Based on individual case description, we will focus on finding cross-case patterns and matching these patterns with our propositions in this section. The findings are summarized in Figure 3. In the figure, a contingency is summarized in a row, and each column indicates the functionality and the relationship it supports, which appear at the bottom of the figure. Since the functionalities are organized according to the combination of the detailed functionality and relationship they support, it is possible that one functionality can appear in more than one column (such as product catalogues in EM A), and two EMs in two different industries can appear in the same column. For example, both EMs D and E offer reverse auctions, and reverse auctions involve both long and short-term relationships, so the combination of reverse auctions and long-term relationships is explained in one column and the combination of reverse auctions and short-term relationships is explained in another. In each cell of Figure 3, the fractional number in the bracket is the number of cases reporting a high/low value/the total number of cases studied for each functionality. The exceptional cases are transaction uncertainty, transaction frequency and non-contractible factors. For transaction uncertainty, we provide a brief description of the uncertainty. For transaction frequency, we provide a brief description as well as a fractional number indicating the percentage of cases reporting such a high/low value. For non-contractible measures, we provide in the bracket the median value of the five items (see Table 2) used to gather data from interviewees on the level of non-contractible factors for each functionality, based on a five point Likert scale.
In each of the following sub-sections, we will explain the propositions according to where in the figure the functionality is located. If it is to the left of column 4, it represents more market-oriented functionalities, and if it is to the right of column 4, it represents more collaboration-oriented functionalities. Finally, column 4 includes reverse auctions with long-term relationships, which represent a transitional area where mixed responses can occur.
We examined three kinds of uncertainties for each functionality: Predictability of the products to purchase, predictability of how many to purchase (exaggerated by order quantities since the bigger the quantity, the worse the consequences of order fluctuation) and market variability (measured by the rate of price and supplier change, which are often seen in fragmented industries because they are more dynamic and less stable than concentrated industries). Row 2 of Figure 3 provides a comprehensive view of the transaction uncertainty involved in each functionality, derived from the more detailed view of Table 3.
Table 3. Transaction uncertainty in EMs
| ||Functionalities||EM||Predictability of what to buy||Predictability of how many to buy|
| ||Product catalogues||A||Unpredictable (Products change from one customer to another)||Unpredictable (Depends on end users; order changes can be 50-60%)|
| ||Forward auctions||B||Predictable (Can predict what machine they will buy)||Predictable (The number of machines they are going to buy is controlled by the budget)|
|Transaction facilitation||C||Predictable (All normally used medical-surgical products)||Unpredictable (Daily or weekly prediction) (Fluctuates by numbers of incoming and outgoing patients) (Order size varies)||Prices and suppliers stable (Concentrated on supply side)|
| ||D||Predictable (Normally used mining supplies such as welding gases, industrial tools)||Unpredictable (Big variation)||Prices and suppliers stable (Fragmented on supply side)|
| ||Reverse auctions (Short-term)||D||Predictable||Predictable|
|Reverse auctions (Long-term)||D||Predictable (6 months)||Some predictable and some unpredictable||Prices and suppliers stable (Fragmented on supply side)|
| ||E||Predictable (Months-years)||Most are predictable (Months-Years out) small order fluctuation, big size orders||Prices and suppliers stable (Slightly concentrated on supply side)|
| ||CPFR||E||Predictable (Months-multiple years)||Less predictable, (Week- 3 months); huge order fluctuations (25%-300%); big order size|
We found that market variability was low for most EMs except for EM B, which sells used heavy construction equipment. The price of used equipment changes from one machine to another, and the supplier set for used heavy construction equipment is not stable. Many companies, such as financial and construction companies, can temporarily serve as suppliers when they have used equipment to liquidate. Dealers also enter and exit this market frequently. Before dealers and end users actually see a used machine, they are not sure what the price will be and from whom they will buy. In the following we will focus on examining the uncertainty caused by the predictability of what and how many to buy.
To the left of column 4, the predictability of what and how many to buy is low except for the EM A product catalogue. In EM A, since each promotional product is custom made, the distributor cannot predict what artwork (e.g., logos) the end user may require. Sometimes, even clients are not sure about the artwork. They may ask for distributor suggestions, and then decide what to buy. This causes great uncertainty for distributors. At the same time, the size of orders fluctuates a lot from one order to the other, and they are “all over the place.”2 This further increases distributor uncertainty. Most of the time, distributors handle demand uncertainty by purchasing from long-term suppliers. When long-term suppliers do not offer the products, distributors purchase from short-term suppliers.
To the right of column 4, the predictability of what to buy is high but companies are uncertain of how many to buy. In EMs C and D transaction facilitation and EM E CPFR, the traded products are either consumables or retail products that are needed on a daily basis. The specifications of these products are set in advance, but the quantities fluctuate according to current needs, production schedules, and seasonality. The high demand for these products also intensifies order fluctuation.
In column 4, when reverse auctions are used to negotiate long-term contracts with suppliers in EMs D and E, in most situations, buyers can predict what and how many they are going to need at least six months out. “In most of the buyers that we dealt with, they pretty much know twelve months out, now its time to buy this, now its time to buy that, and this is when I do this kind of item.” Only one buyer who dealt through EM D suggested that some products were predictable and some were not predictable.
According to Proposition 1, we should see in Figure 3 low transaction uncertainty3 to the left of column 4 and high transaction uncertainty to the right of column 4. Since the trend in Figure 3 matches Propositions 1a and 1b well, this leads to the conclusions that a company is more likely to use market-oriented functionalities when it faces low transaction uncertainty (columns 1 and 3), or when transaction uncertainty is high but no long-term relationships are available to absorb this uncertainty (column 2). In contrast, a company is more likely to adopt collaboration-oriented functionalities when it faces high transaction uncertainty, and long-term relationships are available to absorb this uncertainty (columns 5, 6, and 7).
Proposition 2 indicates that collaboration-oriented functionalities are more likely than market-oriented functionalities to involve transaction specific investments. From row 3 of Figure 3 we see such a pattern with the exception of EM A product catalogues.
In EM A, no participating companies reported transaction-specific investments in long-term relationships with their trading partners. Although promotional products are customized to each order, supplier production is mass customized so that every distributor can access it without supplier-specific investments. This property applied between trading partners with both long- (column 5) and short-term (column 2) relationships.
The data from other EMs matches our propositions well. First, to the left of column 4, when forward auctions (EM B) and reverse auctions are used to support short-term relationships, there is no specific investment made. Especially for EM B forward auctions, buyers and sellers do not necessarily know each other when purchasing online, so of course no specific investment is made.
To the right of column 4, all the participants observed transaction-specific investments of some kind, including proprietary knowledge about suppliers, supplier-specific information system investments, specific production equipment, specific human resources, warehouse investments, and the resources needed to set up CPFR initiatives. “We brought in people to specifically work with them through EM D to make this happen.”“This particular customer is what we call an anchor customer. That is, we have a location in that city because this customer is there… If that customer wasn't there, we probably wouldn't be located there….”
For column 4 reverse auctions, we found mixed evidence of both high and low specific investments. Specific investments in the reverse auction-long-term relationship combination were high, since five out of seven cases reported such investments as warehouses, information systems, specific knowledge, and special production capacity. Among these, four cases were investments made by suppliers. We therefore believe that suppliers seem to be at a greater disadvantage in reverse auctions than normally considered. Reverse auctions not only reduce supplier margins, but also appear to make suppliers even more “captive” to the relationships.
Examining the functionalities across the different columns of row 3 in Figure 3, we see that there is an approximate association between functionalities and transaction-specific investments except for the product catalogues in EM A. Therefore, proposition 2 appears to be confirmed. A company involved in relationships that require high transaction-specific investments (columns 5 and 7) tends to use EM collaboration-oriented functionalities. Otherwise, it tends to adopt market-oriented functionalities (columns 1, 2, and 3).
Row 4 of Figure 3 demonstrates an overview of the entire transaction frequency picture. Transaction frequencies tend to be higher in collaboration-oriented functionalities (columns 5, 6, and 7) than market-oriented functionalities (columns 1, 2, and 3).
To the left of column 4, EM A distributors order from short-term trading partners infrequently when those products are not available from their long-term suppliers. For EM B forward auctions for used heavy equipment, and EM D reverse auctions supporting short-term relationships, both involve one off purchases. Thus the transaction frequency is low.
For row 4 functionalities to the right of column 4, all involve high transaction frequencies. In EM A, the incentive for distributors to purchase from long-term suppliers frequently is to obtain volume discounts. Products are ordered daily or weekly through the transaction facilitation functionality of EMs C and D. One EM C participant called these products “high volume movers.” There are three reasons why such “high volume movers” use transaction facilitation and long-term relationships: (1) This avoids time consuming and frequent negotiations, and saves staff time for more productive work; (2) It consolidates transactions to fewer strategic suppliers; (3) It motivates buyers to follow company purchasing policy and reduce maverick purchases. In CPFR, as conducted through EM E, products involved are either branded products or promotional products. These are high demand consumer products, so they need to be re-ordered frequently.
As for reverse auctions used to support long-term relationships (column 4), we found extreme mixed evidence for transaction frequencies. On one hand, as a buyer commented, “the more we're using them on a daily basis, the easier it is for us to be able to roll it into reverse auctions because we've got bigger volumes associated with it.” On the other hand, some buyers were testing most of the product lines to see if they were suitable for reverse auctions. For them, “everything is possible in reverse auctions,” and the transaction frequency ranged from very low to very high.
Proposition 3 is therefore confirmed, with the conclusion that transaction frequency is a strong indicator of a company's choice of collaboration vs. market-oriented functionalities. Reverse auctions used to manage long-term relationships seem to serve as a transitional area between market-oriented and collaboration-oriented functionalities, in terms of transaction frequency.
Complexity of Product Description
Proposition 4 suggests that products with simple descriptions are more likely to be transacted through market-oriented functionalities, and products with complex descriptions are more likely to be transacted through collaboration-oriented functionalities. Unfortunately, no such pattern was found in the data analysis, since all the functionalities involved products that were standard and easy to describe (row 5 of Figure 3).
The following products were traded through the five EMs: (1) EM A promotional products. Although promotional products tend to be highly customized, the base products advertised are quite standard and easy to describe. (2) EM B used heavy construction equipment. These items are complex from the average consumer's point of view, but they are standard and easy to describe for dealers and other people in this industry. (3) Medical-surgical products in EM C transaction facilitation, welding gases and mining supplies in EM D transaction facilitation, and packaged consumer goods in EM E. Most are standardized and easily to describe.
A few non-standardized products were also traded: (1) In EMs C and D transaction facilitation functionalities, two suppliers offered nonstandard capital goods, normally perceived to be hard to describe. However, suppliers suggested that complex products were not suitable for transaction facilitation functionalities. (2) In EMs D and E reverse auctions, both easy-to-describe and complicated products were involved, such as store construction contracts and employee compensation insurance. But most of the products tended to be simple and standardized and, according to interviewee comments, more complicated products do not lend themselves to reverse auctions as much as easy-to-describe products.
From the general overview of row 5 in Figure 3, the complexity of product description was found to be low in most cases. Proposition 4 about the complexity of product description could therefore not be confirmed. It seems that easy-to-describe products are more suitable for both market-oriented and collaboration-oriented functionalities than are complicated products.
From row 6 of Figure 3, we can see that there is a difference in the level of non-contractible factors, such as trust, quality, information sharing, supplier innovativeness, and supplier responsiveness, between collaboration-oriented and market-oriented functionalities.
To the left of column 4, in EM A product catalogues and EM D reverse auctions supporting short-term relationships, and EM B forward auctions, the level of non-contractible factors is low. One exception is that buyers or sellers working with EM B value quality and supplier responsiveness. In the used heavy construction equipment market, quality is very important for safety and construction delay reasons. However, in EM B, equipment quality is controlled by inspections or onsite checks, not a specific governance structure such as long-term relationships. Suppliers were reported to be responsive to buyer requirements. A preliminary explanation is that in a competitive business environment, where the buyers have more and more power, suppliers are more responsive to customer requirements due to competitive pressures.
Moving to the right of column 4, in EM A product catalogues, suppliers valued non-contractible factors for reasons of trust, assured quality, favours for rush orders, and ease of getting credit. For EM C transaction facilitation, (1) mutual trust and level of information sharing between these business partners are high due to the high transaction volume involved and the use of zero inventory practice; (2) Quality is very important for hospitals since “hospitals are not dealing with products, but patients.” The ordering of low quality or legally unproved products could lead to severe consequences; (3) New medical tools and medicines are introduced frequently due to advances in the medical field (an average of every 18 months). Supplier innovativeness is important to hospitals in order to serve their patients better; (4) Sharing information and synergies with suppliers was important, especially for research-oriented hospitals.
EM E CPFR is used only when the business partners have trusting relationships. Participating suppliers are usually innovative and responsive to buyer requirements, and they supply higher quality products to these buyers. Information sharing is intense in CPFR since it is the basic requirement of CPFR.
In column 4, the highly ranked non-contractible factors were counter-intuitive for reverse auctions used to support long-term relationships, since reverse auctions are often treated as a tool that causes distrust and disincentive between buyers and sellers. The following are some explanations. First, trust and quality is important to buyers, since reverse auctions are often used to buy “quite important things.” Second, suppliers become more flexible and innovative, not because of greater incentives in strategic long-term relationships, but because of competitive pressure and the hope of doing more business with the buyers. In some reverse auctions, buyers took into account the disincentives to suppliers in making their decisions. Buyers seemed to take measures to avoid disincentive problems such as: (1) doing invitation-only reverse auctions; (2) offering longer-term contracts; (3) in evaluating the bids, considering factors other than just pricing.
Considering the pattern shown by row 6 of Figure 3 for non-contractible factors, it is clear that the level of non-contractible factors differs between market-oriented and collaboration-oriented functionalities. The conclusion is, in agreement with Proposition 5, that companies valuing non-contractible factors tend to use collaboration-oriented functionalities. Otherwise, they tend to choose market-oriented functionalities.
Our first finding is that EM functionalities cannot be classified as dichotomous or exclusive categories. They should be treated as being positioned on a continuum between market-oriented and collaboration-oriented functionalities. When a functionality is located closer to the market-oriented extreme, it has a more competitive element to it. An EM, and even a functionality, can support a continuum of relationships. For example, many researchers treat reverse auctions as a pure market-oriented functionality (Mithas et al., 2003), but this research suggests it is a hybrid functionality with elements of both competition and collaboration. Jap (2003) also recognized the role of reverse auctions in supporting long-term relationships. Public product catalogues, such as those appearing in EM A, support more than short-term relationships and search. This finding is also somewhat different from the literature, since the aggregation of many product catalogues is considered to be a reflection of increased market efficiency (Anonymous, 2000; Bakos, 1998).
The case analysis confirmed Proposition 1a and 1b about transaction uncertainty, Proposition 2 about transaction specific investment, and Proposition 3 about transaction frequency. However, the complexity of product description proposition (Proposition 4) was not confirmed. It seems that simple products tend to be transacted in all the functionalities we studied, instead of just market-oriented functionalities. RFP and product co-development are two functionalities that tend to involve complex products. These functionalities appeared frequently, but were not studied, due to the lack of usage among participants. Our disconfirmation of Proposition 4 does not agree with some of the literature such as Mahadevan (2002), who argued that complex products tended to be traded in collaboration-oriented functionalities and simple products tended to be traded in market-oriented functionalities. But our outcome is in accordance with (Garcia-Dastugue & Lambert, 2003) who argued that closed auctions (similar to RFP) involved complex products, and most other functionalities only dealt with simple products.
Non-contractible factors, such as trust, information sharing, quality, supplier responsiveness and innovativeness, were confirmed as strong indicators of which functionalities to choose (Proposition 5). This finding is in accordance with Bakos and Brynjolfsson (1993), but differs from Mithas et al. (2003). Mithas et al. studied the explanation power of asset specificity and non-contractible factors in the choice of using or not using reverse auctions, and concluded, “even after controlling for asset specificity, buyers prefer to avoid reverse auctions if they value supplier investments in non-contractible factors such as quality, supplier innovation, information sharing, trust and flexibility.” However, this study sometimes showed that, even if buyers valued non-contractible factors, they still used reverse auctions. This discrepancy can be explained by:
(1) Reverse auctions are a new type of tool that buyers are in the process of testing. Most users were among the first movers in using reverse auctions, and they could be overly optimistic about the value of non-contractible factors in their relationships with suppliers.
(2) Some buyers may attempt to use reverse auctions due to their benefit of price reduction, even if they value non-contractible factors.
(3) Some buyers appear to be taking cautious measures to control the incentive aspects of reverse auctions, such as providing longer-term contracts and considering factors other than price. As a result, most of the time, incumbent suppliers win the contracts.
(4) Competitive elements are seen more and more to be introduced in the long-term relationships in EMs. Even in CPFR, some buyers introduce competitive bidding for product slots in collaboration with strategic suppliers (mentioned as “direction of auction” by one participant). Reverse auctions can be viewed as a technique used by buyers, both to apply competitive pressure and to nurture long-term relationships, in the purchase of important products. In this study, we have seen that, because of the competitive pressure on suppliers participating reverse auctions, they may have to provide non-contractible benefits to buyers, as well as transaction-specific investment.
Another conclusion from this study, in accordance with Mithas et al. (2003), is that non-contractible factors are stronger indicators of the functionality choice than asset specificity. This conclusion arises from the transaction specificity proposition (Proposition 3) that was confirmed with less confidence.
Supplier Power in the Strategic Choice of Different Functionalities
The data analysis also revealed that supplier power played a role in buyer choice of functionalities. For all the participants studied, supplier power comes from two aspects: brand image and the supplier's volume of sales to the buyer (Porter & Millar, 1985). For example, in CPFR, brand image is a major consideration of buyers with whom to implement CPFR, because branded suppliers are important to competitive advantage. However, in reverse auctions, buyers always appear to choose those suppliers that can be replaced easily without significant loss.
“It's a bit different from that, the online negotiation (auctions) is usually used for things where you've got a couple of different possibilities for who you're going to buy it from. Whereas the more CPFR-like applications are entirely focused on branded goods.”“Ordinarily a big driver is how much of my business is with this supplier, and how important is this supplier to my performance in this category. So there's a revenue number that starts it out, and then within that, the retailer will usually look at how well I collaborate with this supplier today.”
Power is the political aspect of buyer-supplier relationships. Benson (1975) and Reve and Stern (1979) suggested that buyer-supplier relationships have two aspects, economic and political. TCT theory focuses on economic efficiency comparisons, and the political aspect refers to the power and cooperative/conflicting atmosphere among trading partners. Power is recognized as an important factor that affects the adoption of IOSs (interorganizational information systems), such as EMs. The effect of buyer power on EM adoption is not discussed in this paper, since buyer power has more effect on the adoption of an EM than on the strategic choice of the different functionalities, the focus of this paper. Supplier power tends to affect buyer choice of which functionality a buyer would prefer. In fact, supplier power has also been related to some transaction cost-related research, such as Lieberman (1991), Murray, Kotabe, and Wildt (1995), Kraut, Steinfield, Chan, Butler, and Hoag (1998), and Iskandar, Kurokawa, and LeBlanc (2001). Since supplier power is a newly found construct in this research, future research should consider including it as a formal contingency that affects buyer choice of EM functionalities.
Conclusions and Future Research
This research has offered an overview of the strategic choice of different EM functionalities, from the buyer-supplier perspective. We found that transaction frequency and non-contractible factors were strong indictors of which functionalities to choose. As far as transaction uncertainty is involved, firms need to discriminate among the kinds of uncertainty that lead to the choice of market-oriented/collaboration-oriented functionalities. However, the degree of complexity in product description does not appear to affect a firm's choice of functionalities. On average, both simple and complex products are traded through EMs, but simple products are traded in much higher volume than complex products. Some specific functionalities, such as RFP/Q and product co-development, can handle complex products better than other functionalities.
In our research, we noticed that reverse auctions were being used more and more to support long-term relationships (where transaction-specific investments and non-contractible factors are high), and some collaborative functionalities, such as CPFR, can involve competition. This offers some future research opportunities, since reverse auctions are a new type of tool and many users are still experimenting with it. Some possible research questions are “Can reverse auctions support long-term relationships well?” and “In what circumstance can reverse auctions perform well in the support of long-term relationships?” Also, due to the “direction of auctions” in using CPFR, “Does the use of EMs impact market structure by favouring markets instead of hierarchies?” Due to the limited scope of this research and the limitation of case methods, we could not answer these questions, but they are future research topics of interest.