Bidding and Information: Evidence from Gilt-Edged Auctions

This paper looks in detail at pricing in Gilt Auctions. It compares the prices received at all auctions between May 1987 and February 1995 with both the When-Issued price (the pre- auction market for delivery of the auction stock) and the price of a comparable parent stock (when one existed). The paper finds a marked difference in performance between non-fungible stocks (when a parent exists but the tranche cannot be traded identically with the parent until a few weeks after the auction due to part payments and different first coupons) and fully-fungible stocks (when the parent and tranche trade identically from auction day). The average auction price received for non-fungible stock was significantly below both the When-issued price (about 13p per o100) and the adjusted parent price (about 24p per o100). In the case of fully fungible auctions there was no discernable price difference between auction stock, When-issued, or adjusted parent stock price. Auction prices below comparable secondary market ones is a common result for Government debt auctions (similar results have been found for the US, Germany and Mexico). The explanation usually given is that uncertainty about the correct price leads bidders to shade down their bids. The paper examines this explanation and finds that although it can help explain the results for non- fungible auctions (measures of uncertainty seem to be correlated with the degree of price difference) it does not explain the markedly different results observed for fully-fungible auctions since the dispersion of bids for these auctions is no less than for non-fungible ones suggesting that some uncertainty still exists. This is a slightly revised version of a paper that was published in July 1995 as part of the Debt Management Review.