July 31, 2003

Shorting Poindexter

Tom Maguire wonders why it was that proponents of DARPA's Policy Analysis Markets played up the idea of the U.S. government funding trading pits in bioweapons attacks on Israel and assassinations of foreign leaders, rather than talking about forecasting indexes of political stability and economic output:

Semi-Daily Journal: Comment on More on DARPA's Policy Analysis Market: ...Well, props to James S (and others), for attempting to explain what this program was, rather than recycling the hype offered by the critics. A DARPA doc describing the program is here (p B-8 of Appendix, or p. 68-9 of the .pdf file): http://www.darpa.mil/body/tia/TIA%20DI.pdf.

So, sort of related - why the fierce opposition to what is arguably a useful idea? Was it a hit on Poindexter? And why did the plan proponents get so stupid? Letting the critics make headlines with these examples of "assassination contracts" was incompetent. Why did they not have their ideas thought through, with clear, sensible example that they were ready to defend? Why are we dragging the details out of tired old James, when they should have had a press kit available for that hearing?

Or was it a mousetrap, where the subject was sprung on an unsuspecting Wolfowitz? Just wondering...

And then he raises the possibility that only someone receiving weblog RSS feeds through his fillings would raise:

Maybe they were [long] the Poindexter[-Decides-to-Spend-More-Time-with-His-Family] contract on tradesports.com...

:-)

Posted by DeLong at July 31, 2003 05:07 AM | TrackBack

Comments

"Tom Maguire wonders why it was that proponents of DARPA's Policy Analysis Markets played up the idea of the U.S. government funding trading pits in bioweapons attacks on Israel and assassinations of foreign leaders, rather than talking about forecasting indexes of political stability and economic output."

Because there is simply no sensitivity in these folks. None. I do not give a fig whether the idea was "useful" to an odd sort of gambling economist, the idea was barbaric.

Why not focus on keeping American soldiers from coming to harm in Iraq, and then on building more stable political-economic systems in Africa and elsewhere?

Amazing lack of sensitivity!

Posted by: lise on July 31, 2003 08:28 AM

Puckish; I like it. But I'm unconvinced that the idea was as well-thought-through as the case James S. has been making.

To me, it smells like the kind of half-baked idea that could be sold to bureaucratic superiors who a) hate intelligence professionals the way most people hate roaches, and b) salivate like Pavlov's dogs at the words "free market."

Gordon Liddy had an idea like this about 31 years ago, and so did a bunch of people working in Reagan's basement about twenty years ago. Come to think of it, wasn't Poindexter one of them? Gee wilickers, Rumsfeld might say.

Posted by: Altoid on July 31, 2003 08:28 AM

Perhaps economists could focus on honoring American soldiers by citing correct casualty figures for the press.

Casualties:

American soldiers 107
British soldiers 11
---
120 Since May 1

American 248
British 44
---
292 Totals

Note: American forces have risen to 148,000
British forces have been cut from 10,000 to 5,000

Posted by: lise on July 31, 2003 08:34 AM

Tom's questions are the right ones. I think this was, to put it crudely, a hit on TIA and Poindexter (who, let's face it, should never have been allowed back into government in the first place). Incredibly enough, Wyden and Dorgan never talked to Charles Polk or anyone from Net Exchange about PAM before holding their press conference (which you can still see at C-Span.org). That suggests to me that they weren't interested at all in why DARPA was interested in this idea or even in how the market was really going to work. During that press conference, they repeatedly said that they were just going by what was on the Web site. So they weren't even willing to put in a couple of hours of research before demolishing a potentially good idea. Along the same lines, remember that the supposed occasion for the press conference was for them to report on their investigations into government "waste." PAM was supposedly an example of wasteful government spending. Considering that its budget for the next six quarters -- which was how long the pilot was going to last -- was a couple of million dollars at most, it seems likely that were much better targets if what Wyden and Dorgan were really interested in was waste, rather than raking the Bush administration over the coals for its moral insensitivity.

As for why the plan's proponents handled the press so badly, I think it was a combination of things. First, there was what seems to me to be the typical Bush administration response: if it doesn't sell, and it makes you look bad, dump it, even if it's a good idea. (The flip side of this is: if it does sell, keep it, even if it's a bad idea.) Second, DARPA had not done a good job (or any job) of explaining this program to higher-ups at the DoD or elsewhere. And the truth is that once the New York Times put "assassination contracts" on the front page -- doing so, by the way, without ever talking to anyone from Net Exchange or trying to figure out how the market was really going to work -- the game was probably over from a PR point of view.

As for why the "assassination of Arafat" contract was even on the website, well, it seems as if that was a case of ill-conceived advertising. In the wake of the announcement, everyone has been assuming that PAM would have had tens of thousands of traders trading millions or billions of dollars in contracts. But Polk suggested to me that their real concern was just getting people --and especially people who wouldn't normally be interested in trading futures, like policy analysts -- to trade at all. They apparently thought concrete -- and obviously important -- examples of specific-event securities would be more likely to get prospective traders interested than would a long explanation of combinatorial futures contracts (which was, nonetheless, in place on the site). That was clearly a PR mistake.

I do think, though, that the press really did a bad job on this one (especially in the first couple of days). If you watch that Wyden and Dorgan press conference, you'll see that they were not asked a single tough question. The reporters just acted aghast that such an insane scheme could ever have been proposed. The Times articles were completely uncritical and offered what I think was a deceptive picture of how the program would have actually worked. Today, thankfully, they ran a nice defense of PAM by Hal Varian at http://www.nytimes.com/2003/07/31/business/31SCEN.html, but then on their op-ed page there's a piece by Todd Buchholz that repeats the canard that PAM was intended to try to predict random assassination attempts.

Posted by: James Surowiecki on July 31, 2003 08:52 AM

Joseph Stieglitz in the LA Times doesn't seem to share Mr. Surowiecki's view that PAM was a good idea:

http://www.latimes.com/news/opinion/commentary/la-oe-stiglitz31jul31,1,4162448.story?coll=la-news-comment-opinions

Posted by: cafl on July 31, 2003 09:04 AM

Joseph Stieglitz in the LA Times doesn't seem to share Mr. Surowiecki's view that PAM was a good idea:

http://www.latimes.com/news/opinion/commentary/la-oe-stiglitz31jul31,1,4162448.story?coll=la-news-comment-opinions

Posted by: cafl on July 31, 2003 09:09 AM

I was actually wondering if Professor Stiglitz would weigh in on this. Actually, I think his critique -- although rhetorically aggressive -- is actually relatively mild and not especially convincing, given what we know about how PAM would have really operated. By focusing on the "terrorist attack" part of the market, he makes it appear preposterous that any good information could be out there that the experts of the FBI and the CIA don't already have. In the first place, we know that most of PAM's contracts would have had nothing to do with terrorist attacks, and when it comes to questions like "How stable will Jordan be in August of next year?" there's no doubt that there's lots of information out there that the U.S. does not have.

Stiglitz also exaggerates the ease with which the CIA can gather information. He asks "Did [Poindexter] believe that the 1,000 people "selected" for the new futures program would have this information? If so, shouldn't these people be investigated rather than rewarded?" The CIA doesn't always who know who has relevant information. One hope of PAM was that it would give people who wouldn't talk to the agency, and who the agency would never come across, an incentive to disclose what they knew.

Manipulation is a problem in any market, but as the market grew, manipulation would become less of a problem. In any case, Stiglitz, like many others, implies that the CIA would be robotically following PAM's counsel. I don't think that was ever part of the plan. Finally, the insurance critique is pure political grandstanding, and is also an interesting comment on the idea. PAM could only be a good hedge, protecting the wealthy against terrorism while leaving the rest of Americans exposed, if it was successful at predicting the future. So which is it? Bad because it won't work or bad because it will work but will exacerbate political inequities in the U.S.?

A couple of notes on Hal Varian's excellent piece: although trading sums in the pilot market were going to be relatively small, had PAM gotten up and running, there were no plans to limit the amounts people could invest to $100. Also, the question of insider trading is a complicated one. PAM promised participants anonymity from DARPA or any U.S. agency, thinking that without that promise no one from the Middle East would participate. So I don't think it's correct to say that the CIA would have been able to find out who placed a big bet on Abbas being removed from office.

Posted by: James Surowiecki on July 31, 2003 09:24 AM

Funny thing. The broad attentive public, fools that we are, immediately knew just how bizarre and cruel the idea for an options market in mass death was. Economists, some economists, have no idea what moral sensitivity is. Imagine a bunch of these DARPAs spending time thinking about Liberia or Angola. Nah.

Posted by: lise on July 31, 2003 09:34 AM

James S. writes: "Considering that its budget for the next six quarters -- which was how long the pilot was going to last -- was a couple of million dollars at most"

I believe it was $8 million.

Which would pay for 213 airport screeners, assuming a pay rate of $25k/year, which is about what they're paying to start.

Posted by: Jon H on July 31, 2003 09:39 AM

Needless to say, I think PAM would probably have been far more valuable than hiring 213 airport screeners.

In any case, though, the $8 million number is deceptive. $8 million was the budget for 2004 and 2005 for the broader FutureMap program. FutureMap included not just PAM but also experiments into setting up internal markets comprising "15 to 20 participants addressing questions about the probabilities of specific kinds of failure within our national infrastructure." Although DARPA's description is vague, it appears as if the participants would come from within the intelligence community and also, perhaps, from the policy world as well. (DARPA uses the phrase "invited participants.") This would be analogous to that internal market HP set up, and fits with what Charles Robb (?) suggested on a different thread on Brad's site. As far as PAM specifically goes, Polk told me that it was going to cost about $750K over three years (that includes 2003). Even if that's an underestimate, the overall cost would clearly be minuscule.

What's troubling is that Wyden and Dorgan want to defund (actually, they may have already succeeded in doing this) not just PAM but also the FutureMap internal-market program as well. (They want the $8 million authorization -- $3 million for 2003 and $5 million for 2004 -- out of the budget.) I can't see any meaningful argument against using market mechanisms internally with people. But the uproar over PAM probably means that plan gets tossed overboard, too.


Posted by: James Surowiecki on July 31, 2003 10:16 AM

Sorry, that's $3 million for 2004 and $5 million for 2005.

Posted by: James Surowiecki on July 31, 2003 10:17 AM

Where were all these DARPAs and NeoCons when it came to figuring out just how much in life and material it would cost to occupy Iraq? Wonder if they bet on finding WMDs in Iraq. Oh, still looking.

Posted by: arthur on July 31, 2003 10:18 AM

"DARPA had not done a good job (or any job) of explaining this program to higher-ups at the DoD or elsewhere."

Or to me, that matter, although I'm just an ordianry citizen, not a higher-up in any sense. But the whole notion seems like mystical free market crap to me, and I have yet to see anybody explain this in a way that enables me to see how the program would boil down to more than, "Conventional wisdom, even in the absense of actual information, often makes surprisingly good guesses."

Posted by: rea on July 31, 2003 10:26 AM

We don't know if the exchange would yield useful info. On the one hand, there is the notion of the market as a superprocessor of scattered information, on the other, we have a very unusual and restricted type of information we are expecting the market to yield insight into. So, in the spirit of empirical science, we should test the proposal. If it turns out useless, kill the program. If it works--and we would need to define what "works" means in this case--let's use it. (Can be buy options on whether the terror exchange will yield useful info?)

Posted by: maciej on July 31, 2003 11:06 AM

Arthur,

Tradesports has been running contracts on finding WMDs for sometime. Right now the market is giving it a 20 percent chance of happening by the end of Sept.

Oh, and the Poindexter futures just tanked.

Posted by: Justin Lahart on July 31, 2003 11:16 AM

Sorry, that was a response to rea, not Arthur...

Posted by: Justin Lahart on July 31, 2003 11:19 AM

Reuters is reporting that Poindexter will resign "within weeks", which is why his futures market just tanked.

So now both the DARPA offices which Tony Tether established when he came in are leaderless.

Posted by: jam on July 31, 2003 12:35 PM

Reuters is reporting that Poindexter will resign "within weeks", which is why his futures market just tanked.

So now both the DARPA offices which Tony Tether established when he came in are leaderless.

Posted by: jam on July 31, 2003 12:38 PM

Reuters is reporting that Poindexter will resign "within weeks", which is why his futures market just tanked.

So now both the DARPA offices which Tony Tether established when he came in are leaderless.

Posted by: jam on July 31, 2003 12:38 PM

Reuters is reporting that Poindexter will resign "within weeks", which is why his futures market just tanked.

So now both the DARPA offices which Tony Tether established when he came in are leaderless.

Posted by: jam on July 31, 2003 12:41 PM

> Tom Maguire wonders why it was that proponents of
> DARPA's Policy Analysis Markets played up the idea
> of the U.S. government funding trading pits in
> bioweapons attacks on Israel and assassinations of
> foreign leaders, rather than talking about
> forecasting indexes of political stability and
> economic output."

Well, the text in the DARPA proposal is pretty small, and that's the only thing that they talk about in there.

Posted by: Naunihal on July 31, 2003 04:04 PM

Not realisingthere was a thread onBrad's blog onthis, I've mailed this to Hal Varian:

"You condemn the reflexive outrage at the Pentagon's ideas for a “terrorism futures” market - which I share as a matter of fact; would you regard a market in public health futures (say on indicators of AIDS deaths in Africa) as morally acceptable, even though it faces lower technical barriers? But quite apart from the moral issues, I think there are grounds for scepticism that it would actually work in your sense of pooling informed views (see your NYT article of 31 July, http://www.sims.berkeley.edu/~hal/people/hal/NYTimes/2003-07-31.html) .

Even markets that are perfect in the traditional sense of negligible transactions and communication costs will be imperfect as knowledge aggregators. A market brings together several categories of traders: arbitrageurs, informed fundamentalists, noise traders, market predictors relying on second-order information about other traders, and insiders. In catch-phrases, technicians, experts, fools, speculators, and crooks.

I suggest that a market is informationally efficient to the extent that the first two categories dominate the last three. In fact, since the arbitrageurs do not bring in external information, the experts must dominate information input.

The function of the arbitrageurs is to make the market; liquidity depends on their willingness to provide trading capital; they have no quarrel with the experts, fools, and speculators, but will quit if the crooks are too numerous, since their coups will make prices move discontinuously. The equilibrium may be bimodal: thin, opaque markets attract speculators and crooks, and you have say the Russian stock exchange ca. 1993. Arbitrageurs will readily supply capital to honest, transparent, boring markets like that in US Treasury bonds and will tend to keep them that way.

The experts in the market are a fraction of all experts. In the equity markets, a long-term investor like Warren Buffett is not present day to day, though his views are known to traders since his holdings are a matter of public record. It seems plausible that the markets will work better as information aggregators the more representative the sample is of all experts. Since buy-and-hold is demonstrably the best long-term strategy for equities, there would seem to be limits to the possible representativity of active traders.

A market in political security futures would have great difficulty in meeting these tests.

First, most experts are law enforcement officials and would be debarred from participating. Another large group would not participate on ethical grounds. The market would need to be global to be any use, and cultural barriers would reinforce legal ones.

These restrictions on participation by experts would increase the proportion of fools, speculators and crooks in the market. In this case, the crooks - market manipulators - would include terrorists and their sympathisers, as well as their enemies the spooks. Deception is a tool of all war, and the essence of secret war; why should the adversaries refrain from using this lovely new toy? You would end up with an oily black cloud of data smog."

Posted by: James Wimberley on August 6, 2003 06:36 AM

The moral indignation about the PAM (especially on the side contracts on assassination) comes I think from the prohibition of instrumentality made explicit by Kant: never using other people simply as means to one’s own goals. All relations with others should include a minimal concern for their aims as individuals like oneself. A self-interested, arm’s length trading transaction between distantly separated individuals, mediated by impersonal market tools like money, shops and markets, meets the test so long as one may reasonably assume that the other party or parties enter freely into it, like Adam Smith’s butcher and baker. But speculative markets or betting books on events affecting third parties do not treat them as free participants but as objects.

Take a very simple case: of a single patient – say myself – undergoing dangerous surgery. Is it proper for my children, my office colleagues, or strangers such as the Berkeley economics faculty to wager on my survival? Intuition says not. The obligation is to help suffering people, even minimally, not to use their predicament as input to a game. Betting on professional sporting events is distinguished by the facts that the contests are voluntary, the players are putting themselves on public show, they know that their livelihood depends on commercialisation, and the outcome is morally indifferent. The same goes more or less for elections. In spite of the poetic justice, the short-lived book on John Poindexter’s losing his job was not fair.

The surgery wager can I think be made legitimate by either getting the informed consent of the subject, or providing a modest benefit. If the members of the economics department sign IOUs promising to pay $10 towards a funeral wreath if I die, the chits have a negative subjective expected value of -$10<x<$0. Bears would be ready to pay bulls this sum to hedge the liability. The going price would reflect an aggregated view of the risk.

Of course this market has no practical interest, and its design not optimal, but I think its moral characteristics scale up to ones that might be useful. For example, in the Middle East the European Union, the US or Bill Gates could guarantee 1000 grants of €5000 to Palestinian students per year for 10 years subject to a specific level in a transparency index. Donors could securitise these obligations - selling forward contracts - and generate a market reflecting informed opinion in trends in Palestinian governance, taking account of the incentive effect of the programme. Similar markets could be set up for other conditional aid undertakings, with the added benefit of making the undertakings more credible because contractual.

Posted by: James Wimberley on August 13, 2003 05:47 AM
Post a comment