Irma. The divergence between LAE and indemnity has also been noticed with insurance companies who are vigilant in their claims payments and are concentrated on keeping their indemnity low.

We evaluated company performance in the most recent events by combining Indemnity and LAE on a dollar basis. We observed that some of the best performers had the highest LAE as a proportion of indemnity. This highlights that “not all LAE is created equal”. LAE can vary significantly by event size, severity and client. In addition to LAE, the propensity of litigation in Florida should be reflected in a company’s view of risk.

Litigation Trends

The Florida Department of Financial Services (“FDFS”) is a good resource in assessing various companies’ litigation trends. The chart below is a sample of Florida companies based on information from the FDFS which shows litigation rates per 1,000 policies. It is also insightful to assess litigation rates per claims filed. At AXA XL we cross-reference these stats with the type of business that an insurer writes. In addition, we assess how exposed that company is to the Tri-county region of Florida. Normally there is correlation between the extent of coverage in the Tri-county area with a company’s litigation rates.

Since catastrophe models do not capture litigation trends in their methodology, it is important that the industry addresses this deficiency in their view of risk. One method would be to apply an adjustment to a company’s exposure in the Tri-county area or to apply an adjustment on those modeled events that impact the most litigious parts of the state.  An even more granular method would be to formulate a modeled severity curve that is consistent with what has been observed. These same approaches can be used to accurately reflect Florida’s struggles with AOB.

AOB Issue

According to the President and CEO of Florida Citizens, AOB claims in Florida increased from 5,400 in 2013 to 21,000 in 2018. Other sources have indicated that claims with an AOB attached are more than double the cost of the average claim. In addition, there are cases where AOB costs make up 20% of the company’s overall Hurricane Irma total loss. It is very clear that legislation needs to be changed in order to address this issue that seems to be spreading throughout the state. While we are hopeful that change will come, it is hard to predict with a degree of certainty.

The Florida Senate Bill CS/SB 122 aims to address AOB by curtailing a third party’s ability to benefit from one-way attorney fees.  Florida’s one-way provision mandates that insurance companies cover the plaintiff’s legal fees if the court judgement is $1 over the insurer’s initial offer. Under Senate Bill CS/SB 122, only the first party in an insurance contract would be able to benefit from Florida’s one-way provision. This bill passed the Florida Senate’s judiciary committee in March 2019.  Florida Senator Douglas Broxon sponsored the bill and recently stated the following: “Abuse of the AOB process is causing increased property insurance rates for homeowners across the state. If you own a home, you’re paying a higher premium because of bad actors.” In order for this proposal to become reality, it has to pass Florida’s Senate and be reconciled with the House of Representatives companion bill which was recently passed. There is no denying that the aforementioned proposed law has momentum so we will continue to monitor this space.

Hurricane Irma Loss Development

Another issue that reinsurers have been monitoring is the significant loss development from Hurricane Irma. A lot of the development that we have seen can be traced to both unanticipated litigation and AOB cost. In addition, Hurricane Irma development or the lack thereof can be an indicator of an insurance company’s control over their claims process. It is anticipated that companies with significant development will face greater scrutiny by reinsurance underwriters at renewal. We have noticed major discrepancies with insurance companies’ initial loss pick when they only relied on a modeled estimate. Ideally, insurance companies should have other tools to accurately assess their loss from a major event. The following graph highlights the extent of Hurricane Irma development from the second quarter of 2018 to the present for various Florida insurance companies. An additional factor that is contributing to Hurricane Irma development is the issue of tile roofs.

 

Florida Building Codes

The Florida Building Code – Section 708.1.1 states the following: “If more than 25 percent of a roof or section of a roof is repaired, replaced, or recovered then the entire roofing system or roof section must be brought up to code.” Section 502.3 states that “Work on non-damaged components that is necessary for the required repair of damaged components shall be considered part of the repair.” Translation - insurance companies are being forced to replace an entire roof to like kind and quality if they are unable to find a matching tile. This provision can apply even if only a small percentage of the roof is damaged. This provision is unsurprisingly inflating the cost of tile roof claims. We have evaluated how our Florida clients code roof covering and worryingly 40% do not have anything coded as tile roofs. The default coding tends to be set to unknown which has a lower damageability factor in RMS relative to tile roofs. This issue highlights the need to monitor and appropriately address data quality.

Data Quality

How can we better ensure that the data provided by companies is an accurate reflection of the risk being underwritten? Cape Analytics is one tool which uses visual imagery and computer learning to assess the validity of a building’s features. Cape Analytics also has the ability to evaluate an entire portfolio, highlight discrepancies, and update data coding to actuality. We have observed specific instances where inaccurate data coding understates a company’s Average Annual Loss (“AAL”) by upwards of 10%. We have also observed trends in how data is coded amongst various broking houses. As they gain wider acceptance, we believe that Cape Analytics and tools like it will bring tangible value to the overall industry. A more traditional but effective way to evaluate a company’s data is RMS’ Data Quality Tool (“DQT”).

The DQT tool can measures the impact of primary (occupancy, construction, etc.) and secondary modifiers (roof age, square footage, etc.) on an insurance company’s ground up loss. When secondary modifiers are shown to decrease the PMLs by a significant amount, we would look into these results carefully to ensure the credit given is warranted.  As a start, we may set the secondary modifiers to unknown and evaluate those results in our internal pricing model. An indicator that an insurance company’s data may not be accurate is if the Florida only portion of their Hurricane Irma loss sits at a high return period on their Florida only curve.

Hurricane Irma Return Period

Hurricane Irma sits as a one in 10 year event on RMS’ Florida residential curve. The worrying discrepancy lies when an insurance company’s actual Hurricane Irma l loss sits at a one in 20 or 30 year return period. We accept that geographic concentration plays a role but data quality is also a factor. The graph below highlights where Hurricane Irma sits as a return period for various Florida insurance companies.  The trend line shows where Hurricane Irma sits as a return period once the secondary modifiers are set to unknown. This is one indicator of how an insurance company may perform prior to an event. In order to evaluate an insurance company’s performance post event, a market share analysis can be performed.

 

 

Hurricane Irma Performance

Over the last few months we have been amassing data from our portfolio and calculating the average damage ratio by county from Hurricane Irma. We multiply this figure with the actual Total Insured Value by county for the given insurance company. We then sum up the values from each county to come up with an expected market share loss. We compare this figure with their actual loss which indicates how well the insurance company performed in Hurricane Irma relative to their peers. The implied market share loss is based on a composite that provided data. The overall average performance includes clients that were not included in the damage factor calculation.

Conclusion

The (re)insurance industry is now better informed to make improved underwriting decisions due to the plethora of information provided  from the significant events over the last two years.  A better representation of LAE, improved representation of severity costs, and increased scrutiny of modeling data should quell the previous optimistic view of hurricane risk in Florida. The industry will never price for these factors with 100% precision. With that said, it is important to account for these issues as best as possible in order to address the inherent uncertainty that comes with the Florida insurance market.

At AXA XL, we have relayed relative performance metrics from Hurricane Irma to our partners in the industry. Our Florida clients have valued the findings we provide as our interests are aligned in understanding and adequately addressing the range of issues in the Florida marketplace. Ultimately we aim to protect individual policyholders and move from just being a payer to partner with our clients.

  • About The Author
  • Assistant Underwriter, U.S. Property Reinsurance Team, AXA XL
Invalid First Name
Invalid Last Name
Invalid email
Invalid Captcha
 
Subscribe
More Articles