Determinants of Perceived Interfirm Dependence in Industrial Supplier ...

Journal of Management and Governance 2: 213–232, 1998. © 1998 Kluwer Academic Publishers. Printed in the Netherlands.

213

Determinants of Perceived Interfirm Dependence in Industrial Supplier Relations NIELS G. NOORDERHAVEN1, BART NOOTEBOOM2 and HANS BERGER2 1 Faculty of Economics, Tilburg University, The Netherlands; 2 University of Groningen, The

Netherlands Abstract. Studies of industrial buyer-supplier relations mostly focus on structural characteristics of the transactions between parties, and assume a direct relation between these factors and characteristics of the relationships. This paper focuses on the psychological dimension of transaction relations, reflected in perceptions of dependence, from the perspective of an industrial supplier. In the first step of the analysis, three groups of structural determinants of perceived supplier dependence are explored: factors related to goal mediation, factors related to relation-specific assets, and factors related to network embeddedness. After that, the influence of these structural factors and of perceived supplier dependence on the ordering of buyer-supplier relations is investigated. Data come from a study of the micro-electronics assembly industry in the Netherlands. The findings show that sales to a particular buyer as a percentage of the total sales of the supplier and the growth of sales to a particular buyer (two forms of goal mediation) are important determinants of perceived supplier dependence. Human asset specificity is also related to perceived supplier dependence. Network embeddedness variables play only a minor role in explaining perceived supplier and buyer dependence in this study. The data further show that perceived dependence has an effect on the degree of ordering in the relationship, next to structural factors like the extendedness of the relationship beyond the focal transaction and physical asset specificity. The effects of human asset specificity and dedicated assets on ordering are contrary to what was expected on the basis of the literature.

1. Introduction Industrial supplier relations have attracted much scholarly attention in recent years (see, e.g., Fichman and Goodman, 1996; Håkansson and Johanson, 1993; Imrie and Morris, 1992; Lane and Bachmann, 1996; Leuthesser, 1997; Sharma and Sheth, 1997). These relationships are becoming more important because increased competition and higher paces of innovation call for close cooperation between industrial buyers and their suppliers. In these close cooperative relationships investments in relation-specific assets often take place. These relation-specific investments enhance capabilities for effective and efficient supply. For instance, close cooperation may enable a supplier to deliver components ‘just-in-time’ to a manufacturer. In many cases this is only possible if at least one of the firms concerned adapts its production and logistic processes to those of its partner. Close cooperation also helps firms to survive under a ‘regime of rapid innovation’ (Williamson, 1985: 143). If a manufacturer involves prospective

214

NIELS G. NOORDERHAVEN ET AL.

suppliers at an early stage of the process of developing new products, delays in bringing the products to the market are avoided (‘early-supplier-involvement’). However, this kind of cooperation compels the manufacturer to adapt the product developed to the capabilities of the selected suppliers, and these suppliers have to invest resources without knowing with certainty if there will be a quid-pro-quo. These examples demonstrate that close cooperation, while enabling firms to better meet the requirements of increasingly competitive markets, also leads to higher levels of dependence (Håkansson, 1982, 1989; Heide and John, 1988). As a result of relation-specific investments, the transacting firms face the conditions of a bilateral monopoly, in which they are more or less ‘condemned’ to continue doing business which each other. If one of the parties is more dependent than the other, opportunistic rent-seeking behaviour by the less dependent may be provoked (Klein et al., 1978). If both are equally dependent, costly haggling over inputs and outputs may occur (Williamson, 1985: 21). In both cases many of the advantages may dissipate. Thus cooperative buyer-supplier relations inevitably bear the seeds of conflict. This paper focuses on the factors determining the level of supplier dependence in supplier-buyer relationships. In the literature, various factors are suggested, such as investments in relation-specific assets, the importance of sales to a specific buyer relative to the total turnover of a supplier, and the relative position of supplier and buyer in a larger network. It is argued that these structural factors are important, but that also psychological factors have to be taken into account. The extent to which structural determinants lead to consequences for control mechanisms applied in the transaction relation (as hypothesized by, e.g., transaction cost economics) may be expected to be mediated by psychological factors, such as the level of trust in the other party. Extending the analysis of supplier-buyer relations to include also psychological factors is important, because it helps making these theories more realistic and managerially relevant (cf. Noorderhaven, 1996). Management decisions shape interorganizational relationships, and managers act on their perception of the environment rather than on some objective rendering of structural factors. In this paper we will not try to ascertain which psychological factors impact on the relationship between structural situational factors and the way in which managers shape buyer-supplier relations. Rather, we will do some preliminary work to prepare the ground for an inclusion of psychological factors by analyzing the relations between the perception of dependence, structural factors of the relationship, and control measures such as legal and private ordering mechanisms. The ideas developed in this paper are tested in an empirical study of interfirm relationships in the micro-electronics assembly industry in the Netherlands. Section 2 discusses the distinction between structural and psychological determinants of dependence. The perception of dependence is posited as an important variable mediating the relationship between structural characteristics of relationships and control mechanisms employed in these relationships. Section 3 focuses

PERCEIVED INTERFIRM DEPENDENCE

215

on structural determinants of dependence. On the basis of relevant literature, factors that are likely to influence perceived dependence are discussed, and hypotheses are formulated. In Section 4 the empirical study is described. The findings of this study are presented and discussed in Section 5. Conclusions follow in Section 6. 2. Structural and Psychological Determinants of Dependence In much of the literature on industrial buyer-supplier relations, the focus is on explaining the forms these relations assume on the basis of structural characteristics, like investments in relations-specific assets. In the next section, we will discuss the structural factors suggested by three schools of thought. But first we will introduce another type of determinants, viz., psychological determinants. Organizations are no natural phenomena, but man-made artifacts. Therefore, structural situational characteristics can only have consequences for characteristics of interorganizational relations through managerial actions. Managers will act or refrain from action depending on their perception of the situation. Managerial perceptions of dependence may be of crucial importance for the governance of interfirm relations (Berger et al., 1993, 1995; Bresnen, 1996). A supplier may according to objective measures be heavily dependent on a particular buyer, as long as the managers of the firm do not perceive this dependence (e.g., because it has grown very gradually, and not as the result of one or more conscious decisions), no behavioral implications are to be expected (Noorderhaven, 1995, 1996). A certain level of dependence may also be perceived, but psychologically discounted, e.g.: either as a result of generally favourable experiences in the past with the same buyer or in anticipation of being able to manoeuvre into a more favourable position (perhaps through projected growth or planned diversification) (Bresnen, 1996: 135). If psychological variables are accepted as being important in industrial supplier relations, the following conceptual framework can be adopted (see Figure 1). The framework is built on the assumption that the perception of dependence can be distinguished from its antecedents. There is some evidence that practitioners can indeed do so (cf. Smith et al., 1997), and therefore it makes sense also to make this distinction in the conceptual model. In the framework, the structural factors are those mentioned in the literature on industrial supplier relations, e.g., asset specificity. Managerial perception, like a supplier’s perception of being dependent on a particular buyer, are influenced by these structural factors, and lead to managerial action, e.g., in the form of installing control mechanisms or transactional safeguards. But managerial perceptions are also influenced by ‘other factors’. For instance, Joshi and Arnold (1997) found that under conditions of low relational norms dependence and opportunism were positively related, but under conditions of high relational norms an inverse relation was found. Thus, it is likely that the institutionalized norms and rules influence the perception of dependence and its

216


Figure 1. Explanatory framework.

consequences. Empirical evidence of such differences between supplier relations in Germany and Britain is provided by Lane and Bachmann (1996). A two-way relationship between managerial perceptions and relationship characteristics is hypothesized. On the one hand, the perception of dependence based on structural factors may cause a manager to install transactional safeguards. On the other hand, these safeguards, once installed, may alter the perception of dependence, or at least its saliency. A supplier in a situation of dependence may have confidence that the buyer will not behave opportunistically because there is enough trust in the relationship, but also because adequate control mechanisms are present (Das and Teng, forthcoming). 3. Structural Determinants of Dependence Three, partly overlapping, bodies of literature dominate the scholarly discussion of buyer-supplier relations in industry: the literature on distribution channels, transaction cost economics, and the theory of industrial networks. Below we will briefly discuss these three literatures and the structural determinants of (perceived) dependence they suggest. T HE CHANNELS ’

LITERATURE

An important school of thought within the channels’ literature applies Emerson’s (1962) theory of interpersonal dependence in the context of interfirm relations. In this view, the dependence of firm A on firm B is a function of the degree in which firm B mediates goals pursued by A, as well as of the centrality of those goals to A. Very central goals of suppliers are sales and profit, therefore suppliers are relatively dependent on buyers that are responsible for a large fraction of their sales and profit. The extent to which a buyer can be said to mediate these goals of


217

a supplier depends on the presence of alternative buyers and suppliers: alternative suppliers increase, and alternative buyers decrease the capability of goal mediation of the buyer. Frazier et al. (1989) found a significant positive relationship between perceived dependence and four measures of goal mediation by the other party, viz. sales realized in the focal transaction relation as a percentage of total sales, profit realized in the focal transaction relation as a percentage of total profit, expected future sales percentage, and expected future profit percentage. Expectations about future sales may be based on various sources of information, but one very important source will probably be what happened in the past. Therefore, we may assume that growth of sales to a particular buyer in the past leads to the expectation of even more sales in the future, and in this way contributes to perceived supplier dependence. But there is more to exchange than present and future turnover and profit. Some transaction partners may perform their roles more excellently than others, making them more attractive for future interactions. Supplying a particular buyer may for instance be attractive because the buyer provides the supplier with technical and/or market information, or helps the supplier to build up new capabilities. Such transaction relations are more extended than simple buyer-seller relations. Other things being equal (e.g., the same levels of sales percentage and profit percentage), such a buyer is a more attractive partner, and the supplier therefore may feel more dependent on future transactions with it. Frazier (1983) found a positive relationship between the performance of a manufacturer, in the sense of agreeing with the dealer on a marketing strategy and of being perceived to have an interest in the dealer’s welfare, and the dealer’s perceived dependence.

T RANSACTION

COST ECONOMICS

In the logic of transaction cost economics dependence of one party on another is not primarily associated with sales percentage or role performance, but with investment in relation-specific assets. A supplier that has invested in assets which are specific to the production for a particular buyer will feel more dependent than a supplier using general purpose equipment. Perceived dependence is not normally taken into consideration in the transaction cost literature, however. According to transaction cost theory parties to a transaction foresee the dangers associated with dependence resulting from investment in relation-specific assets, and choose a governance form which offers adequate safeguards. These safeguards can have the form of provisions in legally enforceable contracts (legal ordering), or other-than-legal arrangements, for instance, the buyer pays for and possibly owns any tools specific to the production for itself (private ordering). Thus transaction cost economics hypothesizes a positive relationship between the level of asset specificity present in a supplier relation and the stringency of safeguards. This hypothesis has been partially corroborated in empirical studies (for overviews of the relevant literature see Joskow (1988) and Noorderhaven (1994)).

218


Relation-specificity of assets is a general concept covering quite diverse phenomena. Four forms of asset specificity are generally distinguished: physical asset specificity (specialized machinery, equipment or tools), human asset specificity (specialized know how), dedicated assets (investments in assets that are not specialized as such, but incurred specifically for a single large buyer), and location specificity (location of premises near an important transaction partner) (Williamson, 1985: 95–96). The various forms of asset specificity are likely to differ in the strength of their impact on perceived supplier dependence. Investments in specialized physical assets are normally the outcome of conscious management decisions, but that is not necessarily the case with specificity of knowledge (human asset specificity), which may very well originate in largely unconscious learningby-doing processes. The latter form of asset specificity could therefore be assumed to have a less direct impact on perceived dependence (Noorderhaven, 1995). Therefore, it is advisable to analyze the influence of various kinds of asset specificity separately.

N ETWORK

THEORY

Transaction relations between industrial suppliers and buyers are embedded in networks of multiple relations with other firms in the same region or industry (Granovetter, 1985; 1992). Network theory takes the implications of this condition of embeddedness into account. From the perspective of a focal supplier, three types of network embeddedness can be distinguished. Firstly, the focal supplier can be a link in an indirect connection between a buyer and other suppliers. This is for instance the case for main suppliers in the Japanese automobile industry, which coordinate the inputs of second and third tier subcontractors that are not directly in contact with the automobile manufacturers (Asanuma, 1989). This kind of network embeddedness can be called ‘upstream embeddedness’. It may be assumed to increase the dependence of the buyer, because he can less easily dispose of the first-tier supplier (for the buyer would also have to replace the network of contacts maintained by the supplier). Consequently the dependence of the supplier, relative to that of the buyer, will decrease. In the opposite case of ‘downstream network embeddedness’ the supplier does not only sell to a particular buyer directly, but also indirectly. That is, part of the total turnover of the supplier consists of deliveries to third firms, which use these inputs in their production for the same buyer firm. In this situation the supplier is more dependent, for not only the direct but also the indirect sales to the buyer firm are at risk in the case of a conflict (assuming the buyer firm can exert influence on these third firms). Finally, the focal supplier may maintain many other kinds of links: with buyers in other industries, with competing suppliers, or with third firms with which it has neither a competitive nor a transaction relation. An example of the last cate-


219

gory are informal social relationships between the managers of firms which are located within the same community. This third category of relationships may be very important, for instance when informal indirect links can be used to open up new transaction channels. However, because our data contain no information concerning this type of network relations, we will in this paper concentrate on upstream and downstream network embeddedness. H YPOTHESES In the light of the above discussion several research questions are pertinent. Which of the factors suggested by the channels’ literature, transaction cost economics, and network theory contribute significantly to the supplier’s perception of dependence? And, referring to the conceptual model presented in Section 2, how do various structural factors and perceived dependence impact on relationship characteristics? Below we will formulate hypotheses summarizing the expected effects. The first hypothesis pertains to the various forms of goal mediation stressed in the channels’ literature. A high level of sales is an important goal of business firms. Hence, a supplier will feel more dependent on a buyer who is responsible for a larger share of his total turnover (the abbreviated name of the relevant independent variable is between parentheses): H1a: Perceived supplier dependence on a particular buyer is positively related to the share of output taken by that buyer (sales percentage). Apart from present sales levels, future sales levels are important, too. Our dataset contains no information on expected future sales levels per buyer. But, as expounded above, we expect growth of sales in the past to be a good proxy for expectations of growth in the future. For that reason the second hypothesis pertaining to goal mediation is: H1b: Perceived supplier dependence on a particular buyer is positively related to the growth of the output taken by that buyer (growth). Finally, other factors than the quantitative level of sales to a particular buyer may be expected to be of importance. If the relationship is more extended, i.e., consists of more than just transacting goods produced according to specifications, the supplier stands more to loose if the relationship is discontinued. The extra activities may be the exchange of technical or market information, product development, quality improvement, etc. H1c: Perceived supplier dependence on a particular buyer is positively related to the extendedness of the relationship with that buyer beyond the transaction of goods or services (extendedness). The hypothesis based on transaction cost economics is straightforward. Higher investments in relation-specific assets, of any of the four kinds discussed

220


above, may be expected to lead to higher objective as well as perceived supplier dependence (albeit not necessarily to the same extent). Hence Hypotheses 2a–d: H2a: Perceived supplier dependence on a particular buyer is positively related to the level of physical asset specificity (physical assets). H2b: Perceived supplier dependence on a particular buyer is positively related to the level of human asset specificity (knowledge). H2c: Perceived supplier dependence on a particular buyer is positively related to the level of location specificity (location). H2d: Perceived supplier dependence on a particular buyer is positively related to the level of dedicated assets (dedicated assets). Hypothesis 3 focuses on factors described in the network literature. Our data are rather limited in this respect, allowing us only to test hypotheses pertaining to what we called above upstream and downstream embeddedness. As discussed above, we expect upstream embeddedness to make the buyer more, and therefore the supplier relatively less, dependent. Downstream embeddedness, on the other hand, is hypothesized to increase perceived supplier dependence, as the amount of business that is at risk if a relationship turns sour is larger. H3a: Perceived supplier dependence on a particular buyer is negatively related to the level of upstream network embeddedness (upstream embeddedness). H3b: Perceived supplier dependence on a particular buyer is positively related to the level of downstream network embeddedness (downstream embeddedness). Hypotheses 4 and 5 pertain to the relationship between structural factors and perceived dependence on the one hand, and characteristics of the relationship on the other. In this study we focus on the extent of ordering in the relationship, i.e., the extent to which explicit mechanisms for minimizing transactional risks have been installed. Two types of ordering can be discerned, viz., legal ordering and private ordering (Williamson, 1985). Legal ordering refers to the protection given by a complete and enforceable contract. Private ordering refers to other-than-legal arrangements. For the purpose of this study we assume that these two types of ordering are functional equivalents, and in testing hypotheses 4 and 5 we have simply combined them to form a single dimension of ordering. We assume that both structural factors and perceived dependency, impact on ordering, i.e., perceived dependence has an effect independent of that of the structural factors. Although we realize that the relationship between the perception of dependence and the degree of ordering is complex, not in the last place because the supplier cannot unilaterally decide on ordering, we have chosen for a set of relatively simple hypotheses:

221


Table I. Overview of hypothesized relationships

Sales percentage Growth of sales Extendedness of relation Physical assets Knowledge specificity Location specificity Dedicated assets Upstream embeddedness Downstream embeddedness Perceived dependence supplier

H4:

H5:

Perceived supplier dependence

Ordering

H1a: + H1b: + H1c: + H2a: + H2b: + H2c: + H2d: + H3a: − H3b: +

H4: + H4: + H4: + H4: + H4: + H4: + H4: + H4: − H4: + H5: +

Sales percentage, growth, extendedness, asset specificity and downstream embeddedness are positively related to ordering; upstream embeddedness is negatively related to ordering. Perceived dependence of the supplier is positively related to ordering.

The hypothesized relationships are summarized in Table I. 4. A Study of Suppliers’ Perceptions of Dependence S ETTING

OF THE STUDY

The hypotheses formulated above will be tested against data from an empirical study carried out by the authors in 1992–1994. In a pilot project we first analyzed the supplier relations of one particular firm, a Dutch manufacturer of office equipment. In the first phase of this study, we interviewed the manufacturer’s purchasing managers, as well as their counterparts at 12 suppliers. The purpose of these interviews was the drafting and testing of a paper-and-pencil questionnaire. At the second stage we sent this questionnaire to 80 of the manufacturer’s largest suppliers. This pilot project enabled us to construct and refine items and scales for measuring the variables of interest. The study itself focused on the micro-electronics assembly industry in the Netherlands. In this industry components for, e.g., telecommunications equipment and process control devices are produced, often in small series and according to the specifications of the buying firms. Suppliers were approached via the employer’s association for the electronics and metal industry in the Netherlands, FME. Ten suppliers agreed to cooperate in the study. These firms were visited by a member of the research team. These visits took an average of three-and-half hours. During the visit, data pertaining to the relationships with ten of the firm’s most impor-

222


tant customers were collected. Questionnaires were completed by the respondent (either the general manager or the sales manager of the firm) with the researcher clarifying questions when necessary. The questionnaires were completed horizontally, i.e., a question was answered for all 10 relationships with buyers, before moving to the next question.1 In this way data with regard to 97 buyer-supplier relations were obtained. M EASURES The questionnaire used in the study was based on that developed in the pilot study, omitting the items that had proved to be of little value, and adding some new items. In general, our impression is that the data from the study are of excellent quality, given our purposes. Because one of the researchers was present while the respondent completed the questionnaire the possibility that questions would be misunderstood was minimized. This method also had the advantage that there were no missing data. Multiple-item scales were constructed for as many of the dependent and independent variables discussed above as possible. Where necessary, single items were used. In the case of multiple-item scales, Cronbach’s alphas and confirmatory factor analyses were used to assess the reliability of the scale. In an appendix to this paper the items, scales and measures are summarized. All multi-item scales had alphas above 0.7, except growth (0.68). We do not see the low alpha of this scale as a serious problem. Growth, operationalized in our study as a qualitative and quantitative positive development, can be seen as a formative rather than a reflective scale (cf. Heide and John, 1990). As the various items in the scale are not assumed to reflect a single underlying dimension, a high alpha is not expected. As the size of supplier firms may have an effect on its relationships with buyers which is not captured by variables like ‘sales percentage’, we added a control variable for size (measured in terms of total yearly turnover). Table II shows the correlations between all the variables used in the study. 5. Findings P ERCEIVED

SUPPLIER DEPENDENCE

Perceived supplier dependence was regressed on all independent variables simultaneously.2 The results are shown in Table III. A first observation on the basis of Table III is that a substantial proportion (more than 50%) of the variance in perceived dependence of the supplier can be explained on the basis of the independent variables included. Very important explanatory factors are sales percentage and growth. Knowledge specificity also contributes significantly to the explanation of supplier dependence. The effect of physical asset specificity, though not significant, is striking because it is in the direction opposite

Variable

1

1 2 3 4 5 6 7 8 9 10 11 12

0.637∗∗ 0.397∗∗ 0.140 0.275∗ 0.124 0.370∗∗ 0.296∗ 0.289∗ 0.269∗ 0.460∗∗ ∗∗ ∗ ∗ 0.385 0.257 0.247 0.557∗∗ 0.190 0.156 −0.194 0.169 0.529∗∗ 0.648∗∗ 0.187 0.389∗∗ 0.159 0.179 0.169 0.002 ∗ 0.071 0.297 0.097 −0.016 0.327∗∗ 0.255∗ 0.295∗ 0.333∗∗ ∗ 0.107 −0.042 0.319 0.189

Dep. supplier Sales% Growth Extendedness Physical assets Knowl. spec. Location spec. Dedic. assets Upstr. emb. Dwnstr. E. Ordering. Size

2

3

4

5

6

0.598∗∗ 0.314∗ 0.197 0.558∗∗ 0.427∗∗ 0.146 −0.100 0.042 0.120 0.372∗∗ 0.227 0.040 0.373∗∗

7

8

9

10

11


Table II. Correlations between variables

0.318∗ −0.070 0.161 −0.105 0.159 0.040 −0.027 0.091 0.272∗ 0.129 ∗ −0.291 −0.172 −0.093 0.052 0.490∗∗

∗ p < 0.01 ∗∗ p < 0.001

(two-tailed significance levels)

223

224


Table III. Dependent variable: Perceived dependence supplier Sales% Growth Extendedness Physical asset spec. Knowledge specificity Location specificity Dedicated assets Upstream embeddedness Downstream embeddedness Size Constant Adjusted R-square F-statistic Significance F N

0.524 (0.000) 0.331 (0.000) −0.035 (0.703) −0.180 (0.081) 0.232 (0.039) 0.150 (0.073) 0.112 (0.342) 0.059 (0.434) −0.143 (0.059) 0.026 (0.777) 0.831 (0.376) 0.533 11.959 0.000 97

Standardized regression coordinates. Significance levels of T-values between parentheses.

to the hypothesis. Downstream embeddedness is almost significant, but also in the ‘wrong’ direction. O RDERING In the second step of the analysis, we regressed our combined variable ordering (a combination of legal and private ordering) on all dependent variables used in the first analysis, plus perceived supplier dependence. The results are shown in Table IV. Five of the structural variables show a significant effect on ordering, but in no less than three cases (knowledge specificity, dedicated assets, and upstream embeddedness), the direction is opposite to that hypothesized. Furthermore, even if the direct effect of the structural variables is taken into account, there still is a (positive) effect of perceived dependence on ordering. The control variable for size is also strongly significant, indicating that larger suppliers have stronger safeguards. C OLLINEARITY In studies such as these with multiple independent variables which are conceptually close the danger of multicollinearity looms large. We have checked for collinearity by inspecting the tolerances of all the independent variables, defined as 1 − R2i , where Ri is the multiple correlation coefficient when the ith independent variable is predicted from the other independent variables (Norušis, 1990). Supplier depen-

225


Table IV. Dependent variable: Ordering Sales% Growth Extendedness Physical asset spec. Knowledge specificity Location specificity Dedicated assets Upstream embeddedness Downstream embeddedness Perceived dependence supplier Size Constant Adjusted R-square F-statistic Significance F N

0.204 (0.068) −0.136 (0.129) 0.317 (0.001) 0.510 (0.000) −0.438 (0.000) 0.037 (0.659) −0.336 (0.005) 0.204 (0.007) 0.119 (0.113) 0.251 (0.020) 0.564 (0.000) 3.460 (0.189) 0.554 11.853 0.000 97

Standardized regression coordinates. Significance levels of T-values between parentheses.

dence, which entered as dependent in one, and as independent variable in the other equation, was excluded in this analysis. Tolerances varied from 0.758 (upstream embeddedness) to 0.314 (dedicated assets). The last tolerance level, although rather low, is still acceptable. Discussion Only three of the nine (sub)hypotheses pertaining to perceived supplier dependence are corroborated. The findings with regard to sales percentage and growth are consistent with the predictions based on the channels’ literature, but the hypothesized effect of extendedness was not found. The findings with regard to the independent variables based on transaction cost economics are somewhat ambivalent. The positive effect of knowledge specificity is consistent with transaction cost theory, but the expected effect of dedicated assets and location specificity was not found. The (insignificant) negative effect of physical asset specificity is surprising. We return to this observation below. No significant effect on supplier dependence of network embeddedness was found. It is possible that our measures (subjective judgements at the level of one party of a dyad) do not adequately capture characteristics of network structures that influence supplier perceptions and behaviour. Of the ten effects on ordering summarized in hypotheses 4 and 5, only three were supported by our data, i.e., the positive effects of extendedness, physical asset

226


specificity, and perceived supplier dependence. Strikingly, knowledge specificity and dedicated assets are associated with less ordering, rather than more. The effect found for upstream embeddedness is also opposite to that hypothesized. In this case the explanation may be relatively simple: if upstream embeddedness makes the buyer more vulnerable, the higher degree of ordering in the relationship may be his doing. The negative relationship found between perceived supplier dependence and physical asset specificity (although not reaching significance) is problematic from a transaction cost economics perspective. Our finding is not without precedent: Semlinger (1991), in a study of subcontracting relationships in the German automobile industry, found that subcontractors saw themselves as less vulnerable when they themselves invested in relation-specific assets, than when these assets were owned by the buyer. In the latter case, the buyer can more easily switch to another supplier, in the first situation the supplier believes to have some grip on the buyer. We have a comparable explanation for the finding in our study. Our impression from our conversations with respondents is that the absolute level of physical asset specificity in the micro-electronics assembly industry is rather low (this is not reflected in our measures, which capture a subjective assessment by respondents of differences between the investments made for various customers rather than an absolute level). If physical asset specificity in the micro-electronics assembly industry, consisting for instance in test equipment specific to the products made for a particular client, involves only minor costs, the net effect in the mind of the supplier may be that he disregards his relatively small investment that theoretically is at risk, and focuses on his relative advantage vis-à-vis competitors, which has the effect of making the buyer somewhat more dependent. A comparable explanation might be given for the negative relationship we found between knowledge specificity and ordering and dedicated assets and ordering. If a supplier invests in the development of knowledge and skills for a particular buyer, this will make it more difficult for the buyer to replace him, for which reason safeguards become less needed. The case of dedicated assets is more difficult to understand. The only explanation we can think of is that it is very difficult for a supplier to get adequate safeguards from a buyer for investments which are not manifestly specific to that buyer. The fact that we did not find any effect for location specificity can easily be explained by the nature of the industry investigated. Location specificity may be expected to play an important role only in a limited number of special cases, for instance mine-mouth electricity-generating plants (Joskow, 1985). In our study the potential role of location specificity is very limited because of the nature of the components delivered (not very voluminous or weighty, and relatively expensive). Communication over long distances by means of phone or fax (e.g., in the development process) may also be assumed to be relatively easy, given the unidimensionality of the products.


227

6. Conclusions Our findings with regard to structural determinants of perceived supplier dependence confirm only a limited number of the predictions based on the three schools of thought. The effects of sales percentage and growth on perceived supplier dependence were particularly strong, but our findings with regard to network embeddedness are meagre, and those with regard to transaction specificity are not all straightforward. The findings with regard to the determinants of ordering show that a number of the structural factors hypothesized are in fact important, but that there also is an additional effect of perceived dependence, although not very strong. The fact that some of the relationships between structural factors and ordering were in the wrong direction suggests that this relationship is not as cut and dried as is sometimes assumed in the literature. Our study has several limitations that should be borne in mind when interpreting the results. Industry-specific factors may play a role. In the industry investigated the buyer-supplier relations tend to be relatively simple, not manifestating the more complex patterns of reciprocal interdependence expected in the context of ‘intensive technology’ (Bresnen, 1996: 130). Moreover, the procedure of acquiring data concerning almost one hundred supply relations from ten respondents only is less than ideal. A different kind of limitation is the fact that the questionnaires used apparently are unfit to capture important network characteristics, as noted above. Finally, it would be interesting to go beyond the perception of only one of the parties, and confront the views of suppliers with those of buyers.3 Ultimately, the design and development of buyer-supplier relations depends on the perceptions of both parties. On the basis of the considerations expressed above, the outlines of an ideal follow-up study can be sketched. Firstly, the study should also include some of the ‘other factors’ which may be assumed to determine perceived dependence. The level of trust within (Kozak and Cohen, 1997), and the age of the relationship (Leuthesser, 1997) are two obvious candidates, as are relational norms (Joshi and Arnold, 1997). Secondly, the survey should not be confined to one industry only. Thirdly, each supplier should report on only a few relationships with buyers. This means that more suppliers need to be approached in order to get an acceptable sample size. Fourthly, more efforts should be made to also obtain the perceptions of the buyers. Finally, the procedure of visiting suppliers (and possibly also buyers), explaining items when necessary, and with horizontal answering of questions, in our view leads to more complete and reliable data than a mailed questionnaire, and should therefore be maintained. Obviously, there is a tension between the advisability of adhering to this procedure and the desire to limit the number of cases obtained from a single supplier. Here, as always, a balance must be struck between what is theoretically optimal and what is practically attainable.

228


Acknowledgements This paper is a product of a research project sponsored by the Economics Research Foundation, which is part of the Netherlands’ Organization for Scientific Research (NWO). The authors gratefully acknowledge valuable comments on an earlier version made by Ashwin Joshi, as well as by the editor and an anonymous reviewer of this journal. Appendix: Summary of Scales and Measures Perceived Dependence Supplier 2 five-point items (alpha = 0.90): − We cannot afford to loose this customer. − If we loose this customer, it will be very difficult to maintain our current total level of sales.

Perceived Dependence Buyer 2 five-point items (alpha = 0.85): − This customer cannot afford to loose us as a supplier. − It would be very difficult for this customer to replace us with an equivalent supplier.

Sales Percentage Sales to buyer divided by total sales.

Growth of Sales 2 five-point items (alpha = 0.68): − The relationship between our firm and this customer has become better and better over the years. − Our sales to this customer have increased substantially over the years.

Extendedness of the Relationship 5 five-point items (alpha = 0.75): − The fact that we do business this customer gives us the opportunity to buil up technological know how that can also be used for other customers. − Through our business with this customer we get market information that otherwise would be difficult to get access to. − We are an important source of information about new technologies for this customer. − This customer involves us at an early stage in the development of new components (‘early-supplier-involvement’).


229

− This customer involves us in the testing or prototyping of components.

Dedicated Assets 2 five-point items (alpha = 0.75): − Our firm employs substantially more people than if we would not do business with this customer. − Our firm has built up extra capacity in order to be able to do business with this customer.

Physical Asset Specificity 3 five-point items (alpha = 0.75): − For the production for this customer we need very specific machinery, equipment, or instruments. − Most of the machinery, etc., used for the production for this customer can if necessary also easily be employed for other customers (reversed). − We have made investments in order to meet the terms of delivery specific to this customer (e.g., ‘just-in-time’).

Knowledge Specificity 4 five-point items (alpha = 0.71): − We have had to invest much time in learning to know the procedures of supply demanded by this customer (e.g., concerning logistics or quality assurance). − It takes a lot of specific technological know how to supply this customer effectively. − It takes a lot of know how with regard to the internal organization of this customer to supply them effectively. − Our firm has attracted employees with specific know how because of our business with this customer.

Location Specificity 1 five-point item: − The location of our firm is important in the relationship with this customer.

Asset Specificity Buyer 1 five-point item: − This customer has adapted his product design and/or the set up of his production process to the specific character of the inputs supplied by us.

230


Upstream Embeddedness 1 five-point item: − In the production for this customer we make substantial use of subcontractors and components supplied by third firms (other than the purchase of standardized components).

Downstream Embeddedness 1 five-point item: − We have substantial business with firms that in their turn supply this customer.

Legal Ordering (as perceived by the supplier) 3 five-point items (alpha = 0.79): − The contract with this buyer is as complete as possible. − The contract forms the core of our relationship with this customer. − In this relationship it is not very important to us to have a good contract (reversed).

Private Ordering (as perceived by the supplier) 4 five-point items (alpha = 0.71): How important are the following arrangements in the relationship with this customer: − − − −

Co-financing of relation-specific machines by the customer. Co-financing of relation-specific tools by the customer. Guarantees given by the customer for minimal purchases in a specified period of time. Guarantees given by us for delivering during a specified period of time.

Ordering (Legal and Private Ordering Combined) 7 five-point items (alpha = 0.78)

Size Annual turnover (in Dfl. 10,000)

Notes 1 This procedure has the advantage that the respondent focuses on the relative differences between

his relationships with customers, which is exactly what is at issue in this study. However, the absence of an absolute yardstick renders a direct comparison between firms/respondents more difficult. To check for respondent-specific bias, we have included dummy-variables for respondents in our analysis (see below). 2 In the study ten respondents were asked for opinions on in total almost one hundred relationships with customers. Biases may be introduced by this data collection procedure (cf. Hitt and Tyler, 1991).


231

To check for respondent-specific biases, the regression analysis was repeated with a dummy variable for every respondent. First the dummy variables for respondents 1 through 9 were forced into the equation (since the dummies are mutually exclusive, the dummy for the tenth respondent was left out), followed by the independent variables to test the hypotheses. The three independent variables that were significant previously are significant now, too. But controlling for respondent-specific bias, the effect of dedicated assets also becomes significant (in the hypothesized direction), as well as that of physical asset specificity (still in the wrong direction). All in all the findings suggest that the relationships identified are not spurious. 3 Our research design included a mailed survey among the buying firms, but because of the low response rate (only 38 questionnaires were returned) we decided to exclude it from the analysis. However, inspection of the little data we do have suggests important differences in the perceptions of buyers and suppliers.

References Asanuma, B.: 1989, “Manufacturer-Supplier Relationships in Japan and the Concept of RelationSpecific Skill”, Journal of the Japanese and International Economies 3: 1–30. Berger, J., N.G. Noorderhaven, B. Nooteboom and B.J. Pennink: 1993, “Understanding the Subcontracting Relationship: The Limitations of Transaction Cost Economics”, in J. Child, M. Crozier, R. Mayntz et al. (eds.), Societal Change Between Market and Organization (Aldershot: Avebury), pp. 77–98. Berger, J., B. Nooteboom and N.G. Noorderhaven: 1997, “Effects of Trust and Governance on Relational Risk”, Academy of Management Journal 40: 308–338. Bresnen, M.: 1996, “An Organizational Perspective on Changing Buyer-Supplier Relations: A Critical Review of the Evidence”, Organization 3: 121–146. Emerson, R.M.: 1962, “Power-Dependence Relations”, American Sociological Review 27: 31–41. Das, T.K. and B.-S. Teng: forthcoming, “Between Trust and Control: Developing Confidence in Partner Cooperation in Alliances”, Academy of Management Review. Fichman, M. and P. Goodman: 1996, “Customer-Supplier Ties in Interorganizational Relations”, in B.M. Staw and L.L. Cummings (eds.), Research in Organizational Behavior (Greenwich, CT: JAI Press), 18, pp. 285–329. Frazier, G.L.: 1983, “On the Measurement of Interfirm Power in Channels of Distribution”, Journal of Marketing Research 20: 158–166. Frazier, G.L., J.D. Gill and S.H. Kale: 1989, “Dealer Dependence Levels and Reciprocal Actions in a Channel of Distribution in a Developing Country”, Journal of Marketing 53: 50–69. Granovetter, M.: 1985, “Economic Action and Social Structure: A Theory of Embeddedness”, American Journal of Sociology 91: 481–510. Granovetter, M.: 1992, “Economic Institutions as Social Constructions: A Framework for Analysis”, Acta Sociologica 35: 3–11. Håkansson, H. (ed.): 1982, International Marketing and Purchasing of Industrial Goods – An Interaction Approach (Chichester: Wiley). Håkansson, H.: 1989, Corporate Technological Behaviour; Co-operation and Networks (London: Routledge). Håkansson, H. and J. Johanson: 1993, “Industrial Functions of Business Relationships”, Advances in International Marketing 5: 15–31. Heide, J.B. and G. John: 1988, “The Role of Dependence Balancing in Safeguarding TransactionSpecific Assets in Conventional Channels’, Journal of Marketing 52: 20–35. Heide, J.B. and G. John: 1990. “Alliances in Industrial Purchasing: The Determinants of Joint Action in Buyer-Supplier Relationships”, Journal of Marketing Research 27: 24–36.

232


Hitt, M.A. and B.B. Tyler: 1991, “Strategic Decision Models: Integrating Different Perspectives”, Strategic Management Journal 12: 327–351. Imrie, R. and J. Morris: 1992, “A Review of Recent changes in Buyer-Supplier Relations”, OMEGA International Journal of Management Science 20: 641–652. Joshi, A.W. and S.J. Arnold: 1997, “The Impact of Buyer Dependence on Buyer Opportunism in Buyer-Supplier Relationships: The Moderating Role of Relational Norms”, Psychology and Marketing 14: 823–845. Joskow, P.L.: 1985, “Vertical Integration and Long-Term Contracts: The Case of Coal-Burning Electric Generating Plants”, Journal of Law, Economics, and Organization 1: 33–80. Joskow, P.L.: 1988, “Asset Specificity and the Structure of Vertical Relationships: Empirical Evidence”, Journal of Law, Economics, and Organization 4: 95–117. Klein, B., R.G. Crawford and A.A. Alchian: 1978, “Vertical Integration, Appropriable Rents, and the Competitive Contracting Process”, Journal of Law and Economics 21: 297–326. Kozak, R.A. and D.H. Cohen: 1997, “Distributor-Supplier Partnering Relationships: A Case in Trust”, Journal of Business Research 39: 33–38. Lane, C., and R. Bachmann: 1996, “The Social Constitution of Trust: Supplier Relations in Britain and Germany”, Organization Studies 17: 365–395. Leuthesser, L.: 1997, “Supplier Relational Behavior: An Empirical Assessment’, Industrial Marketing Management 26: 245–254. Noorderhaven, N.G.: 1994, “Transaction Cost Analysis and the Explanation of Hybrid Vertical InterFirm Relations”, Review of Political Economy 6: 19–36. Noorderhaven, N.G.: 1995, “Transaction, Interaction, Institutionalization: Toward a Dynamic Theory of Hybrid Governance”, Scandinavian Journal of Management 11: 43–55. Noorderhaven, N.G.: 1996, “How to Make Transaction Cost Economics More Balanced and Realistic”, Academy of Management Review 21: 924–925. Nooteboom, B.: 1996, “Trust, Opportunism and Governance; A Process and Control Model”, Organization Studies 17: 985–1010. Norušis, M.J.: 1990, SPSS/PC+ Statistics 4.0 (Chicago, IL: SPSS Inc.). Semlinger, K.: 1991, Innovation, Cooperation and Strategic Contracting (Paris: International Colloquium on the Management of Technology). Sharma, A., and J.N. Sheth: 1997, “Supplier Relationships: Emerging Issues and Challenges”, Industrial Marketing Management 26: 91–100. Smith, P.M., E.S. Ross and T. Smith: 1997, “A Case Study of Distributor-Supplier Business Relationships”, Journal of Business Research 39: 39–44. Williamson, O.E.: 1985, The Economic Institutions of Capitalism; Firms, Markets, Relational Contracting (New York: Free Press). Address for correspondence: Niels G. Noorderhaven, Tilburg University, Faculty of Economics, P.O. Box 90153, 5000 LE Tilburg, The Netherlands Fax: 31-13-4662875; e-mail: [email protected]