Solo leer esto
- 1. 6 Information and competitive tendering Gian Luigi Albano, Nicola Dimitri, Riccardo Pacini and Giancarlo Spagnolo 6.1. Introduction As stressed in the previous chapter by Bajari and Tadelis, it would be optimal for a procurer wishing to buy a sufficiently standardized, and contractually well-specifiable, good or service to elicit competition among potential suppliers. There are, however, many mechanisms she could use to elicit such competition. This chapter discusses and provides practical indications on how to choose between a sealed bid tendering and a dynamic auction to allocate procurement contracts between competing suppliers. It then suggests simple strategies to keep under control the duration of dynamic procurement auctions when this is a concern for the procurer. A crucial factor to consider in the optimal choice of a tendering format is the nature of uncertainty and the size of different types of costs the selected supplier will face when serving the contract. Therefore we begin with an example of such uncertain costs. Consider a procurement for cleaning services of a large company’s or public administration’s buildings. The contract may specify a variety of services including the cleaning of offices, corridors, halls and more demanding tasks such as the sanitation of laboratories. The contract also establishes that the contractor(s) will be paid a fixed amount of money per unit of surface (e/m2)1 regardless of the nature of the building. Therefore the unit price coincides across categories of surface, whereas the cost of The authors would like to thank Eric Van Damme for discussions and constructive comments on earlier versions of this chapter. 1 This is an example of fixed-price contract. The conditions under which such a contractual form may constitute the procurer’s optimal choice are investigated in Chapter 4. To simplify the exposition, throughout the chapter we shall also assume demand for service to be independent of the price at which the contract is awarded. 143
- 2. 144 G.L. Albano, N. Dimitri, R. Pacini and G. Spagnolo performing the same task in different environments may vary substantially. The sanitation of a laboratory, for example, is presumably more time consuming and requires more expertise than cleaning an office furnished only with a desk and few bookshelves. When estimating the cost of performing the contract in order to place a bid for it, each supplier has to consider at least two different dimensions. The first dimension concerns the supplier’s efficiency in performing each single task specified in the contract. Efficiency results from the interaction of the personnel’s experience in similar tasks, managerial skills and the quality of the cleaning equipment. Thus the supplier’s efficiency captures a private component in his production cost. It is private in that it is entirely firm specific. The second dimension concerns the supplier’s ability to correctly estimate the mix of different tasks in the contract: cleaning few, large buildings with administrative offices requires a different combination of material and human resources than sanitizing a large number of small laboratories. If suppliers are not completely informed about the composi- tion of the demand for cleaning services at the time of bidding for the contract, they face a common uncertainty. Uncertainty about the common component of the cost of serving a contract matters since the contractor may find out that the ‘true’ cost of performing the contract differs from his initial estimate. This may happen if the contractor submitted a bid on the basis of too optimistic a forecast of the common component. More generally, if a supplier does not take this possibility into account at the time of bidding for the contract, he may suffer from the ‘winner’s curse’; that is, he may realize that actual pro- duction costs are higher than estimated ones. On the one hand, the danger of running losses ex post may induce suppliers to bid too cautiously for the contract, which implies potentially high awarding prices for the buyer. On the other hand, the suppliers’ inability to recognize the winner’s curse may generate a too aggressive bidding that results in low awarding prices for the buyer, but may induce the contractor to cut production costs by lowering the quality of the performance. In this chapter, we explain how the buyer can profit by inducing some ‘information production’ when uncertainty about the common component of the cost of serving a contract is relevant and when suppliers’ pieces of private information about the common component are (statistically) linked or correlated. The simple information-producing device is a dynamic auction format (section 6.3). A dynamic auction format, be it increasing in discounts/scores or decreasing in prices, allows each bidder to observe the
- 3. 145 Information and competitive tendering identities of active competitors at different prices2 and, more important, the prices at which competitors quit the competition. Exiting times provide information about cost estimates of those bidders quitting the auction, thus helping remaining bidders revise their own estimates. When the nature of uncertainty concerns almost exclusively the private component of production costs, suppliers elaborate their bidding strategies on the basis of their private information only. Since learning is not an issue, the buyer can then adopt a sealed bid format that requires lower human and financial resources, is less exposed to the risk that suppliers collude, and has a duration that is perfectly determined (section 6.2). Auction length may become indeed a crucial issue in a dynamic format when bidders increase discounts (or lower prices) very slowly. In section 6.4, we will investigate how the buyer can streamline a dynamic auction without losing the benefits of information production. 6.2. Private and common dimensions in the cost function Several factors affect the cost of performing a procurement contract. Some of them are entirely firm specific while some others are common to all participating suppliers. A contract for supplying schools with heating oil involves different distribution costs depending on the distance between any single school and the location where a contractor stocks his oil reserves. Consequently, distribution costs are entirely firm specific. At the same time, when suppliers bid for the contract they are unable to predict the evolution of wholesale price for heating oil throughout the duration of the contract. Such an uncertainty is common in that it affects all suppliers. One simple way of capturing the private and common dimensions in the suppliers’ costs is by using the following general relation Cost = C(Private, Common) The relationship makes it clear that, in general, both components affect production costs, although the design of a procurement competitive ten- dering sometimes requires the buyer to establish which dimension is the more relevant, as we will see in the next two sections. 2 We are implicitly assuming that the open format takes the form of a Japanese auction rather than an English auction. See section 6.5 for the main features of these mechanisms.
- 4. 146 G.L. Albano, N. Dimitri, R. Pacini and G. Spagnolo Table 6.1. PROPER’s costs with known common component A B Reserve price 70 Private value component Estimated cleaning costs: (e/metre2) 40 80 Common value component with no uncertainty Surface to be cleaned: (metre2) 30,000 10,000 6.2.1. The private component We consider again the contract for cleaning services, briefly discussed in the Introduction, and further develop it in order to illustrate how the private component in the suppliers’ production costs may affect their bidding for the contract. The contract for cleaning services comprises two main space categories; (A) offices and corridors and (B) laboratories. Table 6.1 sum- marizes the estimated costs per squared metre for PROPER Ltd (PROPER henceforth), one of the competing suppliers. The table also indicates the exact size of surfaces to be cleaned for both category A and B.l Thus we consider the simplest bidding environment where each supplier perfectly knows the composition of the final demand for cleaning services. Hence PROPER’s bid for the contract will depend only upon his (private) efficiency component and, arguably, upon his conjectures on other com- petitors’ efficiency levels. The cleaning contract for the two types of surface is awarded through a single-lot sealed-bid tendering process, with a reserve price of e70/m2, so that any bid above this level is rejected. PROPER’s cost for performing the contract is simply a weighted average of the two unit costs, where the weights reflect the fraction of each type of space in the contract, Unit Cost ¼ ðe40 · 30000 þ e80 · 10000Þ=40000 ¼ e50=m2 PROPER can safely submit prices between e50 and e70, without losing money. The exact bid will depend on his conjectures about other compe- titors’ bids. For instance, if PROPER faces a group of rivals with large market shares and with an established reputation of high expertise in the business then he may anticipate intense competition for the contract. This would probably induce PROPER to bid closer to e50 than to e70.