Prediction markets are pervasive in finance, especially in modern derivative markets. Someone who is long on the S&P 500 Index is betting that average stock prices in the United States will be going up, while those who are short in this market are betting that they will go down. Price movements in these markets are a good measure of aggregate sentiment, where the aggregation process gives greater weight to those willing to risk larger sums.
The aggregation in online political prediction markets, such as the Iowa Electronic Market (IEM), is more democratic because these markets usually place sharp limits on how much can be bet- the IEM limits bets to no more than $500. Yet as Posner indicates, this and other online political markets have been successful in predicting the outcomes of American elections-more successful than various polls. In the present election, the IEM odds in favor of the Democrats winning the presidency hovered around 60 per cent From May of 2007 to the end of August, but these odds have narrowed considerably since then to about 51-52 per cent for the Democrats to 48-49 per cent for the Republicans. Narrowing has also occurred in various polls. The IEM market is indicating that Senator Obama now has a small lead over Senator McCain.
Since bets on political online markets are small, the motivation of bettors can hardly be the amounts they win or lose. Nor can the usual economic theories of risky choices be of much relevance since the risks to bettors' wealth are rather insignificant. These gambles are made because of utility derived from the gambling itself, not because of the amounts won or lost. This has the very important implication that the positions taken by bettors-for example, whether they bet that the Democratic rather than the Republican presidential candidate would win- is not necessarily determined by which one they expect to win. On the contrary, their betting behavior may be in good measure determined by whom they want to win rather than whom they expect to win.
Studies of betting on sports events show a home team "bias" in the sense that the odds tend to be skewed in favor of home teams relative to the actual winning percentages of home teams. This may be because many local residents bet on their home team, such as Chicagoans betting on the Chicago Cubs, at odds where objectively they should be shifting their bets to visiting teams, and also because individuals in home cities are more likely to bet on games in their cities.
This home team bias is likely to be even more pronounced in political betting markets like the IEM since bets are small. However, if biases of Democratic and Republican bettors are about equally strong, and if a non-negligible fraction of all bettors are making prediction bets, then aggregate betting would tend to give on the whole accurate predictions about who will win, although these predictions would be quite noisy. Predictions rather than hopes may be of relatively large importance in the IEM and other online political prediction markets because the main bettors have been academic economists and financial experts rather than the general public. This type of wishful betting presumably is quite different in betting on other types of events, such as the unemployment rate shown by data to be released on a certain date.
I believe that online political prediction markets, and other online prediction markets as well, should be legal in the United States and elsewhere, even if the amounts bet were quite large. There is no important substantive difference between such online betting markets and the Chicago Mercantile Exchange and other exchanges that allow individuals and organizations to take positions on movements of stock indexes, housing price indexes, and prices of other derivatives. A distinction is sometimes made between political betting markets and derivative markets since participants in derivative markets may be hedging other risks that they face. Yet this distinction has little substance since if larger bets were allowed in online political markets, groups whose welfare depended greatly on political outcomes would make greater use of these markets. For example, if a Republican presidential win would mean greater spending on military weapons, companies in the arms business might hedge their risks by betting on Barack Obama.
If large bets were allowed, some wealthy groups may bet a lot on their candidates in order to exert bandwagon influences on public opinion through their large bets affecting market odds. If so, these markets likely would become less reliable as predictors of outcomes, and hence would have less influence on opinions. To a large extent, therefore, these markets would be self correcting, although online political markets might place various other restrictions on bets, as is common in derivative and other exchanges.
If large bets were allowed, some wealthy groups may bet a lot on their candidates in order to exert bandwagon influences on public opinion through their large bets affecting market odds. If so, these markets likely would become less reliable as predictors of outcomes, and hence would have less influence on opinions.
Could that also be an argument for limiting bets in other types of prediction markets? Has the lack of limits in other markets had the effect you describe - making the markets less reliable and reducing their influence on opinion?
Posted by: Andrew | 09/14/2008 at 11:16 PM
My theory is that voting markets model the costs of voting better than polls. People are far more likely to lie to a poll taker (of course I expect to vote, even though I never have), but you can't lie about placing a bet or not.
Home team bias is an effect which impacts both party candidates. In the absence of other effects, home team bias would make the predictions proportionate to the sum of the bettor's political parties. This is not seen in practice, so other effects must be included which describe why the betting market predicts voter behavior so well.
Voting results are known to be influenced by turnout, i.e. the people desiring to vote for a candidate are sufficiently excited about the candidate to incur the costs of voting.
The same mechanism could be at work in the betting market, i.e. the people desiring to bet on a candidate are sufficiently excited about the candidate to incur the costs/risks of betting.
Suppose a party member's "home team" candidate is not very exciting, so the party member is less likely to place a bet for him/her. The costs of placing the bet mirrors the costs of voting.
If this mechanism is at work, the size of the betting markets (number of bets, total value of bets) should correlate to voter turnout (number of registered voters, number of actual voters).
I hereby allow anyone to perform the follow-up research, attribution not required.
Posted by: DanT | 09/15/2008 at 08:14 AM
Your concluding paragraph suggests that manipulation attempts make the price less accurate. I have a paper forthcoming in Economica that says the opposite - manipulation attempts increase price accuracy.
Posted by: Robin Hanson | 09/16/2008 at 05:59 PM
WHAT IS YOUR PREDICTION FOR OBAMA ?
THANK YOU ...
Posted by: BERNADETTE BROUGHTON | 09/17/2008 at 01:33 AM
One example of manipulation increasing accuracy is when someone bids up a default swap, helping to blow out a company's funding costs, driving them into bankruptcy. Ahem. But you were talking about a model result, and surely the assumptions of the model always hold. Surely.
Posted by: Jason Ruspini | 09/18/2008 at 08:53 AM
Hanson,
I doubt this is true in thinly-traded markets, but I am interested in finding evidence to the contrary. If there is such evidence, I would be very interested in hearing your thoughts on the reasoning behind it.
Thanks.
Posted by: Joe Cisewski | 09/18/2008 at 09:47 AM
I am love this place!!!
Posted by: injection molded plastic | 05/02/2009 at 08:15 PM
nice
مركز تحميل
Posted by: Anonymous | 06/29/2009 at 07:11 PM
thanks
بنت الزلفي
Posted by: Anonymous | 07/06/2009 at 04:54 AM
Thank you, you always get to all new and used it
شات صوتي
Posted by: Anonymous | 07/11/2009 at 06:50 AM
ÿØÿ±ÿØÿ¥ÿ©
___
صور
Posted by: Anonymous | 07/14/2009 at 11:16 PM
AFj15O
Posted by: Anonymous | 07/15/2009 at 09:39 PM
Excuse me. Arguments are to be avoided; they are always vulgar and often convincing.
I am from Tanzania and bad know English, please tell me right I wrote the following sentence: "A culinary center and luxury cabin oasis in the texas hill country small kitchenette with a small refrigerator, microwave oven and coffee maker."
Regards :P Varick.
Posted by: Anonymous | 07/17/2009 at 02:43 AM
Great site. Good info.
Posted by: Anonymous | 07/23/2009 at 02:44 AM
Great. Now i can say thank you!
Posted by: Anonymous | 07/23/2009 at 12:00 PM
thanks for your
ÿ¥ÿßÿ™
دردشه
Posted by: Anonymous | 07/23/2009 at 07:34 PM
Incredible site!
Posted by: Anonymous | 07/24/2009 at 09:41 AM
It is the coolest site, keep so!
Posted by: Anonymous | 07/24/2009 at 10:59 PM
thanks to tell me that,i think thats ao usefully----
tiffanys
links of london
Posted by: Anonymous | 07/25/2009 at 02:54 AM
Very interesting site. Hope it will always be alive!
Posted by: Anonymous | 07/25/2009 at 05:04 AM
Great. Now i can say thank you!
Posted by: Anonymous | 07/30/2009 at 04:58 PM
It is the coolest site, keep so!
Posted by: Anonymous | 07/31/2009 at 09:35 AM
دردشة برق
دردشة الخليج
Posted by: Anonymous | 07/31/2009 at 06:21 PM
ibScDK I bookmarked this link. Thank you for good job!
Posted by: Anonymous | 08/02/2009 at 12:06 PM
ibScDK I bookmarked this link. Thank you for good job!
Posted by: Anonymous | 08/02/2009 at 12:07 PM