How Not to Estimate Losses from Climatic Disasters

In re: Hurricane Maria, what do we know a bit over five years later?

Basically, don’t shoot from the hip if you don’t know statistics, and can’t distinguish survey data from administrative data, preliminary from revised. Consider the following graph.

Figure 1: Cumulative excess deaths from September 2017, for simple time dummies OLS model (blue), OLS model adjusting for population (green), and Quantile Regression model adjusting for population (red) all based on methodology in this post, Milken Institute point estimate (black square) and 95% confidence interval (gray +), Santos-Lozada, Howard letter (chartreuse triangle), Cruz-Cano and Mead (pink squares), Kopits “firm” maximum (teal triangle). Light green band: Kopits estimate of 300-400 for October 2018. Source: author’s calculations, Milken Institute (2018)Santos-Lozada and Howard (2018)Cruz-Cano and Mead (2019), and Kopits (2018a), Kopits (2018b).

From Sandberg, et al. (July 2019) in Epidemiology:

Mr. Kopits’ ire was focused on the Kishore study; while the point estimate for end-2018 (4545) is higher than the others, the confidence interval encompasses the lower estimates. Mr. Kopits’ ad hoc range of 5/31 encompasses no other. (It must be admitted that Mr. Kopits did not at the time understand what a confidence interval was).

18 thoughts on “How Not to Estimate Losses from Climatic Disasters

  1. Moses Herzog

    Just as I predicted, Kopits took down his March 2021 post from his blog. I knew 100% that he would take the entire post down, which is why I copy/pasted his exact words in this comment in the November following that same 2021 summer:

    “This is exactly as we noted in our analysis three weeks ago:

    If the trend holds up, the only barrier between us and $100 oil after Memorial Day will be the mood of Vladimir Putin and the goodwill of Saudi Prince Mohammed bin Salman. Both their treasuries are bare. They will be looking to refill the coffers, and not only that, but to buttress their positions against a Biden team less friendly to poisoners and ax murders than the previous administration.”

    …….. and continuing farther down:
    “At some point, of course, US operators will take the bait. But too late. The Saudi decision to extend the 1 mbpd cut indefinitely can be taken as a declaration of intent — indeed, a thinly veiled declaration of war on the Biden administration — by the Kingdom, and by extension, the rest of the OPEC+ cartel. They are going to keep pushing prices up relentlessly. Pencil in a $10 / barrel rise per month. At that pace, oil prices could reach $100 / barrel during the course of the summer. That’s the message the Saudis and Russians want to send to the Biden administration: ‘Look who has the leverage now.’ ”

    “Gasoline prices have already crested $4 / gallon in California and are flirting with $3 / gallon for regular on the East Coast. It will get worse. Possibly much worse.”

    “And of course the Fed will have to raise interest rates right into the meat of the stimulus program and a still lingering pandemic.”

    “It will get ugly.”

    Kopits’ link no longer works——>>

    It was verified as a 100% slam dunk for me when Kopits’ “brother from another mother” in Harrisonburg VA defended Kopits, thereby verifying my market call it wouldn’t hit $100 in the calendar year of 2021.

    Barkley said: “At the time I made it clear I viewed Steven’s forecast of the crude price hitting $100 per barrel as not all that likely but not at all impossible. I then also made my point that $40 per barrel was also not out of the question, while not all that likely. You found both of those possibilities as just completely ridiculous, beyond the realm of possibility, despite the history I pointed out of much larger oil price changes happening multiple times in the past in much shorter periods of time.”

    It later skipped over $100 in 2022 after Russia’s invasion of Ukraine, a war, which, Barkley Rosser Jr. also predicted would never happen.

    1. Menzie Chinn Post author

      Moses Herzog: I don’t recall Dr. Rosser saying the invasion would never happen. I do recall that he had seen reports that suggested no invasion around the time of “exercises”. If you can point out the comment where he said “never”, I’d be much obliged, and I would revise my opinions.

      1. Moses Herzog

        Well it’s difficult to find, but I know round about the timing, because it was near to the second invasion (after the first into Crimea). It was also very near to the time (possibly??) the last time you saw him in person because I think he was giving a talk at your class or something or other. He pretty definitively said it. There were portions of his personality I strongly disliked, but I wouldn’t lie about the man. I’ll hunt it up here with my morning coffee and Chinese style fried bread and try to have it up before noon.

      2. Moses Herzog

        “Do keep in mind I am the one here with access to Russian media. That has now been blaring for several days that the troops will go home after the exercises are done, and exercises are exactly what they are doing now. This has more recently been reinforced by statements from Putin in press conferences, such as the one just held after the visit of German Chancellor Scholze.”

        “There is not going to be an invasion, even if some of the details of what Zelensky and Ukraine may agree to are not fully settled, and Victoria Nuland has been shooting her mouth off too much, somebody I wish was not part of this administration.”
        “I have just heard from my wife that in Moscow it is being reported that Putin has announced the end of the “successful” exercises in Belarus and that troops will be withdrawn. I suspect a deal was cut on Saturday in the Biden/Putin phone call in which it was agreed that the US advisers in Ukraine would be withdrawn, which is happening right now. Russian media had gone on and on about US troops supposedly being in Ukraine and threatening an invasion.”

        One of the many fascinating things about Barkley Rosser Jr. was…… how if “you” (any person) stroked his ego just a mild amount you became a heck of a guy literally overnight. You’ll kindly notice in that same thread Kopits gushes like a teenage girl going to the prom about his interview with Steve Bannon. Barkley, the self-proclaimed hero and demi-god of racial issues was “oddly” silent on Kopits’ pleasure in his Bannon get-together.

    2. pgl

      Steven Kopits recently told us that collusion to raise the price of oil was AOK. I wonder how Barkley would have responded to the latest nonsense from the world’s worst consultant.

    3. pgl

      ‘the only barrier between us and $100 oil after Memorial Day will be the mood of Vladimir Putin and the goodwill of Saudi Prince Mohammed bin Salman. Both their treasuries are bare. .. Pencil in a $10 / barrel rise per month. At that pace, oil prices could reach $100 / barrel during the course of the summer. ‘

      Given the fact the Brent oil was selling at less than $70 a barrel on Memorial Day 2021, I guess the mood of MBS and Putin was generous to the fault. Pencil in – why pencils? So Stevie can erase his stupid forecasts when reality bites this troll in the rear end. Why does Stevie forecast? To make retarded monkeys look good.

  2. Ivan

    Pgl previously pointed to the FTC report on shale oil producers collusion. Here is a link to a nice digested writeup on it.

    It may have accounted for as much as 25% of recent inflation. Oil is going into the prices of almost every product, so when you inflate oil you inflate everything.

    You are not going to hear that story on Faux news. Nor will you hear how Trump told big oil that they needed to give him a billion if they want to be free of any oversight or restrictions (on pollution, price fixing, etc.).

    1. pgl

      That discussion added a lot of useful information. I also enjoyed the picture of Summers looking a little goofy over the inflation tie.

      Notice something – as soon as we started presenting actual facts into this collusion, Princeton Stevie boy decided to run and hide. I guess he will not be Sheffield’s defense attorney after all.

  3. baffling

    I recall telling Steven at the time, simply look at the example of hurricane Katrina to get a reference point for what hurricane Maria may do. Katrina hit an urban area. Maria struck areas that were not rural by any stretch. it simply was not fathomable that you would expect only a fraction of the deaths from Maria, a much stronger storm, than Katrina, given they both impacted some populous locations. the low numbers steven initially proposed were simply preposterous. analysis without reference to any data whatsoever. to be taken seriously, because, “do you know who I am?” mentality.

    1. pgl

      “do you know who I am?”

      I have this line a lot in my professional life. It is usually after I took some advocacy “analysis” to the woodshed and the “expert” witness for the other side tried to pull rank with me. I often wanted to reply – “yea, a lying idiot” but fortunately refrained and stuck to the real issues. Yea – the real world has way too many arrogant morons. Yet Princeton Stevie pooh somehow managed to be the worst of all time.

    2. Macroduck

      I’ve seen the foundations of the “do you know who I am?” trick from the other side. Here’s how we did it in client meetings:

      – Sales guy introduces me – always.
      – Grey at the temples, in a suit, and I get a whole side of the table to myself.
      – I slip “the last time I was in Beijing/London/Ottowa” into the conversation, no matter how long ago it was.

      Now they “know who I am” even though none of that means the analysis is right. It’s embarrassing how it works.

  4. Steven Kopits

    Given that your time series ends in 2018, I don’t think we know losses five years later.

    Btw, it’s a weather loss, not a climate loss, unless you believe the Caribbean wound never be hit by Category 5 hurricanes before or after.

    Finally, the Harvard study was complete garbage, which was knowable at the time. If your data or methodology is garbage, so are your confidence intervals with respect to real world events (although you can certainly run the numbers through a stats package).

    You seem to love statistics for statistics’ sake. I could care less. I care about statistics with respect to action, behavior or policy. Does a confidence interval, or any statistical measure, change what you would have done or should do? If it doesn’t, then it’s a mere artefact, a curiosity, but of no use to people making real world decisions.

      1. pgl

        Only 400 deaths? My understanding is that as many Puerto Ricans died from Maria and New Yorkers died from the 9/11/2001 attack on the World Trade Center. In other words, almost 3000 deaths in both cases.

        Hey Stevie – come to NYC and tell any of the residents that only 400 people died on 9/11. I dare you.

    1. baffling

      its a climate loss if the Caribbean is more susceptible to Cat 5 storms due to climate change.

      “Given that your time series ends in 2018, I don’t think we know losses five years later.”
      well, unless those folks come back from the dead, its not going to get any better Steven. you are trying to raise doubt with a comment like that, to distract from your poor prediction to begin with.

      “You seem to love statistics for statistics’ sake. I could care less.”
      if you are going to present statistics in your argument, then you should understand what they mean and imply. if you are not willing to understand that, then you should not present or comment on statistics. very simple.

      “If it doesn’t, then it’s a mere artefact, a curiosity, but of no use to people making real world decisions.”
      that is only true if you do not understand how to use statistics. use a different tool, if you can find one of value. but don’t denigrate statistics to cover up your ignorance of the topic, Steven. simply develop another tool that is useful and acceptable by others. wouldn’t that be a worthwhile contribution by you to society? or is the task too difficult?

Comments are closed.