Asymmetrical Information

"Which is to say that if you survey a large group of people on a large number of factual questions, and take the answer that is chosen by the most people for each question, the group is generally right more often than any of its members, even the best performing ones."

Hmm, this must be qualified to apply to situations in which there is a lot of private information that is efficiently aggregated by the market, right?

Because it sure is not the case that if we asked a large group of people a large number of questions on theoretical physics and analytical chemistry we would find that the mode answer is anywhere close to being "..right more often than any of its members, even the best performing ones."

I can think of two reasons why this would work. The first is, like other futures markets, people who participate in it will study the subject. Instead of 500 CIA analysts the terrorism could then have 500,000 analysits all trying to research the groups, politics, motivations etc. This would be a very efficient way to bring a lot of brain power to the problem without having to hire full time analysists who may become corrupted by group-think and other dangers of large organizations.

The second is that groups associated with terrorists will be tipped off so they could place their bets early. Limiting the payoffs to the economic carrying out the attack would not be very useful. Such a policy would be a price control which would only distort the market thereby hurting the reason behind it. You are hoping a market will provide you with answers. More importantly those terrorists could have the usual ideological reasons for engaging in terrorism *but* they could also tip off friendly groups. The market would become a tool for those who are friends to terrorists to be paid off.

Finally another problem is what if some non-terrorist came upon a 'sure thing'? Say he aquires intelligence unavailable to anyone else. The incentive of this market would encourage him not to report the pending attack but to place bets that it would happen. Yes the rise in 'attack' futures would signal a possible attack to the authorities but at the cost of literally paying someone to *not* reveal information that may stop the attack.

Sadly, it's been shut down due to complaints, some valid, some not.

One issue is diplomacy: if the head of some state was killed and the US paid out for it, there'd probably be some hard feelings about it. I wonder what restrictions would have been placed on possible bets. It's funny really - when Jim Bell came up with the somewhat similar Assassination Politics, he got a lot of unfriendly federal attention.

In addition to the benefits of market knowledge, such a system will provide an incentive for would-be terrorists to give away their plans. The market would amount to a public standing bribe to give the authorities advance notice of an attack. If any member of a team succums to the temptation of easy money (perhaps for their family if a suicide attack) then this market system will detect that trade and tip us off to the event.

While military has shutdown this program, there is really nothing to stop a Thinktank orginization like Rand to do something similiar, albeit, with the roster limited to those who are in the business.

First, the original idea was to limit the market to so-called "experts": This solves the "theoretical physics problem," as well as making it highly unlikely that the terrorists themselves would enter the market. Second, it is by no means clear which way the terrorists would bet. They could short the market in order to hedge their plan. If the plan works, I am happy. If the plan fails, I have the money to carry out the next plan.
Third, whether or not it would have worked would have depended on how seriously the "experts" took it; my guess, FWIW, is at the stakes being mooted, it would accomplish little.
Still, to the extent that it differed from the official intelligence assessment, it might have been interesting.
Finally, whether or not it was a good idea, I heartily agree that the political grandstanding was idiotic.

Before the 9/11 attacks there were apparently some unusually large put purchases in several airline stocks. I mean extremely large and, obviously, quite suspicious looking after the fact. I was trading on an options floor at the time and the story was that these trades were being investigated after the attacks.

It is quite likely terrorists or those who are aware of terrorist plans are already using the capital markets to profit from their self-generated events. Adding a futures market does not really add an incentive that is not already there.

well, then, how about a SETI-type distributed application, sort of an AL-QUAIDA-AT-HOME?

This is all solved by making it a no-REAL-money market, like the ones currently played for fun by people all over, for movies and the like.

Yes, the terrorist could short the stock (or buy put options) on the target of the attack, but the terror futures would be a better "pure play," especially if the target is a governmental or non-profit (or non-traded corporate) entity.

Problem: presumably the individual investors are of differing wealth and theoretically have different utility scales for investments in the market. If this is the case, we would see different levels of investment in the market by different players, even given that all have the same information. As all presumably do not have the same information, this creates a problem of distortion: good information input does not necessarily correlate with an ability or incentive to invest in proportion.

Sigivald: making it a no-real-money market eliminates the incentive for accuracy provided by real money.

Re: the political grandstanding, when did the term "progressive" start to mean "opposed to anything new"? Cause that's what it seems to mean nowadays.

"... if you survey a large group of people on a large number of factual questions, and take the answer that is chosen by the most people for each question, the group is generally right more often than any of its members"

Clearly Jane rarely bets on Professional football against the line.

"I think you ought to end the careers of whoever it was thought that up." - Barbara Boxer D-Calif

Wow, we're going to be hearing about this one for a long time. From people who are motivated by an underlying hatred of capitalism, mostly.

Ten grand says it supplants 'yellowcake.'

Too bad we may never know the details of this proposal, it'd be really interesting to see how they were going to try to make this work.

To me, the deepest problem with a terrorism futures market has to do with the complexities that would arise from its interaction with law enforcement, and the way that might distort traders' analyses. Consider -- by virtue of brilliant analysis or leaked information, a small group of noncomplicit people correctly guesses that an event is planned. They bet heavily on the event. Law enforcement watches the market reflect this, intervenes, and stops the event. The smart traders lose their money. The market has to estimate not only the probability of an event a priori, but also the probability that law enforcement will intervene, which in part depends on the market itself. All this would make it very difficult to discern exactly what the prices various contracts would reflect. If the market were efficient, it would have to be useless to law enforcement, which would defeat the larger point of having it. I'm not sure how this kind of problem would be resolved.

Re: direct perverse incentives - all participation in such a market could be public and non-anonymous. Parties might even be asked to agree ahead of time to waiving some rights and permitting increased financial scrutiny in the case of significant payoffs so that money trails can be followed. Such publicity and scrutiny would have the adverse effect of discouraging some participants who might have good information or analytical skills from participating, but this would be outweighed by the advantage in discouraging staged mayhem. Scrutiny would work against insiders whether they were hedging against a failed plot or betting confidently on pulling off their nastiness.

Re: diplomatic fallout - just another reason this kind of thing would be better done by the private sector. Also, it shouldn't be the market orgainzers who determine the content of the futures traded. Market participants should determine what events they wish to base their contracts on. The market organizers should only devise standard forms of contracts that would be appropriate and enforcable, and then organize markets in the particular contracts participants create by giving them standard symbols and providing an infrastructure for convenient and public electronic trading. Given the heterogeny of contracts that would likely develop, no particular foreign government or official would likely have grounds to complain, unless the betting moving strongly in a bad direction, in which case the complainer would have more important things to worry about.

I thought it was a very good idea but it apparently wasn't vetted at the political level.

The value of the idea is self-evident. A futures market would offer Bayesian prior probabilities. These priors could be plugged into Bayes nets for war games, decision analysis, etc.

Criticisms about direct manipulation of the market by hostile forces are weak because they assume the purpose of the market is to use the raw probabilities as naive estimators of the probabilities of various international events.

What the critics miss is that Bayesian priors interact with EMPIRICAL OBSERVATION, yielding an improved posterior probability distribution. The mathematics are self-correcting. We can't confuse probability and likelihood.

p(x | D) := p(x) p(D | x)

Say Bin Laden bet that there would be an attack in the Phillipines in September. Sure, he could drive p(x) the Bayesian prior up. But if intelligence sources in the Phillipines indicated that terror cells appeared demoralized and weak,
p(D | x) would have a very small value; in other words, it is unlikely that terror cells would be demoralized and disbanded conditional on imminent attack, so the posterior probability of attack would converge downward toward accuracy.

I was sick watching members of congress criticize this proposed system, because they are so utterly incompetent to understand, let alone evaluate the proposal; oh, but they were filled with moral outrage!

I'm tired of being preached to by jackasses who have the gall to mock the intelligence of the President when they themselves are clearly illiterate on matters of math and logic, as was demonstrated yesterday.

Uh, before you call out too many people on their math and logic

p(x | D) := p(x) p(D | x) ?????

Jeff,

"I was sick watching members of congress criticize this proposed system, because they are so utterly incompetent to understand, let alone evaluate the proposal; oh, but they were filled with moral outrage!"

The particular member of Congress who most makes me laugh is Sen. Dorgan from North Dakota. Coming from a state whose biggest industry is wheat production, you'd think he'd know something about futures markets, wouldn't you? Or does the DLC playbook tell him all soccer moms are ignorant about futures markets, so he should act accordingly?

The idea of Congress being ignorant about futures markets was especially humorous to me after I spent the past weekend skimming all 300 pages of 7 USC Chapter 1, the Federal statute regulating futures markets.

Personally, I loved the idea. This was very close to being a very large scale delphi study. The delphi technique has been around for a long time and has proven quite useful. Basically, you ask a lot of people who know something about a subject some questions. Then, you feed their collective answers back to them and ask again. You can go around as many times as you like. Strangely enough, while no single person knows what's going on, collectively, we often have a pretty good idea of what's going on. Search Google for "delphi study" and you'll get lots of hits.

John Brunner added betting to this in his novel The Shockwave Rider back in 1975. His fictional future world used "delphi pools" to predict all sorts of numbers useful in making public policy (eg, how much grain will we produce next year? How many people will die in plane crashes? What will be our GDP?). His delphi pools sounded an aweful lot like this Pentagon program. That book was prescient in so many ways...

One again, misplaced political correctness tanks a good idea. We need a national awards program (sponsored by someone big who could really put some money into it and judged by website voting or something) to give trophies each year to the public officials who do the stupidest things in the name of political correctness.

C'mon. This was a ludicrously silly idea. Embarassing, really.

To suggest that a market of professionals would have more insight into the probability of terror attacks than our intelligence services is truly an insult to the CIA and FBI. Relying on the 'invisible hand' works in securities markets, but terrorist risks are one-off events that can be influenced by insiders (terrorists).

This market would only be useful for traders hedging their terrorism peril to their porfolio positions. In no way would it improve intelligence. To suggest so would be to imply that these traders could have inside information and withhold it from the government in order to profit from it. Or more ludicrously, that a party could deliberately plan an attack in a quest for profit - which would mean that this market could be further incentive for potential terrorists.

No thanks, I'll leave my trust the CIA and FBI, in spite of the yellow-cake and 9/11 blunders.

Perhaps we have our first entry here. Then again, if wallster isn't a public official, we should consider expanding the eligibility for the Award Rob proposes.

Uh, before you call out too many people on their math and logic

p(x | D) := p(x) p(D | x) ?????

Careful there achilles. If Jeff was intending for ':=' to be "proportional to" then he is indeed correct (although you could chide him on not explaining what he meant by :=).

The bigger problem is that efficient market theory, while a neat theoretical package, isn't that highly regarded anymore. After all it cannot explain bubbles.

Still I think this program was interesting and did offer some potential. Now, we'll never know.

Killing this program also is not strictly speaking without potential costs. The idea was to use it to predict terrorist attacks. If it turned out to be a good predictor then it could have been used to attempt to thwart terrorist attacks, and thus save lives.

It is similar to the cost the FDA imposes.

"Sure that drug could save your life...or it could kill you. Since we don't know, we aren't going to let you have it, so you will certainly die. Have a nice day."

Since the Professor seems to have gotten the InstaKnickers in a real twist over the demise of DARPA's Terrorism-Futures market plan, I thought I would take a ride on the Jane Galt Line to see what economics-savvy posters thought of the plan (after a station stop at Brad DeLong's).
Unfortunately, this issue seems to have become an ideological train wreck: most of the post/comments I have seen have fallen into just a few pat categorizations:
1) Brilliant innovative idea foolishly aborted by posturing politicians.
2) Idiotic notion concocted by morally blinkered and market-obsessed ideologues.
Personally, I think the real "issue" has been clouded by altogether too much rhetoric (pro and con); but the real question should be, IMO, WILL IT BE EFFECTIVE?
Because, unfortunately, as anyone with any experience in the field will instinctively know - MARKETS ARE NOT ALWAYS RIGHT!
If the underlying plan behind the PAM was to expand the field of terrorism probability analysis to encompass more players and to provide a "money where one's mouth is" incentive to be more rigourous in one's analysis - ie, correct: then, maybe, the basic plan might have some merits.
However, the way it was presented by the Pentagon (thank you, Adm. Poindexter!) - it has been all too easy to characterize it as a sort of "lottery of death" designed to give give big-buck payouts to savvy insiders in case of terrorist disaster.
But, then, what is the purpose of anti-terrorist intelligence? One would hope that it would be the timely detection, and, God willing, PREVENTION of deadly attacks - not just the maintenance of a "betting book" on their likelihood. And, since most terrorists make it a point to keep their plans and preparations secret - what good purpose is served by checking out how many "players" have bet on "coup in Turkey", "Abdullah II deposed" or "Hezbollah attacks Chicago"? Even if the "market" puts a high or low value on the probability of any of these events: aren't these "odds" something our intelligence services should already know?

Jane Galt writes: "But there is a problem, which is that it creates some rather perverse incentives"

That's not the problem.

The problem is that the market would be a publicly accessible notice of where our intelligence agencies and security services are looking, and where they're not looking. That is, after all, the point of the market - high-priced contracts would get extra attention and defensive measures.

Terrorists could use that. I see no reason to believe that terrorists would mount an attack they *know* is expected and being defended against, when they could look at the market data and find a target that *isn't* being defended against.

Look - smart investing is to buy low, sell high.

In order to work, the market assumes that terrorists would "buy high" by striking targets with high prices.

I don't think that's a safe bet.

I don't think they'd waste time investing in the market or trying to profit from their own terrorist attacks. That's stupid.

But they would use the market as excellent intelligence into the activities of the opposition. ie, us.

It's not like, say, weather futures. The atmosphere isn't actively looking to subvert your defenses. If the market says there's going to be an active tornado season, the atmosphere can't decide to ignore tornado alley completely and instead whip up some catastrophic hailstorms.

Steve,

If you know how to render the symbol for "proportional to" on a limited character set, maybe you can help me out.

I don't feel too bad. After all, YOU figured out my notation, didn't you?

Maybe I should have said "the glyph that looks like a little fishie," to dumb it down for left leaning readers. Just kidding of course.

OK seriously then, point well taken. In the future, I'll place an appendix in my posts explaining notation. I wanted to emphasize the concept of likelihood without getting sidetracked on Bayes theorem by writing the denominator.

And to YOU, achilles:

I didn't write the symbol '=' did I? Now THAT would be an egregious error.

No, I wrote a symbol that comes as CLOSE as I could come to the glyph that looks like a little fishie. I was hoping people like you might get it. Apparently not.

How about (= or o= ?

Still, to feign outrage that MY ambiguous symbol was in any POSSIBLE sense morally equivalent to the incoherent grunting of Barbara Boxer and the rest of the democrat brain trust, then you lack proper moral discrimination.

Jeff,

Believe, I know about Bayes Theorem, I'm a huge fan of it, and Bayesian analysis in general.

Regarding your problem with the proportional symbol...try:

[font face="Symbol"]µ[/font]

Replace the brackets with the appropriate greater than or less than sign. Lets see if it works in MT's comments:

µ

Hmmm...it worked, but a tad bit small though.

µ

I think

[font size=4 face="Symbol"]µ[/font]

will do it...or maybe 5.

Nice source for HTMLized math symblos.

Ron Bailey discusses this over at Reason (link opens in new window)

"Why wouldn't terrorists just hop online and start betting if they couldn't either mislead American authorities about their plans or make money to fund more al Qaeda operations?" Wyden asked. Why not indeed? If terrorists were trying to use PAM to make money that "would mean that they are giving up information to gain money," says Hanson. "In other words, we're bribing them to tell us what they are going to do. That's kind of like normal intelligence gathering when we bribe agents for information."

But what about terrorists using PAM bids to mislead our intelligence agencies? Markets are very good at dealing with that problem insists Hanson. "There are already lots of people who have an interest in obscuring information in markets," says Hanson. "There are CEOs who want to make their company's performance look better than it is to boost its stock prices, or the President of a South American country who wants to make his country's prospects look good to keep exchange rates high." In the long run they lose money and provide higher profits to those with superior information.

Consider this scenario. Let's say PAM offers a futures contract on a pool of 50 Middle Eastern leaders and we know that on average one of them gets killed every year. If assassination is random that means that each one has a 2 percent chance of being dead in the next year. Prices that move away from 2 percent indicate that the market participants anticipate a greater or lesser likelihood that any particular leader would be killed. Say terrorists flood the market with bids on assassinating Egyptian President Hosni Mubarak so that the prediction rises to 4 percent. Why would they do that? To direct our attention away from one of the other 49 leaders that they are actually targeting? That doesn't get them a lot of misdirection. And other market participants would still be bidding on other likely candidates for assassination and would take the profits away from the terrorist bidders when Mubarak attends the next Arab summit meeting a year later.

Hah! Tradesports has opened betting on whether Poindexter keeps his job

Dublin, Jul. 30 (TradeSports) -- Tradesports has opened a new 0-100 proposition contract on Admiral Poindexter remaining on the Defense Advance Research Projects Agency(DARPA) payroll by August 31st 2003.

This contract can be found under Current Events>Pentagon.

The contract will expire at 100 if Admiral Pointexter is still officially on the DARPA payroll by 11:59pm ET on the specific date. The contract will expire at 0 if Admiral Poindexter is not on the DARPA payroll at this time.

Tradesports will expire the contracts based on any official US Government/Pentagon announcement and crossed checked in 3 reliable independant media sources.

Thanks Steve, it's perfect!

Sorry achilles, I should have gotten off my lazy ass and figured out how to express proportionality correctly. Touche.

"I was hoping people like you might get it. Apparently not. How about (= or o= ? Still, to feign outrage that MY ambiguous symbol was in any POSSIBLE sense morally equivalent to the incoherent grunting of Barbara Boxer and the rest of the democrat brain trust, then you lack proper moral discrimination."

Whoa there. I don't know what people "like me" are supposed to get or not get, nor did I express 'moral equivalence' between your formula, gruinting and Barbara Boxer - so I have no idea what that tirade was about either.

It certainly was not obvious to me (until Steve pointed out) that you were trying to indicate proportionality. Once I read his post, it all became clear. That's all, don't read too much into it.

I wonder if such a market could be a substitute (or compliment) for terrorism insurance? A large landmark building owner could insure himself against attack by purchasing bets that an attack will happen. Every month he loses the cost will be an insurance premium. If he wins one month, it will be like an insurance payout. On a positive side he can win whether or not the terrorist attack damages his property....

To make such a thing work, however, you have to have a lot of participants. If I'm placing a $100M bet that an attack will happen I'll need people on the other end willing to put down money that it won't.

Also, I'm wondering if anyone ever addressed the issue of the person who becomes aware an attack is going to happen. Today such a person can go to the authorities and be a hero...or at least prevent an attack. With the market such a person with inside information would actually have the perverse incentive to sit on the info and buy.

Yes the prices in the market would signal a *possible* attack but that is not much comfort knowing that without the market the person may have provided the real goods on the attack.

Also limiting the market to 'professionals' or 'experts' would defeat the entire point. Are futures markets limited to farmers?

Also, I'm wondering if anyone ever addressed the issue of the person who becomes aware an attack is going to happen. Today such a person can go to the authorities and be a hero...or at least prevent an attack. With the market such a person with inside information would actually have the perverse incentive to sit on the info and buy.

Right, and to make big bucks chances are he'll have to buy big, unless the odds are heavily against the attack. If he does start buying big (i.e., getting greedy) then he'll probably start to change the odds.

"Hmmm...what does this guy know that we don't! Maybe we should get in on it too."

Then somebody notices this and the authorities start paying more attention.

I don't worry so much about the manipulation (by the way, what is to stop terrorists from doing this now with the stock market?). What I worry about are things like the bubbles we see in financial markets. Still it doesn't mean this would be completely useless.

That terrorism futures market is a silly idea. Many people have good reasons why it's silly, but they are also silly. Like insider trading. Silly

The only reason such large-scale collective prediction works is because the people making the predictions know something about what they are predicting. That's why americans can predict the outcome of american elections. When 278 million out of 280million americans don't know _where_ Jordan is, there is no way they can have any usefull predictive abilities. It's like having a futures market for socio-political events on far away planets. Silly idea. It's still a bad idea with professionals because these people are more arrogant than they are informed.

Fourty white guys in suits watching Fox News or reading CIA yellowcake papers clicking radio buttons to wager pennies on events in far away places is silly.

Now- the idea would be _STELLAR_ if you could some how get the people in the middle east to vote. That's what we need. Real time public opinion polls from the people we're trying to watch.

Dennison,

There are plenty of people in the stock market right now who don't know what they are doing, what they are buying, etc. So your argument:

The only reason such large-scale collective prediction works is because the people making the predictions know something about what they are predicting.

I find weak.

Drawing a parallel between MetLife and this proposal regarding a market for catastrophic loss of life is erroneous. MetLife only sells insurance whose beneficiary has an insurable interest in the person/entity being insured. I can't buy a policy on your life, as your death would make no difference to me (financially speaking).

I wonder what such a security would look like?

"This security shall pay $100 if a terrorist attack or act of war inflicts damage inside the United States exceeding $10M in damage between dates X and Y"

Sounds like a pretty interesting idea that could have some predictive value (even people who don't know much about the Middle East could factor in their knowledge of security...gut feelings etc.) but maybe even more importantly provide a way for businesses to insure themselves against terrorism without direct Federal involvement?

Dennison says:
The only reason such large-scale collective prediction works is because the people making the predictions know something about what they are predicting. That's why americans can predict the outcome of american elections. When 278 million out of 280million americans don't know _where_ Jordan is, there is no way they can have any usefull predictive abilities.

That's ok. Most of those 278 million wouldn't "invest" in that market. Of the remaining 2 million, some will have some clue what's going on, and some will have direct interests in the Middle East. That's enough to give a market predictive value. Clueless punters will lose their money. Clueful ones will drive the market.

Besides, it's way more than 2million who know something about the Middle East - most Jews do, and most Arabs do, and there's probably about as many non-semites in the US who have some interest in the area. Let's say more like 20 million who have a little knowlege.

Please. The reason this idea is absolutely insane is that (unlike a normal futures market) its very purpose is to totally reverse the very outcomes it's trying to predict. If investors use inside knowledge to correctly predict an imminent terrorist attack and the government then utilizes the information from the market to prevent the attack (as is the market's purpose), the investors lose all their money. And since the market is public, terrorists themselves can simply monitor it to find out what the government thinks they will be likely to attack, and then go attack something else.

Reply to Bruce Moomaw:

Perhaps the market will be less useful for predicting an attack as it will be for evaluating the *risk* of an attack. Right now we have a Dept. of Homeland Security that is issuing various levels of alert that no one knows what do so about. More than a few people have even suspected that the alerts serve more to keep everyone in panic mode than to provide real protection.

A futures market will, in fact, serve as a check on gov't. Such a market will also measure risk at any given time. Say right before election the incumbant gov't tells us we are on highest alert. An independent market that doesn't budget would be a pretty damming indication that such an alert was being issued for political reasons. Likewise, an incompentant gov't that tells us everything is fine may be discovered as futures prices increase.

Please forgive the typos in that last post....

B

Comments are Closed.