A new working paper by Robin Hanson observes: “When a controversy erupts in the media, and widely differing views are expressed, it is natural to wonder which opinion is the one more favored by those who are most informed about the topic.” At least it’s natural for me.
If you followed the controversy on the Policy Analysis Market (PAM) – popularly known as “terrorism betting markets” – you might have noticed a pattern in the media coverage: The more immediate the publicity, the more negative. Initial reports by daily media heaped scorn on the idea; the lower-frequency media was more favorable. At least that’s how I saw it.
At long last, there is now hard data confirming my casual impression. In Robin Hanson’s new working paper, The Informed Press Favored the Policy Analysis Market, Hanson reports systematic econometric evidence on media coverage and terrorism betting markets. Independent raters assessed the positivity of 396 pieces of media coverage on a 7-point scale, from least favorable to most favorable. A positive attitude correlated with ALL of the following signs that the journalists involved knew what they were talking about:
1. mentioning someone with firsthand knowledge
2. time since the media firestorm
3. article length
4. a news versus an opinion style
5. the periodical’s prestige
6. the periodical’s frequency
The most natural interpretation is simply that the more the journalists knew, the more favorable their attitude toward terrorism betting markets.
Hanson’s closing remark is particularly fascinating:
Unfortunately,public policy regarding PAM continues to reflect the uninformed opinion, a situation that seems unlikely to change anytime soon.This should give pause to those who think that our democratic processes can effectively produce informed policy, even when the electorate is highly uninformed. In this case, as in many others… policy reflects the uninformed position.
READER COMMENTS
Tony Vila
Mar 21 2005 at 4:39pm
Economists love for PAM is so sadly unrealistic.
To be clear, PAM would only be an analytic tool, and not an active department in of itself.
So the same degree that the public (and then the political bodies involved) distrusted it, so would any results predicted by PAM be distrusted and ignored.
Not to mention that our policy making apparatus is profoundly uninterested in objective analysis of security policies and terrorism (on both sides of the aisle). There are a great many security papers and blogs out there discussing how little cost-benefit analysis goes into security decisions. Our absolutist and emotion driven response might be good for many reasons (game theory speaking, deterrence and comfort), but it’s not something that desires to be informed by “market predictions”.
I suppose one could be hopeful and think PAM means movement in the direction of serious cost-benefit analysis, but there’s a lot of other steps they could possibly be taking before that, which they certainly aren’t.
Lastly, markets has predictors have generally been precocious at aggregating and analyzing known information, not ferreting out hidden information and incorporating that, the latter being what Poindexter and others stated as their desire.
Ronnie Horesh
Mar 22 2005 at 3:49am
Policy Analysis Markets sound a bit frivolous to me. Social Policy Bonds would display just as much information but would also give direct incentives to people to do something about terrorism, or whichever social problem is targeted. Social Policy Bonds are non-interest bearing bonds redeemable for a fixed sum only when a specified goal has been achieved. They would be backed by by government or the private sector. They would be auctioned initially and thereafter would be tradable at any time until redemption.
dsquared
Mar 22 2005 at 3:57am
I wish the PAM people wouldn’t keep claiming that their market was “ready to start trading in 5 months”. It visibly wasn’t. The site consisted of precisely five flat HTML pages. There was no indication at all that they had a secure betting exchange platform ready to go, let alone clearing arrangements, anti-money-laundering measures, etc, etc, etc. Setting up a betting site, particularly one that is meant to be based in the USA, is a difficult regulatory job and there was just no indication that the amount of resources behind the PAM were anywhere near commensurate to their marketing. As far as I can see from Robin’s site half the start-up cash had been spent before the difficult bit had even started. The media firestorm certainly played its part in killing the PAM, but, in my opinion, it is not at all clear that there was a viable project here to begin with.
Robin Hanson
Mar 22 2005 at 7:24am
Dsquared, our webpages on July 28, 2003 were indeed very simple, but this doesn’t mean we weren’t on track to deliver as promised. Net Exchange, the small business responsible, had before and has after built and delivered many other markets as promised. In the case of PAM, I know that we had developed and lab-tested prototype software, and were well on the way to managing the various operational issues you mention. We were only a month away from starting trading with a set of 100 initial users. I’m not sure why you would expect us to have put up incomplete market software on the website well before we wanted users to trade.
dsquared
Mar 22 2005 at 8:20am
The software issues are not particularly the difficult ones (though I note that Net Exchange has delivered plenty of e-procurement auction systems but no large publicly-acessible two-way order books that they are prepared to talk about on their website). What I look for in a betting exchange site (I’m a member of Betfair, tradesports and the IEM, and I’m not too happy with the IEM) is more along the lines of:
1) Financial details about the legal entity standing behind the exchange or its guarantor
2) Clearing arrangements for the exchange, establishing whether there was a single central counterparty or not
3)Regulatory and tax status of the exchange; whether one’s profits would be classified as gambling winnings or income
4) Specific details on who was eligible to join (both internationally and in terms of US state laws)
5) Settlement and delivery criteria for the proposed contracts
6) Rules on market conduct
7) Disputes and/or arbitration procedures for cases where the rules might have been breached
8) Custody arrangements for deposit balances and margin accounts
etc, etc. I think it’s really rather naive to assume that all these could have been sorted out in five months. (Note that these are for the most part issues that don’t arise, or which can be dealth with informally, in a non-public market like an e-procurement system).
As far as I can see, you and Net Exchange are trying to make the quantum leap from a toy “market” which is basically a formalisation of existing procurement arrangements and tendering processes, to a genuine market open to the whole world. I think that’s a huge leap, and I suspect that you’ve massively underestimated it. I’m sure that with sufficient time and money, the problem might be solved in as much as you could get a site up and running, but look at Tradesports and the liquidity on their HS alert level contracts; what makes anyone think that there is a huge untapped reserve of amateur policy analysts over and above those who already deal on Tradesports? Or to put it another way, look at the length of a typical IEM prospectus; did you have anything like as thorough a document ready to go for every contract you planned? If so, I’d be really interested to see it.
Boonton
Mar 22 2005 at 12:31pm
My thought was that ‘terrorism markets’ could function very nicely as a supplement to insurance. After 9/11 there was a lot of worry that the insurance industry would collapse and refuse to offer any coverage for terrorism related diasters. Proposals were floated for government intervention because businesses could not buy coverage at affordable premiums.
Now consider a ‘Terrorism future’ that would pay $10 should an act of terrorism on US soil cause over $100M in damage. A company with $100M in assets could purchase 10 million contracts. If a terrorist event happened they would see a $100M payout (of course, if the terrorist act destroyed the company’s plant then they would just see their loss covered as if they had insurance).
Of course insurance companies themselves could buy these contracts and use them as a way to lower premiums by hedging the risk of a terrorist diaster. However who will be on the other end of the contract and write them? Anyone willing to bet that a terrorist act will not happen. That could be regular investors, speculators and so on.
Robin Hanson
Mar 22 2005 at 10:57pm
Sheez. Dsquared, I think you are just way out of line demanding that over 18 months after our project was cancelled that we provide detailed documentation to prove to you that we could have delivered what we were under contract with DARPA to deliver. (If you had funding to offer if only we could convince you, well then we might dredge up documentation for you.) We had already gone a long way on many of the issues you are concerned about, so we didn’t need to do them all from scratch in 5 months. And Tradesports does fine without detailed prospectuses.
dsquared
Mar 23 2005 at 5:19am
Boonton: you’ve identified the problem in your last paragraph. If there is nobody willing to write $100m of terrorism insurance, then it is highly likely that there is nobody who is willing to be short 10m contracts.
Robin: well, I’m sure it doesn’t really matter to you what I think, but that’s what I think. I’d note that Tradesports doesn’t actually offer the homeland security contracts any more, and the ones you were proposing look even less attractive to traders – particularly the significant lag with which the EIU was proposing to deliver the settlement data.
Bob Knaus
Mar 24 2005 at 7:11am
I have no dog in this hunt, but perhaps there is a simpler explanation? That is, it takes time to develop a rationale for a bad idea.
Think of it as a corollary to Occam’s Razor — if you have two explanations, the one which took longer to synthesize is less likely to be correct.
Boonton
Mar 25 2005 at 10:39am
Good point, perhaps some creativity can resolve the problem. For example, suppose $5B of the Dept. of Homeland Security’s budget will be required to short terrorism contracts. If no terrorism happens then that will become a ‘performance bonus’ for the department.
Also while it may be hard to find an individual or company willing to underwrite a $500M insurance policy against terrorism it would be easier to find lots of investors and companies willing to underwrite tiny contracts. In effect you would be ‘privitizing insurance’ by letting people supply insurance in tiny contracts of only $1,000 or so.
Comments are closed.