Terrorism Insurance ’Underpriced’ in United States

Risk management expert Erwann Michel-Kerjan discusses how the terrorism insurance sector has evolved since September 11.

September 08, 2008

To help readers better understand the nuances of foreign policy, CFR staff writers and Consulting Editor Bernard Gwertzman conduct in-depth interviews with a wide range of international experts, as well as newsmakers.

More on:

United States

Terrorism and Counterterrorism

In the seven years since 9/11, no major terrorist attack has taken place on U.S. soil. As a result, debate over the financial implications of terrorism has gradually faded. In the meantime, the insurance and reinsurance sectors, which took a heavy hit after the World Trade Center attacks, have worked to gird themselves for the next "day after" scenario. Erwann Michel-Kerjan, an expert on terrorism insurance and the director of the Risk Management and Decision Processes Center at the Wharton School of Business, says there is still much work to be done from a policy standpoint. He notes that roughly 40 percent of major U.S. businesses remain entirely uninsured for the possibility of a terrorist attack, despite the devastating threat an attack could pose for their operations. He also argues that government insurance subsidies, combined with the lack of an attack for seven years, have led to a situation in which terror insurance in the United States may now be underpriced.

Could you give us a brief overview of how the business of insuring against terrorism has proceeded since 9/11? There was a fair amount of concern about this right after 9/11, but the urgency of the issue seems to have faded a bit, at least in terms of press coverage. Where do things stand now?

In a sense one has to go back before 9/11 to understand better what happened. On September 10, 2001, insurance companies in most OECD [Organization for Economic Cooperation and Development] countries like the United States, in Europe, and in Asia, were covering terrorism as part of their commercial products, but they were doing it for free. In other words, they didn’t exclude terrorism from their policies, but at the same time they were not pricing specifically for that risk.

On 9/11, obviously, insurance companies realized that the very nature of international terrorism had radically changed but that it was still not excluded from their policy; they would have to pay between $30 [billion] and $40 billion of insured losses. When President Bush stood up the afternoon of 9/11 and said "This is an act of war," insurance companies first thought they would not have to financially cover these losses because war is excluded very specifically in any insurance contract. But the president stepped back afterward and said, "Well, this is a terrorist attack," and basically insurance companies had to pay.

Reinsurance companies [which sell insurance policies to other insurance companies to protect against the risk of massive payout], who paid for about two-thirds of 9/11, most of them being European reinsurance companies, decided to leave the market for terrorism insurance almost everywhere in the world. As a ripple effect, insurance companies were deprived of reinsurance capacity, and also decided to leave the market wherever they could.

So six months after 9/11, in the United States about forty-five states allowed insurance companies to exclude terrorism from their coverage. A year after 9/11, America was mainly uncovered against a terrorist attack. Because of that, there was lots of pressure from some trade associations and lobbyist groups saying, "We need some kind of coverage. We cannot build up new buildings without being covered against terrorism." So there was a lot of pressure on the government to help provide a solution. A new program was passed in December 2002. It’s called TRIA, which stands for Terrorism Risk Insurance Act. TRIA is a complex system but basically what it does is guarantee to corporations operating in the US that they will be covered up to $100 billion of losses resulting from a terrorist attack, and the $100 billion will be spread between insurance companies and the federal government.

Spread evenly?

Not evenly; that’s a point. Year after year, the distinction becomes more and more toward making the private sector pay more, and today it’s about $25 billion that will be picked up by the private sector, and the rest will be picked up free of charge by the federal government (with a possibility to recoup part of it against all commercial insurance policyholders, whether they have terorirsm insurance or not—which raises equity issue as well). Under the current program that’s an important question: Why, or whether, the federal government should provide free reinsurance to insurance companies, or whether it should try to replace some of that federal reinsurance with private sector reinsurance capacity.

Now if I understand correctly, when TRIA came into being it was basically intended as a temporary measure to protect insurers while they developed their own strategy for insuring against terrorism. It’s been extended twice now [in December 2005 and December 2007]. Do you think they’ve made any good headway in terms of developing their own strategy?

Looking back in history, you learn that there is nothing more permanent than the temporary. The Price-Anderson Act, for our nuclear capacity, was supposed to be temporary. The NFIP for flood insurance, which was supposed to be temporary, was established in 1968. The CEA in Californiafor earthquake insurance is the same thing. We have to keep that in mind.

With that said, there has been a change in how TRIA operates. For instance, with the renewal of TRIA for seven years at the end of 2007, we decided to include domestic terrorism in the program. This makes a lot of sense for us, or for people who have been working in the terrorism insurance market for years now. You don’t want to be here the day after another attack saying, "Well, I’m sorry, but with domestic terrorism attacks you don’t get a penny, versus if an international organization is perpetrating the attack you get full coverage."

We have done several studies, mainly with my colleagues here at Wharton, on several aspects of TRIA. Tria has good features in it, and others that are less so. With TRIA in place, what we’ve seen over the last several years is more firms buying that coverage, partly because that coverage has become cheaper and cheaper. Of course, since we talk terrorism threat here, it’s hard to tell the "right" price, because you need the probability for that. But if one looks at what enterprises pay today, the price has significantly decreased over the past five or six years. There are two main reasons for that. One is that you have that free reinsurance [the $100 billion provided cooperatively by the federal government and insurance firms]. The other is just that seven years have passed without new terrorist attack on U.S. soil and there are other things on the agenda.

You say more and more companies are buying terrorism insurance at this point. What sorts of percentages are we talking about? How common is it?

That’s a good question. Surprisingly, to me at least, there is no national data collection. You would expect that, being a nation at war, putting national security as a key concern on Washington’s agenda, you would like to know what proportion of our companies, who are ensuring the strength of our economy, are protected financially speaking. We don’t know that. What we do know is data gathered by what we call "insurance brokers," in the middle between companies and insurance companies. A company will go to an insurance broker to try to get the best deal on insurance coverage. Working with brokers like Marsh & McLennan, we have collected data-—obviously only for [the brokers’] clients—but looking at these clients, which are typically very large companies, we know that about 60 percent of them have terrorism insurance today. So that’s good in a sense, but if you look at it another way, 40 percent of these companies don’t have any terrorism insurance whatsoever.

Presumably some companies have bigger exposure to the risks of terrorism than others, right?


So is there a percentage of companies for which you would say they probably oughtn’t bother with this kind of coverage in the first place?

It’s not so much how many should have coverage, it’s also how much coverage. A company can have insurance but it covers only 10 percent of its assets. There is a geography factor as well. We are about to finish a new study for which we got access to the portfolio of Marsh & McLennan, and what we observe, not surprisingly, is that the market penetration rate in the New York metropolitan area is much higher than in the rest fo the country, it’s about 78 pecent. So there’s definitely a factor here. For the first time, we are also able to quantify corporate insurance price elasticity on the U.S. market. We find companies headquartered in this region to be much more price inelastic than in the rest of the sample; that is they are less sensitive to insurance price when buying terrorism insurance. Our results show that a 10 percent price increase would decrease the quantity of terrorism insurance purchased by only 0.7 percent.

Can you walk us through how insurance firms think about terrorism risk when they’re pricing these policies? How do you guess at the likelihood of an attack, and how do you then price the economic consequences? It seems like a fuzzy endeavor.

That’s the right characterization. We worked with many of them in a large study called "TRIA and Beyond" that we published three years ago, and one of the conclusions was that, as many other stakeholders, they don’t really look at probability. Which might be surprising for those who would expect that insurance companies to study probabilities carefully. Their statement on why they don’t do that is very simple: "There is no way we can know for sure the probability. We don’t have a clue. There’s a lot of unavailable information, and some intelligence services will have more information than we have." They certainly have a point here. If you look at earthquakes, or car accidents, insurers will have a lot of statistics, but not for terrorism. What insurers do is for the latter to look at their concentration of exposure in different places. They say, "I’m going to provide some coverage at that price, but I don’t want to reach more than X dollars of coverage in New York," for instance. So you lower your maximum possible loss doing that.

This might sound like a silly question, but how much is $100 billion? Can you put that in the context of what other sorts of attacks we’ve seen? Is that a tremendous amount of money?

Oh yes, it’s a lot.  On 9/11, the World Trade Center itself was the target, one-third of the losses was for business interruption. Katrina cost $45 billion to the insurance industry, and that’s to date the largest [payout] worldwide in the insurance industry; 9/11 is now second. The surplus for the U.S. insurance industry is about $400 billion, so if you plug in these numbers together, you see that $100 billion payouts would have a major impact on this industry.

Can you talk about where the reinsurance market for terrorism insurance stands right now? Is it still just totally decimated?

Putting on my Wharton cap, I’d like to see more of the private sector coming in. In the United States, there’s a very small reinsurance coverage for terrorism [market]. Over the past few years I have worked with the presidents of different national programs in Europe (Britain, Germany, and France) to better understand how other markets outside of the US have worked out. And what you see is that some of the reinsurance companies have reentered the market as part of these programs, but for a very small slice, and there are like ten, fifteen, twenty reinsurance companies involved (including most of the largest ones). Typically they will provide, let’s say, $50 to $100 million of coverage, which isn’t a lot, but all of that summed up goes to a few billion dollars. We could imagine a similar solution in the US as well: many players, limited individual exposure.

How do terror insurance markets vary internationally? Are there significant differences between how these policies are priced in the U.S. market versus in other developed economies?

It’s a question that everybody has asked for a few years, but up to 2006 nobody had really spent time to do it. In 2006 we did a study that was published by Morgan Stanley. We started by making the assumption that price reflects, in some fashion, the level of exposure. So if one company is paying $100, and another is paying $50, you can make the argument at least that one is twice as exposed as the other. That’s what the market’s telling you. So we tried to compare similar companies, getting data on the U.S. market, German market, French market, and British market, and to compare how much more expensive was terrorism in the United States versus Europe. Our fundamental assumption was that it had to be more expensive, because we are more exposed here, for a large-scale attack at least. Using Germany as an example (because we thought Germany wasn’t too exposed), we found that depending on the sector, industry, and location, the price of terrorism insurance had become two, three, or even four times higher in Germany than it was in the United States. That was really puzzling to us. Given that prices have continued to decrease here, the question for both business and policymakers is, has terrorism insurance become underpriced in the United States? And I think that’s largely the case.