A Detector and Evaluation Framework of Abnormal Bidding Behavior Based on Supplier Portrait

Introduction

Bidding management is an important part of the healthy development of enterprises. At present, most company use two bidding models-invitation bidding and public bidding, which adopted by many power companies (Peng, 2018). The invitation bidding has strong motivation, suppliers can participate in bidding activities only after receiving an invitation from the enterprise. This form is controllable and can clearly control the number, nature and scope of suppliers. Public bidding refers to that the company directly publishes the bidding announcement on the media, and the suppliers participating in the bidding choose to participate according to their own conditions. The model cannot control the number, nature and quantity of suppliers.

Usually, the bidding scoring formula based on the price scoring stage in the bidding behavior can be verified as Equation (1). In particular, P is the scoring bid price and the B value is calculated by averaging the truncated P value. The value of B is named as the best reference value. Therefore, the closer the bidding price of the supplier group is to the mean value, the higher the score will be. Based on this Equation (2), it can be considered that if the bidding prices of group suppliers are all in a range; it is most beneficial for suppliers to control the overall average price near their own bidding prices. Considering the profit-seeking characteristics of businessmen, suppliers are more inclined to raise the b value to improve their profits:

The main object of enterprise bidding is the supplier. Whether the supplier group carries out the bidding behavior well determines whether the bidding work can be carried out smoothly. Therefore, how to accurately locate and detect suppliers' abnormal bidding behavior is a very important scientific issue. There are mainly three types of supplier ‘s abnormal bidding behaviors: enterprise affiliated operation, subcontracting and collusion (Huang, 2019). Enterprise affiliated operation refers to the undertaking of projects by others in the name of other construction units, which will lead to many projects being undertaken by unqualified construction enterprises. Subcontracting means that the enterprise transfers the winning project to other enterprises through subcontracting after it wins the bid, and the enterprises that take over do not have the bidding qualification. Collusion refers to the fact that the owner determines the construction enterprise during bidding, and then turns the open bidding into invitation bidding. In the stage of winning the bid at a low price, such collusion between the construction enterprise and the design institution will be existed.

Due to improper bidding behavior, it will affect the normal work of the bidder and bring great risk of operation and maintenance (W.Z. Wu, 2019; Z.J. Wu, 2019; Wei, 2019; Xia, 2019; He, 2020; Wang, 2011). The relevant literature proposes some auction rules for the bidding of traditional behaviors to prevent collusion, such as establishing a reservation price (Graham & Marshall,1987), selecting an effective auction mechanism according to the correlation between the conspirators(Laffont & Martimort, 2000), and selecting an effective auction mechanism by using the information asymmetry related to the potential conspirators, including the extraordinary probability of not selling the auction object(Che & Kim, 2006,2009), including the effective upper limit And the lowest price to choose an effective auction mechanism(Chowdhury, 2008). However, collusion schemes are always difficult to detect. They are usually negotiated in a strictly confidential way, and they are usually not obvious from the results of an auction, because the bid encirclement must present a truly competitive bidding and effective auction mechanism (Bajari & Summers, 2002). Another problem is that an effective strategy to avoid collusion usually requires the auctioneer to be able to predict the distribution function of the value that the bidder obtains from it, and know which interest group the bidder belongs to and choose an effective auction mechanism (Bajari &Summers, 2002), although obtaining this information in practice is complicated, if not impossible.