31 May 2011

Donald Boudreaux: I'll Take That Bet

Writing in the WSJ last week economist Donald Boudreaux of George Mason University offers to make a bet in order to make a point about human-caused climate change in response to Bill McKibben's silly essay from earlier in the week in the Washington Post:
I'll bet $10,000 that the average annual number of Americans killed by tornadoes, floods and hurricanes will fall over the next 20 years. Specifically, I'll bet that the average annual number of Americans killed by these violent weather events from 2011 through 2030 will be lower than it was from 1991 through 2010.
I am willing to take this bet in order to raise awareness of the fact that both sides of the debate over climate change debate can't see the forest for the trees.  The factors that will drive loss of human life due to weather extremes in coming decades will be increasing vulnerability and exposure.

As a condition of the bet, when I win (which unfortunately will occur long before 2030) I ask that the proceeds go directly to the American Red Cross. (Should I lose the bet come 2030, I'll make out a check to the charity of Prof. Boudreaux's choice.)  A second condition is that Prof. Boudreaux agrees to write an op-ed for the WSJ (or some other venue) explaining the bet and why he lost (of course, I am willing to do the same).

Here are the technical terms of the bet that I will accept.  Prof. Boudreaux refers to three hazards: floods, tornadoes and hurricanes.  The dataset that he proposes using is the official record kept by the US National Weather Service (available here in PDF through 2009, and here in PDF for 2010).  This leads to the following summaries for the base period of 1991-2010 proposed by Prof. Boudreaux:

Tornado Flood Hurricane
1991 39 61 19
1992 39 62 27
1993 33 103 2
1994 69 91 9
1995 30 80 17
1996 26 131 37
1997 67 118 1
1998 130 136 9
1999 94 68 19
2000 41 38 0
2001 40 48 24
2002 55 49 53
2003 54 86 14
2004 34 82 34
2005 38 43 1016
2006 67 76 0
2007 81 87 1
2008 126 82 12
2009 21 53 2
2010 45 103 0

1129 1597 1296 4022

In his WSJ column Professor Boudreaux asks his readers to subtract the deaths from Hurricane Katrina because they were the result of a levee break.  It is not clear whether that was just a rhetorical move for the column or if that calculus also extends to the bet. He can clarify that for me in his response. With Katrina the total is 4,022 deaths and without Katrina the total is about 3,000 (again, Prof. Boudreax can tell me which number he prefers).

So far 2011 has seen 518 deaths from tornadoes.  This means that from today through 2030 the United States could see only 3,500 additional extreme weather deaths, or 180 per year (using the higher baseline that includes Katrina deaths, or 154 per year using the lower number of 3,000).  Such numbers would represent an improvement over 1991-2010, and Prof. Boudreax would still lose the bet.  We should be so lucky, and it would take a lot of luck, to see so few deaths due to extreme weather.

The fact of the matter is that our vulnerability to extreme weather is increasing, due to a combination of a growing population and especially urbanization in locations prone to extreme weather events.  This means that even with the hard work by many professionals in a range of fields that has contributed to the dramatic decrease in the number of deaths over recent decades low death totals are unlikely to continue into the future, as this year's tragic tornado season tells us.  Of course, given expected societal trends a reversal in statistics would not necessarily mean that our disaster policies are failing.  What it means is that our responses to extreme weather require constant vigilance, investment and continued hard work.

In trying to score points in the debate over global warming Professor Boudreax misses what really matters most on this issue.  And that is why my response to his question, "Do I have any takers?" is "Yes."

I will email Prof. Boudreaux with this post and update with his response.


  1. I don't think he means total deaths. I suspect he's taken the total death toll (less Katrina: 3006) and divided it by 20 to get a 20 year average of 150.3 deaths per year.

    I think he feels that same calculation, done 20 years hence, will be a lower number.

  2. Why does the levee breaking not count?

    Many of the tornado deaths are due to structure collapses that would not occur if better built.

    Maybe the tornado deaths should also be reduced.

  3. Boudreaux's point is exactly about "vulnerability and exposure". He expects it to decrease in keeping with the multi-decadal past trend that you both observe. I think you're "reading in" the notion that he's trying to score any point about human-caused climate change other than that he expects humans to adapt to it should it occur.

  4. "Why does the levee breaking not count?"

    Because the residents and the government of New Orleans had several days to evacuate the city.

    Because the hurricane actually missed New Orleans and hit Mississippi.

    Because the levees were known for years by Federal, State and Local governments to be a potential 'disaster waiting to happen.' Every single death in New Orleans during Katrina can be attributed to human error or neglect.

  5. Roger, as it turns out, I read your blog and Don's regularly. Hope my comment (which by its length must be an oversimplification) didn't misrepresent your views:


  6. The table would be better if you included the column for annual total TF&H. Also, doing 2005 both with and without Katrina, to show the effect of one outlying data point skewing perceptions.

  7. "The fact of the matter is that our vulnerability to extreme weather is increasing, due to a combination of a growing population and especially urbanization in locations prone to extreme weather events."

    The U.S. population is growing by about 1 percent per year, and the population growth rate is expected to decrease in the future:


    You would need better define what is meant by "urbanization in locations exposed to extreme weather events" before that rate of growth could quantified.

    Also, I don't see why "urbanization" is bad (rather than good) in "locations exposed to extreme weather events".

    I'd expect that, for a given exposed population, higher rates of urbanization would lead to lower deaths, not higher deaths (because of the greater ability to quickly identify victims and get them to hospitals).

  8. Professor Boudreax is not trying to "score points in the debate over global warming". Global warming has nothing at all to do with his point. If that is the reason you are taking the bet, then you are betting him for the wrong reason.

    His point is that in societies with market economies people have the means and the resources to adapt to change and that our ability to detect and respond to violent weather will continue to get better.

  9. Roger, I think that you are off base with regards to the purpose of the bet. Conversely, Malachi is spot on. This really has little to do with AGW. It has more to do with the assumption by McKibben et al that climate change necessarily leads to ever more negative outcomes. Boudreaux's proxy for negative outcomes is human casualties.

    Boudreaux is a fan of Julian Simon who made a similar bet with Paul Ehrlich on the scarcity of goods. Ehrlich, the author of the Population Bomb, said that the increase in population will necessarily lead to a scarcity of commodities. Simon, having a much more thorough understanding of Economics and human nature, knew that humans adapt to change. Simon won the bet handily.
    Boudreaux asserts that humans will adapt to climate change regardless of whether it is human induced.

  10. I'm a big fan of Cafe Hayek and Don Boudreaux, but I wouldn't have offered this bet for several reasons.

    I also expect progress in adapting to extreme weather, but I expect diminishing returns to increasing investment generally, and I expect smaller investments in the next couple of decades. I expect smaller investments in lots of things, because I expect an aging population to consume more and invest less.

    Also, extreme weather fatalities presumably are Black Swans, i.e. the distribution of these events is not thin tailed. We can't ignore Katrina for this reason. We can't precisely calculate a mathematical "expectation" of events like Katrina, but we can expect them in a broader sense.

    We can't ignore Katrina because "government is to blame" either. Whether Don (and I) like it or not, government measures are a large part of the human adaptation to extreme weather. Maybe state agencies didn't maintain the levees as they should have, but they did build the levees in the first place. Increasing state involvement in these countermeasures should discourage Don, rather than encouraging him.

    I hadn't thought of Roger's objection, increasing population density in areas prone to severe weather, but this objection also rings true.

    On the other hand, I wouldn't accept the bet either.

  11. Why is Bill McKibben's essay "silly" ?

  12. Having skimmed your book The Climate Fix, I am surprised that you are taking this bet. Some of the research you report there is right along the lines of what Boudreaux is arguing. I don't mean this as criticism, just expressing genuine surprise. If you wouldn't mind, would you be willing to explain what seems--on the surface, that is--to be a contradiction?

  13. -12-James Hanley

    I'd be interested in hearing what you see in TCF that you think is contradictory. Thanks!

  14. Niel West,

    Simon won the bet then, but that doesn't mean commodities will always get less expensive or at least stay the same. I seriously doubt that someone taking Simon's side of the bet starting in 2000 would have won. I've tried to look up the data to see, but as far as I can tell, it's all paywalled.

  15. @Roger Pielke,

    Sorry I can't be more precise, as I'm at home and the book is in my office. But don't you discuss some research in there showing that weather-related events throughout the 20th century weren't causing more deaths?

    Or, as I think more about it, was that research showing that the increasing financial cost of storm damage was due just to things like urbanization and not to changes in storm intensity? If that latter remembering is more correct, then your bet doesn't seem contradictory after all.

  16. -15-James hanley

    Gotcha, yes, there is indeed a dramatic global reduction in deaths from extreme events over the past century (see the work of Indur Goklany on this).

    But even at the global level I would doubt that this trend continues indefinitely.

    Today's NYT has a nice article with some data on increasing urbanization in the US:


    Data here:

    Increasing urbanization in regions prone to extreme events -- all else equal -- would mean that when an event does hit a population center, impacts will be larger than in the past for the same event.


  17. "But even at the global level I would doubt that this trend continues indefinitely."

    It doesn't have to continue indefinitely, just until 2030, which isn't that long.

    The increasing urbanization also brings in more focus for better and more accurate early warning systems. The people will want better built buildings and shelters in case of emergencies.

    I agree that there is a potential for a massive loss of life due to a tornado hitting a large populated area, but that is also the reason why people will keep working to negate the effects through technology.

  18. Received by email from Mark Bahner:

    "Hi Roger,

    I didn't get a chance to respond to this in depth last week, but the bet you and Don Boudreaux have bothers me. I think there is a more-than-50/50 chance you'll win, but I don't think the win will demonstrate at all your premise that, "The fact of the matter is that our vulnerability to extreme weather is increasing, due to a combination of a growing population and especially urbanization in locations prone to extreme weather events."

    So I propose a second bet: let's assume that there are no significant additional deaths from tornadoes, hurricanes, or floods this year. Therefore, the death total for this year would be about 520 (assuming there have been almost no flood deaths, and that there will be almost no hurricane deaths).

    If the death toll for this year is 520, the death toll for the 21 years from 1991 through 2011 would be 4542 (including hurricane Katrina). That's an average of 216 per year.

    I propose a bet that the average number of deaths in the next 21 years will below 216 per year. It would be in the form of an annual bet of $50, rather than an amount paid at the end of 21 years. For example, let's say:

    In 2012, there would 150 deaths. I would win. You would pay $50 to the charity of my choice.

    In 2013, there would be 225 deaths. The two-year total would be 375 deaths; an average of 188 deaths per year. You would still lose, because the two year average was below 216. You would pay $50 to the charity of my choice.

    In 2014, there would be 580 deaths. The three-year total would be 955 deaths, an average of 318 deaths per year. I would pay $50 to the charity of your choice.

    In 2015, there would be 120 deaths. The four-year total would be 1075 deaths, an average of 269 per year. I would pay $50 to the charity of your choice.

    The person who was on-net losing for the bet could quit at any time, but the person who was on-net winning would have to let the person who was losing have a chance to catch up.

    If you are really certain that the determining factors in deaths are "growing population and expecially urbanization in locations prone to extreme weather events" you should be happy to make this bet.

    How about it? Do we have a bet?

    Best wishes,