Tag Archives: flood risk

A Way Forward on Flood Insurance?

In the mess that is flood insurance in the U.S., a bright spot emerged late last month when First Street Foundation released a major report on the issue, along with a model that will go a long way toward making assessment of flood risk more accurate and transparent.

The report serves first and foremost as a wake-up call. It says, for instance, that 70% more homes are within a “100-year” flood zone than are designated as such by the the Federal Emergency Management Agency (FEMA). That means 6 million households face flood risks they don’t anticipate, yet aren’t eligible for the National Flood Insurance Program. In Chicago, 13% of properties are at risk, according to First Street Foundation’s report, while FEMA puts that figure at less than 1%. The report says Washington, D.C., and Utah have five times the risk that FEMA sees, while Wyoming, Montana and Idaho have four times the risk.

Those sorts of figures are quite the clarion call, but First Street Foundation goes even further by providing the beginnings of a solution: data. Its model evaluates the risk for 142 million properties in the continental U.S., based on an exhaustive array of different inputs that not only are as accurate as possible for today but that project how risks will develop because of climate change. The model lets you search any address for free.

The model from First Street Foundation, a nonprofit research and technology group, should provide short-term benefits while laying the groundwork for smarter long-term policy decisions.

In the short run, potential buyers will understand their odds better and can either pass on a higher-risk property or can mitigate the risks by buying insurance or retrofitting the building. Banks will see the risks more clearly when writing mortgages — and some 30-year mortgages written today will still be in force in 2050, by which point the report projects at least 11% more properties will be at substantial risk of flooding. Insurers will price more accurately. Government — the 800-pound gorilla on flood policy — will have a better handle on what public works to undertake to protect vulnerable areas and what areas to steer clear of because the flood dangers are just too high.

(My entirely unrepresentative check on homes where I’ve lived over the decades struck me as spot on: All were ranked at the lowest level of risk, except for a condo I owned in Hoboken, N.J., that included the ground floor and that, in fact, flooded twice in the decade I owned it.)

In the long run, better information should allow flood risk to be allocated in a mostly rational manner, with homeowners and insurers mostly splitting the liability, but with government in the background to help with out-of-the-blue catastrophes.

We’ve all heard the stories about homes on the coast that get wiped out by storms, then rebuilt, only to be wiped out again, sometimes more than once. Having more accurate data should lead, in time, to underwriting decisions and government policy that reduce or even eliminate such craziness.

First Street Financial describes its report and model as a necessary but insufficient first step. That sounds right. The report is insufficient on its own because lots of other companies and groups will have to finetune the group’s data and, in general, deepen our understanding of flood risk. At ITL, we’ve long appreciated the work done by reThought and Hazard Hub, among others, but many firms will have to step up. And regulators, not known for turning on a dime, will need to become comfortable with using data that exists for each individual property, rather than thinking in broad, imprecise terms like flood plains.

But the report is a necessary, and very welcome, first step.

Stay safe.


P.S. Here is an intriguing piece from a sister publication, Risk & Insurance, on how insurance could help address systemic problems in police departments. The idea would be to require that police officers carry professional liability insurance. Police departments would cover the average cost of the insurance, but each officer deemed a high risk by actuaries (based on number and type of civilian complaints against them, for instance) would have to cover the additional premium payments. The hope would be to price bad officers out of work before they could do something that would wind up on the news.

I’m not at all sure the idea would work. Institutional forces such as police unions would resist like crazy, and there is surely enough uncertainty about how to weight risk factors that they’d be able to piece together an argument. But I found the idea innovative, so I figured I’d share the article. Maybe there’s a way to build on the idea.

P.P.S. Here are the six articles I’d like to highlight from the past week:

4 Post-COVID-19 Trends for Insurers

It’s not all gloom and doom. A crisis usually functions as a great breeding ground for innovation.

The Case for Paying COVID BII Claims

Is it reasonable to assume coverage for a COVID-19-related BII claim in the absence of a virus exclusion? The answer has to be, yes.

How Risk Managers Must Adapt to COVID

To modernize at the scale and speed required, ​”low-code” application development tools should be incorporated within the enterprise.

COVID: How Carriers Can Recover

Does RFP stand for “Request for Proposal” or “Really Frustrating Process?” Carriers can and must do better.

Strategic Planning in the COVID-19 Era

As insurers develop plans for 2021, the question is, where to start? Traditional processes may need to be supplemented with scenario planning.

ERM Shows Its Worth in Pandemic

Companies with sound ERM practices were better-positioned to deal with the pandemic than those with less sound or no ERM.

Need for Context in Assessing Flood Risk

Florida is the highest-risk state for storm surge, with an estimated 2.8 million single family homes at risk and a replacement value of $581 billion, according to the 2019 Private Flood Insurance Report. Yet, in Florida, there are only 1.7 million NFIP policies reported in force, suggesting a huge opportunity for private insurers.

Even so, total direct written premium in Florida actually declined from $84 million in 2017 to $79 million in 2018, largely within the residential market. So what’s holding insurers back from engaging in the obvious latent opportunities in markets like Florida? Answer: The inability to fully contextualize individual risks, as well as their subsequent impact on an existing portfolio.

Why is the ability to contextualize risk so critical to evaluating flood risk?

To ensure accurate and confident risk selection, underwriting and pricing decisions, insurers must be able to fully understand flood risk in terms of their own portfolio. However, this is only possible when using high-resolution, granular flood data, which provides details of flood type, severity and extent, such as JBA flood maps. Without the ability to assess the full granularity of the risk, other flood map providers, such as FEMA, fail to account for all the nuances required in setting premiums and providing coverage.

See also: Using High-Resolution Data for Flood Risk  

Using granular flood data alongside a geospatial analytics solution, like SpatialKey, further enables insurers to assess flood risk in line with their own decision-making. JBA flood assessments within SpatialKey provide a risk score and overall flood rating. The weighted score methodology provides a normalized measure by which insurers can consistently benchmark the risk and use it to inform their rating. Upon review in SpatialKey, and after consulting with JBA where necessary, an insurer can decide whether to write the risk and ensure that it is applying an appropriate deductible.

Use case: Sherwood Park, Palm Shores, Florida

Using a residential property in Palm Shores, FL, as an example, we can see the importance of contextualizing the flood risk in portfolio terms.

Figure 1: JBA location report within SpatialKey for Palm Shores property across fluvial, pluvial, and coastal flood.

The overall JBA flood rating for the property is medium, based on the likely depth of each flood type at different return periods or probabilities of occurrence. It’s weighted to reflect the fact that different sources of flooding will lead to different amounts of damage; coastal flooding is often more damaging due to the salinity of the water (and therefore has a higher score), whereas pluvial flooding is often cleaner and quick to recede.

The location report in Figure 1 shows that the property under the orange pin has only a 1-in-500-year fluvial flood hazard. This indicates that riverine flooding is likely to be infrequent in the area. There is also minimal coastal (or storm surge) flood hazard. However, there is a pluvial hazard at the 1-in-20-year return period, indicating that flooding from heavy rainfall may be frequent here, to depths of up to nine inches. The flood rating is medium rather than high because pluvial flooding is typically less damaging than the other flood types.

Understanding flood risk in the context of a wider portfolio

To make an informed decision on a policy, the impact of an additional flood risk to an in-force portfolio must also be considered. Decisions cannot be made in isolation. And, while information on the individual hazard is extremely beneficial, accumulations at that location should also drive the decision-making process.

Figure 2: SpatialKey helps inform underwriting decision-making with a view of nearby risks (3 black dots) within this half-mile radius.

In Figure 2, three other risks (black dots) can be seen within a half-mile radius of the chosen location (grey pin). As insurers intelligently grow their flood business, an underwriting rule may dictate that, for example, residential flood should not exceed $1 million in a half-mile radius. Based on that underwriting rule, this particular property could prompt the agent to refer this policy to the home office for additional consideration and underwriting.

See also: How Tech Improves Flood Modeling  

It’s clear that granularity is an asset, especially when selecting and pricing flood risk in the U.S. market. Any data source can be misleading or incomplete when used in isolation. The ability to compare multiple data points and multiple data sources in one place is increasingly important with complex events like hurricanes, for example, which involve multiple perils (i.e. wind, surge, flood).

By leveraging flood data within an advanced analytics platform, you can contextualize risk on a whole new level—helping you make more confident decisions, expand your foothold in the flood market and build a strong market reputation as a champion for superior flood coverage.

How Different Flood Types Affect Risk

For insurers to most effectively understand flood risk, they must have access to data that provides a full picture of the hazard, including the different flood types that might affect a property: fluvial, pluvial and storm surge. Although it may seem that flood is just flood, different types can produce various impacts on a property, causing different levels of damage.

Fluvial, pluvial and storm surge: Why it matters

Much of the U.S. is prone to both fluvial flooding (when rivers overtop their banks) and pluvial flooding (when water accumulates across the surface of the ground as a result of heavy rainfall). However, many coastal regions also experience storm surge flooding, which is a result of increased sea levels caused by weather events.

Storm surge flooding is extremely damaging due to the salinity of the water, while pluvial flooding is typically cleaner and quick to recede, likely resulting in lower-cost claims.

Without a view of these different drivers of flooding, insurers cannot understand the full exposure to their portfolios or fully engage with the private flood insurance market.

Use case: Jacksonville, Fla.

The need to understand all the drivers of flood can be illustrated using a residential property on 2nd Avenue, Jacksonville, Fla. Jacksonville is one of the five most vulnerable cities to hurricanes on the U.S. East Coast and at high risk from flooding, experiencing widespread storm surges and flooding during hurricanes Irma and Matthew.

The residential property shown in Figure 1 originally fell into a FEMA Zone X (designated as minimal flood risk).

Figure 1: Contains data from the FEMA National Flood Hazard Layer.

However, when we look at its location on the JBA flood map, we can see some differences in analysis. The JBA flood map identifies this location as at very severe risk to flood (Figure 2, below), from both fluvial and storm surge flooding, whereas using FEMA data alone would not account for either flood type or differentiate between fluvial and pluvial flood. Accessing data sources in addition to FEMA helps provide a more comprehensive understanding of the risk.

Figure 2

The complex interplay between flood types

The risk is particularly high for hurricane-prone areas like Jacksonville, where storm surges often coincide with inland flooding. It’s important to represent this complex interplay during the mapping process instead of tackling each flood type separately. JBA’s storm surge mapping has been developed in partnership with leading hurricane modelers Applied Research Associates, ensuring that hurricane activity is fully accounted for. Additionally, surge data has been used to modify JBA’s inland flood mapping process to reflect the fact that, during a hurricane, rivers can’t flow out to sea as they can in normal conditions. Flood waters then back up, exacerbating fluvial flooding. For insurers to obtain a complete understanding of the hazard, flood maps must fully represent this relationship.

Even with FEMA recently re-mapping the area as a FEMA A Zone, demonstrating that the area is at risk to flood, the drivers of the flood are not clear. As such, underwriting against the FEMA map alone could misrepresent the insurance coverage required.

See also: FEMA Flood Maps Aren’t Good Enough  

It’s clear that having a view of the different drivers of flood risk is vital for effectively understanding and underwriting the risk, especially in areas where hurricanes can be a major source of flood-driven losses.

Flood Risk: Question Is Where, Not When

Over the past year, flood insurance has become more apparent in the media and trade publications. Normally, only catastrophic events (i.e. hurricanes) capture so much attention, but the combination of some massive floods and the continued progress of private flood legislation has started conversations that are overdue. Both the nature of these storms and floods, and their impact on property owners are getting close attention, and that is welcome because it is changing the way people think about underwriting flood insurance.

Recently,  Jeri Xu of Swiss Re published an article that illustrates such a change of perception. She offers a very useful way to think of the rain events (what NOAA calls 1-in-a-thousand-year rain storms) that have caused some of the most serious recent floods (i.e. 2016 Texas, West Virginia, Maryland and Louisiana). Because these types of flood-causing storms are localized at the county-level (roughly speaking), and there are about 3,000 counties in the country, it is not unreasonable to expect three flood-causing thousand-year rain storms every year. With this insight, Xu has transformed the extremely rare to the commonplace and reconciled the headlines with the stats.

See also: How to Make Flood Insurance Affordable  

A bit of caution is needed when comparing rain events with flood events – for the sake of this argument, let’s assume a millennial downpour does result in flooding (it is not a stretch to say so).

Xu and the headlines are teaching us to stop wondering when a serious flood is going to happen – it is way more important to understand where the damage will be when the serious flood does happen.

The accepted and common way to guess where the flooding will occur is the 100-year floodplain on FEMA’s FIRMs. However, according to this article from David Bull, 85% of the losses in Baton Rouge and Lafayette were outside the 100-year flood plain and uninsured. Clearly, the FIRMs do not help underwriters (or homeowners) understand flood risk (neither where, nor when). Indeed, the FIRMs were never intended for that, as they are rate maps, not risk maps.

Instead, underwriters need information that will help them understand the likelihood of a specific property flooding when there is flooding, because the flood is coming, somewhere.

This approach is comparable to how wind (and, lately, storm surge) is underwritten. Karen Clark & Co. has taken such an approach for hurricane: The software assumes an event (the firm calls them characteristic events, or CEs) and then calculates the expected loss results based on that CE happening. There is good reason for this: Underwriters should assume a handful of hurricanes will land on the coast in a given year, just as they should assume a handful of significant inland flood events should happen annually. Working with that logic makes it less important to wonder when something will happen.

See also: Is Flood Map Due for a Big Data Make-Over?  

It has long been written about how flood losses occur beyond flood zones. Looking at flood risk by where, not when, is an effective way for underwriters to manage their business while considering this fact. More importantly, it is a view of risk that supports the creation of insurance products that can help narrow the protection gap in the U.S., because it is unacceptable to have 85% of damaged homes (in Louisiana of all places) without flood coverage.

6 Lessons From Katrina, 10 Years On

In December 2005, just three months after Katrina savaged the Gulf Coast, we edited On Risk and Disaster, a book on the key lessons that the storm so painfully taught. The book was very different from most of the post-mortems that focused on the country’s lack of preparedness for the storm’s onslaught. It focused sharply on how to reduce the risk of future disasters—and how to understand how to help those who suffer most from them.

One of the most important findings highlighted by the book’s 19 expert contributors was that the storm affected lower-income residents far more than others. Reducing the exposure to potential damage before disasters occur, especially in the most hazard-prone areas, is one of the most important steps we can take. To achieve this objective in low-income areas, residents often need help to invest in measures to reduce their losses. Failing to learn these lessons will surely lead to a repeat of the storm’s awful consequences.

Now, 10 years after Katrina struck, six lessons from the book loom even larger.

1. Disasters classified as low-probability, high-consequence events have been increasing in likelihood and severity.

From 1958 to 1972, the number of annual presidential disaster declarations ranged between eight and 20. From 1997 through 2010, they ranged from 50 to 80. The National Oceanic and Atmospheric Administration reported that the number of severe weather events—those that cause $1 billion in damage or more—has increased dramatically, from just two per year in the 1980s to more than 10 per year since 2010. That trend is likely to continue.

2. Most individuals do not purchase insurance until after suffering a severe loss from a disaster—and then often cancel their policies several years later.

Before the 1994 Northridge earthquake in California, relatively few residents had earthquake insurance. After the disaster, more than two-thirds of the homeowners in the area voluntarily purchased coverage. In the years afterward, however, most residents dropped their insurance. Only 10% of those in seismically active areas of California now have earthquake insurance, even though most people know that the likelihood of a severe quake in California today is even higher than it was 20 years ago. Moreover, most homeowners don’t keep their flood insurance policies. An analysis of the National Flood Insurance Program in the U.S. revealed that homeowners typically purchased flood insurance for two to four years but, on average, they owned their homes for about seven years. Of 841,000 new policies bought in 2001, only 73% were still in force one year later, and, after eight years, the number dropped to just 20%. The flood risk, of course, hadn’t changed; dropping the policies exposed homeowners to big losses if another storm hit.

3. Individuals aren’t very good at assessing their risk.

A study on flood risk perception of more than 1,000 homeowners who all lived in flood-prone areas in New York City examined the degree to which people living in these areas assessed their likelihood of being flooded. Even allowing a 25% error margin around the experts’ estimates, most people underestimated the risk of potential damage; a large majority of the residents in this flood-prone area (63%) underestimated the average damage a flood would cause to their house. It is likely that “junk science,” including claims that climate change isn’t real, leads many citizens to problems in assessing the risks they face.

4. We need more public-private partnerships to reduce the costs of future disasters.

Many low-income families cannot afford risk-based disaster insurance and often struggle to recover from catastrophes like Katrina. One way to reduce future damages from disasters would be to assist those in hazard-prone areas with some type of means-tested voucher if they invest in loss-reduction measures, such as elevating their home or flood-proofing its foundation. The voucher would cover both a portion of their insurance premium as well as the annual payments for home-improvement loans to reduce their risk. A program such as this one would reduce future losses, lower the cost of risk-based insurance and diminish the need for the public sector to provide financial disaster relief to low-income families.

5. Even if we build stronger public-private partnerships, individuals expect government help if they suffer severe damage.

Just before this spring’s torrential Texas rains, there was a huge battle in the Texas state legislature about whether local governments ought to be allowed to engage in advance planning to mitigate the risks from big disasters. Many of the forces trying to stop that effort were among the first to demand help when floodwaters devastated the central part of the state. Even the strongest believers in small government expect help to come quickly in times of trouble. We are a generous country, and we surely don’t want that to change. But jumping in after disasters strike is far more expensive than taking steps in advance to reduce risks. Everyone agrees that the cost curve for disaster relief is going up too fast and that we need to aggressively bend it back down.

6. Hurricanes tend to grab our attention—but there are other big risks that are getting far less attention.

Hurricanes are surely important, but winter storms, floods and earthquakes are hugely damaging, too. Too often, we obsess over the last catastrophe and don’t see clearly the other big risks that threaten us. Moreover, when big disasters happen, it really doesn’t matter what caused the damage. Coast Guard Adm. Thad Allen, who led the recovery effort after Katrina, called the storm “a weapon of mass destruction without criminal intent.” The lesson is that we need to be prepared to help communities bounce back when disasters occur, whatever their cause; to help them reduce the risk of future disasters; and to be alert to those who suffer more than others.

The unrest that rocked Baltimore following Freddie Gray’s death reminds us that Adm. Allen’s lesson reaches broadly. The riots severely damaged some of the city’s poorest neighborhoods and undermined the local economy, with an impact just as serious as if the area had been flooded by a hurricane. Many of the same factors that bring in the government after natural disasters occurred here as well: a disproportionate impact on low-income residents, most of whom played no part in causing the damage; the inability to forecast when a random act, whether a storm surge or a police action, could push a community into a downward spiral; and the inability of residents to take steps before disasters happen to reduce the damage they suffer.


Big risks command a governmental response. Responses after disasters, whatever their cause, cost more than reducing risks in advance. Often, the poor suffer the most. These issues loom even larger in the post-Katrina years.

Natural disasters have become more frequent and more costly. We need to develop a much better strategy for making communities more resilient, especially by investing—in advance—in strategies to reduce losses. We need to pay much more attention to who bears the biggest losses when disasters strike, whatever their cause. We need to think about how to weave integrated partnerships involving both government and the private and nonprofit sectors. And we need to understand that natural disasters aren’t the only ones our communities face.

Sensible strategies will require a team effort, involving insurance companies, real estate agents, developers, banks and financial institutions, residents in hazard-prone areas as well as governments at the local, state and federal levels. Insurance premiums that reflect actual risks coupled with strong economic incentives to reduce those risks in advance, can surely help. So, too, can stronger building codes and land use regulations that reduce the exposure to natural disasters.

If we’ve learned anything in the decade since Katrina, it’s that we need to work much harder to understand the risks we face, on all fronts. We need to think about how to reduce those risks and to make sure that the least privileged among us don’t suffer the most. Thinking through these issues after the fact only ensures that we struggle more, pay more and sow the seeds for even more costly efforts in the future.

This article was first published on GovEx and was written with Donald Kettl and Ronald J. Daniels. Kettl is professor of public policy at the University of Maryland and a nonresident senior fellow at the Brookings Institution and the Volcker Alliance. Daniels is the president of Johns Hopkins University.