Forecasting the Number and Distribution of New Bidders for an Upcoming Construction Auction

Abstract

Estimating the number of new bidders in construction auctions is relevant for both private companies and contracting authorities. For private companies, it allows the total number of competing bidders to be estimated, which may lead to better adjustments of future bids. For contracting authorities, it allows the population size of all potential bidders to be estimated and thus to implement better awarding criteria. Mathematical models for forecasting the number of new bidders and the population size of all potential bidders are, however, very scarce in the construction management literature. In this paper, we propose an exponential model for predicting the average number of new bidders based on an urn analogy. The model allows the number of new bidders to be estimated as a function of new versus total participating bidders observed in previous auctions. The parameter estimates obtained from the model also allow the statistical distribution of the number of potential new bidders to be modeled using a sum of binomial distributions. We validate the exponential model on three published construction auction data sets, showing that the proposed model significantly outperforms the multinomial model—the most advanced model for performing similar tasks found in the literature.