CAT modeling: Which is the best approach for your project?

Learn how trained risk engineers can utilize CAT modeling to help you better understand your organization’s risks and inform your mitigation strategies.

wind turbine sky

The acceleration of the green energy transition is heightening the demand for reliable renewable energy sources. These projects typically require a large amount of space to function properly. Due to the cost of land and its dwindling availability, these projects are often built in locations prone to natural catastrophes. This exposure increases their risk for major losses, which can translate to higher insurance requirements and premiums.

Losses caused by natural disasters continue to rise each year and the current inflationary environment is pushing up replacement costs. Many insurers are reducing their capacity for insuring renewable energy projects located in natural catastrophe zones and becoming more disciplined in their underwriting approach. The potential for damage may lead insurers to make overly conservative risk estimates, which could result in renewable energy developers, owners, and operators purchasing unnecessary and cost prohibitive insurance coverage.

To help prevent this, renewable energy developers, owners, and operators need to have a good understanding of their risks as well as mitigation strategies. Sophisticated risk engineering can help potential insureds secure more optimal coverage for their project, with better limits and scope of cover, at a more cost effective price.

Two assessment methods

The industry typically uses one of two catastrophic risk assessment methods to estimate potential losses due to a natural disaster:

Deterministic assessments
  • Carried out on a per location, per event basis. 
  • Risk engineers determine likely catastrophic events in the given area and measure the consequences.
  • Requires site visits to help risk engineers understand a location’s exposures, allowing them to consider location-specific risks and mitigation measures.
Probabilistic assessments
  • Consider insured’s entire portfolio.
  • Insurers use a computerized process known as CAT modeling to simulate potential catastrophic events that might occur at all of an insured’s locations over a period of time, typically 100 or 500 years.
  • Use historical data, expert knowledge, loss history, and theory to determine both the probability of natural catastrophes during the review period and their potential consequences.
How do these methods compare?

Let’s take flood risk as an example. If employing a deterministic approach, a risk engineer first ascertains a specific property’s 500-year flood exposure, typically by referencing the Flood Insurance Rate Maps provided by the Federal Emergency Management Agency (FEMA). The risk engineer would combine this information with specifics of the client’s premises, such as its elevations, different types of construction, and occupancy. Then, the engineer would use existing data tables comparing depth and speed of water to construction type to estimate the potential damage if a 500-year flood were to hit that specific location.

Conversely, an insurer using a probabilistic approach would plug all of the client’s locations into software models with data on a variety of inputs, including hydrology, terrain, soil type, vegetation, floodplains, and damage. These damage tables incorporate data from various sources, such as loss history and testing, and help establish damage assumptions. The model would then generate thousands of possible flood scenarios and compare depth and speed of water to construction type and occupancy in conjunction with damage table data in order to estimate the potential losses to the client’s whole portfolio. Since the probabilistic approach considers many incidents that are probable within the period under review and their likelihood, it can provide a more accurate measurement of losses over time.

Neither approach provides a complete picture

Deterministic assessments can only consider the probability of a specific catastrophic event happening to a single location and not the probability of all events of any size accumulatively occurring within 100 or 500 years. Additionally, it is typically cost prohibitive to carry out risk assessments on each individual property, making it extremely difficult to determine an insured’s probable maximum loss (PML) across their entire portfolio. Deterministic analysis is informed by a risk engineer’s assumptions and judgement, which can lead to substantive variation in estimated losses. Risk financing based upon those assessments can differ, potentially influencing which engineering firm owners or insurers select.

Probabilistic assessments also have shortcomings. Although more consistent in estimating losses than an engineer’s assessment, without having risk engineers on site, it is hard to know if the models are using the correct locations. A frequent challenge is that models do not recognize the addresses of client sites, potentially leading to inaccurate data or missing data. Probabilistic models were designed for buildings and tend to be less accurate for dispersed assets, such as renewable energy plants that can span across multiple acres. For example, if using a single coordinate for a solar farm to predict tornado risk, the model would assume that all the solar arrays on the farm were hit equally. It’s more realistic to assume that there will be some tangential hits. Locations with several installations and/or buildings dispersed across a wide area may benefit from a method called the campus approach, where the different assets on the site are plugged into the model as separate coordinates, such as each inverter on a solar farm, typically leading to more accurate results.

Advantages of a hybrid approach

A hybrid approach combines the site-specific information obtained from deterministic assessments and client conversations with the loss history and data provided by CAT models. Model accuracy is dependent on the quality of the data plugged into it. Poor data quality may cause the model to make assumptions that are generally conservative, potentially leading to higher insurance costs. Specially trained risk engineers can help improve data quality by factoring into the models multiple considerations, such as redundancies, better construction types, and more resilient foundations and basements. These risk engineers can also consider operational elements in their assessments — for example, the likelihood of equipment failure — that are not typically included in CAT modeling. Note that while natural catastrophes are the top peril for renewable energy companies, operational risks remain a major challenge; considering both helps provide a more complete picture of a property’s overall risk.

Risk engineers trained in using a hybrid approach can pre-screen a client’s properties and predict how the models will represent their risk. Because they understand the way the models work, they can ask clients for relevant information, leading to more accurate assessments without visiting the client’s sites, saving the client time and money. Their experience allows them to determine the accuracy of results provided by the CAT models and how to better apply the models if they do not. This is especially important since specific models are not available for all installation types, requiring the use of proxies.

Early risk engineer involvement is key

Understanding the natural catastrophe exposures of sites shortlisted for a renewable energy project can help owners and developers make better informed decisions. Experienced risk engineers can screen sites to compare each location’s natural catastrophe exposures and give the client an indication of whether a particular site may be more expensive to insure than another. This information can factor into project owner and developer decisions on which sites to buy and develop, as well as make better design and risk measurement decisions.

Renewable energy companies with projects spread across various regions require sophisticated risk mitigation investment to offset risks, prevent losses, and avoid high insurance claims. It is critical to work with risk advisors with experience in the field, who understand how the CAT models are used and are specialists in the technology used to create them in order to best understand your risks and make data-driven decisions.

For more information about how CAT modeling can help you better understand your organization’s risks and inform your mitigation strategies, contact your Marsh representative.

Related insights