Talk:Experimental economics

From Wikipedia, the free encyclopedia

Behavioural biases in auctions

Other experiments went further into the auction experiments with which Smith had begun the discipline. He showed that a naive crowd tends to pay more for any item when it sold via an (increasing price) English auction, rather than a (declining price) Dutch auction.

One explanation of this bias is that a declining price generates a sense of "suspense", whereas an increasing price tends to give the ultimate victor in a bidding war an extra sense of satisfaction. Therefore, the social pressure not to lose may add an extra utility to winning beyond that of the item purchased. Another explanation is the idea in a common value auction that seeing other people bid for the same thing increases its probable value. (See: Classroom experiment example of Dutch auction under-valuation.)

Yet another explanation is that the optimal bidding rule is more transparent in an English Auction than in a Dutch Auction. In an English Auction (for a single unit of a good) the optimal bidding strategy is to continue raising one's bid until doing so would result in a price that exceeded one's value for the unit. (Given infinitely small bidding increments this results in a unit price equal to the second highest value among the bidders.) In a Dutch Auction the optimal bidding strategy is to allow the bid price to fall to a precise fraction of one's value. The fraction is (N-1)/N, where N is the number of bidders in the auction. Naive bidders (and even many experienced bidders) are unlikely to know this bidding strategy, and may therefore allow the price to fall "too low" before purchasing.

I'm not sure this belongs here. I might be alright with restoring it, or moving it to Dutch auction, minus the "naive" comments. But the argument doesn't seem quite right. Suppose you and I are bidding on something we both value at $100. Therefore, according to the above (N=2), you wait for $50. But I accept at $75. How was your strategy optimal? Certainly it's not an equilibrium strategy. 192.75.48.150 17:54, 29 June 2006 (UTC)

I think (N-1)/N applies when valuations are independently and identically distributed across bidders and are draws from a uniform distribution from 0 to some upper value.