Entrepreneur: Start & Grow Your Business

Buyer-supplier contracting: contract choice and ex post negotiation costs.


by Artz, Kendall W.^Norman, Patricia M.
Journal of Managerial Issues • Winter, 2002 •

Structuring partnerships to facilitate long-term, productive relationships is a central concern of researchers in strategy, marketing, law, and economics (Dyer, 1997; Heide and John, 1990; Macneil, 1978; Williamson, 1985, 1991). According to a number of researchers, "the most challenging aspect of developing and maintaining long-term exchange relationships" (Cracker and Masten, 1991: 70) is how to develop efficient (i.e., least costly) contracts to effectively govern exchange (Macneil, 1978; Williamson, 1985, 1991). Formal contracts enable parties in an exchange to coordinate their actions and limit potential opportunistic behaviors (Dahlstrom and Nygaard, 1999; Goldberg and Erickson, 1987). Yet the initial drafting and subsequent maintenance of these contracts can be costly. Depending on the type of contract used, these "transaction costs' can include both the ex ante costs of initially establishing the contract, and the ex post costs of periodically renegotiating and adjusting existing contracts (Cracker a nd Masten, 1991; Williamson, 1985). A critical question to be answered then is "How do transactors manage the process of contract selection and maintenance?" Indeed, while the factors influencing the type of contract chosen and the ensuing costs associated with this choice are seen as critically important to the long-term viability of an exchange, very few empirical studies have examined actual contracting practices and their performance implications.

This article seeks to address this shortcoming in the literature by investigating the process of contract selection and the performance implications related to how those contracts are managed. By doing so, we seek to provide valuable insights into the contracting process that may allow firms to realize the potential gains from their exchange relationship (Williamson, 1996). In addition to the aforementioned contribution, we also seek to directly investigate the costs of contracting. While theoretical predictions provide valuable insights into the nature of contracting, empirical investigation is necessary to evaluate the normative implications of those theories (Dyer, 1997; Williamson, 1996). However, most previous studies have failed to conduct this empirical examination of the cost of transacting.

This study seeks to address this gap in the literature. By doing so, we respond to calls for research that directly measures transaction costs (Rindfleisch and Heide, 1997). It should also be noted that while various types of transaction costs such as negotiating and monitoring costs are incurred during exchanges, we focus solely on the negotiating costs that are incurred to revise incomplete contracts (Williamson, 1985). While the negotiation costs of revising contracts can include a wide variety of costs including travel and computer expenses Dobler et al. (1990) argue that the majority of negotiation costs are labor-related. Thus, we use an approach similar to the one suggested by Anderson and Weitz (1992) and empirically measure the labor costs associated with the renegotiation of incomplete contracts between original equipment manufacturers (OEMs) and their component suppliers.

Thus, we have two main objectives in this paper. First, we develop and empirically test a model of the determinants of contract choice. This model draws on economic and legal theories suggesting that the appropriate type of contract may depend on certain transactional and relational attributes (Crocker and Reynolds, 1993; Williamson, 1985). Second, we examine how exchange partners can effectively manage the costs of adjusting and maintaining incomplete contracts after they have been established. Literature in marketing, social contracting and relational exchange theory suggests that certain behavioral characteristics (i.e., relational norms) may play an important role in contract management (Artz and Brush, 2000; Lusch and Brown, 1996; Macneil, 1980).

In the next section, we discuss the concept of contract completeness. The following section develops hypotheses concerning contract choice and ex post negotiating costs. The methods section describes our data collection, measures, and data analysis and is followed by the results section. The discussion section presents research and managerial implications, suggestion for future research, and limitations of the study.

CONTRACTING

A review of the theoretical and managerial literature in economics, sociology, strategy, and marketing, combined with in-depth field interviews with purchasing personnel responsible for developing contracts, revealed several distinct contract types used in interorganizational exchanges. These contracts can be differentiated based on their degree of completeness.

Contract completeness is the degree to which the obligations of the exchange (e.g., price, quality, delivery, other terms and conditions) are outlined upfront. A totally complete contract is one in which all duties of all parties are completely prescribed for the duration of the contract (Crocker and Masten, 1991). Conversely, a totally incomplete contract places no a priori restrictions on the terms under which subsequent trade may occur (Crocker and Reynolds, 1993). In practice, contracts rarely exhibit the characteristics of purely complete or incomplete contracts. Even in relatively complete contracts, some terms and specifications are left to future determination. Likewise, even very incomplete contracts must specify some terms and conditions upfront to make exchange feasible. Thus, intermediate degrees of contractual completeness are the norm and contract completeness is a relative term.

In relatively complete contracts, the majority of each party's responsibilities and performance expectations are explicitly specified in the document (Lusch and Brown, 1996; Macaulay, 1963). This reduces the potential for opportunistic behavior by transactors after an agreement is reached because terms of the contract are not renegotiable (Dahlstrom and Nygaard, 1999). Since relatively complete contracts attempt to clearly define nearly all terms of the exchange for the life of the contract, specific terms are less likely to be violated (Dyer, 1997). However, while relatively complete contracts may provide protection, it is often difficult to identify all possible contingencies and negotiate mutually acceptable responses. Thus, relatively complete contracts often result in considerable ex ante negotiating costs.

In contrast, relatively incomplete contracts do not attempt to spell out the complete set of terms and conditions for the entire contract term at the time the contract is established. Rather, they seek to define a general process for periodic mutual adjustments of contract items such as price and/or quantity (Goetz and Scott, 1981). Since incomplete contracts do not attempt to stipulate a response to all possible contingencies, they are considerably simpler and less costly to draft than complete agreements (Crocker and Reynolds, 1993). Moreover, incomplete contracts are intended to provide the partners with flexibility to adapt to changing conditions. However, exchange partners using incomplete contracts face greater potential opportunism, or what Williamson defines as "self-interest seeking with guile" (1979: 234), after the initial agreement is reached. In addition, they may require costly ex post bargaining as the parties attempt to periodically negotiate adjustments to the contract (Crocker and Masten, 19 91).

Given the differences between relatively complete and relatively incomplete contracts, transactors face a trade-off between greater safety and predictability (and higher ex ante negotiating costs) offered by more complete agreements, and greater flexibility (and higher risk of opportunism and ex post negotiating costs) offered by less complete contracts (Tirole, 1986). What then are the factors that determine which type of contract is ultimately chosen? We explore this question in the following section.

THEORY AND HYPOTHESES

Contract Choice

A primary contribution of transaction cost economics (TOE) is its insights into the conditions that determine the appropriate structure to govern exchange (Williamson, 1996).

Thus, TCE provides valuable guidance in identifying factors that may influence the type of contract chosen for an exchange. Here, we examine the role of three such factors: environmental uncertainty, transaction-specific assets, and the potential for opportunism.

The first of these, environmental uncertainty, is defined as the inability to predict changes in factors surrounding an exchange (Walker and Weber, 1987). Creating relatively complete contracts may be fairly straightforward in simple environments. The task becomes more daunting as the parties become less confident about future conditions or the variables of concern are less quantifiable (Sutcliffe and Zaheer, 1998). As Goetz and Scott state, contracts "... tend to be less complete where the parties are incapable of reducing important terms of the arrangement to well-defined obligations" (1981: 1094).

In the OEM-supplier relationships examined here, both parties likely develop different goals and expectations for the exchange, and hence for the terms of the contract (Artz and Brush, 2000). For example, if a supplier cannot accurately forecast the price of its inputs, it will be reluctant to enter into a contract in which the sales price for its product is fixed for an extended period (i.e., a complete contract). Rather, that supplier will seek to protect itself from fluctuations in its input costs by insisting on a contract in which the sales price of its product can be periodically adjusted during the contract (i.e., an incomplete contract) (Sutcliffe and Zaheer, 1998; Heide and John, 1990). Similarly, volume uncertainty makes it difficult for an OEM to predict demand for its end products and may be hesitant to commit to purchasing a specified quantity of a supplier's component (Walker and Weber, 1987). However, without the OEM's commitment, the supplier will be hesitant to invest in production capacity f or fear it will be stuck with costly excess capacity if OEM sales do not materialize (Helper, 1991).

Environmental uncertainty makes it more difficult for exchange partners to negotiate complete contracts (Rindfleisch and Heide, 1997). Instead, transactors will be more likely to rely on relatively incomplete contracts that ". . . do not attempt to explore and stipulate responses to every possible event" (Crocker and Masten, 1991: 73). Such agreements are less costly to initially negotiate and provide more flexibility to make adjustments as the future becomes more apparent (Crocker and Reynolds, 1993). Conversely, where economic conditions are relatively simple and static, exchange partners will tend to choose more precise but rigid agreements. Thus, we propose the following hypothesis:

Hypothesis 1: Environmental uncertainty is positively related to the use of incomplete contracts by exchange partners.

The second factor influencing contract choice is transaction-specific assets. The distinguishing feature of these assets is that they have little use or value outside a particular relationship (Williamson, 1996). As a party increases its investment in transaction-specific assets, it becomes increasingly dependent on the other to realize the value of its investment. Thus, the investing party is increasingly vulnerable to opportunistic behavior by its exchange partner (Weiss and Kurland, 1997). To protect itself, a firm often attempts to establish contracts that will safeguard its investment.

While scholars generally agree that a strong relationship exists between transaction-specific assets and the choice of contract type, there is disagreement about the direction of this relationship. For example, Williamson (1979) suggests that transaction-specific assets lead to greater use of incomplete contracts. He argues that asset specificity increases the level of appropriable quasi rent to be divided, and that this motivates long-term affiliations between transactors. However, long-term agreements increase the likelihood that conditions surrounding the exchange will change and it becomes increasingly important that the parties be able to adjust their contractual relationship to prevent any partner from being "locked-in" to a contract that is unprofitable for them. Thus, as Williamson states "What is needed, evidently, is some way for declaring admissible dimensions for adjustment so that flexibility is provided under terms in which both parties have confidence" (1979: 251). Such adjustments can best be made through the periodic renegotiations included in incomplete contracts (Crocker and Masten, 1991).

On the other hand, other scholars argue that the larger the specific investment, the greater the likelihood that transactors will use complete contracts. While asset specificity does increase the amount of quasi rent to be divided, this increases the amount of effort that transactors are willing to spend to negotiate a favorable distribution of those rents and makes periodic contract renegotiations unattractive (Goldberg and Erickson, 1987). Specifically, in their examination of exchange relationships between petroleum coke refiners and aluminum producers (in which large transaction-specific assets are present), Goldberg and Erickson state:

"reopening a contract would mean that the parties would haggle over how to share the pie. The more parties are isolated from alternative trading partners, the larger the size of the pie. The larger the pie, the more resources the parties would devote to pursuing it. That is, the higher reliance interest in the aluminum contract would result in higher renegotiation costs, making frequent renegotiation less attractive" (1987: 391).

Thus, transaction-specific assets may make it less attractive to periodically reopen contracts for renegotiation, leading negotiators to favor complete contracts.

As the above discussion illustrates, there are opposing theoretical arguments about the effects of asset specificity on contract choice. Here, we test for the relationship suggested by Williamson (1991) in the following hypothesis.

Hypothesis 2: Investments in transaction-specific assets are positively related to the use of incomplete contracts by exchange partners.

Crocker and Reynolds (1993) suggest that the probability of future opportunism depends, in part, on the likelihood that a transactor's behavior will lead to a successful outcome. One factor determining success is the ease with which the opportunistic party can be replaced in the exchange. If alternative suppliers are unavailable or difficult to find, a buyer's threat of switching to another supplier is not credible (Heide and John, 1988) and suppliers are more inclined to act opportunistically. If, on the other hand, a buyer can easily replace an opportunistic supplier, then the supplier's attempt at opportunism would likely meet with little success (Joskow, 1987). Thus, the availability of alternative trading partners likely reduces the potential for opportunism and increases the likelihood of using less complete contracts. Stated more formally, we propose:

Hypothesis 3: Greater availability of alternative suppliers is positively related to the use of incomplete contracts by exchange partners.

Ex Post Negotiating Costs

A second aim of this research is to investigate the performance implications associated with differences in how a relationship is managed. Incomplete contracts define the process by which contract terms can be adjusted over time. While this may mitigate some of the potential for opportunism by defining the general terms of the exchange (e.g., renegotiations of price and quantity at specified intervals), enough residual opportunism continues to exist so that considerable scope may remain for exercising more subtle, though still costly, bargaining strategies" (Crocker and Masten, 1991:77). Theoretically, the extent to which transactors using incomplete contracts are able to mitigate opportunism and reduce ex post negotiating costs depends not only on the contract itself, but also on the effectiveness of the process by which periodic contract adjustments are negotiated (Zaheer and Venkatraman, 1995).

To provide insight into the factors that allow the management of ex post negotiation costs associated with incomplete contracts, we turn to the literature in social contracting and relational exchange theory (Macneil, 1980). In contrast with the economic perspective of TCE, relational exchange theory focuses on the sociological characteristics of exchange relationships. Behavioral characteristics, or "relational norms," are argued to be critical determinants of how efficiently periodic contract renegotiations are carried out (Dore, 1983; Macneil, 1978, 1980). Relational norms evolve over time as exchange partners become increasingly familiar with each other and establish behavioral rules for such processes as conflict resolution, monitoring and joint problem solving (Ring and Van de Ven, 1992). Because each exchange dyad evolves in a unique manner, different relational norms emerge (Dwyer et al., 1987). The key question then, for those who choose incomplete contracts, is "What are the process elements that de termine the size of those ex post negotiation costs?"

It should be noted that while two dyads may be governed by similar incomplete contracts, they may differ significantly in the extent and type of relational norms that support the contract (Dore, 1983; Macneil, 1980). Thus, two exchange relationships with similar incomplete contracts may have very different levels of ex post negotiation costs (e.g., Joskow, 1987) and success. Consequently, despite having similar relational contracts, and similar levels of asset specificity and uncertainty, two dyads may incur different ex post negotiation costs because they have different relational norms supporting that contract.

A number of specific relational norms, including trust, information sharing, the expected length of a relationship, partner flexibility, and joint problem solving (Heide and John, 1992; Noordeweir et al., 1990; Zaheer and Venkatraman, 1995), have been hypothesized to impact ex post negotiation costs. Regardless of the specific norm examined, researchers agree that increasing the relational content of an exchange encourages cooperation between partners and thereby discourages opportunistic behavior.

Three such relational norms are considered here: collaboration (e.g., Goldberg and Erickson, 1987), commitment (e.g., Heide and Miner, 1992), and communication strategies (e.g., Dwyer et al., 1987). Preliminary field interviews indicated that these three relational norms have particular relevance in the present context. In general, we expect that as these three relational norms become prevalent, opportunistic behavior will decline and ex post negotiation costs will fall.

Collaboration. Collaboration refers to the willingness of the parties to work together to create a positive exchange relationship. Collaborative actions can enhance a relationship and curtail opportunistic behaviors (Goldberg and Erickson, 1987). For example, joint planning and forecasting allow both parties to participate in determining each other's roles and responsibilities and may foster mutually beneficial expectations (Dwyer and Oh, 1988). Similarly, bilateral efforts to share information (e.g., production requirements and future design changes) implies a more open and complete disclosure of relevant information (Schuler, 1979). Moreover, if maintaining the overall health of the exchange is emphasized, the transactors are likely to be more willing to seek efficient adaptation to unexpected environmental changes (Mohr and Spekman, 1994). Thus, a high level of collaboration provides a context in which contract renegotiations are simplified and ex post negotiation costs are reduced. Thus:

Hypothesis 4: Collaboration by the exchange partners will be negatively related to ex post negotiation costs.

Commitment. Commitment concerns the extent to which exchange partners expect their relationship to continue for the foreseeable future (Dwyer et al., 1987; Heide and John, 1990). Such expectations can encourage cooperation by providing the opportunity for one partner to retaliate if the other behaves opportunistically (Parkhe, 1993). Opportunistic behavior by one party in one period can be matched by opportunistic behavior by the other partner in the next, and cooperation can be met with cooperation. While non-cooperative behavior has been found to be the dominant strategy for discrete exchanges, the expectations of reciprocity encourage partners to cooperate in ongoing exchanges (Parkhe, 1993). In addition, exchange partners in long-term relationships are more likely to consider the impact of current actions on the present negotiation as well as anticipated future interactions. Thus, they may be more willing to incur short-term disadvantages since they anticipate future opportunities to recoup their concessi ons (Heide and Miner, 1992; Parkhe, 1993). Hence, we hypothesize the following:

Hypothesis 5: Commitment by the exchange partners will be negatively related to ex post negotiation costs.

Communication strategy. Communication strategy, which has received considerable attention in the marketing channels literature (e.g., Dwyer et al., 1987; Frazier and Rody, 1991), refers to the type of communications that transactors use to try to influence negotiations. Coercive communication strategies involve applying direct pressure and stressing the adverse consequences of non-compliance (e.g., threats or legalistic pleas, which suggest that compliance is required by the formal contract terms). When one firm uses coercion to gain a more favorable negotiation outcome, the partner is likely to view that firm as exploitative rather than accommodative and to engage in retaliatory behavior (Cool and Henderson, 1998). Coercive communications promote opportunistic behaviors such as deliberately altering information, making promises that are never delivered, and outright lying--all of which lead to interfirm conflict (Frazier and Rody, 1991). Furthermore, coercion causes negotiators to become more rigid in their views, making adaptation to uncertainty more difficult (Cadotte and Stern, 1979). In contrast, non-coercive strategies attempt to persuade rather than demand by focusing on beliefs about business issues and applying little direct pressure. An example is recommendations, which are simple requests made by one party that stress the benefits the other party will receive by complying (Dwyer et al., 1987). Non-coercive negotiation strategies promote flexibility and accommodation and encourage the parties to work together to resolve problems (Frazier and Rody, 1991).

Communication strategies may significantly impact ex post negotiation costs (Cool and Henderson, 1998). Specifically, the use of non-coercive strategies reduces the likelihood of opportunism and improves the exchange partners' adaptive capabilities. Thus, we hypothesize the following: Hypothesis 6: Non-coercive communication strategies by the exchange partners will be negatively related to ex post negotiation costs.

METHOD

This study focused on the exchange relationship between OEMs and their component suppliers. Traditionally, these relationships have been governed by relatively complete contracts. Recently, a trend toward greater supplier partnering and an increase in environmental uncertainty has prompted greater use of incomplete contracts (Dyer, 1997). Thus, OEMs and suppliers now use a wide range of different contract types and provide a good context to test our model.

Using the OEM-supplier dyad as the unit of analysis, this study examined the OEM's view of its relationship with a self-selected supplier. Respondents filled out the survey with reference to a major component supplier with whom they had an exchange relationship for at least three years and who was either the sole contact or the leader of the team responsible for the supply relationship.

Consistent with Parkhe (1993), we targeted the highest level individuals who play a significant role in managing supply relationships rather than the most senior managers in each firm. Since these individuals are located in the purchasing function (Bhote, 1989), purchasing agents have frequently been used as respondents in research examining interorganizational exchanges (Perdue et al., 1986; Rubin and Carter, 1990). Presurvey interviews with OEM purchasing managers indicated that in the majority of cases one individual is responsible for strategic and operational issues associated with a supplier. Thus, only that person possessed the detailed knowledge necessary to respond to the survey questions. Consequently, responses were gathered from only one informant. This approach is generally consistent with the recommendation to use the most knowledgeable informant (Huber and Power, 1985).

Surveys were sent to a random sample of 1,400 purchasing managers from a national mailing list of purchasing agents. These individuals were from OEMs in the two-digit SIC groups 35, 36, 37 and 38 (industrial and machining equipment, electronic and electrical machinery, computer equipment, and transportation machinery). Manufacturing firms in these industries have been restructuring their relationships to increase their competitiveness and are heavy purchasers of component parts (Kalwani and Narayandas, 1993). In addition, using several industry codes increases the number of observations and increases the generalizability of findings. Four hundred and fifteen responses were received from this group. After eliminating surveys with incomplete information and those with respondents who indicated they had insufficient knowledge of the purchasing relationships, the final sample size was 393.

Because collecting data from one informant introduces the possibility of common method variance, steps were taken to minimize potential problems. To assess informant competency, survey items measured the respondent's knowledge of, and involvement in, the focal issues (Kumar et al., 1993). Only respondents reporting at least a four (on a five-point scale with "5" representing the highest knowledge and involvement) on both dimensions were retained in the final sample. Mean scores for respondent's knowledge and involvement were 4.5 and 4.7, respectively. Thus, it appears reasonable to assume that the respondents possessed considerable expertise concerning the particular supply relationship. To reduce the potential of biased performance assessments, the measures for contract choice and negotiation costs were placed after the measures of transaction-specific assets, environmental uncertainty and the relational variables (Podsakoff and Organ, 1986). In addition, several factors emerged in a factor analysis with all variables, none of which accounted for the majority of variance. These results indicate that common method variance is not a significant problem (Harman, 1967).

We assessed non-response bias by using t-tests to compare early to late respondents on all variables in the model and other variables such as annual purchases and length of relationship with the supplier. The only significant difference (p < .05) was a tendency for early respondents to be smaller than late respondents. Thus, non-response bias does not appear to be a significant issue.

Measures

When constructing our measures, we examined factor loadings of the individual items and item-to-total correlations in order to determine which of the individual items were meaningfully correlated with the overall factors. Using the cutoff suggested by Pedhazur and Schmelkin (1991), only those items with loadings above 0.5 were included in the final scale.

Contract Type (Dependent). Contract type indicates the degree of completeness. The most restrictive and complete contract is the FirmFixed Price (FFP) contract, which specifies all terms upfront and allows no ex post adjustments (Garrett, 1997). The Fixed Price with Economic Price Adjustment (FP/EPA) is less complete and generally uses labor or material indices to determine prices according to a negotiated formula. Some ex post negotiations are required to adjust the labor or material measures (Garrett, 1997). Fixed-Price Incentive (FPI) contracts are the most incomplete of the contract types studied. In FPI contracts, performance targets are successively negotiated as events unfold and the parties acquire more information (Garrett, 1997). These three contract types were coded so that higher values of contract type imply more incomplete contracts (FFP = 0, FP/EPA 1, FF1 = 2).

Environmental Uncertainty (alpha =.77). Six items measured environmental uncertainty. The six items address both price uncertainty (the inability to estimate prices for the focal supplier's product) and volume uncertainty (difficulty in forecasting expected demand) (Noordeweir et al., 1990).

OEM Transaction-specific Assets (alpha = .84). Seven items measured the OEM's specific assets. These include specialized physical assets as well as personnel, knowledge and training that is specialized to the focal supplier (Anderson, 1985).

Availability of Alternatives (alpha -.72). We constructed a measure of the ease with which the OEM could find an alternative supplier. Three items assessed how dependent the OEM was on the supplier, how easily they could get components from another supplier, and how much the OEM's performance would suffer if the current relationship were terminated.

Ex post Negotiation Costs (Dependent) (alpha = .83). A negotiation occurs whenever parties bargain on essential contract elements (e.g., price, warranties) to reach an agreement (Rubin and Carter, 1990). We define Negotiation Costs as the resources expended by the OEM in negotiating contract changes with its supplier. Nine items measured ex post negotiation costs. These items assessed the number of OEM personnel involved and total hours spent in contract preparation and negotiation, the number of bargaining sessions, preparation time and the time required to reach agreement, and the amount of conflict in the relationship (Kutschker, 1985).

Collaboration (alpha = .86). Collaboration was measured by 9 items, adapted from scales used by Heide and John (1992), Mohr and Spekman (1994) and Noordeweir et al. (1990). Items assessed the following: OEM and supplier efforts to share information, the extent to which the exchange partners make efforts to assist one another, the amount of joint planning and problem solving, and the extent to which the parties share the responsibility for maintaining the relationship.

Commitment (alpha = .83). Commitment was measured by three items. These items assessed the extent the exchange partners expect the present relationship to continue for the foreseeable future and the degree to which each party takes actions to maintain the relationship (Heide and Miner, 1992).

Non-coercive Communications (alpha = .84). A three-item scale was developed to measure recommendations, one specific type of non-coercive communications. Specifically, these items measured the extent to which communications stressed how requests would benefit each party and to what extent communications were absent of threats.

Control Variables. Three control variables are included in the analysis. First, relationship length was measured by the log of the total number of years that the parties have had a purchasing relationship. Over time, behavioral routines affecting negotiation processes develop and may influence ex post negotiation costs (Madambi and Helper, 1998). Second, relationship importance, the extent to which the supplier can impact OEM's performance, was assessed by the potential impact of the contract negotiation on OEM performance and the impact of the supplier's component on OEM end-product quality. Suppliers providing components that have a relatively large impact on a buyer's product cost or quality warrant more attention than other partners (Spekman, 1988). One way this attention can manifest itself is in the effort expended in negotiations. When the potential impact of a contract renegotiation is large, we expect more resources to be devoted to insuring that the contract is appropriately structured than when the potential impact is small. Third, we use the log of the number of employees in the OEM divided by the number of employees in the supplier to assess the impact of organization size. This relative size metric allows us to examine whether size differentials impact our model variables.

Data Analysis

The first phase of this research used ordered probit analysis to examine the factors associated with choice of contract type. Contract type is a categorical dependent variable that is ordered with respect to an underlying latent variable (i.e., contract completeness). Higher values of contract type indicate more incomplete contracts. The second phase of this research used OLS regression to predict ex ante negotiation costs, a continuous dependent variable. Because ex ante negotiations are applicable only to relationships governed by incomplete contracts, this analysis used only the 133 dyads using incomplete contracts. Diagnostic analysis revealed that regression assumptions were met (e.g., normality of the residuals) and that no problems existed with outliers (Neter et al., 1990) or multicollinearity.

RESULTS

Descriptive statistics and correlations are shown in Table 1. The results of the probit (Column 1, Table 2) and regression (Column 2, Table 2) analyses generally supported the hypotheses. Hypothesis 1 predicted a positive relationship between environmental uncertainty and incomplete contracts. Increases in environmental uncertainty (p<.001) increased the use of incomplete contracts. Support was also found for hypothesis 2, which stated that investments in transaction-specific assets increase the likelihood that relatively incomplete contracts will be used (p<.05). Hypothesis 3 was also supported. Greater availability of alternative suppliers increased the use of incomplete contracts (p<.01).

The regression results support all hypothesized relationships (hypotheses 4.6) between relational norms and ex post negotiation costs. Specifically, collaboration (p<.001), commitment (p<.0l) and non-coercive communications (p<.05) lowered ex post bargaining costs. Relationship importance was the only significant control variable and increased ex post negotiation costs (p<.001).

DISCUSSION

Although numerous authors have investigated the impact of incomplete contracts and the incentives created by these contracts (e.g., Crocker and Masten, 1991), here we have sought to provide insight into the incentives that encourage transactors to enter into contracts that are left intentionally incomplete. Agents face a tradeoff between the cost of drafting more complete agreements and the exposure to opportunism from incomplete contracts. Understanding contract choice, and the conditions that influence it, may provide important insights into effective contracting practices and successful interfirm collaboration.

Research Implications

The results of this empirical work strongly support the predictions of the TCE framework, and validate that the degree of contractual completeness is a function of the characteristics surrounding the exchange (Crocker and Reynolds, 1993; Williamson, 1996). Complex and uncertain conditions, in the form of price and volume uncertainty, increase the costs of creating more complete agreements, and prompt transactors to select the greater flexibility of incomplete agreements. Similarly, the finding that OEM investments in specialized assets lead to the negotiation of incomplete agreements strongly suggests that the complexity of writing contracts associated with such assets is prohibitively expensive.

While the findings concerning contract choice provide valuable insights into the process of contract selection, the insights from examining factors affecting ex post negotiation costs shed light on how transactors manage incomplete contracts so that rent-dissipating activities are minimized. Indeed, each of the relational norms--collaboration, commitment and noncoercive communications--were found to reduce ex post negotiation costs, supporting the theoretical predictions of relational exchange theory. These findings clearly support the argument that these norms play a significant role in the performance of purchasing relationships (Macneil, 1980; Ring and Van de Ven, 1992). These norms can improve relationship flexibility and increase the ability of exchange partners to respond to unexpected environmental contingencies. This research supports that of Goldberg and Erickson (1987), who in their seminal case study of petroleum coke found that the costs of renegotiation need not be prohibitively high if appropria te relational norms exist between exchange partners.

It is worth noting that relationship importance, a control variable, was also found to increase ex post negotiating costs. This confirms the expectation that OEMs place more emphasis on managing important suppliers. It is also interesting that, in our findings, relationship length was not related to bargaining costs. One could reasonably expect that over time, OEM-supplier negotiations would become more accommodative as the exchange partners become more familiar with the others' routines and preferences. It may be, however, that purchasing relationships can be built on stable histories of non-cooperation as well (Madambi and Helper, 1998; Heide and Miner, 1992). Thus, contract negotiations may not become more accommodative simply because the parties have worked together longer.

Managerial Implications

Many firms are turning to long-term, buyer-supplier relationships. Developing close buyer-supplier relationships is presented as a way to increase customer value, while lowering costs (Cannon and Homburg, 2001). These benefits, however, are not automatic. The findings of this study show that incomplete contracts may be necessary under certain conditions, such as high uncertainty. When incomplete contracts are used, relational norms are important in reducing renegotiation costs. Given the importance of relational norms, particularly when incomplete contracts are used, partner selection decisions should consider a partner's likely adherence to relational norms.

This study has shown that investing in the development of relational norms can lower the cost associated with future negotiations. Thus, firms should also be willing to invest in activities that foster the development of relational norms. The cost of these investments are likely to be recovered, at least in part, from the savings in negotiation costs. While we examined the renegotiation of existing contracts, these same benefits are likely to accrue in the subsequent negotiation of new contracts as well. Therefore, investments made in developing cooperative relationships should be more than offset by the reduced costs of managing ongoing relationships.

Future Research and Limitations

Although our hypotheses were supported, other questions need to be answered. For example, even though relational norms were found to reduce an OEM's ex post bargaining costs, questions remain about the robustness of the benefits derived from relational norms (Artz and Brush, 2000). Commitment, for instance, lowered bargaining costs, because of the OEM and supplier's expectations of a long-term relationship (Hill, 1990). In contrast, if the exchange were in a vertically integrated firm, negotiating costs would be lowered by the force of organizational authority (Dow, 1987). However, suppose that the exchange partners continued to invest in specific assets, which would increase the potential gain from opportunistic behavior. Would the relational norms continue to be strong enough to deter opportunistic behavior, or at some point would basic OEM-supplier incompatibilities arise, thereby restricting relational norms' ability to deter self-serving behaviors (Walker and Poppo, 1991)? Similarly, other than relationa l norms, what other mechanisms exist that can be used to discourage opportunistic behavior and encourage productive interfirm exchange (Anderson and Welts, 1992)?

It should also be noted that the measure of ex post negotiation costs in this study focuses only on OEM bargaining costs. Future studies should consider other types of transaction costs (e.g., monitoring costs). Another useful advance would be to consider negotiation costs on both sides of the dyad. While we argue that OEM negotiation costs are a critical determinant of the exchange success, a more complete picture of the total costs associated with contract maintenance in the relationship would be provided by examining costs on both sides of the dyad. Indeed, as Dyer (1997) points out, potential differences between an OEM and supplier in terms of their commitment to the exchange, the type of monitoring systems and the like may lead to different levels of negotiating costs for the two parties. Moreover, the measure of negotiation costs in this study is based primarily on examining the labor component of these costs. While the literature argues that these labor-based costs are an important component of transac tion costs, other factors such as monitoring and overhead costs may also need to be considered (Noordeweir et al., 1990).

We have drawn on TCE and social exchange theory for this study. By drawing on other theories, future studies may provide additional insights into the relationship between exchange partners. For example, the procedural justice literature may shed light on relational norms. Tyler and Bies (1990) note that relational norms that emerge during ongoing relationships have procedural properties. Thus, the success of interfirm exchange may be affected by whether the partners perceive that the procedures used in the relationship (e.g., problem solving, performance reviews) are fair.

Future research would also benefit from identifying and avoiding some other limitations of this study. The cross-sectional nature of the data prevented tests for causality. For example, although the results support the hypothesis that the presence of specific assets leads to greater use of incomplete contracts, our analysis supports only the existence of a relationship between the constructs. It is plausible that a reverse sequence of events is operating and that the presence of an incomplete contract is an antecedent to investing in specific assets. While our results are generally consistent with the theoretical specification of events (Williamson, 1985), alternative explanations of the results cannot be completely eliminated with cross sectional data. Moreover, research would also benefit from examining both the buyer and supplier perception of the relationship. While steps were taken to minimize the problem of only collecting data from one side of the dyad, different results may have been obtained had the supplier's perceptions also been integrated into the analysis. However, as others have argued (e.g., Provan and Skinner, 1989), this limitation should not invalidate our results. Another limitation of our study is that we were unable to assess response bias using waves as suggested by Armstrong and Overton (1977). We did compare early and late respondents; however, a stronger test of non-response bias would compare waves of respondents to detect differences that may affect the results of the study. TABLE 1 Descriptive Statistics and Correlations Variables Mean s.d. 1. 2. 3. 4. 1. Negotiation Cost 11.58 4.35 1.00 2. Contract Type 1.13 0.26 0.15 1.00 3. Environment. Uncertainty 15.30 1.84 0.24 0.24 1.00 4. Specific Assets 14.79 5.33 0.27 0.09 0.25 1.00 5. Alternative Suppliers 6.87 1.86 0.11 0.27 0.06 0.06 6. Collaboration 17.03 4.43 -0.16 0.21 0.15 0.18 7. Commitment 7.65 1.85 -0.13 0.15 0.06 0.05 8. Non-coercive 6.86 2.26 -0.16 -0.10 0.09 0.07 9. Relationship Length 2.24 0.67 0.07 -0.08 0.12 0.11 10. Relationship Importance 8.68 1.09 0.10 0.11 0.09 0.08 11. Relative Size 6.22 4.02 0.04 0.08 0.09 0.05 Variables 5. 6. 7. 8. 9. 10. 11. 1. Negotiation Cost 2. Contract Type 3. Environment. Uncertainty 4. Specific Assets 5. Alternative Suppliers 1.00 6. Collaboration -0.26 1.00 7. Commitment -0.22 0.16 1.00 8. Non-coercive -0.12 0.03 0.23 1.00 9. Relationship Length -0.23 0.23 0.18 0.20 1.00 10. Relationship Importance 0.25 0.21 0.34 0.29 0.11 1.00 11. Relative Size 0.07 0.04 0.08 0.03 0.09 0.14 1.00 Correlations with absolute values greater than 0.09 are significant at the .05 level. TABLE 2 Results Independent Variables Column 1 Column 2

Contract Type Negotiation Costs Intercept 1.291 *** 10.610 ***

(0.297) (1.620) Environmental Uncertainty 0.316 ***

(0.084) Specific Assets 0.176 *

(0.114) Alternatives 0.614 **

(0.227) Collaboration -0.549 ***

(0.047) Commitment -0.247 **

(0.104) Non-coercive Communication -0.263 *

(0.124) Relationship Length 0.314 0.357

(0.287) (0.511) Relationship Importance 0.772 1.186 ***

(0.566) (0.228) Relative Size 0.111 0.132

(0.161) (0.254) Chi-squared 15.64 Adjusted R-squared 0.291 F 4.078 *** * p<.05 ** p<.01 *** p<.001

References

Anderson, E. 1985. "The Salesperson as Outside Agent or Employee: A Transaction Cost Analysis." Marketing Science 4: 234-254.

_____ and B. Weitz. 1992. "The Use of Pledges to Build and Sustain Commitment in Distribution Channels." Journal of Marketing Research 24: 18-34.

Armstrong, J.S. and T.S. Overton. 1977. "Estimating Nonresponse Bias in Mail Surveys." Journal of Marketing Research 14: 396-402.

Artz, K.W. and T. Brush. 2000. "An Examination of Coordination Costs in Collaborative Strategic Alliances." Journal of Economic Behavior and Organization 41: 337-362.

Bhote, KR. 1989. Strategic Supply Management. New York, NY: AMACOM.

Cadotte, E.R. and L. Stem. 1979. "A Process Model of Interorganizational Relations in Marketing Channels." In Research in Marketing. Ed. J. Sheth. Greenwich, CT: JAI Press.

Cannon, J.P. and C. Homburg. 2001. "Buyer-Supplier Relationships and Customer Firm Costs.,, Journal of Marketing 65(1): 2943.

Cool, K.. and J. Henderson. 1998. "Power and Firm Profitability in Supply Chains: French Manufacturing Industry in 1993." Strategic Management Journal 19:909-927.

Crocker, K.J. and S.E. Masten. 1991. "Pretia Ex Machina? Prices and Process in Long-Term Contracts." Journal of Law and Economics 34: 69-99.

_____ and K.J. Reynolds. 1993. "The Efficiency of Incomplete Contracts: An Empirical Analysis of Air Force Engine Procurement." Rand Journal of Economics 24: 126-146.

Dahlstrom, R and A. Nygaard. 1999. "An Empirical Investigation of Ex Post Transaction Costs in Franchised Distribution Channels." Journal of Marketing Rese