How aggressive or conservative any given bidder is relative to his competitors in a federal offshore oil and gas lease sale frequently varies considerably from sale to sale. In spite of the variability, tracing the relative aggressiveness through time of various bidders strongly suggests that widely different bidding strategies are being followed. However, in any sale given the measure of relative aggressiveness defined in this paper and called the bidder's bias, the results achieved in a sale (such as frequency of winning and average money left on the table) can be predicted fairly accurately from a statistical model of the bidding process. When the same measure of relative aggressiveness is calculated for the federal government's presale estimates, they are found to be consistently conservative. Furthermore, the average placement of these estimates relative to the bids submitted in a sale is predicted accurately by the same statistical model. The clear implication is that the bidding process can be considered to be a statistical game and that results obtained by a player are determined by how aggressive he is relative to his competitors.
We define as unbiased any bidder for federal offshore oil and gas leases whose bids split the population of other bids into two equal subpopulations, population of other bids into two equal subpopulations, one all greater and one all less than his bids. The value of fB for such an unbiased bidder will be 0.5. Observe that a bidder's bias thus is measured relative to the bids submitted by a jury of peer bidders.
If a bidder is aggressively biased, more than half the jury will bid lower. The value of fB for an aggressive bidder will exceed 0.5; the more aggressive the bidder, the higher his fB. For a conservatively biased bidder, less than half the jury will bid lower. Conservative bidders will have a value of fB less then 0.5; the more conservative the bidder, the lower fB.
The essential thrust of this study was to analyze the bias exhibited by bidders for federal offshore oil and gas leases offered by sealed, competitive bonus bidding. (We delete from consideration any low, noise bids as per the algorithm used previously.) Also, the presale estimates prepared by the federal government starting in 1968 are treated. Throughout, we consider all leases receiving any bids regardless of whether the lease subsequently was issued or not.
In Section I we study the aggressive and conservative bias shown by bidders in six recent sales, finding great variability in bidding characteristics. In Section II we examine for the same sales how bidders' actual outcomes compare with statistically predicted results; the good agreement supports the plausibility of our statistical model. In Section III we examine how the aggressive and conservative stance of seven major bidders varied through time; we find great variability from sale to sale, but detect clear evidence of different bidding philosophies. In Section IV we make for these seven philosophies. In Section IV we make for these seven major bidders through time a similar comparison with that made in Section II for all bidders in six recent sales and find similar results. In Sections V through VII we present results of similar analyses for federal presale estimates. In Section V we present the presale estimates. In Section V we present the results of applying our statistical model of bias to an estimator whose values are not used in computing bid parameters; we conclude that such estimates can be expected to deviate widely from bids submitted, particularly when relatively few bids are submitted. particularly when relatively few bids are submitted. In Section VI we examine the bias of federal presale estimates and find it to be consistently conservative. In Section VII we compare the observed relationship between federal presale estimates and bids submitted in all recent sales to that predicted by our statistical model and find the agreement to be about the same as the agreement presented for the bids themselves.
Five supporting appendices, A through E, are referred to that are not included in the text. Copies of these are available from the authors.