Insurers look to AI as FEMA rates rise

Aerial imagery provides one of the best options for insurance companies to stay profitable in high risk markets like flood.

Insured losses from floods doubled to $83 million during 2011-2020 as compared to the previous decade and global flood losses reached $20 billion in 2021 alone, according to Swiss Re. (Photo: bilanol/Adobe Stock)

Billions of people worldwide face flood risk, and this number is steadily increasing with the effects of climate change. Consider that more than a third of Pakistan was underwater this summer as a result of intense rainfall. Although that extent of flooding has not yet been experienced in the United States, it certainly indicates a concerning trend. With global warming resulting in severe drought in some places and heavy rainstorms in others, U.S. infrastructure is having to withstand higher volumes of water in a shorter period of time. Drought can also accentuate the effects of flooding through reducing the permeability of ground, causing water to remain on the surface. With flood defenses overwhelmed, many U.S. states are seeing a rise in flooding.

Other manmade trends such as increased urbanization are also having a concerning impact on the rate of flooding, with vast amounts of concrete being used for roads, parking lots and developments. This has led to a rapid reduction in the total permeable surface area, putting drainage infrastructure under significant strain.

The recent increase in the scale of flooding presents significant challenges for insurance providers. Unfortunately, with increased urbanization in flood zones, it is becoming increasingly difficult to predict the extent to which infrastructure will be able to handle runoff water. Consequently, insurers are put in a position whereby they do not have enough information to determine whether to issue policies. Meanwhile, insurance businesses that do choose to insure risk can experience unsustainable losses. Consider that insured losses from floods doubled to $83 million during 2011-2020 as compared to the previous decade and global flood losses reached $20 billion in 2021 alone, according to Swiss Re.

According the U.S. Federal Emergency Management Agency (FEMA), just one inch of floodwater can cause up to $25,000 in damage, and the cost is rarely covered by homeowners’ insurance. To avoid serious and largely irreversible financial consequences, it has become critical for homeowners in high risk areas to obtain specialized flood insurance.

The flood insurance market

You might presume that these circumstances mean the number of flood insurance policies issued would be on the rise, but this is not the case. Unfortunately for the millions living in potential flood zones, the cost of insurance premiums has been increasing in recent years, forcing some homeowners to drop policies due to the financial burden. This is even true for lower cost alternatives such as those offered through FEMA’s National Flood Insurance Program. Over the last year, however, there has been a decline in FEMA policies, from 4.96 million in September 2021 to 4.54 million in June 2022. This can be attributed to rising insurance rates.

As people have been opting out of FEMA flood coverage, insurance companies have been presented with a new opportunity to target a gap in the market. However, as the rising FEMA rates were a result of increasing risk, and given that flooding already presents a significant challenge for assessing costs, many insurance companies have been avoiding flood insurance entirely, citing high risk.

Nevertheless, there are still ways that risk can be calculated at a relatively low cost, and large numbers of properties can be simultaneously screened for indicators that suggest a particularly high risk. This could include a variety of factors such as building construction or the permeability of the surrounding ground and the presence of nearby bodies of water.

AI and 3D property data

Image provided by GeoX

In the last decade, artificial Intelligence has emerged as the game-changing technology of the insurance sector. AI models are becoming more effective in the analyzing and processing of insurance claims, especially in the case of natural catastrophes. Using 3D modelling of high resolution aerial imagery (pictured at right) can provide rich and accurate data at scale. Visual data can accelerate the process of inspection, underwriting, and claims, whilst simultaneously helping with risk prediction and preparedness. In some cases, due to the speed of automated analysis it may be possible to identify risks and pre-emptively warn property owners so that damage can be avoided entirely.

Many insurers are using aerial imagery to support the underwriting process, especially when determining the condition of the roof or scanning for any potentially hazardous debris nearby or measuring the elevation of the ground around the property. In this way, they become more efficient, trim costs, and are more able to offer affordable flood premiums. Finding individual images of a specific property and analyzing them in person takes time, and can limit the number of properties that can accurately be insured with up-to-date information. Therefore, implementing AI recognition with aerial imagery can significantly reduce the time required to create insurance products and vastly increase the number of properties that can be surveyed. With computer vision and machine learning, insurers can identify risk factors in all the images across a database and automatically categorize properties by their risk level. This is particularly useful given the high number of properties that are insured and reinsured for flood insurance.

With AI taking the industry by storm, new aerial imaging capabilities are constantly being developed in the P&C space, meaning predicting and preventing loss is more accessible than ever. By using 3D property intelligence derived from aerial imagery, insurers can obtain more data points for more commercial and residential properties nationwide – for example by pinpointing impermeable urban areas in flood risk zones. Considering the complexity of the underwriting process, any company that still relies solely on governmental property records instead of using AI-derived property databases for underwriting is significantly increasing the risk of using outdated information and making incorrect risk management decisions. From a cost saving and a risk averse perspective, using AI-derived databases of aerial imagery is one of the best options for insurance companies to stay profitable in high risk markets like flood.

With such increasing environmental pressures, insurers need to better protect their customers, improve business performance and differentiate by adding value beyond product. With the industry experiencing a rapid digital transformation, innovation is now cherished by insurers. Climate change may be outside the industry’s control, but failure to respond to climate challenges and subsequent catastrophes like flooding is partly a result of the current insurance model. And the solution is there: In a world where digital is king, pinpointing and predicting the probability of complex risks before they happen can reduce the number of causalities and strengthen the business performance of Insurers.

Guy Attar (Info@geox-group.com) is co-founder and executive chairman of GeoX.

See also: