6. Study 3: Comparison of Auctions and Negotiations
6.5 Data analysis and results
6.5.4 Group comparisons
The analysis and results of hypotheses testing also indicate potential impacts from: (1) the experimental setting (laboratory vs. online), (2) the outcomes (winners vs. non-winners), and (3) the buyer’s role and behavior in exchanging different types of information. Thus, further analyses were conducted to examine their effects.
Comparison 1: laboratory setting vs. online setting
made significantly more bids and greater concessions in the laboratory than online (T1 vs. T3). The suppliers in negotiations did not behave differently between the two settings (T2 vs. T4) in terms of number of offers and total concession. The transactions were converged significantly faster in the laboratory than online for both auctions and negotiations (p<0.01). On average, it took two times longer when the transactions were conducted online.
In terms of the outcomes, there was no significant difference in negotiations between the two settings (T2 vs. T4). In auctions, the buyer’s profit was slightly decreased from the online setting than in the laboratory (66.94 vs. 75.82, p<0.1), while the supplier’s profit was significantly improved when they played online (3.94 vs. -7.82, p<0.05). On average, the suppliers were able to reach profitable contracts when they were given a longer time in round duration and total transaction length. They also achieved more efficient contracts with higher joint gains and outcome equity (p<0.05), though the Pareto optimality did not differ from the laboratory setting.
It was also found that the suppliers in auctions reported a higher level of assessment of the process and outcomes when they were bidding online. There were no significant differences in their assessment of the systems between the experiments, neither in auctions nor in negotiations.
Overall, the results were consistent between the two experiments, whereas the effects of mechanisms on process, outcomes and assessment were weakened in the online setting. This was expected as stronger effects could be observed in a more controlled setting (i.e. laboratory experiment). It is also possible that the time pressure and competition level were higher in the laboratory setting, which caused larger concessions from the suppliers in auctions and thus led to worse contracts for them.
Comparison 2: winning suppliers vs. non-winning suppliers
The winning suppliers were awarded with the contract and they may perceive themselves to be more successful than other suppliers (i.e. non-winners). Since the convergence speed and the outcome
variables are measured at the transaction level (i.e. an agreement was reached between buyer and supplier), they are not applicable to those who did not reach a contract. Thus, the comparison was performed on the suppliers’ behavior in the process and their assessment of the process, outcomes and system. Table 6-4 shows the results of a group comparison between winners and non-winners.
Table 6-4. Comparison between winning suppliers and non-winning suppliers
Experiment 1 Experiment 2 T1 T2 T3 T4 Groups WS NWS WS NWS WS NWS WS NWS No. of suppliers 28 81 23 49 17 49 38 88 Process No. of bids/offers 8.50* 5.41 3.09 3.16 5.35^ 4.04 3.58* 2.75 Total concession 70.64* 51.51 16.70 20.43 49.65^ 39.16 29.61^ 18.14 Assessment Process 3.60 3.40 4.52 4.13 4.26* 3.41 4.23* 3.66 Outcomes 3.26* 2.52 3.99* 2.79 4.56* 2.75 4.18* 2.57 System 3.89* 3.33 4.42* 3.55 4.37* 3.29 4.41* 3.71 Note: (1) “WS”–winning suppliers, “NWS”–non-winning suppliers; (2) numbers are mean values; (3) significance for comparison in each treatment: * p < 0.01, ^ p < 0.05.
The winning suppliers submitted greater number of bids/offers than the non-winning suppliers in three treatments: laboratory auction (8.50 vs. 5.41, p<0.01), online auction (5.35 vs. 4.04, p<0.05) and online negotiation (3.58 vs. 2.75, p<0.01). They also made larger concessions than the non- winners in those three treatments. This indicates that the winners were more actively participating in the transaction and competing against other suppliers.
Overall, the winners had significantly positive assessment of the process, outcomes and system than the non-winners (p<0.01). This suggests that it is the outcome (i.e. winning or losing the contract) and not the concession-making that affects their assessment. In turn, this may indicate the low payout from making large concessions that result in loses. The only exception was their assessment of process in the laboratory experiment, which might be due to the shorter time frame and faster convergence. The results indicate that the suppliers’ outcomes indeed affected their perception and evaluation of the transaction process, outcomes and systems. When they won the competition and were awarded with the contract, they had more positive attitude towards their assessment.
Comparison 3: public information vs. private information
It was noted that one of the differences between the two mechanisms is the buyer’s role and behavior. In auctions, the buyer does not participate during the process; thus, the suppliers receive only public information about the winning bids and admissible bids that are automatically generated and updated by the mechanism. There is no private information nor offers from the buyer to the suppliers.
In negotiations, the buyer can not only send messages to the suppliers but also make offers or counter-offers. If the buyer only sends messages, then her role is similar to the buyer in auctions; there is no explicit concession and obligation from the buyer side. However, when the buyer makes counter-offers, it often requires reciprocity and concessions and thus decreases their gains. Then, the concession-making pattern is different from auctions (two-way vs. one-way).
The buyer’s different role and behavior, particularly in exchanging different types of information (e.g. message only, reciprocal offers), may affect the transaction process and outcomes. Thus, a further analysis on the transaction process was conducted to examine the effects of different information conveyed from buyer to suppliers. The result shows that different types of interactions could be identified based on the information types. Specifically, three types of information were identified:
(1) public information in auction with winning bids and admissible bids;
(2) private information in negotiation without buyer’s offers (messages only); and, (3) private information in negotiation with buyer’s offers (messages attached to offers).
The effects were examined by analyzing the suppliers’ bids or offers that were followed or replied to the different information. Considering the differences between buyers (e.g. experience, strategies), the analysis was conducted at the transactional level, i.e. the suppliers’ responses to the information from the same buyer in each auction or negotiation.
There were a total of 2,307 bids, offers and messages recorded from the 434 participants (373 suppliers and 61 buyers) in the two experiments. All the bids were included in the analysis as the bids were submitted following the public information announced by the buyer through the auction mechanism (N1=965). The buyer’s offers and messages in negotiations were only used to categorize the suppliers’ subsequent offers. The opening offers made by suppliers were excluded as they were not replying to buyer’s offer or message. The remaining records included 81 offers replying to buyer’s message (N2=81) and 372 offers replying to buyer’s offer (N3=372). Table 6-5 shows the results of a comparison between the different types of information.
The results confirm that different types of information provided by the buyer affected the transaction process in terms of number of bids/offers and concession-making. The buyer’s concessions were calculated with the value change between the supplier’s offers but based on the buyer’s preferences. Thus, it increases the buyer’s gains resulting from the concessions made by the suppliers.
Table 6-5. Comparison between different types of information
Experiment 1 Experiment 2
Auction Negotiation Auction Negotiation Information Public no B-offer Private, Private, B-offer Public no B-offer Private, Private, B-offer No. of bids/offers per transaction 24.14*# 1.19# 6.48 17.00*# 1.47# 6.21 Supplier’s concessions 10.81# 9.36# 3.76 9.19^ 5.47# 11.42 Buyer’s concessions 11.21# 10.90# 4.41 10.39^ 7.08+ 11.92 Note: (1) numbers are mean values; (2) significance when comparing to private information without offer: * p < 0.01, ^ p < 0.05; (3) significance when comparing to private information with offer: # p < 0.01, + p < 0.05; (4) three types of information: public information in auction (public), private information in negotiation without buyer’s offers (Private, no B-offer), and private information in negotiation with buyer’s offers (Private, B-offer).
The public information revealed in auctions led to a greater number of bids submitted by the suppliers. The private messages without buyer’s offer did not motivate the suppliers in making counter-offers, whereas the suppliers made more offers when responding to the buyer’s offers. Similar results were found in the two experiments, though fewer bids and offers were made in the
online setting.
The suppliers made significantly larger concessions in their bids following the public information in auctions. In negotiations, buyer’s messages also led to greater concessions in supplier’s subsequent offers, while buyer’s offers resulted in smaller concessions in suppliers’ counter-offers. There were no significant differences in suppliers’ concession-making between the public information in auctions and the buyer’s message in negotiations. However, different results were found in the negotiations in Experiment 2. The suppliers made much larger concessions when replying to buyer’s offers and smaller concessions to buyer’s messages. The suppliers’ concessions led to an increase in the buyer’s gains. In the laboratory setting, the buyers benefited from their public information in auctions and messages in negotiations; in the online setting, their gains were improved by making reciprocal offers to the suppliers.
This indicates that the private messages in negotiations had the same effect as the public information in auctions in the laboratory setting, which may be due to the similar environment with high time pressure. In the online setting, the suppliers had longer time to verify the information from the buyer. The public information in auctions is more transparent and valid for all suppliers than the private information in negotiations. In particular, a message without an explicit offer becomes difficult to verify. Thus, the suppliers may make their bids/offers based on the public information or buyer’s offers, in which case they may feel it is worthwhile to make concessions.
6.6 Discussion
E-procurement has advanced with the adoption of information technologies and various market mechanisms, leading to cost savings, strategic advantages and enhanced business relationships. Effective procurement depends not only on the proper selection of products and/or services but also on the appropriate selection and use of market mechanisms. Despite the number of general guidelines that have been formulated, there is lack of empirical evidence that can assist and suggest strategic
choices of various mechanisms.
Business procurement often involves multiple parties and multiple attributes, which requires advanced market mechanisms. Auctions and negotiations are traditionally two different classes of market mechanisms. With the advancement of information technologies, these two different mechanisms have recently been extended to facilitate and govern such procurement transactions: from single-attribute auctions to multi-attribute auctions and from bilateral negotiations to multi- bilateral negotiations. This study takes a further step to experimentally compare two such mechanisms in e-procurement: multi-attribute reverse auctions and multi-bilateral multi-attribute negotiations. Their differences in the process, outcomes and suppliers’ assessment were investigated in two experiments, one in the laboratory and one on the Internet. Group comparisons were also conducted to further examine the effects of experimental setting, suppliers’ outcomes and buyer’s behavior.