Tuesday, November 02, 2004
Always at the last minute
I've pointed out before my doubts regarding the Iowa Electronic Markets. Chief among the is the impression that participants don't get serious about their "investments" until the very lst day. IEM performs about as well as a met-analysis of polls (which is, in a sense, what it aspires to be), but only on the day before the election. Before that, my impression for several elections of observation, is that it underperforms pretty much any intelligent reading of the polls.
So right on schedule, after months of leaning absurdly towards a large Bush victory, the IEM converges on what has seemed likely since the debates started---a Kerry win:
My explanation is simple: IEMers are playing with small sums (<$500), in a shallow but heavily watched market. Up to the last minute, the rational use of your money is to push the appearance of an advantage towards your preferred candidate (think of it as a form of campaign contribution). But on election eve, it's too late to sway the electorate, and all that is left is to try to salvage your investment by trading efficiently. Hence the last day convergence on the "truth".
So right on schedule, after months of leaning absurdly towards a large Bush victory, the IEM converges on what has seemed likely since the debates started---a Kerry win:
Symbol Bid Ask Last Low High Average
DEM04_G52 0.214 0.257 0.230 0.130 0.258 0.186
DEM04_L52 0.351 0.370 0.370 0.301 0.370 0.329
REP04_L52 0.255 0.300 0.300 0.300 0.398 0.358
REP04_G52 0.094 0.095 0.095 0.050 0.163 0.129
My explanation is simple: IEMers are playing with small sums (<$500), in a shallow but heavily watched market. Up to the last minute, the rational use of your money is to push the appearance of an advantage towards your preferred candidate (think of it as a form of campaign contribution). But on election eve, it's too late to sway the electorate, and all that is left is to try to salvage your investment by trading efficiently. Hence the last day convergence on the "truth".
