Friday, May 15, 2015

New York Times Op Ed

I had an Op-Ed in the May 14, 2015 New York Times.  It summarizes the work I did in Canada on a benefit/cost analysis of the Moore Enhanced Building codes.  The academic paper came out last month in Weather, Climate and Society and is co-authored with Greg Kopp at the University of Western Ontario and Paul Kovacs, Executive Director for the Institute for Catastrophic Loss Reduction.  The Op-Ed can be found here.

Friday, April 3, 2015

Straw, Sticks or Bricks: An Empirical Test of the the Three Little Pigs

In May 2013 two violent tornadoes roared through central Oklahoma.  The first, struck Moore on May 20, the third time in 14 years the town experienced a violent tornado.  An estimated $3 billion[1] in damages resulted from that storm alone as well as causing 24 fatalities and hundreds of injuries.  The most tragic result was the death of 7 children at Plaza Towers Elementary School.  Tragedy can bring change as residents struggle with how to respond to such an event.  For the city of Moore, it was time to consider revising their building codes raising the wind load standard from 90 to 135 mph.  This change was designed to build homes that would remain essentially intact for all but the most extreme tornadoes thus lowering the cost to rebuild and perhaps providing a safer home for residents.  It would be difficult to estimate the effect these code changes may have on casualties but a benefit/cost analysis can determine if the code changes will provide financial benefits to the community.  This became the focus of my appointment as a Fulbright Scholar in Canada last year.  I collaborated with Greg Kopp, a Wind Engineer at the University of Western Ontario and Paul Kovacs, the Executive Director of the Institute for Catastrophic Loss Reduction in Toronto to undertake a benefit/cost analysis of the recently adopted new building codes for Moore, OK.  The paper is published by Weather, Climate and Society and can be found here.
To conduct the benefit/cost analysis, three pieces of information must be estimated.  First, how much does the new code add to the cost of construction?  Second, how much damage from future storms can be mitigated by the new code and finally, how much potential damage can be expected across the life of the structure?  The first question was answered by the Moore Association of Home Builders and the engineering consultants hired by Moore.  Their estimate is that implementation of the new codes will cost $1 per square foot.[2] 
A more challenging question is to provide an estimate of the reduction in damages that can be expected from the new code.  The new code required roof rafters to be closer together, hurricane straps for the roof/wall connections, anchor bolts to tie the exterior walls to the foundation, better exterior sheathing and wind rated garage doors.  While the code was designed for wind fields through EF-2, fully 70% of all tornado[3] related damage comes from tornadoes rated EF-3 and above.  But the rating given to a tornado refers to damage consistent with the highest wind detected within the path and does not reflect the wind field for the entire life or area of the tornado.  So a home subjected to an EF-5 tornado may benefit from the new code since almost 90%[4] of the wind field in an EF-5 tornado is EF-2 and less, on average.  Once damage is examined by wind field, rather than tornado rating, 46% of the overall damage comes from wind fields that are EF-2 and less.  Our estimate is that 65% of the EF-0 through EF-2 damage can be mitigated by the new code reducing overall tornado damage by 30%.
The final variable needed is an estimate of future tornado damage.  Since the area of one city is small, it is better to make this estimate for an entire state.  Oklahoma has experienced almost $32 billion in tornado damage from 1989-2012.[5]  The Oklahoma Department of Insurance reports that 65% of that damage is residential so an annual average of $832 million is our estimate of annualized residential damage in current dollars.  Extrapolating that estimate across the 50 year life of the structure, adjusting for inflation, then discounting the estimate back to current dollars provides an estimate of almost $36 billion in expected residential losses from tornadoes for Oklahoma.[6]  If all homes were built to the new standard, a reduction in damage of $11 billion can be expected. 
For the new code to be justified, the increased cost of construction must be less than $11 billion.  To get an estimate of the overall cost of the new code, we need to know the average size of homes in Oklahoma and the current number of homes.  Census[7] and Zillow[8] provide data indicating that the average sized home in Oklahoma is 2,000 square feet and there are currently 1.67 million homes.  This means that if every home in the state had been built to the new code, the state would have increased construction cost by $3.3 billion.  But this increased cost is lower than the estimate of reduced damages by a factor of more than 3 to 1.  So the decision to change the building code in Moore appears to be a good decision and one that should be considered for adoption by the entire state of Oklahoma.

[1] Swiss RE, (2014), “Natural Catastrophes and Man Made Disasters in 2013”, Sigma, No. 1, 2013.
[2] Hampton, Joy (2014), “Moore city council considers storm-resistant building code upgrades, Norman Transcript, March 17, 2014.  Available online at:
[3] Estimate based on damage estimates from the Storm Prediction Tornado archive which can be found at:
[4] Ramsdell, J. V., and Rishel, J. P., (2007), Tornado climatology of the contiguous United States, Tech. Rep. NUREG/CR-4461, Nuclear Regulatory Commission, Washington, D. C.
[5] Estimate comes from state level data provided by the Oklahoma Department of Insurance and national data provided by the Insurance Information Institute.
[6] Estimate uses a 2% inflation rate and a discount rate of 2.5% based on the yield of the 10 year U.S. Treasury Notes.
[7] Census (2012), Selected Housing Characteristics, American Community Survey, 2012 5 year estimates,  American Fact Finder, available online at:
[8] Zillow Real Estate (2014), available online at:

Thursday, May 22, 2014

Better Construction Matters!

Later this summer, a journal article Jeff Czajkowski of Wharton and I co-authored will appear in Land Economics.  But a shorter version of the research was published last week in Visualize, an insurance industry publication.  The link to the article is here:

We were interested to see if better enforcement of building codes could translate into lower damage when a hazard strikes a community.  The hazard we studied was hail, which is one of the costliest hazards for insured losses.  To determine the degree to which building codes were enforced we used the Building Code Effectiveness Grading Schedule (BCEGS) which is administered by the Insurance Services Office (ISO).  Damage data came from two sources, industry level damage from ISO and exposure level damage from Travelers Insurance.  We found that the difference in insured losses between communities with the higher ratings and those with no rating was a reduction in losses of about 20%.   

Monday, May 5, 2014

Building a House of Straw when a Wolf is in the Neighborhood

The Little Rock office of the National Weather Service has released some information on the damage survey conducted after the tornadoes of April 27.  The disturbing finding is that many of the homes suffering the worst damage did not have the exterior walls anchored to the foundation properly.  Anchor bolts are L or J shaped bolts inserted into the foundation before it cures leaving a threaded portion allowing the contractor to bolt the baseplate to the foundation.  Instead the baseplate of some of the homes were attached simply with cut nails.  Their summary can found here:

It may be faster and slightly cheaper to fasten the exterior walls without the anchor bolts but the cost and time saved makes little sense.   I checked prices for anchor bolts online and found that, when purchased in bulk the cost is less than $1.00 each for an 8 inch L shaped bolt with a diameter of ½ inch.  Spaced at 2 foot intervals, the material cost of a 2500 square foot house would increase by less than $100.  The American approach to production has been make it faster and make it cheaper.  That may be fine if the product in question is one that is largely disposable.  But if you are building a residential structure that may encounter a wind storm it seems to me that it is incumbent on the community and on the contractors to ensure that it will perform better than this.

It should not be expected that all residential homes have the capacity to survive a direct hit by an EF-5 tornado.  But the path of most EF-5 tornadoes only reach that strength for a small part of the path itself, meaning that many structures will only be exposed to winds in the lower level of the Enhanced Fujita scale.  Better construction will mean that the residents of those homes have a higher chance of survival and overall, the damage from the storm will be less.

A tension exists between providing affordable housing versus requiring that all homes be built to anticipate any possible hazard.  Communities want developers to choose their town for the next development and if lax standards make development more likely it’s tempting to water down the requirements.  But the cost differential on this one element can’t possibly be justified.

Monday, April 28, 2014

Tragedy Visits Again

Once again a storm roared through the country’s mid-section dropping powerful tornadoes that destroyed communities leaving grief and sorrow for the families and friends of the storms victims.  I’ve witnessed this scene many times both by living in Tornado Alley but more recently from the research I do.  The question that pops up in the aftermath is almost always the same, “What can be done?”

The morning after the Arkansas tornadoes I was watching the morning news shows and as the networks were covering the event I watched a reporter pick up a piece of debris and claim that these homes were well constructed, they were clad with brick.  I’m not an engineer but I work alongside them and I can testify that a home clad with brick is not necessarily an indication of strong construction.  The strength required to withstand tornadic winds comes from features that lie behind the brick and can’t usually be seen from the outside.  It involves how the various components of the home, foundation, walls and roof are tied together, what engineers call a “continuous load path”.  If the connections between the roof and the wall are weak, the roof will fail causing a failure of the walls as well, regardless of what material it is clad with.  Decisions to include these features must be done when the homes are first constructed.  Which brings up a second question, “Should these features be required by building codes?”

That question cannot be answered easily.  First, the most important element when thinking about tornadoes is preserving life and avoiding injuries.  The best advice, when confronted with an imminent tornado is to go to a central part of the structure and put as many walls between you and the tornado as possible.  If you have a shelter, use it!  And don’t get in a car and leave your home unless you live in a very vulnerable structure and have a safe place to go to quickly.  Tornadoes and cars are a deadly combination.  But beyond the issue of life safety, there is a secondary discussion of how to minimize the damage from tornadoes and this is where enhanced building codes come into the conversation.  For this question, as in most public policy debates, we must examine the costs of better construction to the expected benefits.

The key elements of the “continuous load path” are anchoring the exterior walls to the foundation and solid connections (hurricane straps) between the roof structure and the exterior walls.  Stiffening the exterior walls with stronger sheathing further increases the ability of the wall to resist the stress from the wind pressure and using a reinforced garage door reduces the probability of it failing which can cause a cascade of structural failures.  These changes cost money and if we ask all homes to be built in this way, the overall cost to society is large.  But as urban areas expand the chances of multi-billion dollar storms increases so it may be cost effective in some areas.  Oklahoma has experienced over $5 billion in tornado and wind storm damage since 1996.  The damage data I’m using here is found in the Storm Prediction Center’s tornado archive which comes with a caveat, that the damage data is not very reliable.  The true amount of the damage is likely above that amount.  So what would these added construction features add to the cost of a home?  It depends on who you ask but the material cost for the anchors and the roof/wall connections is small.  Better sheathing adds some cost but the home must be sheathed with something so it’s just a question of the added cost of better materials.  The largest component would be the reinforced garage door.

If the mitigation had increased the construction costs by $2,000 per home, total costs to the state of Oklahoma would have been about $2.4 billion for all permanent homes in the state.  And if these features would have resulted in a 30% reduction in damage, using the SPC damage figures, the damage over the last 18 years would have been $1.7 billion.  Not enough to cross the benefit/cost threshold.  But homes are an investment that will be on the plains of Oklahoma for more than 18 years.  So we need to adjust our analysis to a longer time frame.  Plus we need a better assessment of the actual damage than the SPC archive.  There is some evidence that the actual damage is much more than shown on the archive.  If it’s off by 40% then actual damages experience are closer to $8 billion meaning that a 30% reduction in damages is now $2.4 billion.  As a side note, the SPC archive has total damages from tornadoes in 2011 at less than $10 billion whereas the National Climatic Data Center has it at over $20 billion.

There is also empirical evidence that home buyers are willing to pay more for better construction.  To study this question economists use models that parse out the effect on the selling price of homes feature by feature.  These models are called hedonic pricing models and studies conducted in areas vulnerable to natural hazards show that selling price goes up when hazard mitigation features are present.

Better structures will not ensure that casualties will be eliminated but it’s reasonable to assume that if we can built better homes, those residing in the homes are more likely to escape harm or at least minimize the injuries and suffering that inevitably follow in the wake of nature’s most powerful storm.  And if we reduce the monetary damages in the process, well that’s just a bonus.


Tuesday, May 28, 2013

The Dallas Morning News ran an op-ed I wrote about where to spend tornado safety funds.  The most tragic element of last week’s tornado in Moore was that an EF-5 tornado, an extremely rare event, hit two elementary schools with full force.  Few structures are designed for that type of stress but engineered buildings would be about the best choice you could have.  Most residential structures would be leveled.

The public policy question here is what, if anything, can we do to minimize casualties?  And how should we fund those efforts?  Currently, FEMA and the state of Oklahoma has a program to pay for most of the costs to engineer saferooms in new Oklahoma schools which would provide protection from even the largest tornado.  Currently, only 100 schools in Oklahoma have received this funding and regrettably, Plaza Towers Elementary in Moore was not one of those.  If you consider that Oklahoma has 1,780 campuses and estimates to provide a safe room run from $500,000 to $1,000,000 per campus, the outlay to protect all schools would be immense, $1 to $2 billion dollars.  FEMA’s HazardMitigation Grant Program (HMGP) provides grant money to areas declared by the President of the United States to be disaster areas.  This designation releases money that can be used by residents to install a FEMA approved safe room or communities to designate for other mitigation uses like sheltering options for mobile home parks.  The most recent iteration of this program in Oklahoma was so oversubscribed that grantees were determined by lottery.

In general, my research suggests that using public money to save lives from tornado fatalities fails the benchmark that spending is considered reasonable if the cost per avoided fatality is less than $10 million.  In our book Economic andSocietal Impacts of Tornadoes and the follow up to that book DeadlySeason:  An Analysis of the 2011 TornadoOutbreak, Dan Sutter and I show that these programs are well outside that benchmark for most uses of the program.  Using the funds to reduce fatalities in mobile homes is the one exception for some states.

These events define tragedy and emotions run high, as they should.  My suggestion as we go forward to implement these programs would be as follows:  1)  For school districts, instead of outright grants, use Federal money to provide the financing at 0% interest to cover the additional costs in new schools to provide a safe room.  For existing schools who wish to modify campuses to provide a safe room, provide financing at 0% interest as well.  Each district and its residents must decide whether or not this expenditure is the optimal use of limited funds, but for those that do, make it possible by providing the money at no interest.  2)  Grants for use in residential housing should be strictly limited to the most vulnerable structures, mobile homes, or to low income families.  The HMGP currently addresses this by providing a higher percentage of the safe room covered by the grant, from 75% to 90%.  But, again, funds are very limited and it seems to me that more affluent families should be able to make the choice to build a safe room without help (see my previous post on the safe room my wife and I installed) and that if we do anything, it should be directed at the most vulnerable structures.

Monday, May 27, 2013

I was interviewed on Bloomberg Friday.  The video link is here.  One of the questions was about saferooms and given the horrible tornado in Moore last week it is an important topic.  I live in North Texas and have seen my share of tornadoes.  Surviving one depends on many factors but the most important are the size of the storm, the amount of time you have to prepare and the type of structure you are in when the storm hits.  A well-built engineered structure is the safest.  The worst is a mobile home.

Permanent homes provide reasonable shelter from small tornadoes and with modest enhancements in construction can also provide reasonable shelter from many strong storms.  That would account for almost 98% of all tornadoes (EF0-EF2).  The Insurance Institute for Business and Home Safety suggests building homes to meet their Fortified Home Program.   Some communities are adopting aspects of the program in their building codes.  Moore OK, is one of those.  But that would not apply to homes built prior to the enactment of enhanced codes.

For residents of permanent homes to get protection from violent tornadoes some type of shelter is required, in residence or underground, and the options for those are growing every year.  We recently converted our hall closet to a shelter.  Retrofitting an existing structure is not ideal and if an EF-5 tornado took a direct hit on our house, it’s not a sure bet we would survive.  But since 1950, there have only been 59 EF-5 tornadoes, including the one last week, out of more than 60,000 tornadoes.  So we are talking about an extremely rare event.  And also remember that to receive the EF-5 rating, the tornado does not have to exhibit EF-5 winds for the entire path.  The tornado that hit Joplin 2 years ago was rated EF-5 but that rating applied to about 6 miles of the 21 mile path.  The tragedy there, like the tragedy in Moore is that the strongest winds occurred over a populated area.

The public policy question is whether or not we should require the installation of shelters by homeowners.  They do add a significant cost to the price of a home.  Some homeowners, myself included, are willing to pay the added cost but I’m not sure that most people would, even those that live in tornado alley.  A second policy option is to offer a subsidy to homeowners for a shelter.  Dan Sutter and I have written extensively about this.  This is a very expensive program and will reduce fatalities but at a higher cost than programs to reduce casualties from other risks we face.  Oklahoma has such a program but the recipients are decided by a lottery.  My state, Texas, has a similar program and we live in a county that participates in it.  But I feel strongly enough about this that I did not apply for the subsidy.  (That was a popular decision in my house.)  If public money is to be used in this way, it should go to vulnerable structures like mobile homes.  And if it is decided that public money will be used to subsidize the installation of shelters in permanent homes, it should be means tested so that people with the income to bear the cost are not receiving the subsidy at the expense of those who cannot.