Analysis of Bidding Data on Offshore Oil Lease Sales

摘要

Several types of statistical analyses are reported on the bidding data for offshore oil leases in the five sales from 1954 to 1967, in an attempt to obtain a method for identifying leases that become productive. These methods include using standardized variables, nonstandardized variables, stepwise regression on the variables, stepwise regression on the ranks of the variables, maximum likelihood estimation of the parameters, PRESS, and metroglyphs. While the methods using ranks seem to identify the producers a higher percentage of the time, the methods using the data are more accurate in identifying nonproducing leases. The percentage of correct classification was usually in the 60 to 70 percent range for all methods attempted. No single variable seems to be a consistently accurate predictor of whether a lease will be a producer or a nonproducer. (ERA citation 05:012036)