Performance Outcomes of Purchasing Arrangements in Industrial …€¦ · Administration, University of Vermont. George John is Associate Pro- fessor of Marketing, Carlson School

Performance Outcomes of Purchasing Arrangements in Industrial Buyer-Vendor RelationshipsAuthor(s): Thomas G. Noordewier, George John and John R. NevinReviewed work(s):Source: Journal of Marketing, Vol. 54, No. 4 (Oct., 1990), pp. 80-93Published by: American Marketing AssociationStable URL: http://www.jstor.org/stable/1251761 .Accessed: 13/08/2012 14:49

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access toJournal of Marketing.

http://www.jstor.org

http://www.jstor.org/action/showPublisher?publisherCode=ama

http://www.jstor.org/stable/1251761?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

Thomas G. Noordewier, George John, & John R. Nevin

Performance Outcomes of Purchasing Arrangements in

Industrial Buyer-Vendor Relationships

Purchasing arrangements for repetitively used industrial supplies assume many different forms. Drawing on transaction cost analysis, the authors advance a conceptual framework that organizes these arrangements along a continuum of relationships. They use data from a survey of 140 OEM purchasers of bearings to demonstrate that performance in terms of acquisition costs is enhanced when, under conditions of uncertainty, firms introduce more relational elements into their purchasing arrangements. Possession cost performance improved when larger volumes of bearings were purchased. Implications for theory and practice are discussed.

PURCHASING performance is an important deter- minant of a firm's competitiveness. On average,

industrial firms spend more than half of every sales dollar on purchased products (U.S. Bureau of Census 1985). Because each dollar saved in procurement is equal to a dollar of new profit (Dobler, Lee, and Burt 1984), purchasing's influence can "easily spell the difference between leadership in an industry and an untenable competitive position" (Heinritz and Farrell 1981).

In practice, industrial firms distinguish between two very different kinds of purchases: those for items repetitively needed in production or in maintenance (e.g., fasteners, bearings, paint, etc.) and those for capital equipment (e.g., milling machines, power

Thomas G. Noordewier is Assistant Professor, School of Business Administration, University of Vermont. George John is Associate Pro- fessor of Marketing, Carlson School of Management, University of Min- nesota. John R. Nevin is the Grainger Wisconsin Distinguished Profes- sor, Graduate School of Business, University of Wisconsin-Madison. This work was supported in part by a grant from the Richard D. Irwin Foun- dation. The comments of the anonymous JM reviewers are appreciated.

generating devices, etc.). Though much attention has been directed toward understanding capital equipment purchases (e.g., how to apply payback analysis and discounted cash flow models to the equipment purchase decision, the economic merits of buying vs. leasing equipment, whether to purchase new or used equipment, etc.), the acquisition of repetitively used items (or RUIs) has received very little scrutiny from scholars and practitioners. As a result, there are some crucial gaps in the literature on RUI procurement. One of the most critical of these gaps is the lack of empirical research pertaining to how buyer purchasing performance is affected by the organizational form of the vendor interface.

Transaction cost analysis (TCA) sheds valuable light on this question. A distinctive feature of TCA is that it is based on the assumption that firms are motivated to craft efficient organizational forms or governance structures. It is a normative model, implying that firms that follow its prescriptions will have better performance, from lower transaction costs, than those that do not. The theory identifies environmental uncertainty as a critical predictor affecting this relationship.

Despite this normative aspect of the model, vir-

80 / Journal of Marketing, October 1990

tually no empirical tests of TCA's performance implications have been reported in the literature of marketing or other disciplines. What few empirical TCA studies have been conducted generally have examined only whether firms organize as predicted by TCA (ex- ceptions include studies by Heide and John 1988, who examined the performance implications of safeguarding specific assets, and Anderson 1988, who examined the efficiency implications of conforming to TCA- based rules about using a direct salesforce vs. independent representatives).

The extant tests of TCA also have been limited largely to explaining global shifts in organizational form (e.g., employee salesforces vs. independent representatives, Anderson 1985; resellers vs. forward vertical integration, John and Weitz 1988). Changes in organizational form within a class of relationships have received little attention.

We empirically examine the relationship between the organization of the buyer-supplier interface and performance in RUI procurement. On the basis of TCA reasoning we offer, from the buyer's perspective, a model of purchasing arrangements for repetitively used items. Refutable predictions linking performance to organizational form and uncertainty are tested with data from a sample of industrial firms.

After discussing the concept of buyer performance in RUI procurement, we outline the theory relevant to understanding the structure of buyer-vendor relations. Then we present the empirically testable hypothesis relating purchasing performance to theory-mandated governance structures. The sampling and data collection procedures are described next, followed by the operational measures. We then report the results of the data analysis. Finally, we summarize the findings and provide some ideas for future research and managerial practice.

Performance in RUI Purchasing In theory, all purchasing costs can be classified, either directly or indirectly, into one of three categories: invoice costs, possession costs, and acquisition costs. Purchasing performance can be defined as the minimization of these three costs. As Hannaford (1983) points out, however, in the case of repetitively purchased supplies, "hidden" inventory (possession) and administrative (acquisition) costs typically far exceed invoice costs. Because neither possession nor acquisition costs involve the actual invoice price, they are basically "resource losses incurred due to imperfect information" (Dahlman 1979), or "transaction" costs. That is, they are costs (losses) due to imperfect coordination between buyers and sellers attempting to transact in an imperfect world. Consider each type in turn.

Possession Costs Possession costs arise in RUI purchases because, un- like capital equipment, recurrently ordered and used items must be stockpiled by user firms. Unless buyer and vendor can achieve perfect just-in-time logistical coordination, temporal discrepancy is inevitable between the time when an order is placed and the subsequent delivery of ordered items. It therefore becomes necessary to create and maintain storage areas, control inventory, pay insurance and taxes, and incur pilferage and other expenses.

A major tangible indicator of possession cost is inventory turnover. In purchasing, turnover is the ra- tio of dollar purchases over some time period to average in-stock inventory levels for that period. Turn- over is an excellent indicator of possession cost because it measures the extent to which capital is tied up in stock. Heinritz and Farrell (1981) illustrate the point:

If the approved policy is to carry a sixty-day supply, the inventory should amount to one sixth of the amount of annual expenditures.... If, by efficient planning, scheduling of purchases, and stores management, continuity of operation is maintained with a turnover of inventory every forty-five days instead of every sixty days, average inventory is reduced by 25 percent ..

The monotonic relationship between increases in turnover and decreases in average stock levels makes turnover a good possession cost indicator (in the example, an increase in turnover from six to eight turns annually results in a 25% decrease in average inventory). The cost of carrying inventory decreases with increases in turnover, and therefore higher turnover implies superior inventory performance.

Acquisition Costs

Acquisition costs stem from the need to originate req- uisitions, interview salespeople, expedite deliveries, receive and edit invoices, follow up inaccurate and late deliveries, and perform other activities. Conven- tional purchasing procedures (the standard purchase order) are accompanied by numerous controls and generate a substantial amount of paperwork. When such procedures are applied to the purchase of major capital items, both the purchase order paperwork and the time devoted are small in relation to the value of the item being purchased. Applied to scores of repetitively purchased items, however, conventional meth- ods of supplies acquisition involve a predictably great administrative cost.

As is true of transaction costs in general, indicators of these acquisition costs are not likely to be found in accounting records. However, we can identify ob- servable indicators of acquisition costs by considering the "expediting" or followup work done by buyers. Such work is widely recognized as one of the most

Performance Outcomes of Purchasing Arrangements / 81

important and expensive aspects of purchasing (Dowst 1972). It involves checking on the status of goods shipped and taking action when deliveries are either not on time or inaccurately filled (or defective). De- lays in delivery often necessitate putting pressure on suppliers in the form of letters, telegrams, phone calls, and even personal visits. Similar efforts are evoked by inaccurate deliveries and substandard products. These activities constitute costs of the execution and enforcement variety. Consequently, we use lateness of deliveries and extent of unacceptable items as ob- servable indicators of acquisition costs.

Crafting RUI Purchase Relationships

Transaction cost analysis (TCA) has attracted considerable interest recently in marketing. Consisting of a blend of economics and organizational theory, it was developed chiefly by Williamson (1979, 1985) and is concerned with the issue of crafting efficient "governance structures." Because of its broad scope of ap- plicability, the approach has been incorporated into various areas of marketing, including the analysis of vertical integration in channels (e.g., Anderson and Weitz 1983), the organization of marketing activities (e.g., Ruekert, Walker, and Roering 1985), and buyer- seller relations (Ford 1980; Dwyer, Schurr, and Oh 1987). As we see shortly, it also has tremendous relevance to the purchase of repetitively needed items.

TCA is a normative model, which implies that firms following its prescriptions will perform better (in the sense of having lower transaction costs) than other firms. Basically, the theory proposes that performance will be enhanced when there is congruence (or a "match") between the governance structure employed and the underlying dimensions of exchange. According to Williamson (1979, 1985), three dimensions (or attributes) of transactions potentially determine the most appropriate governance structure: (1) the degree to which transactions are supported by transaction-specific investments (asset specificity), (2) the frequency with which transactions recur, and (3) the uncertainty surrounding the exchange.

In TCA, the "object is to match governance structures to the attributes of transactions in a discriminating (i.e., transaction cost-economizing) way" (Williamson 1981). The governance structures origi- nally conceptualized by Williamson consist of the two discrete alternatives: markets and hierarchies. Implicit in TCA is the assumption that the context in question will affect the relative importance of the three attributes in crafting an appropriate governance structure.

Our study is concerned with the acquisition of repetitively used items. Because RUIs are standardized commodity-type items, which by definition are pur-

chased recurrently, and because such items do not generally require specialized investments of much consequence, their procurement affords little scope for variation on asset specificity and frequency.1 There- fore, specificity and frequency do not provide an ex- planation about the type of governance structure needed to minimize transaction costs of RUI purchases. Un- certainty, in contrast, has considerable variability across most recurrent purchase contexts, and is therefore the principal attribute TCA implicates as determining the appropriate governance structure for repetitive purchases.

Environmental uncertainty in RUI purchasing. Environmental uncertainty is defined as unanticipated changes in circumstances surrounding an exchange. Williamson argues that uncertainty has a profound effect on the appropriate governance mechanism for a given exchange because unanticipated changes render it more difficult, if not impossible, to spell out all possible contingencies beforehand. As uncertainty increases, the limited cognitive capabilities of human agents put a strain on their ability to craft a priori agreements that take all relevant contingencies into account. At the same time, the need to adapt to changes becomes greater. Forward planning processes via a priori agreements become increasingly replaced by mutual adjustment procedures. Thus, the goal is to craft agreements with good continuity and adaptation properties. As we show subsequently, not all governance structures are equally well suited for effecting adaptation.

One should note that uncertainty is a construct that has a long history in the organizational research literature. Unfortunately, it has been defined and operationalized in rather different ways in various studies. The contradictory results that have been observed are due to these inconsistencies. Balakrishnan and Wernerfelt (1986) provide a thoughtful discussion of the problems with the construct, and argue for more narrowly construed definitions and usage. Consistent with their position is our use of the relatively narrow conceptual definition offered by Williamson as unanticipated changes in relevant factors surrounding the exchange.

The domain of this construct in our study warrants some elaboration. Unanticipiated changes in circumstances may arise in many ways in RUI purchasing. For instance, buying firms face markedly varying de- grees of unanticipated changes in their forecasted volume requirements and the mix of items needed. Other

'Some nonzero amount of specific assets is assumed to be present. If the level of such assets were truly zero, market-based exchange would suffice regardless of uncertainty levels.


matters include unanticipated changes in the production or technology involved as well as in market factors such as product availability. Marketing choices by firms also can affect the uncertainty felt in the purchasing area. For instance, firms that focus on stable downstream markets will have more certain needs for RUI purchases. Likewise, firms with high volume, short product line strategies have less uncertain purchasing needs. Our measure of environmental uncertainty samples this domain and relies heavily on the pioneer- ing effort of Anderson (1985).

Governance Elements of Purchasing Relationships

Expanding on Williamson's initial description of markets versus hierarchies, researchers recently have described several elements of organizational form. Williamson (1985), for example, drawing on the work of Macneil (1978) and Stinchcombe (1985), has sug- gested that governance structures can be arrayed on a continuum of "relationalism," anchored by the market (discrete) and hierarchy (relational) at the polar ex- tremes.

To speak of discrete exchange anchoring one end of an organizational continuum and relational exchange the other begs the following question: To what characteristics do these labels refer? Though Williamson has not provided much detail about the operational characteristics of relational exchange, other researchers have expanded on the basic idea. Macneil (1978) provided the earliest discussion of relational exchange by drawing on legal theorizing about contractual exchange. Consistent with his approach is the organizational-theory-based analysis presented by Stinchcombe (1985). He considers characteristics that "simulate the operation of hierarchies." In the marketing literature, Dwyer, Schurr, and Oh (1987) provide a detailed discussion of Macnneil's ideas adapted to marketing exchange. In all of these theoretical ac- counts, various characteristics are described whereby a relationship can move away from the discrete end of the continuum. This discrete pole is best described by Goldberg's (1976) account as being a transaction in which ". . no duties exist between the parties prior to formation [of the exchange], and in which the duties of the parties are determined [completely]" up front. Simply put, the buying firm specifies its needs and the selling firm meets the terms if they are acceptable. Increasing the relational content of an exchange makes the interaction much more complex.

Despite these conceptual analyses, empirical work based on the ideas of a discrete-relational continuum is scarce. Part of the difficulty is that the conceptual analyses do not offer a clear set of operational dimensions. Williamson only hints at the operational dimensions of governance whereas Macneil offers a large

set of about 28 overlapping dimensions. Stinchcombe offers a smaller set of dimensions, but they are cen- tered around "simulating the operation of hierarchies." Palay (1984) and Kaufmann and Stern (1988) have built operational measures of the relational content of interfirm exchange. Palay measured a set of five dimensions to characterize the relational content of rail shipper-carrier interactions. Though his reli- ance on single items and open-ended questions can be questioned, the article is valuable in that it provides a start in defining relationalism in an operational way. It shows that Macneil's set can be distilled down to a more manageable number. Kaufmann and Stern built on Palay's work to measure a larger number of dimensions in their study of interfirm litigation. We draw on these two studies to specify the dimensions of relationalism measured in our study, but we do not offer these dimensions as a definitive measurement model. Though our preliminary field interviews, supported their relevance in our study context, other contexts might yield somewhat different dimensions.

Supplier flexibility. Suppliers often are called upon to react to unforeseen (and unforeseeable) changes- contingencies that could not have been predicted beforehand. This element defines the flexibility dis- played by suppliers toward buyer-requested adjustments to the extant relationship. Buyer requests for adjustments (in price, maintained stock levels, emergency deliveries, etc.) constitute opportunities for a supplier to display flexibility. At the discrete end of the continuum, buyers expect the terms of exchange with suppliers to be "binding and specific" (Macneil 1981). As firms move away from this extreme, buyers expect suppliers to display more flexibility in response to requests for changes. One should not con- fuse flexibility as described here with the degree of a priori formalization (or lack thereof) of the exchange. It is not the degree to which agreements have been tightly worded ex ante that is of concern; rather, it is the reaction toward change requests that matters.

Supplier assistances. Closely related to the element of supplier flexibility, but conceptually distinct, is the position suppliers take toward assisting buyers in a relationship. At the discrete end of the continuum, ". .. benefits and burdens are sharply divided; each party has his benefits, all his, and his burdens, also all his" (Macneil 1980). As firms move away from this pole, the supplier displays an increasing willingness to provide the buyer assistances, and is even willing to make sacrifices for which there is no im- mediate or explicit compensation. Examples of assistances include notifying buyers in advance of shipment delivery problems and recommending stock substitutes when delivery troubles develop. In each case, the supplier is going beyond some minimally acceptable level of conduct agreed to a priori.


Information provided to supplier. All purchase transactions involve information exchange. However, the quantity and type of information buyers provide vendors vary greatly. At the discrete end of the continuum, buyers are concerned with minimal amounts of information. Typically, this information would consist of the product specifications, prices, delivery schedules, and the like. However, as firms move away from this end, other types of information begin to be communicated, particularly long-term forecasting, proprietary, and structural planning information, including future product design information, production planning schedules, and so on (Palay 1984).

Monitoring of supplier. This element consists of the monitoring or supervisory actions that the buyer undertakes to ensure supplier performance during the execution of the exchange agreement. At the discrete end of the continuum, the market mode of governance implies little control over the activities of an independent supplier. Rather, the supplier delivers the specified performance in reaction to the market price that is offered for its services. As firms move away from this extreme, active supervision is used by the buyer to a greater degree to ensure specified performance. As Stinchcombe (1985) puts it, this supervision "sim- ulates the operation of hierarchies" by performing the control and enforcement function normally thought of in connection with vertically integrated hierarchies.

Expectation of continuity. The last element of governance structures describes the expectation of future exchange between buyers and sellers. At the discrete end of the continuum, the parties expect that the "... transaction commences sharply by clear, instanta- neous performance; sharp in, sharp out" (Macneil 1981). 'Spot sale" exchanges are examples of this end of the continuum. As transactions become more relational, they occur over longer periods of time, have less definite termination dates, and are generally neither sharp in nor sharp out. There is also a greater expectation of repeat business with the exchange partner.

Relational syndrome. Though we discuss the five elements independently of one another, it is clear that they are related. As Stinchcombe (1985) has argued persuasively, the elements tend to support one another and thus constitute a syndrome of functionally related elements. For example, it seems clear that increased scope of communication involving information about future product design changes would not be very useful if there were little expectation of continued future business. For this reason, we model the five elements as comprising a syndrome of relational governance (or relationalism), which can be thought of as a single higher order construct. Relationalism does not exist in a context-free vacuum. It must be operationalized

in the context of a specific exchange. In RUI purchasing, some of the dimensions relate to actions of one party (e.g., information provided to supplier) and other dimensions relate to actions of the other party (e.g., flexibility of supplier). Preliminary fieldwork is used to establish the proper source for each dimension as it is not specified a priori in the basic theory.

Hypothesis TCA Effects TCA implies a discriminating match between the type of purchasing arrangement employed and the level of uncertainty. This discriminating match is readily specified by examining the effects of increased uncertainty on the governance elements supporting the exchange. Increased uncertainty surrounding the exchange renders adaptability more important. If the governance structure used for the exchange were not very adaptive, the unforeseen contingencies would result in low levels of buying transactional performance.

The basic theoretical postulate is that the adaptation capabilities of governance structures are enhanced by increasing the relational aspects of the structure. An examination of each of the purchasing governance elements shows this link more readily. To begin, consider the impact of increased supplier flexibility on adaptability. When suppliers are more in- clined to respond favorably to buyer requests for changes, buyers are obviously able to adjust more readily to changes in circumstances. Likewise, an increase in the willingness of suppliers to provide assistances makes buyer adaptation to uncertainty more effective. Similarly, information provided to suppliers clearly contributes to buyer adaptability. In a more uncertain environment, information is vital because needs are changing more rapidly. Buying firms in tur- bulent environments become increasingly aware of the need to exchange information about production plans, more completely share usage information, allow supplier representatives to view their operations, and so on.

The beneficial effect of monitoring suppliers in a more uncertain environment is not as obvious as that of some of the other governance elements. However, as Stinchcombe (1985) notes, more operating controls are ". .. to make sure, as far as one can with documents, that the contractor has followed routines . creating a trail of paper for auditing if necessary." These checks are useful in ensuring that the increased information provided to suppliers in more relational exchange is not misused by opportunistic vendors. Thus, monitoring suppliers supports the other governance elements in enhancing adaptability.


The final governance element, continuity, makes possible more effective adaptation to uncertainty for the following reason. Rather than having to make (possibly unrealistic) assumptions about an uncertain future, buyer and seller can wait for the relevant information to emerge as the future unfolds, and then make suitable adjustments to the current set of circumstances. When the parties know they are in an ex- tended arrangement, they are more willing to accept short-term disadvantages because things will even out in the long run. There is less perceived need to max- imize short-term gain.

Matching governance and uncertainty. Though the TCA literature holds that performance will be enhanced when there is a "match" between the governance structure employed and the uncertainty surrounding the exchange, Williamson does not clearly specify the match across all possible levels of uncertainty. Performance is thought to improve when more relational structures are introduced in response to high levels of uncertainty, but the optimal (performance- maximizing) structural response under conditions of low uncertainty is not clearly indicated.

In empirical research with manufacturers' agents, Heide and John (1988) found that actions undertaken to safeguard specific assets improved performance at high levels of such assets but had no effect (detrimental or otherwise) at low levels. Similarly, Anderson (1988) found that a sales district's efficiency increased as it conformed to the industry rule's recommendation (about the direct salesforce vs. repre- sentative decision) in unpredictable environments but conformity did not increase efficiency in predictable environments. Drawing on these studies, we posit the following refutable prediction:

When environmental uncertainty is high, buyer inventory and administrative performance is enhanced by effecting a corresponding increase in the relational content of the governance structure. Under conditions of low uncertainty, performance is not enhanced by such changes.

Figure 1 depicts the hypothesized effect. Purchase relationships are categorized on the basis of combinations of values of relational governance (the composite construct) and uncertainty. Cell 1 represents the low uncertainty/low relational governance condition. Cell 2 represents the low uncertainty/high relational governance condition. Cells 3 and 4 represent the high uncertainty/low relational governance and high uncertainty/high relational governance conditions, respectively. The hypothesis states that buying firms observing the requisite "match" between uncertainty and governance should have higher purchasing performance. Let Pi represent the level of purchasing performance associated with cell i. The primary hypoth-

Low

Low

Uncertainty

High

Purchasing Performanc

High

Cell 1 Cell 2

Cell 3 Cell 4

Low

Uncertainty P1i . P2

P4

High Uncertainty

Low High Relationalism

'P, is the level of buyer purchasing performance in cell i.

esis can be operationalized as the two following contrasts:

P4-P3 should be significantly positive. P2-PI should be not be significantly different from zero.

These contrasts are depicted graphically in the lower panel of Figure 1.

Other Effects A long-standing principle in channels research holds that the more powerful firm can influence its exchange partner to carry out its wishes (e.g., Stern and El-Ansary 1982) and hence improve its own performance. For example, if the buyer can shift some of

Performance Outcomes of Purchasing Arrangements /85

FIGURE 1 Interaction Effect of Relationalism and

Uncertainty on Buyer Purchasing Performance'

Relational Governance

e

j

the inventory task to the supplier because the supplier is more dependent (Buchanan 1986; Porter 1980), higher inventory turnover will be observed within the buying firm. Similarly, a buyer with clout over a vendor might insist that the latter undertake more predelivery inspections of shipped items to ensure that orders are filled more accurately. Or, a buyer might insist that a vendor build a warehouse nearby and maintain better inventories, thus reducing the possibility of late deliveries. Note that in each instance buyer performance improves. However, the performance enhancements occur for reasons unrelated to the achievement of a discriminating match between uncertainty and governance structure. Rather, the improvements reflect the buyer's ability, or power, to extract more favorable terms of trade from the supplier.

Increases in buyer inventory turnover also may occur because the buyer is willing to pay a higher price for shifting the inventory task to the supplier. Simi- larly, improvements in the acquisition cost indicators may occur because the buyer is willing to pay for better service. In the latter case, it is clear that if a firm is willing to pay for, say, the cost of more predelivery inspection, the number of inaccurately filled orders and defective items will diminish. The result is lower acquisition transaction costs, but higher invoice prices. The number of orders (or releases) issued by the buyer also may affect buying performance. For example, a buying firm can improve turnover simply by issuing orders more frequently (in smaller lot sizes).

Because dependence (or clout), price paid, and number of orders issued each may affect the dependent variables, we account for them in our empirical investigation. The actual measures of these potential influences are described in detail subsequently.

Method Sample and Data Collection To test the hypothesis, we need a context in which an industrial commodity-type item is purchased repetitively. Several supply items were examined in a preliminary field study and a decision was made to select a single category of products, ball and roller bearings, as the research context. One of the considerations that prompted this decision was the diversity of firms that purchase these items, which ensures variation in the focal variables. Furthermore, bearings are an important class of supplies for the OEM customers considered here. The saliency of these purchase arrangements makes it relatively easy to collect the needed data about the vendor relationships. Finally, by fo- cusing on the purchase of a single category of supplies, we reduce extraneous variation and enhance the statistical power of the tests.

A broad cross-sectional sample of OEM purchasers of ball and roller bearings was developed. Us- ing the four-digit SIC code representing the manufac- ture of ball and roller bearings (SIC 3562), we examined the Census of Manufacturers publication to determine the identity, by industry, of the major users of ball and roller bearings. From their data, we identified (by four-digit SIC code) the top 20 industries consuming ball and roller bearings. Together, they account for the vast majority of purchasers of these products. With the assistance of a commercial list broker, we com- piled a list of 483 companies across these industries. This list constituted the sampling frame.

Data collection was accomplished by mail questionnaire from a random sample of companies drawn from the sampling frame. To ensure that the person who completed the questionnaire was the individual primarily responsible for, and hence most knowl- edgeable about, the relationship, the companies were contacted in advance by telephone and the individual buyer to whom the questionnaire was to be mailed was identified. This individual generally had the title of Director of Purchases, Senior Buyer, or Materials Manager. Generally, phone contact with three to four individuals in a snowball approach was necessary to select an informant for each firm.

The informants were asked at the outset of the questionnaire to identify their primary bearing supplier. This primary supplier then served as the referent for all remaining questions. The final response (after followup) was 150 questionnaires (31%). Of these questionnaires, 140 ultimately proved usable. Table 1 provides some descriptive statistics for the sample.

Measures Elements of purchasing relationships. The five relational governance elements are supplier flexibility (FLEX), supplier assistances (ASSIST), information provided to supplier (INFO), monitoring of supplier (MONIT), and expectation of continuity (EXPECT).

TABLE 1 Sample Characteristics (140 firms)

Average Values of Variables

Number of full-time employees 935

Number of employees in purchasing 7.6

Annual purchases of bearings (all suppliers) $2,034,930

Annual purchases of bearings (primary supplier) $1,287,069

Order size (primary supplier) $42,352


Each of the five elements was operationalized as the mean of a set of Likert-type items. All of the items consisted of 5-point strongly agree/strongly disagree scales. The composite relational governance measure (REL) was constructed as the sum of the five separate elements (equally weighted). The items used for the measures are listed in the Appendix.

Uncertainty. Environmental uncertainty (UNCER- TAINTY) assesses aspects of market turbulence and vendor-related uncertainties. Price and volume uncertainties are key aspects of this construct's domain. This measure comprises five items consisting of 5-point strongly agree/strongly disagree scales. The mean of the five items was used as the measure of uncertainty.

Buyer transaction performance. Buyer transaction performance is assessed by using indicators of possession costs and acquisition costs associated with exchange relationships. Our measure of possession cost is inventory turnover (TURNOVER). Buyers were asked to indicate the annual inventory turnover (# turns/year) for bearings purchased from their primary supplier. As with all the other measures used in the study, pretests revealed that the needed information was readily available at this level of disaggregation. Higher numbers for this measure indicate improved purchasing performance.

Two measures of acquisition cost were used in the study. In the first, each respondent was asked to report the percentage on-time delivery (%ON-TIME) record of the primary supplier. It was reasoned that activities such as followup and expediting are dimin- ished when late deliveries are less frequent. Higher numbers for this measure indicate better purchasing performance.

The second measure of acquisition cost asked the respondents to assess the percentage of acceptable bearings delivered (%ACCEPTABLE) by the supplier (i.e., not defective or substandard). Again, the reasoning behind the choice of this measure was that re- medial acquisition activities increase when delivered items are unacceptable.

As we discussed previously, a buying firm could improve turnover and on-time and acceptable delivery performance by extracting more favorable terms of trade from its exchange partner. Because such transaction performance improvements are unrelated to the causal process of interest, they must be accounted for in any test of the focal hypothesis. Five such variables follow.

Amount. This measure consists of the total dollar amount of the buyer's purchases (AMOUNT) from the primary supplier. It serves as a proxy for possible scale economies.

Dependence. The second measure is the buyer's assessment of dependence (DEPBUY) on the supplier

for the bearings. Respondents rated their dependence on a 5-point totally dependent/not at all dependent scale (increases in the magnitude of this scale correspond to a decreased dependence on the supplier). This behavioral construct has been used extensively in previous channels studies (e.g., Buchanan 1986).

The third measure describes the dependence of the supplier (DEPSUP) on the buyer. Respondents rated the supplier's dependence on a 5-point totally dependent/not at all dependent scale (increases in the magnitude of this scale correspond to an increased dependence on the buyer).

Price. Recall that we considered the possibility that buyer transaction performance might be improved if the supplier were willing to provide better service, al- beit for a higher price. Hence respondents were asked to assess the price paid (RELPRICE) to the supplier for bearings in relation to the prevailing market price (as a percentage of the prevailing market price).

Distance. We asked respondents to assess the distance (DISTANCE) of the supplier's warehouse from the buyer (in miles). Presumably, a closer warehouse permits improved delivery times (etc.), but maintain- ing decentralized inventories probably increases the supplier's overall costs.

Frequency. The final measure describes the frequency of orders (or releases) issued by the buyer. A buying firm may improve transaction performance simply by issuing orders more frequently. In the case of turnover, for instance, improvements can be real- ized by ordering more often in smaller lot sizes. To account for this possibility, buyers were asked to indicate the number of orders for bearings issued annually (HOWMANY) from their focal supplier.

Analysis and Results Measure Validation Each variable that was measured with multiple items was subjected to a scale development and purification procedure. The initial sets of items composing each of the five purchasing governance elements and uncertainty were subjected to item analysis procedures to identify a unidimensional scale for each set (Churchill 1979). On the basis of item-total correlations, ill-fitting items were dropped. The subsequent reduced sets of items were subjected to confirmatory factor analysis using LISREL VI (Joreskog and Sorbom 1986).

In the case of the uncertainty items, the measurement structure consisted of the usual one-factor model. The fit statistics showed an adequate fit to the data. The internal consistency of the 5-item scale (UNCER- TAINTY) was calculated to be .64.

The items composing the five governance ele-


ments were subjected to a different analysis because of the more complex measurement structure implied in the theory. Recall that these five elements were considered to constitute a syndrome of relational governance. This syndrome is modeled as a second-order factor model in LISREL, where the observed items arise from five first-order factors (ASSIST, MONIT, INFO, EXPECT, and FLEX). These five factors themselves arise from a single second-order factor (REL). In algebraic terms, it is:

Yt = A(rF + ) + et

where:

y A r

is the 22 x 1 vector of scale items, is the the 22 x 5 first-order loading matrix, is the 5 x 1 second-order loading matrix,

Indicator ASSIST First-Order Loadings (X,j)

AS1 .863b AS2 .338 (7.0) AS3 .662 (7.3) AS4 .591 (7.3) AS5 .674 (8.3) MO1 M02 M03 M04 M05 M06 IN1 IN2 IN3 IN4 EX1 EX2 EX3 FL1 FL2 FL3 FL4

Second-Order Loadings (Yjk) First-order construct ASSIST MONIT INFO EXPECT FLEX

Goodness-of-Fit Statistics X2 (204 d.f.) GFI AGFI RMS residual Bentler and Bonnett p Bentler and Bonnett 8

at-values are in parentheses. bFixed parameter.

is the second-order factor, J is the 5 x 1 vector of first-order factors, and e is the 22 x 1 vector of item residuals.

The parameter estimates (after deletion of ill-fitting items) and associated fit statistics are reported in Ta- ble 2. The hypothesized second-order structure ap- pears to represent the data adequately. All of the loadings are large and significant and the fit indicators, such as the Bentler and Bonnett (1980) indices of fit, as well as the other indices, suggest a reasonable model. Using Joreskog's formula, we calculated the reliabil- ity of the second-order construct as .74.

Hypothesis Test

Recall that the hypothesis corresponds to the two statistical contrasts described previously. We estimated

TABLE 2 Standardized LISREL Estimates'

MONIT INFO EXPECT FLEX

.629b

.648 (6.1)

.777 (6.6)

.519 (5.2)

.442 (5.2)

.676 (6.4) .686b .790 (9.2) .755 (8.7) .573 (7.1)

.366b

.879 (5.7)

.730 (6.1) .373b .743 (5.6) .648 (5.2) .617 (5.2)

REL .964 (7.8) .552 (4.7) .528 (4.9) .356 (3.1) .534 (4.1)

= 328.2, p < .05 = .83 = .79 = .08 = .80 = .90


a linear model of the following form for each of the three dependent variables (TURNOVER, %ON-TIME, and %ACCEPTABLE).

Y = bo + blC1 + b2C2 + b3C4 + b4AMOUNT

+ b5DEPBUY + b6DEPSUP + b7DISTANCE

+ b8RELPRICE + b9HOWMANY + e

The dummy variables C1, C2, and C4 represent the combinations of values of uncertainty (UNCER- TAINTY) and composite relational governance (REL) corresponding to cells 1, 2, and 4, respectively (Fig- ure 1). We split the firms into the high and low categories according to the mean for the composite governance scale and the uncertainty scale. Notice that one of the dummy variables (C3, corresponding to cell 3) is omitted for the usual reason that the complete set of four dummy variables would constitute a lin- early dependent set. The results of the estimation are reported in Table 3. Several diagnostic checks for multicollinearity and heteroskedasticity were undertaken.2 No problems were evident.

Tests of contrasts. In Table 3 are the estimated coefficients and the effect sizes associated with each contrast, as well as their significance levels. The critical theoretical predictions, stated in terms of cell performance differences (P4 - P3) and (P2 - P1), are tested in the following manner. The estimated coefficient corresponding to each effect is first pinpointed. For instance, in the regression with %ON-TIME as the dependent variable, the effect size for the performance difference between cell 4 and cell 3, (P4 - P3), is provided by the coefficient of C4. In Table 3, we find this coefficient, b3, to be 8.00. To assess its statistical significance, we divide the estimate into its estimated standard deviation to obtain the observed t- value, which then is compared with the critical t for significance.

2Several checks were carried out to verify model adequacy. First, "tolerance" estimates were inspected to assess whether multicollinearity was a problem. No evidence was found. Then the residuals were inspected for heteroskedasticity. We plotted the residuals from each equation against the sorted predicted measure and then against the sorted dependent variable. No particular pattern was evident. We also plotted the residuals against each of the independent variables. No patterns were found in these plots to indicate any problem of heteroskedasticity.

As a further precaution, two variables were dropped from the model (HOWMANY and AMOUNT) because their means were very different from their medians. The equations then were reestimated. The tests of the hypotheses on the reestimated model were consistent with those of the original model.

Specifically, the contrast (P4 - P3) in the reestimated model is greater than zero for the %ON-TIME and %ACCEPTABLE equations and is insignificant for the TURNOVER equation. This result is the same as that observed in the original version. Also consistent with the original results, the contrast (P2 - P,) is not different from zero for any of three dependent variables.

As another example, consider the performance difference effect, (P2 - Pi), for the %ON-TIME regression. Table 3 shows the estimated size of this effect as 1.33, which is calculated from the coefficient estimates (b2 - bl = 9.18 - 7.85 = 1.33).

How well do these contrasts support our expectation? To begin, the (P4 - P3) effect, estimated by b3, is positive (as hypothesized) for all three dependent variables. For the %ON-TIME measure, the contrast shows the expected positive sign (8.00) and is significant. As for the %ACCEPTABLE measure, the contrast value is positive (.90) and also significant. Finally, the contrast for the TURNOVER measure shows the expected positive sign (4.78), but is not significant (p = .14). In sum, increased relational governance under high uncertainty increases transaction performance for acquisition costs, but not possession costs.

Turning to the (P2 - PI) effect, estimated by b2 - bl, we find that as hypothesized none of the three effects are different from zero. In the case of %ON- TIME, we find that the contrast is positive (1.33), but not significantly different from zero. Likewise, the contrasts for %ACCEPTABLE (.07) and TURN- OVER (4.42) are insignificantly different from zero. In sum, increasing relational governance has no effect on transaction performance at lower levels of uncertainty.

These results tell us that the pattern of effects consists of two nonparallel lines, as shown in Figure 1. Evidently, the beneficial effects of increased relationalism on transaction performance occur only when levels of uncertainty are relatively high.

Effects of nonfocal variables. Among the non-TCA variables, TURNOVER increases significantly as the total dollar value of bearings purchased (AMOUNT) increases. Possibly buffer stocks increase less than proportionally with amount purchased because of smaller swings in demand, enabling the buying firm to turn its inventory faster.

Performance as measured by %ON-TIME increases as the buying firm becomes less dependent on the supplier (DEPBUY) for bearings. Decreased dependence may give buying firms more leverage with suppliers. The result is greater buyer ability to influence the performance of suppliers in delivery timeli- ness.

Finally, for the model involving %ACCEPT- ABLE, the estimated coefficient to RELPRICE (-1.60) is negative and significant at the .10 level. Thus, as the price paid in relation to the prevailing market price increases, the percentage of bearings delivered that are acceptable declines. This result is counterintuitive and warrants further investigation in future studies.

An alternative set of models also was estimated to


TABLE 3 Estimates of Coefficients

Independent Dependent Variable Variable TURNOVER %ON-TIME %ACCEPTABLE

CONSTANT 20.058 76.84a 99.48" C1 .82 7.858 1.03" C2 5.24b 9.18a 1.10a C3 (not in equation) C4 4.78 8.00a .908 DEPBUY -1.61 2.15" .05 DEPSUP -.80 1.28 -.04 DISTANCE -.002 -.002 2.97E-04 RELPRICE -9.98 -.05 -1.60b HOWMANY -5.62E-04 -.003 -2.10E-04 AMOUNT 1.13E-06" 2.99E-09 2.62E-08 Radj .27 .07 .09 F(9, 130) 6.68 2.088 2.49a

Effect Size Constrast TURNOVER %ON-TIME %ACCEPTABLE Cell 4 vs. cell 3 4.78 8.00 .90a Cell 2 vs. cell 1 4.42 1.33 .07 aIndicates significant at p < .05 (2-tail). blndicates significant at p < .10 (2-tail).

assess the robustness of the results. Because the mean split approach is only one of many possible ways of categorizing continuous data, and involves some degree of inherent arbitrariness, an alternative specification was used to assess robustness. Regressions were run with an uncertainty x relational governance product term, the uncertainty and governance terms indi- vidually, and the remaining variables from the pre- ceding model. The results3 (Table 4) were similar to

3For each of the three performance measures, %ON-TIME, %AC- CEPTABLE, and TURNOVER, regressions of the following form were estimated.

Y = bo + bjUNCERTAINTY + b2REL + b3UNCERTAINTY*REL + b4AMOUNT + ... + b9HOWMANY + e

In this specification, the proposition that the effect of relationalism on performance is contingent upon (or moderated by) uncertainty is supported if b3 > 0. The results of this estimation are reported in Table 4. We find that b3 is, in fact, significantly positive for both %ON-TIME (b3 = 2.09; p = .03) and %ACCEPTABLE (b3 = 2.16; p = .03) and insignificant for TURNOVER (p = .34). These results are consistent with those of the previous tests reported for the mean split model.

A set of checks similar to those in footnote 2 were carried out for this specification as well. As before, "tolerance" estimates were inspected for multicollinearity, and residual plots were inspected for heteroskedasticity. No problems were found. As before, HOWMANY and AMOUNT were dropped from the model and a trimmed model was estimated. The hypothesis test on the interaction coefficient was compared across the two models and consistent results were obtained for all three de-

those from the previous specification, and hence offer evidence of the robustness of the effects.

Concluding Remarks To the best of our knowledge, our study is the first empirical attempt to test the performance implications of TCA by using a multidimensional discrete-relational continuum. We examine whether the structure- performance relationship is a contingent one, based on the TCA framework. The context chosen for study is the relationship between industrial buyer and seller of repetitively purchased ball bearings. Despite TCA's normative implications for how buying firms should organize the vendor interface, and its relevance in a world characterized by environmental uncertainty, few empirical tests of TCA's performance predictions have been reported in the marketing literature or elsewhere. This lack of evidence about the normative aspects of TCA is a crucial gap in the literature, because managerial decision rules based on TCA are meaningful only if they can be demonstrated to enhance performance. The basic finding can be summarized in the following statement.

Increasing relational governance in an industrial buyer- seller relationship when the level of uncertainty is

pendent variables. Specifically, the interaction coefficient in the trimmed model is positive and significant for %ON-TIME and %ACCEPTABLE, and is insignificant for the TURN- OVER equation. This result is the same as that observed in the original version.

90/ Journal of Marketing, October 1990

TABLE 4 Continuous Form Estimates (standardized coefficients)

Independent Dependent Variable Variable TURNOVER %ON-TIME %ACCEPTABLE

CONSTANT .00 .OOa .00O ENV .63 -1.95a -2.00a REL .63 -.86b -.98b REL*ENV -.84 2.09a 2.16a DEPBUY -.130 .20a .04 DEPSUP -.06 .09 -.05 DISTANCE -.07 -.03 .12 RELPRICE -.09 6.68E-04 -.18a HOWMANY -.03 -.08 -.04 AMOUNT .47a -.01 .14 R2dj .28 .12 .09 F(9, 130) 6.99a 3.11 2.45a "Indicates significant at p - .05 (2-tail). blndicates significant at p < .10 (2-tail).

relatively high improves buyer purchasing performance in acquisition cost terms. Such changes have no effect on transaction performance under conditions of relatively lower levels of uncertainty.

In terms of theory development, these results support the prescriptive or normative validity of TCA-that firms adhering to the prescriptions of the model have enhanced purchasing (transactional) performance. The results shed some light on the long-standing issue of whether one should observe detrimental effects at low levels of the predictor variables (uncertainty and/or specific assets). Our results are consistent with those of other empirical studies. Heide and John (1988) found that actions undertaken to safeguard specific assets improved performance at high levels of such assets but had no effect (detrimental or otherwise) at low levels. Anderson (1988) found that a sales district's efficiency increases as it conforms to the industry rule's recommendation (about the direct salesforce vs. rep- resentative decision) in unpredictable environments. In predictable environments, however, conformity did not increase efficiency. Our study shows that actions undertaken to effect adaptation, understood in terms of a discrete-relational continuum, are beneficial at high levels of uncertainty but have no effect (detrimental or otherwise) at low levels.

The results also highlight some needed theory development. In contrast to our view that greater uncertainty warrants shifting away from market modes of exchange, Gatignon and Anderson (1988) found support for their contention that uncertainty in the form of greater country risk evokes greater use of market modes of entry by exporting firms. These differing results are probably due to the different definitions of the uncertainty construct in the studies. Uncertainty in the form of country risk is very different from unanticipated changes (in the prices, volumes, etc.) in

RUI purchases. Notice that country risk uncertainty does not originate within the task environment of the exchange. This type of uncertainty is not mediated by either party to the exchange. Hence, mutual adjustment under increased relational governance does not mitigate the problem. Rather, such risk can be reduced by simply placing fewer investments at risk. One way to accomplish this is to rely on less durable ties. In effect, one relies more on market-based exchange rather than an ongoing relationship, which is the result observed by Gatignon and Anderson. We hope future work will clarify these distinctions between the different aspects of uncertainty.

Another important line of inquiry for future research is integration of the efficiency-oriented TCA analysis with the behavioral approach that emphasizes power and dependence. Our study's explicit focus is buyer performance, defined in terms of efficiency. Thus, dependence (or influence) variables, which are outside the scope of TCA, are introduced only as controls. No attempt is made to integrate them fully into a theoretical framework.

From a managerial perspective, our findings un- derscore the benefits of a theoretical framework for understanding exchange (purchase) relationships. Substantively, the most interesting conclusion is the notion that buying firms can realize enhanced performance by crafting an "appropriate" governance structure (defined multidimensionally and in terms of a continuum), and that this sort of enhancement can be observed after controlling for better performance achieved via improvement in the terms of trade.

However, these results should be viewed in light of the constraints of the study, such as the single-industry focus and the use of the same data to develop measures and test relationships. The cross-sectional data also limit the validity of the causal inferences. It


is possible that relationalism is a response to better transaction performance.

Note also that significant TCA effects are observed only for acquisition costs and not possession costs. The significant effect of scale on the possession cost measure (turnover) suggests that transaction-related factors are less relevant than production-related factors such as size for some performance outcomes. It cautions us that managers should not focus exclu- sively on transaction cost minimization given the multidimensional nature of performance.

Appendix Items Used to Operationalize Relational

Governance Elements

Supplier Flexibility (FLEX) 1. Supplier is flexible in response to requests we make. 2. Supplier can readily adjust its inventories to meet un-

foreseen needs that might occur. 3. Supplier handles change well. 4. Supplier can provide emergency deliveries.

Supplier Assistances (ASSIST) 1. Supplier calls in advance to advise us of shipment (de-

livery) problems. 2. Supplier makes an effort to help us during emergencies. 3. Supplier recommends stock substitutes when delivery

troubles develop. 4. Supplier helps us in value analysis ideas, cost reduc-

tions, problem solving, etc. 5. Supplier advises us of potential problems in meeting our

needs.

Information Provided to Supplier (INFO) 1. We give supplier usage information to help him plan

for our needs. 2. We keep our supplier informed of production plans. 3. We regularly provide supplier with long-range forecasts

of supply requirements. 4. We inform supplier in advance of impending design

changes.

Monitoring of Supplier (MONIT) 1. We advise supplier of its performance in relation to that

of other suppliers. 2. We monitor supplier's inventory levels. 3. We assess supplier's performance through a formal ven-

dor evaluation program. 4. In this arrangement, supplier must provide summary

usage reports, tally sheets, or some similar kind of report (on a quarterly or monthly basis).

5. We conduct quality training for vendor personnel. 6. The relationship we have with supplier makes use of

many controls.

Expectation of Continuity (EXPECT) 1. We expect our relationship with supplier to last a long

time.

2. The relationship we have with supplier is essentially "evergreen."

3. Renewal of relationship with supplier is virtually au- tomatic.

Uncertainty Elements (UNCERTAINTY) 1. Availability of bearings in the market is highly uncer-

tain. 2. Uncertainties in production or distribution of bearings

in the market are a real problem. 3. The market in which we buy bearings is complex. 4. Supply of bearings in the market is not stable. 5. Prices for bearings are volatile.

Inventory Turnover (TURNOVER) On average, what is the annual inventory turnover of bearings bought from this supplier? turns/year

Percentage On-Time Delivery Record (%ON-TIME)

What percentage of orders (deliveries) of bearings from this supplier are late? % (reverse-scored)

Percentage Acceptable (oACCEPTABLE) What percentage of bearings delivered by this supplier are defective, not up to specifications, the wrong items, or otherwise unacceptable? % (reverse scored)

Dollar Amount of Buyer's Purchases (AMOUNT) What is the annual total dollar value of your company's purchases of bearings from this supplier? $

Buyer's Dependence on Supplier (DEPBUY) In purchasing bearings from this supplier, what is your assessment of the following: How dependent are you on this supplier for bearings (reverse coded)?

Totally dependent 5 4 3

Not at all 2 1 dependent

Dependence of Supplier on Buyer (DEPSUP) In purchasing bearings from this supplier, what is your assessment of the following: How dependent is this supplier on you as a customer?

Totally dependent 5 4 3 2

Not at all 1 dependent

Relative Price Paid (RELPRICE) What is the price you pay this supplier for bearings in relation to the prevailing market price? (check appropriate response)

Over 30% below market price 21% to 30% below market price 11% to 20% below market price 1% to 10% below market price Market price 1% to 10% above market price 11% to 20% above market price 21% to 30% above market price Over 30% above market price


Distance of Warehouse From Buyer (DISTANCE) How close is this supplier's warehouse to your company?

miles

Number of Orders Issued Annually (HOWMANY) On average, how many orders (or releases) for bearings are issued annually from this supplier?

REFERENCES Anderson, Erin (1985), "The Salesperson as Outside Agent or

Employee: A Transaction Cost Analysis," Marketing Sci- ence, 4 (Summer), 234-54.

(1988), "Strategic Implications of Darwinian Eco- nomics for Selling Efficiency and Choice of Integrated or Independent Sales Forces," Management Science, 34 (May), 599-618.

and Barton A. Weitz (1983), "Make or Buy Deci- sions: A Framework for Examining Vertical Integration Is- sues in Marketing," Wharton Working Paper No. 83-001, University of Pennsylvania.

Balakrishnan, Srinivasan and Birger Wernerfelt (1986), "Technical Change, Competition and Vertical Integration," Strategic Management Journal, 7 (July-August), 347-59.

Bentler, Peter M. and Douglas G. Bonnett (1980), "Signifi- cance Tests and Goodness of Fit in the Analysis of Co- variance Structures," Psychological Bulletin, 88 (3), 588- 606.

Buchanan, Lauranne (1986), "The Organization of Dyadic Re- lationships in Distribution Channels: Implications for Strat- egy and Performance," unpublished dissertation, Stanford University.

Churchill, Gilbert A., Jr. (1983), Marketing Research: Meth- odological Foundations. Chicago: The Dryden Press.

Dahlman, Carl J. (1979), "The Problem of Externality," Jour- nal of Law and Economics, 22 (April), 141-62.

Dobler, Donald W., Lamar Lee, and David N. Burt (1984), Purchasing and Materials Management: Text and Cases. New York: McGraw-Hill Book Company.

Dowst, Somerby R. (1972), Basics for Buyers: A Practical Guide to Better Purchasing. Boston: Cahners Books.

Dwyer, F. Robert and Sejo Oh (1988), "A Transaction Cost Perspective on Vertical Contractual Structure and Inter- channel Competitive Strategies," Journal of Marketing, 52 (April), 21-34.

, Paul Schurr, and Sejo Oh (1987), "Developing Buyer-Seller Relationships," Journal of Marketing, 51 (April), 11-27.

Ford, David (1980), "The Development of Buyer-Seller Re- lationships in Industrial Markets," European Journal of Marketing, 14 (5/6), 339-53.

Gatignon, Hubert and Erin Anderson (1988), "The Multina- tional Corporation's Degree of Control Over Foreign Sub- sidiaries: An Empirical Test of a Transaction Cost Expla- nation," Journal of Law, Economics and Organization, 4 (Fall), 305-36.

Goldberg, Victor (1976), "Toward an Expanded Economic Theory of Contract," Journal of Economic Issues, 10 (March), 45-51.

Hannaford, William J. (1983), Systems Selling: A Marketing Guide for Wholesaler-Distributors. Washington, DC: Dis- tribution Research & Education Foundation.

Heide, Jan and George John (1988), "The Role of Dependence Balancing in Safeguarding Transaction-Specific Assets in Conventional Channels," Journal of Marketing, 52 (Janu- ary), 20-35.

Heinritz, Stuart F. and Paul Farrell (1981), Purchasing: Prin- ciples and Applications. Englewood Cliffs, NJ: Prentice- Hall, Inc.

John, George and Barton Weitz (1988), "Forward Integration Into Distribution: An Empirical Test of Transaction Cost Analysis," Journal of Law, Economics and Organization, 4 (Fall), 337-55.

Joreskog, Karl G. and Dag Sorbom (1986), LISREL VI: Anal- ysis of Linear Structural Relationships by the Method of Maximum Likelihood. Mooresville, IN: Scientific Soft- ware, Inc.

Kaufmann, Patrick J. and Louis W. Stem (1988), "Relational Exchange Norms, Perceptions of Unfairness, and Retained Hostility in Commercial Litigation," Journal of Conflict Resolution, 32 (September), 534-52.

Macneil, Ian R. (1978), "Contracts: Adjustment of Long-Term Economic Relations Under Classical, Neoclassical, and Re- lational Contract Law," Northwestern University Law Re- view, 72, 854-905.

(1980), The New Social Contract. New Haven: Yale University Press.

(1981), "Economic Analysis of Contractual Rela- tions: Its Shortfalls and the Needs for a Rich Classificatory Apparatus," Northwestern University Law Review, 75, 1018- 63.

Palay, Thomas (1984), "Comparative Institutional Economics: The Governance of Rail Freight Contracting," Journal of Legal Studies, 13 (June), 265-87.

Porter, Michael (1980), Competitive Strategy: Techniques for Analyzing Industries and Competitors. New York: The Free Press.

Ruekert, Robert W., Orville C. Walker, Jr., and Kenneth J. Roering (1985), "The Organization of Marketing Activi- ties: A Contingency Theory of Structure and Performance," Journal of Marketing, 49 (Winter), 13-25.

Ster, Louis W. and Adel El-Ansary (1982), Marketing Chan- nels. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Stinchcombe, Arthur L. (1985), "Contracts as Hierarchical Documents," in Organization Theory and Project Man- agement, A. Stinchcombe and C. Heimer, eds. Oslo: Nor- wegian University Press.

U.S. Bureau of Census (1985), Annual Survey of Manufac- turers. Washington, DC: U.S. Government Printing Of- fice.

Williamson, Oliver E. (1979), "Transaction-Cost Economics: The Governance of Contractual Relations," Journal of Law and Economics, 22 (October), 3-61.

(1981), "The Economics of Organization: The Transaction Cost Approach," American Journal of Soci- ology, 87 (November), 548-77.

(1985), The Economic Institutions of Capitalism. New York: The Free Press.

Reprint No. JM544106
