A correlated bidding model for markup size decisions

Abstract

Whereas competitive bidding models have been studied for more than five decades with many factors being considered and statistical methods proposed, the correlation among bids of different companies and its effects on markup decisions have not been explored. Through a multivariate competitive bidding model, the significance of the correlation is investigated in this paper. Mechanistic arguments and probabilistic analysis based on a breakdown of cost estimates show that bid ratios are positively correlated to one another. This fact is then incorporated as a priori information into a Bayesian statistical method to estimate the correlation coefficients from historical data with missing values. The effectiveness of the proposed Bayesian method has been demonstrated through a case study. The proposed bidding model has a flexible mathematical structure, which allows one to better characterize actual varying bidding patterns. It also includes the Friedman and Carr models as its special cases. Moreover, through the use of the streamlined Bayesian method, the new model can be implemented easily in practice.