Presentation to Arkansas legislative committee on economic development
on earthquake safety provisions in building code
Seth Stein, William Deering Professor of Geological Sciences, Northwestern University
January 28, 2008
I am happy to have the opportunity to speak to you today about the important question you are facing, how to prepare for possible future large earthquakes in Arkansas. This is a complicated question involving hard choices, to which there is no right answer.
A way to see this is to think of yourselves – the state - as a homeowner whose insurance agent is trying to sell him a new policy. He says the policy will cost more, but offers better coverage. You ask how much it costs, and he says he doesn't know. You ask him how much the coverage is, and he says he doesn't know. Most people would tell him to figure out the costs and benefits, then come back and you'll discuss whether to upgrade the policy.
In a nutshell, that's my advice to you.
This comes from my perspective as an earthquake seismologist interested in the kinds of earthquakes that occur in the Midwest. I have been studying them for 30 years, and been working on central U.S. earthquakes specifically since 1990. During most of that time I have been on the faculty of Northwestern University, where I am currently William Deering Professor of Geological Sciences. I have published more than 100 scientific papers, written a widely used seismology textbook, and edited a variety of books including one this year about the topic we're discussing: science, hazard, and policy issues for earthquakes in the middle of continents. I've won a number of awards and held many positions in the geological community including editing the top journal and serving as director of the consortium of universities using GPS for earth science. In addition to research I am active in scientific education, including a national lecture tour on behalf of the Incorporated Research Institutions for Seismology and Seismological Society of America. I have also advised financial institutions on earthquake hazards.
The question you face is how much your communities should spend over the next 500-1000 years to reduce the damage from large earthquakes that may happen. Deciding this involves comparing benefits and costs. The more strongly you build; the better off you will be if a major earthquake happens. However, the more strongly you build, the more you spend resources that won't be available for other uses. This is like buying insurance: the more you buy, the better you're covered but the more it costs.
The specific question you're facing relates to a new building code favored by the Federal Emergency Management Agency (FEMA). This would raise the earthquake resistance of buildings to withstand the maximum shaking expected in 2,500 years. It requires building to similar levels as in California.
I think the alternative being proposed here makes more sense. It is to follow a more modest and less expensive standard now, designing for the maximum shaking expected in 500 years, and arrange for an objective outside study of whether the California one makes more sense. If the California standard turns out to make more sense, then you should do it.
What to do is a policy decision that you have to reach. There are three kinds of information that can help you. The first is seismological: what we can say about the hazard - how often future earthquakes will happen, how large they will be, and how much shaking they will cause. The second is engineering: how much will it cost to build buildings with different degrees of earthquake resistance and how much would these reduce damage in a future earthquake. The third is economic: what are the costs and benefits to the community of different construction standards.
From my view, I would like to make two points.
First, although there is an earthquake hazard in the New Madrid seismic zone – which includes parts of Arkansas – it is much smaller than you sometimes hear. The argument in favor of the new building code is the idea that the earthquake hazard here is about the same or greater than in California, so you need to build to California standards. That probably sounds unlikely to you, and it is. It's based on some very questionable assumptions. I think it's more useful to think of the earthquake hazard here as about 1/3 - 1/10 that in California.
Second, the case for the new building code, the 2500-year standard, is very weak. FEMA has been pushing for it without any serious study of the costs and benefits involved. No one really knows if this makes sense. I think there's a good chance that the costs are much greater than the benefits. Thus I recommend that you adopt a less stringent criterion at present. I also recommend that you, ideally with neighboring states, arrange for an objective outside study to assess the seismic hazard and costs and benefits of different policies including the 2500-year one. The results of such a study would help you make long-term policy. What FEMA is pushing for now could turn out to be getting the wrong answer as fast as possible. As the old line asks, "Do you want it right, or do you want it now?"
I'd like to amplify on these points. There's more detail, including graphics, in the printed material.
I suspect most of you were surprised to hear the FEMA and U.S. Geological Survey (USGS) claim that your earthquake hazard is the same or bigger than California's. I certainly was when I heard this and thought it wasn't right. Almost all the seismologists I talk to are also skeptical, simply because earthquakes are a lot less common here.
There's a lot of hype about earthquakes in the area, so it's useful to put some numbers on this, We're talking about the entire central U.S. or New Madrid seismic zone, which covers more than 100,000 square miles including parts of Arkansas, Tennessee, Kentucky, Missouri, Illinois, Indiana, and Mississippi. Damaging earthquakes aren't common and they're typically a nuisance rather than a catastrophe.
Specifically, in this entire area, there's an earthquake with magnitude 5 or greater on average about every 15 years. These do at most minor damage, depending on whether they're close to a populated area. Since the bigger an earthquake the less common it is, there's one with magnitude greater 6 about every 150 years. These can do some more damage, but usually aren't that large a concern. The largest earthquake in the past century, 1968 (M 5.5) Illinois earthquake, caused no fatalities. Damage consisted of fallen bricks from chimneys, broken windows, toppled television aerials, and cracked or fallen brick & plaster.
The real concern is large earthquakes like those that occurred in 1811 and 1812. The size of these earthquakes has been hyped enormously. You'll hear things like that they were the biggest in the U.S., caused church bells to ring in Boston, etc. That's not true, but they were magnitude 7 earthquakes that caused shaking across much of the area. Most of the area, places like St. Louis, Louisville and Nashville, had only minor damage, typically a few fallen chimneys. However, if large earthquakes like those occurred again, they could be very destructive.
We don't know if these will happen again on a 500-1000 timescale. Geological studies find that New Madrid earthquakes comparable to those of 1811-1812 occurred about 1450 and 900 AD, suggesting that these could happen every 400-500 years. It doesn't look like they're going to happen again soon. Precise measurements using the Global Positioning System (GPS) show that motion across the Seismic Zone currently is either very slow or zero. Earthquakes occur because the earth acts like a spring – motion is stored up over hundreds of years and is then released. Because motion has to accumulate for many years to cause a large earthquake, it will be at least hundreds of years, and perhaps much longer, before another large earthquake happens.
In other words, we're not talking about a disaster that will happen tomorrow. We're talking about a potentially serious problem on a few hundred-year time scale. It.s worth thinking about, but there's plenty of time to think it through and develop a sensible strategy.
It.s important to realize that this is very different from California. Southern California earthquakes of a given magnitude are about 30-100 times more common than in the central U.S. That's because California's result from the approximately 2 inches per year (46 mm/yr) motion between the Pacific and North American plates, whereas New Madrid is inside the North American continent, where GPS data show the motion is less that 1/20 of an inch (1 mm/yr). There's a complication because rock in the Midwest transmits seismic waves better, so a given size earthquake causes more shaking. This reduces the effect of the difference in earthquake rates.
So the seismology tells us that in any year, a building in southern California is 3-10 times more likely to be seriously shaken than is one in the central U.S. In fact FEMA did their own study using a different method and came up with about the same answer. They calculated the annual earthquake loss ratio, which is the ratio of annualized earthquake loss to the replacement cost of all buildings in the area. They got values for Memphis and St. Louis 1/5 - 1/10 of those for San Francisco and Los Angeles. Memphis was 32nd among major U.S. cities; St. Louis was 34th. Since ratios are equivalent to the fractional risk of building damage, this predicts NMSZ buildings 5-10 times less likely to be damaged during their lives than ones in California.
How is it, then, that FEMA and the Geological Survey are telling you that the central U.S. is as dangerous as California and should build to the same standard? It's because of the way they've defined the earthquake hazard.
Earthquake hazard is not something we measure or know. Instead, it's something we calculate using a computer model by making assumptions about where and how often future earthquakes will occur, how large they will be, and how much ground motion they will produce. These aren't things we know well, especially in the central US where large earthquakes are rare. Thus it's been said that estimating earthquake hazard is like playing "a game of chance against nature of which we still don't know all the rules."
Depending on what you assume, you can get the hazard to be much higher or much lower. The U.S.G.S. maps got a hazard in Memphis as high as in California by choosing high values for every parameter of the model: how often big earthquakes will occur, how big they'll be, and how much shaking they will do. We've tried choosing lower values – which I think make more sense – and the hazard comes out a lot lower.
For example, they assume that right after a big earthquake, another one on the same fault is just as likely. So if they are 500 years apart on average, then in the next 50 years there's a 50/100 or 10% chance. A lot of us think it makes more sense to assume that the fault has to store up energy for the next earthquake, so the risk starts off very low and then grows with time. Calculating this way the probability is much less, about 1-2%.
But there's an even bigger problem than what you assume about the future earthquakes. It's how they defined the hazard. The hazard is defined as maximum shaking that the model predicts will happen in some time period. Traditionally in the U.S. and the rest of the world, this has been the maximum shaking predicted on average once every 500 years, which is the same as what the model predicts has a 2% chance of happening once in 50 years. Using this criterion, the hazard in Memphis is much less than in California.
However the new code uses the maximum shaking predicted once in 2500 years, or what there's predicted to be a 2% chance of experiencing once in 50 years. Calculated this way and using the high-risk model parameters I just discussed, the hazard in Memphis is predicted to be similar or greater than in California.
I don't think this make sense. Typical buildings have a 50-100 year lifetime, so the issue is what's the strongest shaking they are likely to experience during their life. This is much higher in California than in the Midwest. The 2500-year value is the maximum shaking predicted at a point on the ground in 2500 years – but the same building won't be there.
An extreme analogy might be whether a building is in more danger from a flood or a meteor impact, like the one that killed the dinosaurs. In 100 years, a flood is by far the bigger risk. In a billion years, a meteor might be a greater danger since it could do more damage. The reason we don't design buildings for meteor impacts is that they don't last a billion years.
That's why traditionally in the U.S. and the rest of the world, buildings are designed for the maximum shaking predicted on average once every 500 years. That's actually tougher than the way buildings are designed for many other hazards. They're designed for the maximum flood expected every 100 years. This is for normal buildings: special ones like nuclear power plants or hospitals can be designed to higher standards.
To summarize, I think it's clear that the earthquake hazard in the Midwest is a lot less than in California.
One could say, all right – even accepting the earthquake hazard is really much less than in California, why not build to that standard anyway? Wouldn't that be the safest thing to do?
The answer is that it would be the safest thing to do. But it might not be the best thing for the community. That's because earthquakes aren't the only problem communities face.
Let's talk about the problem nationwide and then in the Midwest. Earthquakes kill on average about 6 people a year in the US, but that number varies. Most years no one is killed, about every 20 years an earthquake kills more than 50 people, and the 1906 San Francisco earthquake killed 3,000 people. Hence on average earthquakes are about as dangerous as skateboards (10 deaths per year) or football (20 deaths per year). They're a lot less dangerous than bicycles (700 deaths per year), bad weather (500 deaths per year), or fires (3200 deaths per year). As far as we know, no one has been killed by an earthquake in the Midwest to date.
However, earthquakes can cause major property damage. The 1994 Northridge earthquake under the Los Angeles metropolitan area caused 58 deaths and $20 billion in property damage. The building damage results in most of the deaths. This principle is often stated as "earthquakes don't kill people; buildings kill people".
Thus the primary defense against earthquakes is a building code whose goal is designing structures that should not collapse. The trick is that to decide on a sensible building code.
A community's choice of building codes reflects a complicated interplay between seismology, engineering, economics, and policy. The goal is to assess the seismic hazard and chose a level of safety that makes economic sense. That's because such design raises construction costs and diverts resources from other uses.
As the later speakers will discuss, there are large sums involved. Estimates vary, but they're in the range of 1-10% of building costs, depending on the building and what you do. Since building costs are about $2 billion annually in the Memphis area alone, even 1% is significant. In fact, retrofitting the Memphis VA hospital to bring it to California code, including removing nine floors, cost $100 million. You're spending a lot of money every year for hundreds of years, in hope of reducing losses if a large earthquake comes.
This means striking a tricky balance that calls for careful thought. You don't want the building code to be too weak and permit unsafe construction and undue risks. On the other side, if it's too strong you're imposing unneeded costs on the community. There's also the risk that if the code requires costs that people think don't make sense, they'll evade the code, so you can end up with weaker buildings than if people followed a less stringent code.
Deciding where to draw this line is a complex issue for which there is no unique answer. Although national codes offer overall insight, local jurisdictions are under no obligation to accept them, and can modify them to balance local hazards and costs. This is increasingly happening around the Midwest as you'll hear, for example in Shelby County and Kentucky.
It's not clear that California is the best model. California's building codes evolved from decades of damaging earthquakes there. Although this process involved more trial and error than benefit/cost analysis, they are accepted because they seem to strike a sensible balance consistent with the public's experience.
The situation here is different in that damaging earthquakes are rare, so there's not much experience to guide you. Hence you've got a tough problem. I'd like to offer some general suggestions. I am not an engineer and so will defer specifics to others.
I think there are two useful ideas that can be helpful.
The first is "There's no free lunch."
The resources used for earthquake safety aren't available for another purposes. This is easy to see in the public sector, where there are direct tradeoffs. Funds spent strengthening schools aren't available to hire teachers, upgrading hospitals may mean covering fewer uninsured, stronger bridges may result in hiring fewer police and fire fighters, etc.
A similar argument applies to saving lives - the proposed code might over time save a few lives per year, whereas the same sums invested in other public health or safety measures like health care or highway upgrades could save many more. Deciding on the best balance is tricky.
The second is "There's no such thing as other people's money."
It's easy to say that business will pay for most of this. But since the private sector provides most of our goods, services, and jobs – directly or indirectly - costs are ultimately borne by society as a whole. Imposing costs on the private sector affects everyone via reduced economic activity (firms don't build or build elsewhere), job loss (or reduced growth), and resulting reduction in tax revenue and thus social services.
I'm amazed that FEMA and the USGS are urging communities to adopt the new building code, without analyzing whether the increased cost is justified by the benefits. It's a classic unfunded mandate. I don't think that makes sense.
I think the course of action proposed here is more sensible.
It would be to build to the 500-year standard, not 2500 years. From what we know now, this seems like a sensible balance. I don't think there are serious risks involved for several reasons.
First, as I discussed earlier, the USGS hazard maps are computed using values that predict a hazard on the high end of what's plausible, so you're pretty safe even with the 500-year ones.
Second, time is on your side. Any big earthquake is probably hundreds of years away. Whichever standard you chose, the natural cycle of replacing older buildings by safer ones will make communities more earthquake resistant. This is especially true when brick buildings are replaced by concrete. This happens naturally in the private sector for economic reasons. I would make sure that this also happens for important public facilities like schools, hospitals, fire stations, etc.
In addition to adopting the 500-year standard at present, I also recommend that that you, ideally with neighboring states, arrange for an objective outside study to assess the seismic hazard and the costs and benefits of different policies including the 500- and 2500-year standards. The results of such a study would help you make long-term policy.
There's an example in the written material, which is a recent study by the National Institute of Standards and Technology for the U.S. Fire Adminstration. The study looked at the costs and benefits of putting sprinkler systems to suppress fires in new homes. They argue that although analysis in 1984 found that these were not cost-effective, improvements in their performance and cost now make them cost-effective. Insurance companies are now offering discounts for homes with sprinkler systems, and communities are starting to require them.
There are several ways to do such a study. One is via the National Academy of Sciences / National Academy of Engineering study process. They have a committee on seismology that could sponsor it. This is the usual way of coming up with recommendations on policy issues that complicated scientific and technical aspects. Their process is described in the written material. It.s very carefully set up to bring in a range of expertise and reflect the full spectrum of views. They work with the organization that sponsors the study to define the study's goals, but make sure that that organization does not influence its outcome. It is the best way we have of coming up with sensible results that will be generally accepted. For example, this is the approach being taken for hurricane protection policy for New Orleans.
There are also reputable consulting firms who could do a good study for you. I've worked with several and found their work very good.
I think the costs would be reasonable, especially if some of the other states participate. You also might ask FEMA and the USGS to also cover part of the cost. After all, if they think the 2500-year criterion makes sense, they should be confident that an outside evaluation would agree.
I think this alternative makes much more sense. It.s to do something reasonable now and take the time to think through carefully whether it's worth doing something more.