Estimating the number of new and repeated bidders in construction auctions

Abstract

The number of new bidders – bidders from whom there is no previous registered participation – is an important variable in most bid tender forecasting models, since the unknown competitive profile of the former strongly limits the predictive accuracy of the latter. Analogously, when a bidder considers entering a bid or when an auctioneer is handling a procurement auction, assessing the likely proportion of experienced bidders is considered an important aspect, as some strategic decisions or even the awarding criteria might differ. However, estimating the number of bidders in a future auction that have not submitted a single bid yet is difficult, since there is no data at all linking their potential participation, an essential requirement for the implementation of any forecasting or estimation method. A practical approach is derived for determining the expected proportion of new bidders to frequent bidders as a function of the population of potential bidders. A multinomial model useful for selective and Open tendering is proposed and its performance is validated with a dataset of actual construction auctions. Final remarks concern the valuable information provided by the model to an enduring unsolved bidding problem and the prospects for new research continuations.

Keywords:

• Bidding
• forecasting
• modelling
• multinomial
• new bidders
• tendering

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

The supplementary material for this paper is available online at http://dx.doi.10.1080/01446193.2016.1231408.

Notes

1. This is not to mention the known, but dubious, practice of informal intercommunication between bidders and their contacts – the bidders ‘grapevine’.

2. Until Equation 1 is simplified later into Trinomial and Binomial models, this tentative multinomial model only considers the promotions, not the desertions (only positive changes or increases of x_r), of bidders from the r group to the r + 1 group. Otherwise the sum of outcomes x₁, x₂, …, x_i would be zero, since the promoted r-bidders would the same as the leavers from the r − 1-bidders category; if not for the new bidders (r = 1 bidders, that is, x₁) who would be the only ones that would add (they would be the only new ones), but not subtract.