How Not to Bet

by on June 26, 2014 at 7:10 am in Economics, Science | Permalink

Tim Harford writes of bets:

Pundits who make wagers may look grubby but at least they are accepting a cost for failure. A more subtle advantage is that betting encourages forecasts that are specific and quantifiable.

Exactly, but not all bets are well considered. Consider the following from Christopher Keating:

I have heard global warming skeptics make all sorts of statements about how the science doesn’t support claims of man-made climate change. I have found all of those statements to be empty and without any kind of supporting evidence. I have, in turn, stated that it is not possible for the skeptics to prove their claims. And, I’m willing to put my money where my mouth is.

I am announcing the start of the $10,000 Global Warming Skeptic Challenge. The rules are easy:

1. I will award $10,000 of my own money to anyone that can prove, via the scientific method, that man-made global climate change is not occurring;

…5. I am the final judge of all entries but will provide my comments on why any entry fails to prove the point.

This is a poorly constructed bet. Notice first that the preamble, “empty”, “without any kind of supporting evidence”, “prove,” is simply unscientific bluster. Scientific thinking is Bayesian. (Appropriately, I first saw Keating’s bet flagged on the blog Prior Probability.) In contrast, this talk is suggestive of someone who doesn’t weigh evidence, a point of some importance given that he writes “I am the final judge of all entries…”. Keating is really betting that no one will change his mind and that’s not a bet that I would want to take.

Second, what counts as proof? And what does it mean to say that man-made global climate change is not occurring. How much is man made? How fast is it occurring? What are the costs? What are the benefits? Keating’s challenge is vague and poorly worded.

For an example of a much better bet consider the Caplan-Bauman bet on temperature change over the next 15 years. Caplan has agreed to pay Bauman $333.33 if the average temperature over the period 2015-2029 is more than .05C greater than the average temperature over 2000-2014 as measured by the National Climatic Data Center. If the average temperature increase over that time frame is less than .05c then Bauman will pay Caplan $1000. 

The Caplan-Bauman bet is quantifiable and specific and it contains odds, as it should since Bauman expresses much greater certainty in his belief than Caplan.

Keating seems even more certain in his beliefs than Bauman. But how certain? Will he accept my offer of the same bet at 5:1 odds?

A. Pedant June 26, 2014 at 7:32 am

‘Climatic’

BC June 26, 2014 at 7:56 am

The Caplan-Bauman bet is a step in the right direction, but I think it would be even better for global warming acceptors to make futures markets for global temperature (instead of offering odds). Give their predictions for global temperature change and let skeptics take short positions on temperature. The more the predictions exceed actual temperature outcomes, the more the skeptics will make (and vice versa if actual temperature exceeds prediction).

Forcing someone to make a two-sided futures market is the best way to determine the mean of their true probability distribution. Although in this case, the acceptors will know that the skeptics will want to take the short side, that will be counter-balanced by the fact that the acceptors will not want to deliberately under-predict temperature because that would undermine their case for actions to reduce global warming.

JWatts June 26, 2014 at 9:40 am

+1

Doug June 26, 2014 at 11:00 am

“Forcing someone to make a two-sided futures market is the best way to determine the mean of their true probability distribution”

Agree with most of what you say, but this is not necessarily true. When you make a market, particularly a two-sided market, you’re opening yourself to any counter-believers. Conditional on some other informed party wanting to take a certain direction on the bet your priors will be different. For example I can think the chance of rain tomorrow is 50%, but if some monomaniacal bettor comes and wants to wager thousands that it will be sunny, well then the chance of rain is probably less. And vice versa the other way.

mulp June 26, 2014 at 11:08 am

Warren Buffett and all his corporate peers are making billion dollar bets on the harm of climate change and forcing you to take the other side.

They are the property insurers.

When the property insurers think the odds are really in their favor, that seems to be when the climate skeptics run for government protection, like passing laws stating climate change can not be recognized by government agencies.

In other words, insurance regulators are supposed to deny insurance rate hikes.

Or seeking government bailouts for those who refused to bet with the insurance companies that took the position that climate change is going to cost you.

Brian Donohue June 26, 2014 at 11:14 am

Doesn’t underwriting property and casualty insurance amount to a bet against climate doom-mongering?

Finch June 26, 2014 at 2:09 pm

+1

In the absence of the reasonable proposed futures market, a bet on the reinsurance business is about as direct a large scale bet as you can make against AGM being a major worry. Which, admittedly, is not at all the same thing as a bet that AGM is not a real phenomenon. Buffet is not a dumb guy – he might reasonably believe “global warming looks like a real thing, but it’s coming in under prediction and our ability to cope with it is better than most people realize.” He is on the record saying things about global warming not really having the weather event consequences many attribute to it.

enoriverbend June 26, 2014 at 2:00 pm

“Warren Buffett and all his corporate peers are making billion dollar bets on the harm of climate change”

What? Here’s Buffett in March this year:

“The effects of climate change, “if any,” have not affected the insurance market, billionaire Warren Buffett told CNBC on Monday—adding he’s not calculating the probabilities of catastrophes any differently. ”

“It has no effect … [on] the prices we’re charging this year versus five years ago. And I don’t think it’ll have an effect on what we’re charging three years or five years from now.”

“The public has the impression that because there’s been so much talk about climate that events of the last 10 years from an insured standpoint and climate have been unusual,” he continued. “The answer is they haven’t.”
From CNBC, Squawk Box

Dustin June 26, 2014 at 4:56 pm

Isn’t coastal Real Estate a futures market for global temperature? And as others have said, investments in insurance companies would be another.

jtf June 26, 2014 at 8:16 am

Yeh, even though I’m one of the ones that does place my trust in mainstream climate science that’s a pretty piss-poor bet. Even the great skeptic bugbear/acceptors bible, the IPCC reports, are couched in probabilistic statements and expected values. In fact, there’s an idea: call the IPCC a revenue-neutral bookmaker by using its various degrees of statistical confidence as odds.

ThomasH June 26, 2014 at 1:53 pm

How else should IPCC couch its expectations except as probabilities?

jtf June 26, 2014 at 2:42 pm

The IPCC is doing it right and Keating is doing it wrong.

Engineer June 26, 2014 at 8:22 am

Instead of betting cash, have the AGW proponents make estimates about the climate 3-5 years from now and stipulate that funding for their research depends on the accuracy of their predictions.

Tony June 26, 2014 at 10:05 am

AGW has nothing to do with the climate 3-5 years in the future. Try 30-100 years into the future and then you might have a bet. In the meanwhile, you might go back and read the 1990 IPCC report and note that it is almost entirely consistent with what has happened since then.

JeffJWatts June 26, 2014 at 10:44 am

“you might go back and read the 1990 IPCC report and note that it is almost entirely consistent with what has happened since then. ”

That statement is wrong. Annex of the 1990 IPCC report predicts (figure A10) that is we had capped emissions at 100% of 1990 level that global temperatures would have risen roughly 0.6C from 1990 to 2014.
https://www.ipcc.ch/ipccreports/far/wg_I/ipcc_far_wg_I_annex.pdf

Instead, global emissions have risen substantially since 1990 levels. However, adjusted temperature data shows a rise of less than 0.3C during that period of time. Furthermore, it’s been roughly flat for over a decade.
http://upload.wikimedia.org/wikipedia/commons/f/f8/Global_Temperature_Anomaly.svg

Tony June 26, 2014 at 11:39 am

How the heck do you get “less than 0.3C” from that graph? Fit a line and from 1980 to 2014 the overall trend is at least 0.5C.

XVO June 26, 2014 at 11:59 am

Because if you could read the graph you would see that in 1990 the 5 year mean was at .3 and then in 2010 it was at .6….. .6-.3=.3

Tony June 26, 2014 at 12:39 pm

I misspoke in the previous comment but the prediction is still pretty close to observed. The prediction graph shows a rise of about 1.9C per century. The actual temperature graph shows a plain as day trend over the past ~50 years of 1.5C per century. That’s not “uncomfortably wrong”.

Dan Weber June 26, 2014 at 12:49 pm

I think the predicted number was 0.5C, not 0.6C ( http://i.imgur.com/HDs1KVV.png ) while CO2 output has gone up roughly 50%, although it was kind of flat for the 1990s ( http://www.energytrendsinsider.com/2012/07/02/global-carbon-dioxide-emissions-facts-and-figures/ ) as Russia deindustrialized.

What were the error bars on the IPCC’s prediction? If they had to hit on the dot they definitely whiffed, but I don’t think it was that. OTOH, if they had 95% confidence the temp rise would be, say, between 0.4C and 0.6C, then it’s fair to say they were wrong.

So, what were the error bars?

Finch June 26, 2014 at 2:46 pm

> So, what were the error bars?

The chart doesn’t show any error bars, though that might have been your punchline.

My knowledge is relatively shallow and I have been a mild global warming believer, but if I remember The Economist correctly we are at or below the low-end 95% confidence interval for a broad swath of the major models. There is considerable rejiggering going on with the latest concern being that the proposed corrections over-correct.

The Other Jim June 26, 2014 at 8:46 am

I think it’s simpler just to point out that it’s impossible to prove a negative.

Also, the burden of proof is on those who assert the theory. It’s absurd to claim AGW is real, demand hundreds of millions of dollars a year to “fight” it, and then say “Well, you can’t prove it ISN’T real.”

Might as well be fighting the giant space unicorn floating behind Mars, out of view. (Or solar fluctuation, for that matter.)

Chip June 26, 2014 at 9:20 am

Exactly.

With the vaunted models becoming so uncomfortably wrong, it’s almost as if climate scientists are suddenly realizing that they can not in fact take a small sample of data from an incomprehensible complex planetary climate and know what is going to happen in the future.

So, confused and unsure, they throw their theory back at skeptics and say (we can’t prove we are right so) prove we are wrong.

Tony June 26, 2014 at 10:11 am

Please show us what models were “uncomfortably wrong”. Svante Arrhenius first calculated a value for climate sensitivity in the 1890s, and by 1906 had settled on a value that is consistent with what we believe today. No computers, no climate lobby, and a belief that human activity couldn’t possibly burn as much fossil fuel as we actually have. The resulting temperature increase was pretty much what he expected it would be, it just happened much sooner. I’d call that “uncomfortably correct”.

http://en.wikipedia.org/wiki/Svante_Arrhenius#Greenhouse_effect

Baphomet June 26, 2014 at 2:26 pm

Oh, you know about the unicorn, do you? Well, we still have other secrets.

Z June 26, 2014 at 8:53 am

I think you’re missing the point. The bet is not intended to be taken up by anyone. What he is creating is a weapon to wield against the Cathars. “Why don’t you put your money where your mouth is?!” Any attempt to explain why it is a sucker’s bet will either be ignored or come off poorly. Of course, he will attract some lesser lights that he can easily dismiss, thus “proving” he is right and the heretics are wrong.

lemmy Caution June 26, 2014 at 9:41 am

not a bet

andrew' June 26, 2014 at 8:57 am

It is perfectly worded if your objective is to be a hack.

Andrew' June 26, 2014 at 9:02 am

Like this: Prove damage from global warming. See? Mine is even better. In 100 years I will definitely win while he could lose.

Brian Donohue June 26, 2014 at 9:29 am

Yup. I don’t know this Chrisopher Keating guy, perhaps he is a Very Important Person, but this bet is so poorly-constructed in so many ways as to be unworthy of critique even.

beamish June 26, 2014 at 9:02 am

It’s a fine bet for his rhetorical purposes. It may be based on one offered by Joseph Mastropaolo to defend creationism, though Mastropaolo is a little less crude in his selection of a biased jury.

Is your five to one offer open to the public?

Z June 26, 2014 at 9:18 am

I was going to mention this in my response, but I could not recall the name. It is telling how the AGW folks have appropriated the language and tactics of the religious. Anyone familiar with the Hebrew Bible would recognize the dynamic, but that’s a dwindling number these days, particularly amongst those debating AGW.

beamish June 26, 2014 at 9:38 am

It’s just like when Solomon’s wives turned his heart towards foreign gods.

T. Shaw June 26, 2014 at 9:07 am

Keating is a glib idiot. That “bet” is an insult to our intelligence.

To quote a far-more intelligent public figure, “I’m insulted!”

cfh June 26, 2014 at 9:10 am

Keating is offering a prize, not a bet.

lemmy Caution June 26, 2014 at 9:40 am

Right. No one would accept a bet on Keating’s terms.

JWatts June 26, 2014 at 9:46 am

Yes, and “2. There is no entry fee;”.

However, with the structure of the ‘prize’ there’s no reasonable expectation he’ll ever pay any money out. And since, at least some people will submit an entry, it’s mostly a ploy to bash skeptics over the head.

derek June 26, 2014 at 10:39 am

How about a bet that any decreases in carbon emissions in the US is due to natural gas discoveries and low prices, as well as the export of manufacturing elsewhere.

How about a bet that carbon taxes will increase the use of firewood for home heating? I’m seeing evidence of that around me.

Or another. A local college got a bunch of money to upgrade their heating plant, installing a new boiler and water to water heat pump system. I would bet that the carbon used in manufacture and installation will be offset by carbon saved at about the same time as the equipment is due for replacement the next time around.

Chris S June 26, 2014 at 5:21 pm

“carbon taxes will increase the use of firewood for home heating”

Sounds like nothing more than a carbon-tax enforcement issue.

Tom Donahue June 27, 2014 at 4:11 pm

Well, it took a little digging, but I found the answer to your bet. You lose.

In general, the embodied energy of a boiler is about 6% of total life-cycle energy use. Meanwhile, a modern boiler is about 33% more fuel efficient than an old one. You can put those two facts together for the answer.

But I’m just curious here. What was your estimate of the energy embodied in a boiler? Did you stop to consider that a boiler is a machine for turning fuel into heat, which is about as energy intensive as it gets? It wouldn’t take that much for your local college to come out ahead.

Hook June 26, 2014 at 10:50 am

Keating’s offer is similar to saying, “I will give $10,000 to anyone who convinces me to give them $10,000″ which is actually an offer that anyone with $10,000 is offering all the time.

T. Shaw June 26, 2014 at 11:00 am

Keating should do his compositions before 8AM while he is sober.

mb June 26, 2014 at 11:49 am

This is the most bizarre post I’ve seen here. Alex takes something which is clearly not meant to be a bet, bashes it as a “poorly constructed bet” and proceeds with a discussions of good and bad ways to do bet on various things.

Urso June 26, 2014 at 11:54 am

Tabarrok’s got bets on the brain recently. “A tax on bullshit!” he cries. Not much of one. If Caplan is wrong he’ll pay a whopping $333, in 2029 dollars, effectively amortized over 15 years. I’m guessing that won’t exactly break the bank. And I guarantee that whoever loses still won’t admit he’s wrong, he’ll just say, as yes I was wrong about THIS TIME but just you wait 15 more years, you’ll see.

mb June 26, 2014 at 1:35 pm

Well, there is nothing wrong with a well-designed bet even if the tax on BS might be on the low side. However it is odd to criticize a bull for low milk production.

Jeff June 26, 2014 at 1:15 pm

Keating is betting $10,000 that no one will prove that man-made global warming is false.

Sean Kelleher June 26, 2014 at 12:36 pm

If I was a well informed global warming skeptic who could rapidly assemble a decent power point deck on the topic, I would take Keating’s bet. I would be exchanging minimal effort for the possibility of a 10K windfall. Even if I lost, I would gain a good story to share with my skeptical friends.

triclops41 June 26, 2014 at 1:16 pm

This challenge is either trolling, our Keating knows literally nothing about the scientific method. Asking someoneg to prove a negative with the scientific method would get you an F in high school science.

Nathan W June 26, 2014 at 1:46 pm

99% chance of outcome A.

He’s saying “If you’re stuck on the 1% probability of no outcome, make your case.”

Science is no longer Newtonian. More like Heisenbergian. But if you didn’t take chemistry or physics beyond grade 11 or 12 there’s a pretty good chance you have no idea what I mean.

Science routinely applies probability, and then relies on human discussion of methods, etc. to figure out the intersubjective truth of the matter.

For example, we pay more attention to people who propose that we should act on things of high cost and 99% probability than focusing on the 1% probability that they are wrong.

JWatts June 26, 2014 at 2:20 pm

He’s asking someone to prove a negative with him being the final arbiter on whether they have or not. It’s a fools errand, designed to bash skeptics over the head with, since no one will ever be able to claim the prize.

andrew' June 26, 2014 at 1:33 pm

The sun is doing it.

Where do I collect my $?

Nathan W June 26, 2014 at 1:43 pm

If there is a 1% chance of a $100 trillion cost and $1 trillion of spending can avert that outcome, then a risk-neutral individual would invest $1 trillion to avert that outcome.

If there is a 99% chance of a $10 trillion cost, and $1 trillion of spending (say, in wind and solar) can avert that cost, then … all the biggest lobbies in the world get together to make sure we never doing anything about it???? Wait, something doesn’t make sense here.

enoriverbend June 26, 2014 at 2:10 pm

“If there is a 1% chance of a $100 trillion cost and $1 trillion of spending can avert that outcome, then a risk-neutral individual would invest $1 trillion to avert that outcome.”

No. A risk-averse individual might.

A risk-neutral individual would have no preference on that bet, as long as the individual didn’t take into account the time value of money. Taking into account the time value of money, if he had to put up $1T now to avoid a 1% chance of $100T later, he wouldn’t put up the $1T.

Chris S June 26, 2014 at 5:23 pm

Finally, someone who passed finance 101.

li June 27, 2014 at 1:42 pm

And don’t forget if MY risk is practically static (whatever its level) but someone 50 years from now could have a significantly different (reduced) risk for every dollar I spend now, I’d still rationally spend nought.
And then we can transform the argument to the cost (of intervention) as a % of GDP per Capita. Shall we project what some person’s GDP will be 50 years hence and figure which is a more “fair” strategy: spending now or spending later? The policy argument continues to be confused (imho) with the scientific debate. Obviously the policy has to be (partially) based on the science, but I know of no scientific policy. We’re all dead in the long run – so eat, drink, and be merry! (The future will take care of itself (or not)).

Rusty Synapses June 26, 2014 at 2:36 pm

If you want to bet about something (and I agree it’s not really a bet or even much of an offer), why not just pick sea level? Very easy to measure.

mofo. June 26, 2014 at 4:05 pm

Because that is easy to measure. Betting (or whatever this is) on a measurable phenomenon means you could lose. Betting that no one will change your mind is risk free, all he has to do is say “nope, im not convinced” no matter what is said.

Finch June 26, 2014 at 4:19 pm

FWIW, I don’t think sea level is considered terribly easy to measure.

jdm June 27, 2014 at 11:32 am

On what basis did Caplan make this bet?

li June 27, 2014 at 2:32 pm

What a very poor example to use. The Caplan-Bauman bet is based on the continued existence of the “same” time-series (of average surface temperatures) for the next 14 years. LOL! Man plans and the gods laugh. Recently (a month or two ago) researchers (Science magazine (AAAS)) published their report that Arctic temperatures had been underestimated and needed upward revision. The reason what the paucity of samples and the incorrect “smoothing” algorithm used to extrapolate temperatures in the Arctic. If this is correct, it is virtually certain that the time-series going forward will be changed. Perhaps I missed it, but while the data set might seem to be adequately defined (in my opinion it is not) what “average” means is NOT defined. It seems it means the average 2028 global temperature vs that of 2014 (or 2015?). This is NOT a bet on climate! It is a bet on weather. If you don’t know the difference, find out. One of the major issues offered by the climate science apologists is that SST is NOT a good measure of climate change. (sea surface temp) and that the deep ocean must be considered. If this turns out to be true, then almost certainly the current time-series will be discarded for better measures. There are also a number of serious and probably uncontrollable problems with the data in the time series (lack of calibration, arbitrary data smoothing algorithms (used on the raw data BEFORE inclusion in the time-series) and the global trend in local environments (who here believes that the ratios of wild, agricultural, and urban land will remain fixed over the next 15 years??). Many of the samples are taken “under the street light”, although there are efforts underway to correct that bias. The data itself is described as “a blend” of two different data sets, man you have so many different groups handling this that it is delusional to expect that a comparison of single year averages will mean anything. Is this really the best social scientists can do in finding examples of good bets? Why not something reasonable like “the USA will not win the 2014 World Cup”? (or is it too late to make that bet?) Or wheat production for the USA in 2019 will be less than 2014 on a grams of total edible protein basis.

li June 27, 2014 at 2:52 pm

Sorry for the typos.
BTW, Anthropogenic global warming is almost certainly correct. The alternative is that the atmosphere will be modified so that more or the same energy is emitted/reflected as has been true with lower [CO2]. I am quite certain (I’d bet on it!) that the biologists/environmental scientists have either no models or completely inadequate models on the bioaerosols which would most likely change atmospheric optics and result in a cooler Earth under a higher [CO2]. (I ignore the possibility that the Sun’s emissions will be reduced). What does the IPCC have to say on that? Take a look! My argument is simple: put take a grey box and paint it black and it will absorb more heat (ceteras paribus). The climate models seem to be wrong. No one should be too surprized that the “true believers” don’t care and want to tax Civilization into the stone age. Separate policy (what should we do about it?) from the science (what is it we are talking about?). I still like the idea of a carbon tax: start it off at $0.01 per megaton of C emitted and double it each year that the average is 3 s.d. is above the 20 yr average. (and halved in those yrs it is 3 sd under). (Or more appropriate time spans, tax basis, and multipliers, I haven’t analyzed this scenario…)

Tom July 1, 2014 at 6:38 am

What the “not even wrong” of betting?

Comments on this entry are closed.

Previous post:

Next post: