Week-12

ServiceProvider-3.pdf

Home >Business & Finance homework help >Week-12

How Service Provider Dependence Perceptions Moderate the Power–Opportunism Relationship with

Professional Services

Sean M. Handley* Darla Moore School of Business, University of South Carolina, 1014 Greene Street, DMSB 401M, Columbia, South Carolina 29208, USA,

sean.handley@moore.sc.edu

Jurriaan de Jong School of Management, University at Buffalo, 326D Jacobs Management Center, Buffalo, New York 14260, USA, jurriaan@buffalo.edu

W. C. Benton Jr. Fisher College of Business The Ohio State University, 2100 Neil Avenue, 610 Fisher Hall, Columbus, Ohio 43210, USA,

benton.1@osu.edu

I n this study, we develop a novel theoretical model of how the relationship between the buyer’s use of power and the service provider’s opportunism is moderated by the provider’s perceptions of dependence advantage. Analyzing a dya-

dic dataset of 109 professional service relationships, we find that the extent to which the buyer’s use of mediated and non-mediated power aligns with the service provider’s perceptions of relative dependence is germane to the power–op- portunism relationship. The notion that firm A’s opportunism, in response to firm B’s use of power, is in part influenced by firm A’s perceptions of the relative dependence in the relationship, constitutes a significant theoretical contribution to the power-dependence literature. Our findings underline the importance of (i) understanding the relative dependence in an inter-organizational relationship, (ii) managing the other firm’s perception of this dependence, and (iii) ensuring that the use of power is aligned with the other firm’s perception of the relative dependence in the relationship.

Key words: dependence advantage; inter-organizational power; opportunism; professional services History: Received: February 2018; Accepted: January 2019 by Edward G. Anderson, after 2 revisions.

1. Introduction

Outsourcing has emerged as a prominent business strategy with firms resorting to the use of external ser- vice providers for a myriad of activities. Initially, firms reserved outsourcing for those activities that were non-core, routine, and transactional in nature (Disher et al. 2005, Gottfredson et al. 2005, PriceWaterhouseCoopers 2009). However, increas- ingly organizations are pushing the boundaries of outsourcing and are utilizing external service provi- ders for activities that involve specialized, knowl- edge-based capabilities (Gereffi and Fernandez-Stark 2010, Sako 2006). These outsourced professional services are characterized as being more intangible, cus- tomized, and requiring specialized skill sets of profes- sionals often licensed in their trade (Chase 1981, Lewis and Brown 2012, Von Nordenflycht 2010). The outsourcing of these professional services presents exacerbated management challenges when compared to the traditional outsourcing of more routine,

transactional activities. These challenges with out- sourced professional services raise theoretically and managerially interesting questions about inter-organi- zational power and opportunism, which this study begins to address. One elevated concern with outsourced professional

services is opportunism, or “self-interest seeking with guile” (Williamson 1975, p. 9). Professional services tend to exhibit more measurement difficulty or per- formance ambiguity (Bowen and Jones 1986, Mayer and Nickerson 2005) and involve expert professional workers who have a strong preference for autonomy (Greenwood and Empson 2003, Von Nordenflycht 2010). These characteristics make attempts at formal control and influence tenuous, and heighten concerns of opportunistic behavior (Bowen and Jones 1986, Mayer and Nickerson 2005). Indeed, the literature has repeatedly emphasized the prevalence of service pro- vider opportunism with outsourced technical profes- sional services (e.g., engineering, architectural, construction) such as those examined in this study

1692

(Chang and Ive 2007, Masten et al. 1991, Unsal and Taylor 2010). For example, through their detailed account of the Channel Tunnel project, Chang and Ive (2007) demonstrate that it is not uncommon for sup- pliers (providers) on complex technical service pro- jects to opportunistically “hold-up” the customer and threaten to delay work if favorable post-contract con- cessions are not made. These authors also highlight the limited ability of contracts to safeguard against this risk in part due to poor third party visibility into the relationship, observing “the nominal powers the contract gives the client to force the contractor to pro- gress works without delay regardless of change orders or disputes thereon . . .are ineffectual in practice [emphasis added], because of the poor line of visibil- ity of the adjudicator/judge” (Chang and Ive 2007, p. 397). An experienced buyer of technical professional services, who we interviewed at the outset of this study (see Appendix A), further suggested that per- formance visibility issues frequently arise due to the interdependency of the work of multiple service pro- viders (e.g., architects, general contractors, etc.) He observed that this task interdependency can make assigning responsibility for performance shortfalls, and therefore contract enforcement, challenging. A second heightened challenge in outsourcing pro-

fessional services is assessing each party’s (i.e., buyer and service provider) dependence on the relationship. Existing literature recognizes multiple factors (e.g., magnitude and concentration of exchange, availabil- ity of alternative partners, switching difficulty) that contribute toward a firm’s dependence on the inter- organizational relationship. As any of these factors become equivocal or subject to misinterpretation, the potential arises for the two exchange partners to per- ceive relative dependence differently. In the majority of the dependence-power literature, the less depen- dent and therefore structurally more powerful party was abundantly clear (e.g., dealer or franchise arrangements, exclusive distribution agreements, and supplier relationships in the automotive industry). In contrast, we argue that outsourcing relationships for professional services are rife with opportunities for dependence perception misalignment. Due to the cus- tomized nature of highly professional services and the specialized skill sets involved, professional service buyers and providers “often will have different infor- mation about the inputs (time and resources) neces- sary to produce the service or different abilities to evaluate the outputs” (Bowen and Jones 1986, p. 432). Our preliminary interviews with experienced buyers (i.e., clients or project owners) and providers of tech- nical professional services were consistent with this argument. These interviews revealed that the service providers often have a technical knowledge advan- tage over the buyer. This may place the buyer in an

ill-prepared position to accurately assess viable alter- native service providers and switching difficulties. Additionally, whereas the providers of technical pro- fessional services are regularly competing for busi- ness against other firms, a buyer may infrequently need to assess the market and basis for competition. Collectively, these conditions confer upon service providers an information advantage as it relates to the competitive structure of the supply market. More- over, professional services tend to be more intangible, knowledge-based, and labor intensive. Therefore, idiosyncratic investments in the relationship by either party are more likely to be in the form of human capi- tal rather than physical assets (Masten et al. 1991, Von Nordenflycht 2010). Such investments are likely to be more difficult to identify and value. To the extent that relationship-specific investments determine switching cost and thus dependency, this makes dependence perception differences more likely. So, why is this important? Transaction cost theory

links perceived dependence imbalance to the risk of opportunism (Handley and Angst 2015, Wathne and Heide 2000). Similarly, traditional power logic associ- ates a power source’s perception of relative depen- dence advantage with its use of power to influence and control counterparts in exchange relationships (Emerson 1962, Frazier and Rody 1991, Gulati and Sytch 2007). Therefore, ambiguity in assessing relative dependence and the misaligned perceptions of the relationship that may result should logically influence the relationship between power use and oppor- tunism. The potential dependence perception misalignment between two parties (that is likely to exist between professional service buyers and provi- ders) could lead to a situation where one party (i.e., the buyer or project owner here) attempts to exert influence and control in a way that is unexpected by the other party (i.e., the professional service provider here). With the current study, we contribute to the litera-

ture on the use and impact of power in inter-organiza- tional relationships. In this literature, the concepts of power and dependence are inextricably linked as “The power of A over B is equal to and based upon the dependence of B upon A” (Emerson 1962). Grounded in the seminal framework presented by French and Raven (1959), an organization can attempt to exert its influence through reliance on mediated and/or non-mediated power. The mediated and non- mediated power constructs are explicated in detail later in our theoretical background. Briefly, however, mediated power (i.e., coercive, reward, and legal legitimate) refers to the power source’s explicit use of extrinsic motivation to bring about some direct action on the part of the power target (Brown et al. 1995, Maloni and Benton 2000).1 Non-mediated power (i.e.,

referent and expert) is not explicitly exercised; rather influence is achieved due to the power target’s per- ception that the power source is an expert and the tar- get’s identification with the power source (i.e., pride in association) (Brown et al. 1995). A couple of prior studies examined the direct relationship between power use and opportunism (Handley and Benton 2012a, John 1984). However surprisingly, no previous work has empirically examined the influence of the power target’s dependence perceptions on this relation- ship between the power source’s use of power and the power target’s opportunism. In fact, we are una- ware of any prior study examining moderators of the power use-opportunism relationship. Prior work on the relation between dependence perceptions and power has largely focused on how the power source’s perceptions of dependence serve as an antecedent to power (e.g., Handley and Benton 2012b). Here, our interest lies with the power target’s perceptions of dependence and how these perceptions moderate the target’s reaction to the source’s power use. This dis- tinction between the power source’s and power tar- get’s perceptions of dependence is vital, especially in the professional services context where perception misalignment is plausible for the reasons outlined previously. While the power source’s perceptions influence the use of power, it is the power target’s perceptions that potentially influence the response or reaction to power use. We address this opportunity by investigating the following research question: How does the professional service provider’s perception of depen- dence advantage or disadvantage moderate the relationship between the professional service buyer’s use of power and the provider’s opportunism? In the following sections, we build a novel theoreti-

cal framework that addresses the role of a service pro- vider’s (i.e., power target’s) dependence perceptions in the buyer power-service provider opportunism relationship (see Figure 1). We empirically evaluate our theoretical model utilizing a dyadic dataset of 109 outsourced technical professional service relation- ships, including architectural, engineering, and gen- eral contracting services. These services represent prototypical examples of highly skilled and knowl- edge-based professional services (Schmenner 1986, Von Nordenflycht 2010). Specifically, after we analyze the direct relationship between buyer power exertion and professional service provider opportunism, we investigate how this relationship is moderated by the service provider’s perception of its dependence advantage or disadvantage. Our results depict a complex picture of the relation-

ships between power, dependence perceptions, and opportunism. We find that the buyer’s mediated power is associated with a higher risk of provider opportunism, whereas non-mediated power

corresponds to a lower risk of opportunism. These baseline findings complement prior evidence in the literature on the relationship between power use and opportunism, which had not been previously exam- ined in the unique context of technical professional services. However, as our main contribution to the lit- erature, we find that the service provider’s perception of dependence advantage or disadvantage signifi- cantly moderates these first-order relationships in some complex ways. First, as the service provider’s perception of either the buyer’s or the provider’s dependence advantage increases, the positive rela- tionship between buyer-mediated power and provi- der opportunism is exacerbated. However, when the provider perceives the firms to have balanced depen- dence (or close to it), the relationship between buyer- mediated power and provider opportunism is not sig- nificant. Second, as the service provider’s perception of the buyer’s dependence advantage increases, the negative relationship between buyer non-mediated power and provider opportunism is strengthened. On the contrary, the relationship between buyer non- mediated power and provider opportunism is not moderated by the provider’s perception of their own dependence advantage. In sum, we find that the rela- tionship between power, dependence, and oppor- tunism is much more nuanced than previously recognized by other studies. In doing so, we advance the inter-organizational (supply chain) dependence and power literature as well as scholarly work on pro- fessional service relationship management.

2. Theoretical Background

2.1. Inter-Organizational Power Inter-organizational power is commonly defined as “the ability of one individual or group to control or influence the behavior of another” (Emerson 1962, Hunt and Nevin 1974). Research on the use of inter- organizational power is most frequently grounded in a framework developed by French and Raven (1959). This framework includes five bases of power: reward, coercive, legal legitimate, referent and expert. Reward, coercive and legal legitimate power sources are typically utilized jointly (Frazier and Summers 1984). Legal legitimate power, based on the use of for- mal contracts, normally accompanies reward and

Service Provider Opportunism

Buyer Power

Service Provider Dependence Percep�on

Figure 1 Conceptual Model

coercive power, and these three power types are cate- gorized as mediated power (Brown et al. 1995, Maloni and Benton 2000). Mediated power involves the overt use of extrinsic motivation to influence a specific action (Benton and Maloni 2005, Brown et al. 1995, Frazier and Summers 1984), and is generally associ- ated with competitive and negative uses of power. Coercive and reward power are difficult to distin- guish conceptually, and can be considered “two sides of the same coin” (Handley and Benton 2012b, p. 255). For example, a reward may be given to a supplier for going along with a firm’s wishes or requirements. However, if a buyer threatens to withhold a reward from a supplier, it may be interpreted as punishment or coercion. Thus, in a number of empirical studies, reward and coercive power are grouped together along with legal legitimate to represent mediated power (Boyle and Dwyer 1995, Frazier and Rody 1991, Frazier and Summers 1986). The other two power bases, expert power and refer-

ent power, are classified as non-mediated power. Non- mediated power is not a form of explicit action but instead a cognitive state that develops over time with the power target (Frazier and Summers 1984, Maloni and Benton 2000). Stated differently, non-mediated power is not overtly exercised by the power source, but rather originates from the power target’s willing- ness to fulfill the power source’s requests (Brown et al. 1995, Frazier and Summers 1984, Maloni and Benton 2000). Referent power is where one firm desires identification with another firm for recogni- tion by association (Frazier and Summers 1984, Mal- oni and Benton 2000). Expert power implies that one firm holds information or expertise that is valued by another firm. Similar to the three mediated power bases, referent and expert power often times occur congruently (Kasulis and Spekman 1980). It has been shown that extrinsic attempts at influ-

ence and control (i.e., mediated power) are negatively associated with relationalism (Brown et al. 1995, Fra- zier and Summers 1984, 1986, Handley and Benton 2012b, John 1984, Maloni and Benton 2000, Skinner et al. 1992). While mediated power can gain the com- pliance of another party in the short term, it is appar- ent that in the long term the effects of mediated power are damaging to relationship commitment. In contrast, empirical evidence exists for a positive asso- ciation between non-mediated power and relational factors such as relationship satisfaction, commitment, cooperation, conflict resolution and trust (Benton and Maloni 2005, Brown et al. 1995, Frazier and Rody 1991, Maloni and Benton 2000, Spekman 1979).

2.2. Opportunism Along with bounded rationality, opportunism is one of the core behavioral assumptions underlying

transaction cost theory (Williamson 1985). Oppor- tunism has been defined as “self-interest seeking with guile” (Williamson 1975). Opportunistic actions are deliberately self-serving, and can include distortion of information and reneging on commitments (Jap and Anderson 2003). Exchange relationships at a high risk of opportunism require elevated expenditures on for- mal control mechanisms (e.g., contracts and monitor- ing devices), which threaten the financial viability of a market form of governance (Parkhe 1993, Williamson 1991). For these reasons, developing a deeper under- standing of the conditions and management practices which exacerbate or attenuate the risk of opportunism is practically and theoretically important. As previously described, the risk of provider

opportunism is particularly salient with outsourced technical professional services (e.g., engineering, architectural, construction). In our exploratory inter- views with four experienced practitioners (see Appendix A), several specific examples of how opportunism can manifest in this context were pro- vided. These include: (i) service providers covertly assign inexperienced technical staff while billing the customer at a higher rate corresponding to more experienced technical team members; (ii) service pro- viders increase their margin by substituting inferior materials, while charging the buyer for higher quality materials; (iii) service providers can “shave scope” by skipping certain agreed upon steps in the project that they view as redundant or unnecessary; and (iv) ser- vice providers try to exaggerate the progress made on a project, thereby inflating the required progress pay- ments to be made by the buyer. Closely related to the current study, only two prior

studies examined how the use of power influences the power target’s level of opportunism. John (1984) evaluated power use and opportunism in the oil dis- tribution channel, while Handley and Benton (2012a) considered the effect of both exchange hazards and power use on provider opportunism in business pro- cess outsourcing relationships. Both studies found that the aforementioned negative effects of mediated power on the relationship manifest in the power tar- get acting more opportunistically, while influence and control achieved through non-mediated means results in less opportunism. With the current study, we advance this existing literature on the relationship between power and opportunism in two important ways. First, we examine the direct relationship between power and opportunism in the unique con- text of technical professional services. In doing so, we extend this literature into contexts where it is more difficult for managers to unambiguously assess rela- tive dependence, and hence relative vulnerability. In a managerial context such as this, knowledge asym- metries regarding viable alternative service providers

and switching difficulties can lead to one firm attempting to use power in a manner that is misa- ligned with the other firm’s perception of the relation- ship. As outlined in the introduction, this may be particularly problematic with professional services where elevated performance measurement difficulty and resistance to formal control can increase the risk of opportunism. Testing the power–opportunism relationship in a context with idiosyncratic manage- ment challenges contributes to a better understanding of the generalizability of existing results in the litera- ture (Bettis et al. 2016). Second, and perhaps more importantly, we advance existing literature by consid- ering a potentially salient moderator of the power– opportunism relationship—the power target’s per- ception of relative dependence in the inter-firm rela- tionship. To our knowledge, no prior study examines contingencies of the power–opportunism relation- ship. Thus, existing literature paints the universalistic picture that mediated power is always associated with a higher risk of opportunism and non-mediated power is always associated with a lower risk of opportunism. We extend this stream of literature by advancing and testing the argument that these first-order relationships are conditional on the power target’s perception of rel- ative firm dependence. By examining contingencies, we can provide more nuanced prescriptions for the management of inter-firm relationships than can be offered by prior studies.

2.3. Inter-Organizational Dependence According to the resource dependence theory, interde- pendence between actors “exists whenever one actor does not entirely control all of the conditions necessary for the achievement of an action or for obtaining the outcome desired from the action” (Pfeffer and Salancik 1978, p. 40). Dependence can be operationalized using the concepts of essentiality and substitutability (Jacobs 1974). Essentiality, the importance of the exchanged resource, can be captured by the magnitude of the exchange (El-Ansary and Stern 1972, Pfeffer and Salan- cik 1978, Pugh et al. 1969) and by the concentration of the exchange, or the number of exchange partners (Burt 1982, Kumar et al. 1998). The concept of substi- tutability can be captured by the availability of alterna- tive exchange partners (Brass 1984, El-Ansary and Stern 1972, Kumar et al. 1998), the relative ease of switching exchange partners, or the magnitude of switching costs or relationship-specific investments (Buchanan 1992, Mudambi and Helper 1998). Much of the scholarly research linking dependence

to power traces its lineage back to Emerson’s (1962) seminal work. According to Emerson (1962), if an actor’s net dependence was negative, then the actor was believed to have a dependence advantage and thus be in a position of relative power. Traditional

power logic, and subsequent empirical work that draws upon it, posits that in a situation of dependence asymmetry, and of resulting power disparity, adver- sarial action is more likely (Frazier and Rody 1991, Geyskens et al. 1996, Gulati and Sytch 2007). Specifi- cally, this literature argues that dependence advan- taged exchange partners are more likely to use coercive and punitive actions (i.e., mediated power) to control and influence their relatively more depen- dent counterparts (e.g., Blau 1964, Gundlach and Cadotte 1994, Kumar et al. 1998).2 Alternatively, tra- ditional power logic holds that as a party to an exchange becomes more dependent on its counter- part, it is more likely to rely on non-mediated means of control and influence (e.g., Brown et al. 1995, Han- dley and Benton 2012b). An important consideration absent from this litera-

ture is the possibility that the two parties to the inter- organizational exchange may perceive the relation- ship differently. Increasingly, the supply chain litera- ture more broadly is recognizing that buyers and providers (i.e., suppliers) may differ in terms of how they perceive the relationship (Roh et al. 2013). For example, Carter (2000) studies perception differences regarding unethical behavior in buyer–supplier rela- tionships. Ellram and Hendrick (1995) observe that buyers and suppliers perceive risk sharing differently. More directly related to the current study, Svensson (2004) examines how buyers and suppliers may have different views of dependency. For the reasons out- lined previously, we believe that this is particularly true in the context of outsourced professional ser- vices. Thus, while the existing dependency-power lit- erature treats relative dependence as an antecedent to power use, it does not consider that the power target may view the balance of dependence differently. In considering the power target’s perceptions of depen- dence, we argue that it is more likely to moderate how the power target responds to the power source’s use of power rather than serve as an antecedent to the source’s use of power. We can think of no direct rea- son for why the power target’s perception of the rela- tionship would be an antecedent to the power source’s approach to influence and control. Instead, it seems more logical that the power target’s perception shapes how it reacts or responds to the power source’s approach to control and influence. Applying this reasoning to the context of outsourced profes- sional services, our overarching thesis is that while the buyer’s perceptions of dependence may influence its use of power, the provider’s dependence percep- tions will moderate how it reacts or responds to the buyer’s approach to influence and control. Since the power target’s dependence perceptions are the con- sideration missing from existing literature, it’s role as a moderator of the power–opportunism relationship

is the focus of our theoretical model. However, in our analysis, we do control for factors perceived by the power source (i.e., buyer) and shown by prior research to influence the use of power.

3. Hypothesis Development

Our primary theoretical contribution lies in examin- ing the manner by which the service provider’s per- ception of dependence advantage or disadvantage moderates the relationship between the buyer’s use of power and the provider’s opportunism. However, we first articulate our expectations for how the buyer’s mediated and non-mediated power directly relate to the service provider’s opportunism. Evaluating these baseline relationships allows us to establish a clear connection with prior literature and to develop a solid foundation on which we can interpret the more intri- cate moderation effects. To our knowledge, no prior study has evaluated these relationships in the context of technical professional services outsourcing. As detailed previously, outsourced professional services present unique managerial and operational chal- lenges relative to more routine, transactional services. Therefore, by testing the first-order relationships between power and opportunism we can also comple- ment the limited evidence that currently exists in the literature. Our theoretical model is graphically repre- sented in Figure 2.3 In the sections that follow, we articulate the theory underlying each hypothesized relationship reflected by the model.

3.1. Power and Opportunism As discussed earlier, significant empirical evidence exists for the power source’s use of mediated power having a negative impact on the power target’s assessment of relationship quality and performance (Benton and Maloni 2005, Boyle et al. 1992, Brown

et al. 1995). These negative effects of mediated power on the power target’s assessment of the relationship have been found to manifest in the target acting more opportunistically (Handley and Benton 2012a, John 1984). Mediated power involves the use of overt influ- ence and extrinsic forms of motivation, which damage the power target’s trust in and commitment to the power source (Benton and Maloni 2005, Frazier and Summers 1986). Work in behavioral economics sug- gests that extrinsic motivation “crowds out” the intrin- sic motivation to cooperate (Falk and Kosfeld 2006, Kessler and Leider 2012). As perceptions of oppor- tunism are negatively associated with a history of cooperation (Das and Teng 1998, Parkhe 1993) and col- laboration (Helper et al. 2000), one can expect damage to cooperation to correspond with an increased risk of opportunism. Net, these arguments indicate that the buyer’s use of mediated power should correspond to a higher risk of provider opportunism.

HYPOTHESIS 1. The buyer’s reliance on mediated power is associated with an increase in the provider’s opportunism.

While the extant literature negatively associates mediated power with the power target’s assessment of relationship quality, this literature presents equally compelling evidence that non-mediated power is pos- itively associated with the power target’s assessment of relationship quality. A power source relying on non-mediated power is able to control and influence its counterpart through the target’s view that the source is an expert and through the target organiza- tion identifying with the power source (Brown et al. 1995, Kelman 1958). Influence achieved in this man- ner has been related to stronger relational norms, commitment, and power target satisfaction with the relationship (Brown et al. 1995, Frazier and Rody

Provider opportunism

Buyer Non-mediated power

Provider-perceived buyer dependence advantage

Buyer Mediated power

Provider-perceived provider dependence

advantage

H1 (+)

H2 (-)

H3 (-)

H4 (-)

H5 (+)

H6 (+)

Figure 2 Hypothesized Relationships

1991, John 1984, Maloni and Benton 2000). As rela- tional norms and satisfaction strengthen, the risk of opportunism is expected to subside (Handley and Benton 2012a, Helper et al. 2000). In total, these argu- ments indicate that the buyer’s non-mediated power should be related to a lower level of provider oppor- tunism. Therefore, we hypothesize:

HYPOTHESIS 2. The buyer’s reliance on non-mediated power is associated with a decrease in the provider’s opportunism.

3.2. The Moderating Influence of the Service Provider’s Dependence Perceptions With our first two hypotheses, we posited that buyer- mediated power corresponds with a higher level of service provider opportunism, whereas buyer non- mediated power is associated with a lower level of service provider opportunism. We now turn our attention to the manner by which we expect the ser- vice provider’s perceptions of relative dependence to moderate these first-order effects. Our moderation hypotheses represent four power use-dependence perception scenarios: (i) the buyer relies on mediated power when the service provider perceives the buyer to have a dependence advantage, (ii) the buyer relies on non-mediated power when the service provider perceives the buyer to have a dependence advantage, (iii) the buyer relies on mediated power when the ser- vice provider perceives that it (the service provider) has a dependence advantage, and (iv) the buyer relies on non-mediated power when the service provider perceives that it (the service provider) has a depen- dence advantage.

3.2.1. Service Provider Perception of Buyer Dependence Advantage. When the service provider perceives that it has a dependence disadvantage and thus the buyer to have an advantage, we would expect the positive relationship between the buyer’s use of mediated power and the service provider’s opportunism to be weakened. A service provider that perceives it is in a dependence disadvantaged posi- tion due to relationship-specific investments, the buyer representing a large share of the provider’s business, the buyer having ample alternative service providers to utilize, etc., would have a heightened level of concern that if it were to act opportunistically, the buyer could relatively easily replace them with an alternate service provider (Brown et al. 2000). It is more likely in this situation that the service provider will seek to resolve differences with the buyer through non-adversarial means (Kumar et al. 1998). Therefore, while we would still expect the service provider to be dissatisfied with the buyer’s reliance

on mediated power, their vulnerability puts them in position of having to tolerate it rather than responding with self-interest seeking behavior (i.e., opportunism) (Gundlach and Cadotte 1994, Stump and Heide 1996). In other words, the anticipated relationship between mediated power and opportunism previously outlined would be suppressed due to the service provider’s fear of buyer retaliation up to and including relationship termination. Indeed, this logic is consistent with trans- action cost literature associating the risk of oppor- tunism with vulnerability in a relationship (Wathne and Heide 2000). Therefore, we posit:

HYPOTHESIS 3. When the provider perceives that the buyer has a dependence advantage, the positive associa- tion between the buyer’s mediated power and the provi- der’s opportunism is weakened.

The next hypothesis reflects the second scenario described above whereby the service provider per- ceives the buyer to have a dependence advantage and the buyer relies on non-mediated power. From the provider’s perspective, the buyer has a dependence advantage and does not have to rely on non-mediated means of influence but elects to do so anyway. This implicit expression of empathy and concern for the provider’s satisfaction is received favorably by the provider. Moreover, the buyer’s reliance on non- mediated power serves as a signal to the service provider that the structurally powerful buyer is not simply interested in getting its way with short-term issues but is interested in developing a longer-term relationship with the service provider (Benton and Maloni 2005, Brown et al. 1995). In such a situation, it stands to reason that it is even more unlikely that the service provider would jeopardize the payoffs from follow-on opportunities with the buyer by acting opportunistically on the current project (Parkhe 1993, Poppo et al. 2008). Previously, we expressed our expectation, based on prior research, that non- mediated power and opportunism are negatively related. As a result of the factors detailed here, we hypothesize that the provider’s reaction to the buyer’s non-mediated power will be stronger, resulting in even lower opportunism by the provider. Formally:

HYPOTHESIS 4. When the provider perceives that the buyer has a dependence advantage, the negative associa- tion between the buyer’s non-mediated power and the provider’s opportunism is strengthened.

3.2.2. Service Provider Perception of Service Provider Dependence Advantage. The next two hypotheses reflect scenarios where the service provi- der perceives that it has a dependence advantage.

With the fifth hypothesis, we consider the scenario where the service provider perceives that it has a dependence advantage and the buyer uses mediated power to influence the provider. When the provider perceives that it has a dependence advantage and thus that the buyer is the more vulnerable party to the relationship, its opportunism is not as restrained by fear of retaliation from the buyer (Barney 2002, Wathne and Heide 2000). Even if the service provider is not initially inclined to be opportunistic, the buyer’s use of mediated power frames the relation- ship in a short-term, adversarial mindset (Brown et al. 1995). This relationship framing, instigated by the buyer’s use of mediated power, may provoke the service provider to resort to short-term, self-serving actions as well. Stated differently, the buyer’s unex- pected reliance on mediated power may prompt the service provider to downwardly revise its expected payoff from future interaction and undermine its own motivation to behave cooperatively (Parkhe 1993). This, coupled with the service provider’s perception that the buyer is more reliant on the relationship than it (the provider) is, should strengthen the positive association between mediated power and oppor- tunism described previously. Thus, we have the following:

HYPOTHESIS 5. When the provider perceives that it (the provider) has a dependence advantage, the positive association between the buyer’s mediated power and the provider’s opportunism is strengthened.

The final scenario we consider is where the service provider perceives that it has a dependence advan- tage and the buyer relies on non-mediated power to influence the provider. With non-mediated power, influence is achieved through reliance on expertise and the power target’s pride in association with the power source (Brown et al. 1995, Handley and Benton 2012a). While the service provider may view these approaches in a positive light, their impact in terms of altering the provider’s behavior is anticipated to be more muted when the service provider perceives it has a dependence advantage. This could, for example, be due to the provider believing that it could easily replace the buyer’s business and/or perceiving that the buyer would have great difficulty in switching to another service provider. In this situation, the service provider perceives that it is less reliant on the rela- tionship than is the buyer. Therefore, the relational impact of non-mediated power in this context may not be sufficiently strong to deter the opportunistic tendencies of a provider that believes it can exploit the buyer’s vulnerability. For example, the provider may be attracted to how the buyer operates and con- ducts its inter-organizational relationships, but it does

not view the economic disincentives to opportunism (e.g., due to follow-on opportunities) to be terribly compelling (Parkhe 1993). For this reason, we argue that the provider’s favorable reaction to the buyer’s non-mediated power is likely to be tempered, result- ing in an attenuated negative impact on target opportunism.

HYPOTHESIS 6. When the provider perceives that it has a dependence advantage, the negative association between the buyer’s non-mediated power and the provider’s oppor- tunism is weakened.

4. Research Methods

4.1. Data Collection and Sample The empirical context for this study is the outsourcing of technical professional services. The hypotheses were tested using matched dyadic data collected from 109 pairs of buyers4 and providers of technical profes- sional services such as engineering, architectural, gen- eral contracting, etc. These types of services were chosen as our focus because they align very well with characteristics of professional services commonly noted in the extant literature (e.g., require specialized and knowledge-based skills to make expert judge- ments, are frequently customized and require exten- sive buyer–provider interaction, and involve professionals often licensed in their trade). Thus, these are excellent examples of professional services. Due to the challenging dyadic nature of our research

design, a rigorous data collection protocol was fol- lowed. First, we conducted preliminary interviews (see Appendix A) with multiple buyers and providers of technical professional services to deepen our knowl- edge of the business context, obtain feedback on our questionnaire, and identify the most appropriate pro- fessional contacts for the data collection. These expert interviewees indicated that individuals with titles such as facility manager, facility engineer, and so forth would be directly involved in managing engineering, architectural, and general contracting projects involv- ing external technical service providers. These same experts directed us to the International Facility Man- agement Association. By joining this association, we were able to compile the preliminary contact list for the “buyers” of technical professional services. In total, we obtained contact information for 8661 potential buyers. However, to be eligible for participation in the study, potential respondents were informed that they had to have been directly involved in the management of a technical service project involving an external ser- vice provider5 and that these projects: (i) had to have been completed within the past 24 months, (ii) had to have been located and executed in the United States,

and (iii) had a project budget exceeding $75,000. The buyer’s questionnaire contained a battery of questions related to a specific technical service project and the firm’s relationship with the major external service pro- vider on the project. An additional follow-up survey was conducted with a subsample of non-respondents for two purposes: (i) to ascertain the proportion of ini- tial contacts that were actually qualified for the study based upon our stringent criteria for participation, and (ii) to collect some basic demographic information (i.e., firm size and industry) to compare respondents to non-respondents. Of the 271 respondents to this non- respondent survey, 41% indicated that they did not participate because they were unqualified given our stated requirements. Accounting for this, we estimate the effective number of potential (qualified) buyer respondents to be 5110. The unit of analysis for our study is the matched

dyad. In addition to the research design calling for data from buyers and service providers, our prelimi- nary interviews suggested that our phenomena of interest (i.e., the use of power and opportunism) are sensitive topics on which many professionals would be reluctant to report. Furthermore, requests for con- tact information for relationship counterparts are observed to discourage potential respondents from participating (Handley and Angst 2015, Schilke and Cook 2015). In fact, Carson (2007) abandoned a matched dyadic design due to this challenge. In the face of these obstacles, we are very pleased with obtaining 400 responses (7.8%) from buy-side experts. Comparing respondents to non-respondents in terms of firm size (p-value of t-test = 0.28) and industry rep- resentation (p-value of v2 test = 0.45) did not reveal statistical differences.6 Moreover, our buyer respon- dents reported that they have been in their current or a similar role for 14 years on average; making them extremely well-qualified key informants. The buyer survey also requested that the respondent

provide the research team with the name and contact information for their key contact at the technical service provider. Primarily due to concerns of confidentiality and company policy, of the 400 completed buyer sur- veys, only 239 provided service provider contact infor- mation. Subsequently, the research team sent these 239 service providers a similar online questionnaire asking them to report on the same technical service project and on their own perspective of the relationship with the buyer’s organization. After multiple rounds of con- tact, we received 109 service provider surveys (45.6%) that could be matched-up with a corresponding buyer survey and form a complete dyad. Using the provider firm names and contact information provided in the buyer surveys, we obtained data from Dun & Brad- street to compare respondents to non-respondents for the provider survey and evaluate whether or not there

was a concern of a significant response bias. The differ- ence tests for firm size (p-value of t-test = 0.44) and industry representation (p-value of v2 test = 0.29) again did not reveal statistical differences. Additionally, pro- vider-side informants indicated that on average they have been in their current or a similar role for approxi- mately 13 years. Collectively, this leaves us confident about the representativeness of our sample of provider responses and the qualifications of the provider key informants. Table 1 provides an overview of the dyadic sample used to conduct our analysis.

4.2. Measurement The measurement for all variables is now described. The specific items used to measure the multi-item constructs along with their source of measurement (i.e., either the buyer or provider survey) are pre- sented in Appendix B. All other items are sufficiently described within the text below or in Table 1. The dependent variable, provider opportunism, is based on a previously validated (Carson et al. 2006, John 1984) eight-item scale. Our measures for the buyer’s reli- ance on mediated power (reflecting Reward, Coercive, and Legal legitimate power bases) and non-mediated power (reflecting Referent and Expert power bases) are also multi-item scales. The measurement for each power construct is based on previously established scales used in multiple studies (e.g., Benton and Mal- oni 2005, Gaski 1986, Handley and Benton 2012a). Construction of the buyer dependence advantage and

provider dependence advantage constructs was a multi- ple step process modeled after that used by Kumar et al. (1998) and Gulati and Sytch (2007). First, we developed composite measures of the buyer’s depen- dence and provider’s dependence. The specific items used to assess each party’s level of dependence (see Appendix B) were adaptations of items used in prior studies of inter-organizational dependence (Frazier and Rody 1991, Gulati and Sytch 2007, Kumar et al. 1998) to represent the concentration of the exchange, the availability of alternatives, and the level of rela- tionship-specific investments. Several prior studies have modeled inter-organizational dependence as a formative rather than a reflective construct (e.g., El- Ansary and Stern 1972, Frazier et al. 1989, Howell 1987, Kumar et al. 1998), and we follow this prece- dent. In deciding whether to model a construct as reflective or formative, it is advised that researchers ask two related questions (Bollen 2011, Bollen and Lennox 1991, Diamantopoulos and Winklhofer 2001, MacCallum and Browne 1993): (i) Do the indicators (i.e., items) represent a unidimensional concept? and (ii) Do the indicators cause the composite construct or are they effects of the underlying construct? If the con- struct is not conceptually unidimensional and/or the indicators cause the composite construct (rather than

being caused by it), then modeling the construct as formative is appropriate. The studies listed above argue, and we agree, that dependence is not a concep- tually unidimensional construct but rather is multi- dimensional. The different items making up the dependence measures each contribute toward the overall level of dependence but are not necessarily required to inter-correlate with one another (Howell 1987, Kumar et al. 1998). Further, since the individual items are causal indicators of dependence (i.e., they determine the level of dependence) as opposed to being caused by (an effect of) an underlying unidi- mensional construct, they are appropriately conceptu- alized as formative (Bollen and Lennox 1991, Diamantopoulos and Winklhofer 2001). Thus, consis- tent with prior research on inter-firm dependence, we model buyer and provider dependence as formative constructs. Since buyer dependence (BD) is measured using five

items and provider dependence (PD) is measured using three items, the composite scores for each repre- sent the average value across the items making up BD and PD, respectively. This facilitates the interpretation of the difference between BD and PD as described below. Consistent with prior measures of dependence (Frazier and Rody 1991, Kumar et al. 1998), before averaging the individual items to form the overall dependence measure for each party, all items with a percentage response were transformed onto a 1-to-7 scale for uniformity.7 Using these composite scores for BD and PD, we utilized a spline specification to create

separate measures for buyer dependence advantage (BDA) and provider dependence advantage (PDA) (John- ston 1984). Using separate measures for each party’s dependence advantage is advantageous over a single measure with each firm’s advantage representing opposite extremes in that it allows for the investigation of effects that “go beyond the diametrically opposite hypothesized effects and that would not be uncovered with a single variable approach” (Gulati and Sytch 2007, p. 46). Buyer dependence advantage is specifically calculated by: PD-BD, if PD > BD and 0 otherwise. Likewise, provider dependence advantage is calculated by: BD-PD, if BD > PD and 0 otherwise. In addition to these variables used to test our

research hypotheses, we include several relevant control variables. Incorporating a robust set of con- trols to account for factors previously identified in the literature as influencing inter-firm power, oppor- tunism, or both is imperative to minimize the threat of omitted variable bias. First, we include buying firm size and provider firm size (both in terms of annual- ized revenue in US$) to account for the size of the firms. The former is measured from the buyer survey and the latter from the provider survey. To proxy for the magnitude and complexity of the technical ser- vice project, we include controls for project size (in US$) and project duration (in months); both are assessed by the buyer. These factors are also indica- tors of the strategic importance of the service, which prior research has related to the use of power (Hand- ley and Benton 2012b). Previous studies of inter-firm

Table 1 Sample Overview

Type of services rendered (check all that apply) Customer firm size (annual revenue) Project budget (US$)

Engineering services 30% Less than $1.0 million 4% Between $100k and $249.9k 15% Architectural services 40% Between $1.0 and $24.9 million 20% Between $250k and $499.9k 15% Contracting services 47% Between $25.0 and $49.9 million 6% Between $500k and $749.9k 7% Other 11% Between $50.0 and $99.9 million 13% Between $750k and $999.9k 6%

Between $100 and $499.9 million 24% $1 million or more 57% Between $500 and $999.9 million 7% $1.0 billion or more 24%

Customer firm industry Provider firm size (annual revenue) Project duration (months)

Utilities 2% Less than $1.0 million 8% Less than 6 months 26% Construction 2% Between $1.0 and $24.9 million 46% 6–12 months 45% Manufacturing 9% Between $25.0 and $49.9 million 10% 13–18 months 13% Retail trade 3% Between $50.0 and $99.9 million 11% 19–24 months 8% Transportation and warehousing 3% Between $100 and $499.9 million 13% More than 24 months 8% Information (e.g., Publishing, software, etc.) 3% Between $500 and $999.9 million 3% Finance and insurance 12% $1.0 billion or more 10% Real estate and rental and leasing 7% Professional, scientific, and technical services 7% Education 20% Health care and social assistance 11% Arts, entertainment, and recreation 3% Public administration 12% Other 9%

power and the transaction cost literature have also recognized that formal control difficulties and buyer switching difficulty (Handley and Benton 2012b), as well as contractual compensation structure (Provan and Gassenheimer 1994, Skinner et al. 1987), relate to both power and opportunism. To isolate these poten- tially confounding effects, we include three addi- tional control variables measured from the buyer’s perspective. The variable measurement difficulty is a four-item scale reflecting how difficult it is for the buyer to accurately evaluate the service provider’s performance and work procedures (see Appendix B for specific items). Measurement difficulty serves as an indicator of the efficacy of formal control mecha- nisms such as contracts and performance monitor- ing. A single-item measure of buyer switching difficulty captures the degree to which the buyer feels locked-in to using a particular provider: “It would have required much trouble and expense for our firm to switch service providers for this service.” To control for the type of compensation structure speci- fied in the contract, buyers were asked whether their contract with the service provider was firm fixed price, pure cost plus, or a negotiated hybrid compen- sation system (Bajari and Tadelis 2001). Two dummy variables are included to represent firm fixed price contracts and pure cost plus contracts. Negotiated hybrid contracts represent the baseline (i.e., scored as zeros on both contract type dummy variables). Next, we include previous projects with provider (num- ber of projects) from the buyer survey to control for prior relationship history. Finally, to account for the overall level of relational embeddedness (as opposed to each party’s advantage), we control for joint depen- dence (Gulati and Sytch 2007). Joint dependence is sim- ply a summation of the provider’s assessment of BD and PD as described previously. Three additional variables are utilized solely as

instrumental variables to evaluate the potential endo- geneity of mediated and non-mediated power (de- tailed in section 5). Technical skill similarity represents the extent to which the buying firm’s project manager had skills that overlapped with those of the profes- sional service provider, and is measured by a single- item from the provider survey: �Buyer’s≫ project manager had the same training and technical back- ground as our people on the project (1 = strongly disagree; 7 = strongly agree). Project volatility is a three-item scale, again measured from the provider survey, indicating the frequency of changes during the project to the project’s schedule, budget, and scope (1 = never; 7 = all the time) (a = 0.85). Third party management is a dummy variable captured from the buyer survey reflecting whether or not the buying firm engaged an independent third party to oversee project management activities (i.e., management of

key performance indicators, change requests, etc.) The descriptive statistics and correlations for all vari- ables are presented in Table 2.

4.3. Measurement Model Assessment The validity and reliability of the multi-item con- structs used in the main analysis was evaluated with a confirmatory factor analysis (CFA). The CFA included provider opportunism, the power constructs, and measurement difficulty. Consistent with prior research, Mediated power was modeled as a second- order factor reflecting Reward, Coercive, and Legal legitimate power bases. Non-mediated power was speci- fied to be a second-order factor reflecting Referent and Expert power bases. Although multi-item in nat- ure, the buyer and provider dependence measures were not included in the CFA due to them being for- mative composite constructs. It is not a requirement that formative composite constructs be conceptually unidimensional. That is, “composite indicators might be combined into a composite as a way to conve- niently summarize the effect of several variables that do not tap the same concept although they may share a similar ‘theme’.” (Bollen 2011, p. 366). With formative constructs, the individual items are thus not expected to inter-correlate strongly with one another and therefore assessing internal consistency is not appropriate (Bollen 2011, Bollen and Lennox 1991, Diamantopoulos and Winklhofer 2001). The validity and reliability of the measurement model (including second-order factors) are supported by the CFA results. First, strong overall model fit for the CFA is exhibited by multiple metrics: RMSEA = 0.059 (90% CI: 0.045; 0.072); v2/df = 1.38; SRMR = 0.073; CFI = 0.94; TLI = 0.93. The conver- gent validity for each construct is evidenced by each manifest item loading significantly (all loadings >0.50; p < 0.001) on its specified first-order factor (see Table 3 for details). To evaluate discriminant validity, a series of v2 tests comparing constrained and unconstrained models was conducted. Each of the constrained models specifies a single inter-factor correlation to be one, while all inter-factor correla- tions are free in the unconstrained model. The v2 test for each model comparison was highly significant (p < 0.001), offering a strong indication of discrimi- nant validity. The constructs included in the CFA also demonstrate strong reliability with all Cron- bach’s alpha values and all composite reliabilities exceeding 0.71. Specifying mediated and non-mediated power as sec-

ond-order factors reflecting their respective power bases is validated by each power base loading signif- icantly on its intended second-order factor. Addi- tionally, we conducted a separate exploratory factor analysis with the five power bases as indicators. This

T a b le

2 D e sc ri p ti ve

S ta ti st ic s a n d C o rr e la ti o n s

M ea n

S D

M in

M ax

C o rr el at io n s

[1 ]

[2 ]

[3 ]

[4 ]

[5 ]

[6 ]

[7 ]

[8 ]

[9 ]

[1 0 ]

[1 1 ]

[1 2 ]

[1 3 ]

[1 4 ]

[1 5 ]

[1 6 ]

[1 7 ]

[1 8 ]

[1 ]

P ro vi d er

o p p o rt u n is m

(B )

1 .7 7

0 .7 7

1 .0 0

6 .0 0

1 .0 0

[2 ]

B u yi n g

fi rm

si ze

(B )

4 .5 3

1 .8 8

1 .0 0

7 .0 0

0 .0 2

1 .0 0

[3 ]

P ro vi d er

fi rm

si ze

(P )

3 .2 3

1 .7 8

1 .0 0

7 .0 0

0 .0 4

0 .1 2

1 .0 0

[4 ]

P ro je ct

si ze

(b u d g et ) (B )

3 .7 2

1 .6 1

1 .0 0

5 .0 0

�0 .0 3

0 .0 5

0 .2 4 *

1 .0 0

[5 ]

P ro je ct

d u ra ti o n

(m o n th s)

(B )

1 1 .7 1

9 .1 4

1 .0 0

3 7 .0 0

0 .0 6

0 .1 2

0 .3 0 *

0 .4 2 *

1 .0 0

[6 ]

P re vi o u s p ro je ct s

w it h p ro vi d er

(B )

6 .4 1

7 .5 4

1 .0 0

3 1 .0 0

0 .0 4

0 .3 3 *

0 .0 5

0 .0 9

�0 .0 4

1 .0 0

[7 ]

M ea su re m en t

d if fi cu lt y (B )

2 .4 3

0 .8 2

1 .0 0

5 .0 0

0 .3 4 *

�0 .0 3

0 .0 3

0 .0 9

�0 .0 8

1 .0 0

[8 ]

B u ye r sw

it ch in g

d if fi cu lt y (B )

4 .6 8

1 .7 2

1 .0 0

7 .0 0

�0 .0 1

0 .0 5

0 .0 4

�0 .0 3

0 .1 6

�0 .1 6

0 .1 1

1 .0 0

[9 ]

Fi rm

fi xe d p ri ce

co n tr ac t (0 ,1 ) (B )

0 .4 6

0 .5 0

0 .0 0

1 .0 0

�0 .1 0

�0 .2 0 *

0 .0 0

�0 .0 9

0 .0 9

�0 .2 9 *

�0 .0 9

0 .0 4

1 .0 0

[1 0 ]

C o st

p lu s co n tr ac t

(0 ,1 ) (B )

0 .1 9

0 .4 0

0 .0 0

1 .0 0

0 .0 3

0 .1 5

0 .0 3

0 .0 7

0 .0 4

�0 .0 8

0 .1 4

0 .0 0

�0 .4 5 *

1 .0 0

[1 1 ]

Jo in t d ep en d en ce

(P )

7 .2 8

1 .2 0

4 .1 8

9 .9 1

0 .0 5

0 .1 7

0 .1 2

�0 .0 1

0 .2 6 *

0 .0 1

�0 .0 3

0 .0 4

1 .0 0

[1 2 ]

B u ye r’ s d ep en d en ce

ad va n ta g e (B D A ) (P )

0 .1 9

0 .3 3

0 .0 0

1 .5 3

0 .0 4

0 .1 2

�0 .1 7

0 .1 5

0 .0 5

0 .2 2 *

�0 .0 8

0 .0 6

�0 .1 0

0 .0 0

�0 .0 4

1 .0 0

[1 3 ]

P ro vi d er ’s

d ep en d en ce

ad va n ta g e (P D A ) (P )

0 .6 5

0 .7 7

0 .0 0

3 .2 8

�0 .0 4

0 .0 4

0 .1 1

�0 .3 0 *

�0 .1 4

�0 .2 2 *

0 .0 1

0 .0 8

0 .0 4

0 .0 2

0 .0 7

�0 .4 8 *

1 .0 0

[1 4 ]

B u ye r’ s m ed ia te d

p o w er

(B M P ) (P )

2 .4 9

1 .0 1

1 .0 0

5 .0 0

0 .2 5 *

0 .1 3

0 .1 0

0 .0 8

0 .1 1

�0 .0 3

0 .0 9

0 .1 4

�0 .1 1

0 .0 9

0 .1 0

0 .2 2 *

�0 .0 1

1 .0 0

[1 5 ]

B u ye r’ s n o n -m

ed ia te d

p o w er

(B N M P ) (P )

5 .8 4

0 .9 8

1 .0 0

7 .0 0

�0 .3 7 *

�0 .1 8

�0 .1 2

0 .1 1

�0 .1 3

�0 .0 7

�0 .0 5

�0 .0 3

0 .0 3

0 .0 0

0 .1 5

�0 .1 1

0 .0 1

�0 .2 6 *

1 .0 0

[1 6 ]

T ec h n ic al

sk ill

si m ila ri ty

(P )

4 .6 7

1 .5 3

1 .0 0

7 .0 0

�0 .1 9

�0 .1 7

0 .0 5

�0 .1 5

0 .0 0

�0 .0 5

�0 .0 9

0 .1 9 *

�0 .2 5 *

0 .0 3

�0 .0 9

�0 .1 3

�0 .2 5 *

0 .4 5 *

1 .0 0

[1 7 ]

P ro je ct

vo la ti lit y (P )

3 .2 4

1 .3 8

1 .0 0

7 .0 0

0 .1 3

0 .0 8

0 .3 2 *

0 .1 8

0 .1 7

�0 .0 9

0 .1 7

0 .1 0

�0 .1 3

0 .0 6

0 .0 5

0 .0 2

�0 .0 5

0 .2 2 *

�0 .2 7 *

�0 .0 7

1 .0 0

[1 8 ]

T h ir d p ar ty

m an ag em

en t

(0 ,1 ) (B )

0 .3 3

0 .4 7

0 .0 0

1 .0 0

�0 .0 3

0 .0 9

0 .1 1

0 .0 4

0 .1 1

�0 .2 5 *

�0 .1 8

0 .0 4

0 .1 0

0 .0 5

0 .0 6

�0 .0 2

0 .0 3

0 .2 2 *

0 .1 1

0 .0 9

0 .1 7

1 .0 0

N o te : * S ig n ifi ca n t at

p < 0 .0 5 .

(B ) = S o u rc e o f m ea su re m en t is

fr o m

th e b u ye r su rv ey ; (P ) = so u rc e o f m ea su re m en t is fr o m

th e p ro vi d er

su rv ey .

resulted in a clean two-factor structure with Reward, Coercive, and Legal legitimate power bases loading onto one factor; and Referent and Expert loading on the other factor. Again, this is consistent with much prior research (Boyle and Dwyer 1995, Frazier and Rody 1991, Handley and Benton 2012a). Although Reward is commonly included in the mediated power construct and its second-order factor loading is sta- tistically significant, the loading may be considered on the low side when compared to some “rules of thumb.” Therefore, we repeated our main analysis (detailed in section 6) excluding Reward from the me- diated power construct. Doing so yields results sub- stantively consistent with the main analysis.8 Thus, given the literature precedent of grouping Reward (sometimes referred to as promises) with Coercive (sometimes referred to as punishments, penalties, or threats) and Legal Legitimate (e.g., Brown et al. 1995, Busch 1980, Frazier and Summers 1984, Handley and Benton 2012a, Wilkinson 1979), we retain it in our study. Our dyadic research design allows us to largely cir-

cumvent a major threat to validity with survey-based research: common methods variance (CMV). For all hypotheses, our dependent and independent vari- ables come from a different source (i.e., the buyer vs. the provider, respectively). This is a significant strength of our analysis. Additionally, our four mod- eration hypotheses involve interaction terms. Previ- ous research has determined that significant interaction effects cannot be systematically due to CMV (Siemsen et al. 2010). Therefore, CMV is of lim- ited concern in this study.

5. Endogeneity Check

An empirical challenge to which attention is increas- ingly being given in the sourcing literature is endo- geneity. Specifically, in the context of the this study, it can be argued that the buying firm’s reliance on medi- ated and non-mediated power are potentially endoge- nous decisions. That is, there could be factors unobserved by our model that influence both the reli- ance on mediated and non-mediated power as well as the provider opportunism. To the extent that this is true, the results using a traditional OLS regression would be biased (Greene 2003, Wooldridge 2012). An extensive review of prior studies in the supply chain and marketing literatures employing mediated and non-mediated power as independent variables did not reveal any study that has treated these constructs as endogenous. Therefore, we do not anticipate that endogeneity is a substantive threat to the validity of our results. Notwithstanding the precedent in the lit- erature of treating mediated and non-mediated power as exogenous however, we wanted to err on the side of caution and formally test the null hypothesis that they are indeed exogenous. If we fail to reject this hypothesis, the use of traditional OLS regression is the appropriate and valid analytic approach (Kennedy 2003). We conducted this specification test using Stata’s

two-stage least squares (2sls) instrumental variables procedure. In the first stage, we regressed the poten- tially endogenous variables (mediated and non- mediated power) on a set of instrumental variables (technical skill similarity, project volatility, and third

Table 3 Confirmatory Factor Analysis Results

Construct Items Factor loading p-value Construct Items Factor loading p-value

Provider opportunism Opport1 0.764 0.000 Buyer non-mediated power a = 0.90 Opport2 0.758 0.000 Referent Ref1 0.840 0.000 Composite reliability = 0.91 Opport3 0.836 0.000 a = 0.79 Ref2 0.573 0.000

Opport4 0.742 0.000 Composite reliability = 0.79 Ref3 0.797 0.000 Opport5 0.611 0.000 Opport6 0.791 0.000 Expert Exp1 0.881 0.000 Opport7 0.773 0.000 a = 0.91 Exp2 0.923 0.000 Opport8 0.735 0.000 Composite reliability = 0.91 Exp3 0.836 0.000

Buyer-mediated power Measurement difficulty MD1 0.546 0.000 Reward Rew1 0.847 0.000 a = 0.71 MD2 0.500 0.000 a = 0.78 Rew2 0.708 0.000 Composite reliability = 0.73 MD3 0.769 0.000 Composite reliability = 0.79 Rew3 0.683 0.000 MD4 0.712 0.000 Coercive Coerc1 0.896 0.000 Second-order factor loadings a = 0.87 Coerc2 0.915 0.000 Buyer-mediated power Coercive* 1.000 — Composite reliability = 0.88 Coerc3 0.703 0.000 Legal Legit. 0.922 0.002

Reward 0.340 0.017 Legal legitimate Legal1 0.843 0.000 Buyer non-mediated power Referent* 1.000 — a = 0.95 Legal2 0.976 0.000 Expert 0.870 0.001 Composite reliability = 0.96 Legal3 0.999 0.000 Overall model fit: v2/df = 1.38; RMSE a = 0.059; SRMR = 0.073; CFI = 0.94; TLI = 0.93

Note: *The second-order factor loadings for Coercive and Referent were specified to 1.0 for model identification purposes.

party management) and control variables. The second stage regression, with provider opportunism as the dependent variable, included the control variables along with the instrumented mediated and non- mediated power variables, but did not include the instrumental variables. Valid instrumental variables must meet two conditions: (i) they must correlate sig- nificantly with the potentially endogenous variables of mediated and non-mediated power, but (ii) only corre- late with provider opportunism through their relation- ship with mediated and non-mediated power (Angrist and Krueger 2001). The latter is referred to as the exo- geneity condition. For the first instrument, technical skill similarity, when the buyer’s project manager has similar skills and technical background as the service provider, they are more likely to rely on their exper- tise and shared understanding to address issues in a cooperative manner (Ko et al. 2005). As for project volatility, frequent changes to the scope, budget, or schedule present more issues on which the buyer will need to persuade or influence the service provider. This could arguably affect the buyer’s decision to rely on either mediated or non-mediated power. Mediated power, relative to non-mediated power, can be deployed more quickly (Benton and Maloni 2005, Boyle and Dwyer 1995, Frazier and Summers 1984). When the buyer relies on third party management for the outsourced project, this may indicate that the buyer views its relationship with the professional ser- vice provider as not too strategic; otherwise the buyer would want to directly control and develop the rela- tionship. Important strategic purchases should be associated with the development of long-term collab- orative relationships (Cannon and Perreault 1999, Parker et al. 2008, Schoenherr and Mabert 2011), which runs counter to the use of mediated power. Thus, the reliance on third party management for the outsourced project is likely to influence the buyer’s reliance on mediated and/or non-mediated power. In total, we argue that from a conceptual perspective, our three instruments should clearly correlate with the potentially endogenous power constructs. Yet, we have no conceptual rationale for expecting these instruments to have a significant unambiguous rela- tionship with the dependent variable (service provi- der opportunism) after accounting for their relation with the power constructs. However, as detailed below, it is important to empirically validate these conceptual arguments. To more formally evaluate the validity of these

instruments and to test the null hypothesis that medi- ated and non-mediated power are exogenous, we con- ducted a 2sls analysis in Stata. Post-estimation tests associated with this analysis indicate that our set of instrumental variables (technical skill similarity, project volatility, and third party management) sufficiently meet

the two conditions for valid instruments. Supporting the first requirement for the strength of the instru- ments, the first-stage F-test was statistically significant for both the mediated power model (F = 2.90; p = 0.039; R2 = 0.21; Shea’s partial R2 = 0.07) and the non-mediated power model (F = 10.54; p < 0.001; R2 = 0.36; Shea’s partial R2 = 0.20); rejecting the hypothesis of weak instruments. Sargan’s v2 (0.29; p = 0.59) and Basmann’s v2 (0.25; p = 0.62) are insignificant indicating that the second requirement for valid instruments (i.e., the exogeneity condition) also holds. Net, these results indicate that our set of instrumental variables are empirically valid and can be utilized to test for the endogeneity of mediated and non-mediated power. Upon conducting the aforemen- tioned 2sls analysis, the Durbin v2 (0.35; p = 0.84) and the Wu–Hausman F-test (0.15; p = 0.86) were each non-significant. These results fail to reject the null hypothesis that mediated and non-mediated power are exogenous variables; indicating that traditional OLS regression is a valid approach to analyzing our model. The preceding endogeneity assessment using the

2sls estimator suggests that mediated and non-mediated power can be treated as exogenous rather than endoge- nous. However, while the F statistics for the instru- ments associated with both mediated and non-mediated power are statistically significant at p < 0.05, they are not as strong for mediated power as they are for non- mediated power. Therefore, we also conducted an endo- geneity assessment using the limited-information maximum likelihood (LIML) estimator as it has been shown to be less biased than 2sls when concerns for instrument strength exist (Poi 2006, Stock et al. 2002). Conducting our endogeneity assessment using the LIML estimator results in conclusions similar to using the 2sls estimator. The F tests for the strength of instruments are significant, the tests for overidentify- ing restrictions do not indicate that the instruments correlate with the error term in the second stage, and the tests for the exogeneity of mediated and non- mediated power are not rejected (a traditional Hausman test is used for this in the case of using the LIML esti- mator—v2 = 10.88; p = 0.70. Once again, this suggests that mediated and non-mediated power can be treated as exogenous rather than endogenous. In such cases, it is advised that traditional OLS be used as it is a more efficient estimator (Baum et al. 2003, Kennedy 2003, Wooldridge 2012).

6. Analysis and Results

6.1. Analysis Our primary research objective is to elucidate how the service provider’s dependence perceptions mod- erate the relationship between the buyer’s power

exertion and the service provider’s opportunism. To evaluate these relationships, we employ OLS hierar- chical regression using ordinary standard errors. Hierarchical regression is the commonly suggested approach to examining moderation effects. Variables were sequentially entered into our model in three stages. First, we included the control variables and main effects for the buyer and provider dependence advantage. Next, the main effects for mediated and non- mediated power are added. The final stage incorporated the four terms reflecting the interactions between the perception of dependence advantage and the power variables. All variables were mean-centered prior to the computation of the interaction terms. The largest variance inflation factor was 2.14; well below the desired cutoff of 10 (Cohen et al. 2003). Thus, multi- collinearity does not appear to be meaningfully bias- ing our results.

6.2. Results The hypotheses were evaluated using the results pre- sented in Table 4. The first hypothesis (H1) is sup- ported by the significant positive effect of buyer mediated power on provider opportunism (0.16; p < 0.10) in the Stage 2 model. Similarly, the significant nega- tive effect of buyer non-mediated power on provider

opportunism (�0.28; p < 0.01) in the Stage 2 model supports the second hypothesis (H2). However, the significant interaction effects in the Stage 3 model suggest that we should be cautious about directly interpreting these first-order effects. The Stage 3 model presents evidence that the effects of buyer- mediated and non-mediated power are contingent on the provider’s perception of dependence in the inter- firm relationship. The interactive effect of the provi- der’s perception of the buyer’s dependence advantage and buyer-mediated power is significant and positive (BDA 9 BMP: 0.72; p < 0.05). Although statistically different from zero, this effect is in the opposite direc- tion of that hypothesized. Therefore, the third hypoth- esis (H3) is not supported. The significant interactions between the provider’s perception of the buyer’s dependence advantage and buyer non-mediated power (BDA 9 BNMP: �0.50; p < 0.05) and the provider’s dependence advantage and buyer-mediated power (PDA 9 BMP: 0.25; p < 0.10) support hypotheses four (H4) and five (H5), respectively. Finally, the non-sig- nificant interaction between provider’s dependence advantage and buyer non-mediated power (PDA 9 BNMP: �0.03; p > 0.10) fails to support our sixth hypothesis (H6). The support or non-support for each hypothesis is summarized in the final column

Table 4 Regression Results

DV: Provider opportunism (B)

Hypothesis Stage 1 Stage 2 Stage 3

Coeff (SE) Coeff (SE) Coeff (SE)

Constant 2.05 (0.36)*** 2.16 (0.34)*** 2.19 (0.33)*** Buying firm size (B) 0.00 (0.04) �0.02 (0.04) �0.01 (0.04) Provider firm size (P) 0.02 (0.05) �0.01 (0.04) �0.01 (0.04) Project size (budget) (B) �0.06 (0.05) �0.02 (0.05) �0.02 (0.05) Project duration (B) 0.01 (0.01) 0.00 (0.01) 0.00 (0.01) Previous projects with provider (B) 0.00 (0.01) 0.00 (0.01) 0.00 (0.01) Measurement difficulty (B) 0.31 (0.09)*** 0.28 (0.09)*** 0.27 (0.08)*** Buyer switching difficulty (B) �0.03 (0.04) �0.03 (0.04) �0.04 (0.04) Firm fixed price contract (B) �0.15 (0.18) �0.11 (0.17) �0.17 (0.16) Cost plus contract (B) �0.10 (0.22) �0.09 (0.20) �0.14 (0.20) Joint dependence (P) 0.04 (0.07) 0.06 (0.06) 0.05 (0.06) Buyer’s dependence advantage (BDA) (P) 0.13 (0.27) �0.07 (0.26) �0.24 (0.25) Provider’s dependence advantage (PDA) (P) �0.04 (0.12) �0.06 (0.11) �0.06 (0.10) Buyer’s mediated power (BMP) (P) 0.16 (0.10)* 0.17 (0.10)** H1 (supported) Buyer’s non-mediated power (BNMP) (P) �0.28 (0.08)*** �0.22 (0.08)*** H2 (supported) BDA 9 BMP 0.72 (0.36)** H3 (opposite) BDA 9 BNMP �0.50 (0.27)** H4 (supported) PDA 9 BMP 0.25 (0.16)* H5 (supported)

PDA 9 BNMP �0.03 (0.14) H6 (not supported)

F 1.24 2.51*** 2.86*** R2 0.136 0.274 0.366 Adjusted R2 0.027 0.165 0.238 Change in F 8.88*** 3.23**

Note: ***p < 0.01; **p < 0.05; *p < 0.10. One-tailed tests are used for directionally specific hypotheses. Two-tailed tests are used otherwise. (B) indicates the variable was measured from the buyer’s survey. (P) indicates the variable was measured from the provider’s survey. Bold indicates statistically significant values.

of Table 4. To further evaluate the validity of our findings, we also analyzed our model using an error-in-variables regression with the Cronbach’s alpha values as estimates for the reliability of our multi-item constructs. The results of this supplemen- tal analysis were largely consistent with the main analysis results in that our conclusions for the hypoth- esis tests are unchanged. Similarly, to ensure that heteroscedasticity is not significantly threatening the validity of our results, we also conducted the analysis using cluster-robust standard errors that cluster on industry as well as heteroscedasticity robust standard errors. In both cases, the findings are substantively similar to those presented with the main analysis. To further understand the four moderation effects

and offer additional insight into the values of de- pendence advantage over which mediated and non-mediated power have a significant effect on opportunism, marginal effect plots were developed (Figure 3). These plots show the marginal effect (dy/dx) of either mediated or non-mediated power on pro- vider opportunism at different levels of the moderators (i.e., the provider’s perception of buyer or provider dependence advantage). The range of dependence advantage values over which we show the effects of

mediated and non-mediated power correspond with the range of these values in our data. There are four plots (a–d) corresponding with hypotheses 3–6, respec- tively. In each plot, the vertical axis represents the marginal effect of buyer power (either mediated or non-mediated) on provider opportunism and the hori- zontal axis depicts the level of either buyer or provider dependence advantage. Finally, the dashed lines in each plot represent the 90% confidence bands associated with these effects while the dotted lines represent the 95% confidence bands. These plots offer additional insight into the levels of dependence advantage over which the effects of mediated and non-mediated power on opportunism are significant. Consistent with the results in Table 4, panels (a)

and (c) of Figure 3 show that the positive effect of mediated power on opportunism strengthens as the provider’s perception of either a buyer or provider dependence advantage increases. This runs counter to H3 but supports H5. Moreover, the 90% confidence bands for these effects in panels (a) and (c) demon- strate that buyer-mediated power has a significant posi- tive effect on provider opportunism when the provider’s perception of buyer dependence advantage exceeds approximately 0.20 (on a 0.0 to 1.50 scale) and when

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

M ar

gi na

l E ff

ec t

on P

ro vi

de r

O pp

or tu

ni sm

Buyer Dependence Advantage

90% and 95 % Confidence Bands for Effect of Buyer Mediated Power

Marginal Effect of BMP 90% CI 95% CI

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40

M ar

gi na

l E ff

ec t

on P

ro vi

de r

O pp

or tu

ni sm

Buyer Dependence Advantage

90% and 95 % Confidence Bands for Effect of Buyer Non-Mediated Power

Marginal Effect of BNMP 90% CI 95% CI

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0.00 0.40 0.80 1.20 1.60 2.00 2.40 2.80 3.20

M ar

gi na

l E ff

ec t

on P

ro vi

de r

O pp

or tu

ni sm

Provider Dependence Advantage

90% and 95 % Confidence Bands for Effect of Buyer Mediated Power

Marginal Effect of BMP 90% CI 95% CI

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0.00 0.40 0.80 1.20 1.60 2.00 2.40 2.80 3.20

M ar

gi na

l E ff

ec t

on P

ro vi

de r

O pp

or tu

ni sm

Provider Dependence Advantage

90% and 95 % Confidence Bands for Effect of Buyer Non-Mediated Power

Marginal Effect of BNMP 90% CI 95% CI

(a) BDA x BMP [H3] (b) BDA x BNMP [H4]

Figure 3 Marginal Effect Plots [Color figure can be viewed at wileyonlinelibrary.com]

www.wileyonlinelibrary.com

the provider’s perception of its own dependence advantage exceeds approximately 0.70 (on a 0.0 to 3.20 scale). In combination, panels (a) and (c) show that the positive relationship between buyer-mediated power and provider opportunism is only non-significant when the provider assesses dependence to be bal- anced (or close to it) in the inter-firm relationship. Further illustrating support for H4, panel (b) of Figure 3 clearly shows that the negative effect of non- mediated power on opportunism strengthens as the provider’s perception of a buyer dependence advan- tage increases. The 90% confidence bands in panel (b) also indicate that the negative relationship between buyer non-mediated power and provider opportunism is significant when the provider’s perception of buyer dependence advantage exceeds approximately 0.10. Finally, consistent with the non-significant result for H6 in Table 4, the relatively flat plot line in panel (d) of Figure 3 shows that the negative association between buyer non-mediated power and provider oppor- tunism does not decrease or increase with the provi- der’s perception of its own dependence advantage. Although the 95% confidence bands are slightly wider, the insights are largely similar to those with the 90% bands with respect to the range of values over which the marginal effects are significantly different from zero. The difference between the 90% and 95% bands in panel (c) is slightly more pronounced far out on the horizontal axis. This is simply due to the very limited number of observations in our data that have a perceived provider dependence advantage that extreme.

7. Discussion

7.1. Mediated Power and Opportunism Considered collectively, our results paint an intrigu- ing picture of the relationship between the buyer’s mediated power and the professional service provi- der’s tendency to act opportunistically. First, we find that the professional service buyer’s use of mediated power corresponds to greater service provider oppor- tunism, ceteris paribus. This finding is consistent with prior studies on power and opportunism (Handley and Benton 2012a, John 1984). However, the current study is the first to establish this relationship using truly dyadic data and the first to demonstrate this relationship in the unique context of outsourced tech- nical professional services. In light of the widespread recognition in the literature regarding the idiosyn- crasies in managing professional services, this exten- sion to the power–opportunism literature is notable. More interesting than this main effect of mediated

power, however, is the manner by which the profes- sional service provider’s perception of dependence advantage moderates the relationship between the

buyer’s mediated power and the provider’s oppor- tunism. The service provider’s perception of the buyer’s dependence advantage significantly exacer- bates the positive effect of the buyer’s mediated power on provider opportunism (see panel (a) of Figure 3). This result is exactly the opposite of our expectation. We theorized that when the service pro- vider perceives the buyer to be in an advantaged posi- tion, its opportunistic response to the buyer’s use of mediated power would be tempered due to its vul- nerability to the buyer retaliating up to and including relationship termination. Conversely, we find that the provider views this use of mediated power as the buyer exploiting the provider’s relative weakness, resulting in a visceral negative response (i.e., higher propensity to act opportunistically). It does so, per- haps, because it feels that the powerful buyer has backed them into a corner and it needs to respond in a self-preserving manner. This finding offers strong empirical verification for some anecdotal comments from managers of powerful automotive OEMs pre- sented in Gulati and Sytch (2007, p. 60). One manager concisely stated, “I have found that bullying the sup- plier does not pay off over the long run.” Our finding that the professional service buyer’s use of mediated power is positively associated with the professional service provider’s opportunistic behavior, especially when the provider perceives that the buyer is depen- dence-advantaged, gives credence to this anecdotal observation. The results further demonstrate that the positive

relationship between the professional service buyer’s mediated power and the professional service provi- der’s opportunism is also stronger when the provider perceives that it has a larger dependence advantage (see panel (c) of Figure 3). This result is more intu- itive. When the provider believes that the buyer is more reliant on the relationship than it is, the provi- der is caught by surprise when the buyer uses heavy- handed approaches to influence the provider because the provider views itself as the more powerful party. Thus, the buyer’s use of mediated power is met with a more pronounced response in terms of opportunis- tic behavior. Moreover, the buyer’s use of mediated power frames the relationship as short-term focused and adversarial. When the service provider perceives that the buyer is more dependent on the relationship, this relationship framing may instigate the provider to match the buyer’s adversarial approach by being more opportunistic because it believes the buyer has little recourse.

7.2. Non-Mediated Power and Opportunism The analysis offers support for our expectation that the professional service buyer’s reliance on non- mediated power corresponds to lower levels of

professional service provider opportunism, ceteris par- ibus. Thus, the negative first-order relationship between non-mediated power and opportunism holds. Again, this finding contributes additional empirical evidence in support of this relationship, which has been suggested in prior studies (Handley and Benton 2012a, John 1984), but had not been demonstrated in the context of technical professional services. The moderation results indicate that when the ser-

vice buyer relies on non-mediated means of influ- ence, the associated effect on service provider opportunism is negative; no matter how large the provider perceives its own dependence advantage to be (see panel (d) of Figure 3). However, we find that the provider’s perception of the buyer’s dependence advantage very much moderates the relationship between the buyer’s non-mediated power and provi- der opportunism. The provider’s assessment of the buyer’s dependence advantage strengthens the nega- tive relationship between the buyer’s non-mediated power and provider opportunism (see panel (b) of Figure 3). This supports our theorizing that when the professional service provider perceives the buyer to be dependence-advantaged, it will be pleasantly surprised when the buyer conducts itself in a refer- ent manner and utilizes its expertise to gain the pro- vider’s cooperation. In this case, the buyer’s reliance on non-mediated power further strengthens the pro- vider’s identification with the buyer and the provi- der’s perception that the two parties have congruent value systems (Kelman 1958). Furthermore, the buyer’s reliance on non-mediated means of influence signals that it is concerned with the longer-term rela- tionship beyond the immediate project (Benton and Maloni 2005, Brown et al. 1995). It is unlikely that a service provider that believes it is relatively depen- dent on the relationship would act in manner that would jeopardize these future opportunities. Ulti- mately, these effects further suppress the profes- sional service provider’s tendency to act opportunistically.

7.3. Theoretical Implications The primary theoretical contribution of our study relates to the manner by which a power target’s depen- dence perceptions moderate the power–opportunism relationship. However, our findings for the first-order effects of power on opportunism also represent an important advancement in the literature. To the best of our knowledge, only two prior studies exist that empirically examine the relationship between power use and opportunism (Handley and Benton 2012a, John 1984). Given the limited existing statistical evi- dence, there is a need for additional studies to evaluate the robustness and generalizability of prior findings.

Our empirical context of technical professional ser- vices is distinct and presents idiosyncratic manage- ment challenges relative to the context of these prior studies (i.e., transactional business processes and franchise relationships in the oil industry, respec- tively). Prior literature (Bowen and Jones 1986, Green- wood and Empson 2003, Mayer and Nickerson 2005), as well as our exploratory interviews with industry experts, suggests that with professional services it is more challenging to unambiguously assess factors of dependence, performance measurement is more diffi- cult, and professional service providers are less recep- tive to formal means of control. In light of these factors, whether or not prior findings on the relation between inter-firm power and opportunism extend to professional services was an unanswered question open to empirical investigation. Hence, our study makes a contribution by re-examining the first-order power–opportunism relationship in a novel industry context. Our results more broadly demonstrate that the rela-

tionship between power, dependence, and oppor- tunism are much more nuanced than previously recognized in the literature. We argued and found support for the professional service provider’s per- ceptions of dependence advantage moderating the buyer power–service provider opportunism relation- ship. The service provider’s perception of the firms’ dependence on the relationship, and therefore relative vulnerabilities, shapes its assessment of the risk of the buyer retaliating (e.g., relationship termination) to its opportunism and the consequences if this were to occur. Our technical professional services context was instrumental in affording us these perspectives. The services studied are often customized, involving knowledge-workers with highly specialized skills such as engineers, architects, general contractors and other professionals licensed in their trade. In these relationships, accurately evaluating factors contribut- ing to dependence (e.g., qualified alternative provi- ders, switching difficulties) may be complicated, leading the two parties to have misaligned percep- tions of relative dependence. As our results show, when the buyer’s influence approach is misaligned with how the provider views the relationship, the effects on opportunism can be profound. This impor- tant consideration of the power target’s (provider’s) perception of the relationship has thus far been absent from the power and dependence literature. Further, by examining situations in which the actual applica- tion of power deviates from traditional power logic, this study joins an emerging literature recognizing that some powerful organizations may forgo the use of mediated power for the sake of building collabora- tive, long-term relationships (e.g., Gulati and Sytch 2007).

Finally, by utilizing a spline specification and creat- ing separate variables for the degree of buyer and provider dependence advantage, we determined that the moderating effects of these two do not always operate in completely opposite directions. In regard to the positive relationship between buyer-mediated power and provider opportunism, we found that a large provider-perceived buyer dependence advan- tage and large provider-perceived provider depen- dence advantage each strengthen the positive relationship. This implies, at least in this relationship, that the direction of advantage is less important than the sheer magnitude of the dependence imbalance. Prior research arguing that asymmetric relationships are inherently unstable is consistent with this finding (Buchanan 1992, Kumar et al. 1998). As a result, we echo other scholars and encourage future research to specify each party’s dependence advantage as sepa- rate variables as opposed to opposite extremes of a single continuum.

7.4. Managerial Implications This study has important managerial implications. Most fundamentally, we see that the approach to influence and control (i.e., use of power) needs to be aligned with each party’s relative dependence. When misapplied, power can lead to a greater risk of opportunism. From the professional service pro- vider’s perspective (the power target here), the use of mediated power is typically assumed to be nega- tive and reliance on non-mediated power to be pos- itive for the relationship. Our study shows that these commonly held beliefs need to be conditioned on the perceived dependence situation in the rela- tionship. The most effective means of influencing an exchange partner, either via mediated or non- mediated power, is not a one-size fits all proposi- tion but needs to be tailored to the relationship. Moreover, managers of the dependence-advantaged party cannot presume that the disadvantaged party will simply acquiesce to their mediated power with- out consequence. Our findings suggest that weak power targets exhibit more opportunistic behav- ior when they perceive a strongly dependence- advantaged partner to be exploiting their power using mediated methods of influence. Thus, we rec- ommend that managers of inter-organizational rela- tionships do their homework to understand factors contributing to each party’s dependence. Further, we advise boundary-spanning personnel to consider the long-term implications of their attempts to con- trol the other party to the exchange. They may be able to gain compliance in the short run, but may also be sowing the seeds for opportunistic behavior, which ultimately undermines the objectives of the relationship. This is a particularly salient concern

when managing outsourced professional services where the risk of opportunism is elevated due to performance measurement difficulties and profes- sional workers who are less receptive to being tightly controlled (Bowen and Jones 1986, Green- wood and Empson 2003, Mayer and Nickerson 2005). Our suggestion for managers or project owners to

ensure that they clearly understand the factors con- tributing to each firm’s dependence on the relation- ship seems rather straightforward on the surface. However, as outsourced services become increas- ingly complex, specialized, and knowledge-based, assessing relative dependence will correspondingly become more challenging. For instance, determining the viability of alternative sources of supply or the costs of switching exchange partners may be an exercise fraught with ambiguity when these factors are grounded in more intangible considerations such as specialized technical expertise. Our exploratory interviews with industry experts revealed that often- times buyers do not have the requisite technical knowledge to appropriately evaluate service provi- ders. This raises the risk that the parties will per- ceive dependence differently, which could result in one firm attempting to utilize its perceived power in a manner that is inconsistent with the other’s expec- tation. Our study demonstrates that such a misalign- ment can increase the risk of opportunism. Therefore, we also recommend that professional ser- vice managers take steps to minimize the opportu- nity for dependence perception misalignments to occur. Potential actions include: (i) communicating with exchange partners (e.g., through various pre- qualification activities) to understand how the other party perceives the governance structure in the rela- tionship; (ii) planning pre-bid meetings with all potential providers; (iii) devoting resources toward maintaining a deep knowledge of the supply mar- ket; (iv) making sure that the buying professionals have technical skill sets and expertise similar to that of current and potential providers, or at minimum, are engaging with those individuals within their organizations who do have the requisite expertise; and (v) if the appropriate technical expertise does not reside within the firm, consider partnering with a third party consultant to provide technical support for the buyer or project owner. Indeed, these latter two recommendations came through strongly in our exploratory interviews of experienced practitioners as important factors in appropriately managing the unique challenges of outsourced technical profes- sional services. Taking these actions should help mitigate the potential for dependence perception misalignment and the associated relational risk that could ensue.

8. Conclusions and Future Directions

Notwithstanding the contributions made to theory and practice, the limitations of this study present mul- tiple opportunities for future research to validate and extend our work. First, our research design is cross- sectional in nature; placing the usual constraints on the extent to which we can assert causality in the rela- tionships studied. Further, the relational constructs of dependence and power are dynamic by nature. Thus, we echo the call of others in this area (e.g., Gulati and Sytch 2007) for scholars to apply a longitudinal frame- work and examine how the relationships among power, dependence, and opportunism unfold over time. Second, here, we focused on how the depen- dence perceptions of the professional service provider moderated the power–opportunism relationship. It is also plausible that dependence perceptions can mod- erate the relationship between power use and other aspects of exchange performance. Examining this possi- bility would be a natural extension to this study. Third, this is the first study to evaluate a moderator (i.e., dependence perceptions) of the power–opportunism relationship. It is certainly possible that other relevant factors (e.g., the buyer’s history of power use, the pro- vider’s experience with other buyers in the industry, the nature of the estimating and bidding process, etc.) could influence the relationship between power use and opportunism as well. These factors are worthy of future investigation. Finally, our research represents one of the first dyadic empirical studies of the chal- lenges in sourcing technical professional services. This line of inquiry could be extended by exploring how specific facets of professional services (e.g., skill specialization, knowledge intensity, degree of cus- tomization and required customer interaction, diffi- culty in ascertaining quality and capabilities, and so forth.) differentially impact the use and consequences of inter-organizational power. At the outset, the objective of this study was to

refine our understanding of the relationships between power, dependence, and opportunism. Our findings on the moderating influence of the power target’s (here service provider’s) perception of dependence advantage add relevant insight and much needed nuance to our understanding of these working rela- tionships. We hope that future scholarly investiga- tions can continue to advance this stream of research and build on the foundation that we have established with this study.

Notes

1Here, the generic terms “power source” and “power tar- get” are used to identify the party attempting to exert influence and the party that is the recipient of the influ- ence efforts, respectively. We use these terms to remain

true to the general nature of traditional power logic. How- ever, to be clear, in the present study, the buyer is the power sources and the service provider is the power tar- get. 2Some studies in the supply chain literature have found that powerful firms sometimes refrain from using coercive tactics even though they are structurally in a position of power (cf., Gulati and Sytch 2007). While we recognize these prior findings and appreciate that not all firms fol- low the prevailing logic that we outline, in this study our theoretical framework is grounded in the traditional power perspective. Thus, we align our argumentation with the traditional logic advanced. 3Figure 2 includes two separate boxes for perceived dependence advantage (i.e., buyer advantage and provi- der advantage). This was done to accurately reflect our analysis which employs a spline specification used in prior studies of inter-organizational dependence and includes separate measures for buyer and provider dependence advantage. Please see section 4.2 for addi- tional details. 4In the context of technical professional services such as those we study here, the buyers are sometimes alterna- tively referred to as the “project owners.” For simplicity, we only use the term “buyers.” 5To be clear, these facility management professionals were requested to report on outsourced technical work (e.g., engineering, architectural, contracting) requiring special- ized skills; not routine facility management tasks. 6Non-respondent data for buyer firm size and industry was obtained through the follow-up survey of non-respon- dents described previously. 7On the questionnaire, respondents could provide percent- age responses in 5% increments (e.g., 5%, 10%, 15%, etc.) To match the 1–7 scale of the other dependence items, these percentage responses were rescaled from 1–20 (for each 5% increment) to 1–7 using the formula: 0.685 + [(re- sponse score/20) 9 6.315]. 8The only notable change is that the support for H4 is p < 0.10 instead of p < 0.05.

Appendix A. Preliminary Exploratory Interviews Prior to conducting our main data collection, we inter- viewed four industry experts with extensive experi- ence (i.e., more than 15 years each) with outsourced technical professional services (e.g., engineering, architectural, general contracting, etc). Each interview was approximately one-hour in duration. Collec- tively, the interviews provided the research team with both buyer and provider perspectives of the manage- ment of outsourced technical professional services. These exploratory interviews were conducted for three primary purposes: (i) to ensure that our study was appropriately grounded in practice and was managerially relevant, (ii) to receive feedback on our online questionnaire, and (iii) to identify the most appropriate professionals to target with our data

collection. The table below provides a brief back- ground on the interviewees.

Role(s) Context

Interview 1 Owner, business development, lead civil engineer

Provider of civil engineering and contract management services to municipal, state, and federal DOT

Interview 2 Facilities manager (manages outsourced construction, engineering services, etc.)

Multi-national industrial company specializing in elevator systems

Interview 3 Director of procurement for indirect services (engineering, plant construction, etc.)

Experience in consumer packaged goods and energy sectors

Interview 4 Engineering and construction professional with experience on both buyer- and provider-side of outsourced technical services

Experience with food and beverage processing facilities, healthcare facilities, educational facilities, etc

Appendix B. Measurement Items for Multi-Item Constructs

Construct Measurement items

Provider opportunism (Source = buyer)

(Opport1) <Provider> sometimes exaggerated the necessity of changes they wanted to the plan or the budget

(Opport2) <Provider> sometimes altered facts to get what they wanted

(Opport3) <Provider> would try to renegotiate to their own advantage

(Opport4) <Provider> exaggerated the costs they actually incurred

(Opport5) Cost estimates provided by <Provider> tended to escalate as the project progressed

(Opport6) <Provider> was less and less cooperative as the project progressed

(Opport7) <Provider> would do anything within their means to get a larger share of the gains from our relationship

(Opport8) It was hard to get <Provider> to accept changes without us making certain concessions and compromises

Reward power (Source = provider)

(Rew1) <Buyer> offered incentives to our firm when we were initially reluctant to cooperate with a new program

(Rew2) <Buyer> would favor us on other occasions if we went along with their requests

(Rew3) <Buyer> offered us rewards so we would go along with their wishes

Coercive power (Source = provider)

(Coerc1) If we did not do as they asked, we would not receive very good treatment from <Buyer>

(continued)

Construct Measurement items

(Coerc2) If we did not agree with <Buyer>’s suggestions, they could make things difficult for us

(Coerc2) <Buyer> made it clear that failing to comply with their requests would result in penalties against us

Legal legitimate power (Source = provider)

(Legal1) <Buyer> referred to the terms of our contract to gain our compliance on particular requests

(Legal2) <Buyer> made a point to refer to our legal agreement when attempting to influence us

(Legal3) <Buyer> used sections of our formal agreement as a “tool” to get us to agree to their demands

Referent power (Source = provider)

(Ref1) We admired the way that <Buyer> ran their business

(Ref2) We often did what <Buyer> suggested, because we were proud to be affiliated with them

(Ref3) We talked up <Buyer> to our colleagues as a great business with which to be associated

Expert power (Source = provider)

(Exp1) We saw <Buyer> as an expert in their industry

(Exp2) We respected the judgment of <Buyer>’s representatives

(Exp3) Our firm believed that <Buyer> retains business expertise that made them likely to suggest the proper thing to do

Buyer dependence (Source = provider)

(BD1) It would have required much trouble and expense for <Buyer> to switch providers for this service

(BD2)There were enough potential service providers to ensure adequate competition among the current providers. (reverse coded)

(BD3)There were satisfactory alternate, immediately available, providers for this service. (reverse coded)

(BD4) <Buyer> had made significant investments specific to our relationship

(BD5) Please estimate what % of the total project budget was contracted to your firm? (expressed as a percentage)

Provider dependence (Source = provider)

(PD1) Our firm had made significant investments specific to our relationship with <Buyer>

(PD2) Please estimate what % of your firm’s billable hours came from <Buyer> during the last 12 months. (expressed as a percentage)

(PD3) Our firm would have faced a serious financial crisis if <Buyer> withdrew their business from us

Measurement difficulty (Source = buyer)

(MD1) To what degree was it easy to measure the collective performance of those individuals who performed the key tasks for this project? (1 = very easy; 4 = neutral; 7 = very difficult)

(MD2) Evaluating the performance of <Provider> required extensive inspection and monitoring effort

(continued)

Construct Measurement items

(MD3) For the service provided by <Provider> on this project, it was difficult to ascertain if a good job was being done

(MD4) It was difficult to determine if <Provider> followed agreed-upon standards, specifications, and procedures

Note: Unless otherwise indicated, the scale for all questions is: 1 = strongly disagree; 4 = neutral; 7 = strongly agree.

References Angrist, J. D., A. B. Krueger. 2001. Instrumental variables and the

search for identification: From supply and demand to natural experiments. J. Econ. Perspect. 15(4): 69–85.

Bajari, P., S. Tadelis. 2001. Incentives versus transaction costs: A theory of procurement contracts. Rand J. Econ. 32(3): 387– 407.

Barney, J. 2002. Gaining and Sustaining Competitive Advantage, 2nd edn. Pearson Education, Upper Saddle River, NJ.

Baum, C. F., M. E. Schaffer, S. Stillman. 2003. Instrumental vari- ables and GMM: Estimation and testing. Stata J. 3(1): 1–31.

Benton, W. C., M. Maloni. 2005. The influence of power driven buyer/seller relationships on supply chain satisfaction. J. Oper. Manag. 23: 1–22.

Bettis, R. A., S. Ethiraj, A. Gambardella, C. Helfat, W. Mitchell. 2016. Creating repeatable cumulative knowledge in strategic management. Strateg. Manag. J. 37(2): 257–261.

Blau, P. M. 1964. Exchange and Power in Social Lifeed. Wiley, New York, NY.

Bollen, K. A. 2011. Evaluating effect, composite, and causal indi- cators in structural equation models. MIS Q. 35(2): 359–372.

Bollen, K., R. Lennox. 1991. Conventional wisdom on measurement: A structural equation perspective. Psychol. Bull. 110(2): 305–314.

Bowen, D. E., G. R. Jones. 1986. Transaction cost analysis of ser- vice organization-customer exchange. Acad. Manag. Rev. 11(2): 428–441.

Boyle, B. A., F. R. Dwyer. 1995. Power, bureaucracy, influence, and performance: Their relationships in industrial distribution channels. J. Bus. Res. 32: 189–200.

Boyle, B., F. R. Dwyer, R. A. Robicheaux, J. T. Simpson. 1992. Influence strategies in marketing channels: Measures and use in different relationship structures. J. Mark. Res. 29: 462–473.

Brass, D. J. 1984. Being in the right place: A structural analysis of individual influence in an organization. Adm. Sci. Q. 29(4): 518–539.

Brown, J. R., R. E. Lusch, C. Y. Nicholson. 1995. Power and rela- tionship commitment: Their impact on marketing channel member performance. J. Retail. 71: 363–392.

Brown, J. R., C. S. Dev, D.-J. Lee. 2000. Managing marketing chan- nel opportunism: The efficiency of alternative governance mechanism. J. Market. 64: 51–65.

Buchanan, L. 1992. Vertical trade relationships: The role of depen- dence and symmetry in attaining organizational goals. J. Mark. Res. 29(1): 65–75.

Burt, R. S. 1982. Toward a Structural Theory of Action: Network Mod- els of Social Structure, Perception, and Actioned. Academic Press, New York, NY.

Busch, P. 1980. The sales manager’s bases of social power and influence upon the sales force. J. Market. 44: 91–101.

Cannon, J. P., W. D. Perreault. 1999. Buyer–seller relationships in business markets. J. Market. Res. 36: 439–460.

Carson, S. J. 2007. When to give up control of outsourced new product development. J. Market. 71(1): 49–66.

Carson, S. J., A. Madhok, T. Wu. 2006. Uncertainty, opportunism, and governance: The effects of volatility and ambiguity on formal and relational contracting. Acad. Manag. J. 49(5): 1058–1077.

Carter, C. R. 2000. Ethical issues in international buyer–supplier rela- tionships: a dyadic examination. J. Oper. Manag. 18: 191–208.

Chang, C.-Y., G. Ive. 2007. The hold-up problem in the manage- ment of construction projects: A case study of the Channel Tunnel. Int. J. Project Manage. 25(4): 394–404.

Chase, R. B. 1981. The customer contact approach to services: theo- retical bases and practical extensions. Oper. Res. 29(4): 698–706.

Cohen, J., P. Cohen, S. G. West, L. S. Aiken. 2003. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciencesed. L. Erlbaum Associates, London.

Das, T., B. Teng. 1998. Between trust and control: Developing con- fidence in partner cooperation alliances. Acad. Manag. Rev. 23: 491–512.

Diamantopoulos, A., H. M. Winklhofer. 2001. Index construction with formative indicators: An alternative to scale develop- ment. J. Market. Res.: American Marketing Association 38: 269–277.

Disher, C., C. Teschner, A. Kaul, M. Wright. 2005. Next Generation Outsourcing and Offshoring—Capturing Value in Financial Ser- vices. Booz Allen Hamilton, Chicago, IL.

El-Ansary, A., L. W. Stern. 1972. Power measurement in the dis- tribution channel. J. Mark. Res. 9: 47–52.

Ellram, L., T. E. Hendrick. 1995. Partnering characteristics: A dya- dic perspective. J. Bus. Log. 16: 41–63.

Emerson, R. M. 1962. Power-dependence relations. Am. Sociol. Rev. 27(1): 31–41.

Falk, A., M. Kosfeld. 2006. The hidden costs of control. Am. Econ. Rev. 96: 1611–1630.

Frazier, G. L., R. C. Rody. 1991. The use of influence strategies in interfirm relationships in industrial product channels. J. Mar- ket. 55: 52–69.

Frazier, G. L., J. O. Summers. 1984. Interfirm influence strategies and their applications within distribution channels. J. Market. 48(3): 43–55.

Frazier, G. L., J. O. Summers. 1986. Perceptions of interfirm power and its use within a franchise channel of distribution. J. Mark. Res. 23: 169–176.

Frazier, G. L., J. D. Gill, S. H. Kale. 1989. Dealer dependence levels and reciprocal actions in a channel of distribution in a developing country. J. Market. 53: 50–69.

French, J. R., B. Raven. 1959. The bases of social power. D. Cart- wright, ed. Studies in Social Power. Ann Arbor, MI: University of Michigan Press, 311–320.

Gaski, J. F. 1986. Interrelations among a channel entity’s power sources: Impact of the exercise of reward and coercion on expert, referent, and legitimate power sources. J. Mark. Res. 23: 62–77.

Gereffi, G., K. Fernandez-Stark. 2010. “The Offshore Services Glo- bal Value Chain”: Center on Globalization, Governance & Competitiveness, Duke University.

Geyskens, I., J. B. E. M. Steenkamp, L. K. Scheer, N. Kumar. 1996. The effects of trust and interdependence on relationship com- mitment: A trans-Atlantic study. Int. J. Res. Mark. 13(4): 303–318.

Gottfredson, M., R. Puryear, S. Phillips. 2005. Strategic sourcing from periphery to the core. Harv. Bus. Rev. 83: 132–139.

Greene, W. H. 2003. Econometric Analysised. Prentice Hall, Upper Saddle River, NJ.

Greenwood, R., L. Empson. 2003. The professional partnership: Relic or exemplary form of governance? Organ. Stud. 24(6): 909–933.

Gulati, R., M. Sytch. 2007. Dependence asymmetry and joint dependence in interorganizational relationships: Effects of embeddedness on a manufacturer’s performance in procure- ment relationships. Adm. Sci. Q. 52(1): 32–69.

Gundlach, G. T., E. R. Cadotte. 1994. Exchange interdependence and interfirm interaction: Research in a simulated channel set- ting. J. Mark. Res. 31(4): 516–532.

Handley, S., C. M. Angst. 2015. The impact of culture on the rela- tionship between governance and opportunism in outsourcing relationships. Strateg. Manag. J. 36(9): 1412–1434.

Handley, S., W. C. Benton. 2012a. The influence of exchange haz- ards and power on opportunism in outsourcing relationships. J. Oper. Manag. 30: 55–68.

Handley, S., W. C. Benton. 2012b. Mediated power and out- sourcing relationships. J. Oper. Manag. 30: 253–267.

Helper, S., J. P. MacDuffie, C. Sabel. 2000. Pragmatic collabora- tions: Advancing knowledge while controlling opportunism. Ind. Corp. Change 9(3): 443–488.

Howell, R. D. 1987. Covariance structure modeling and measure- ment issues: A note on “interrelations among a channel entity’s power sources”. J. Mark. Res. 24: 119–126.

Hunt, S. D., J. R. Nevin. 1974. Power in a channel of distribution: Sources and consequences. J. Mark. Res. 11: 186–193.

Jacobs, D. 1974. Dependency and vulnerability: An exchange approach to the control of organizations. Adm. Sci. Q. 19: 45– 59.

Jap, S. D., E. Anderson. 2003. Safeguarding interorganizational performance and continuity under ex post opportunism. Man- agement Sci. 49: 1684–1701.

John, G. 1984. An empirical investigation of some antecedents of opportunism in a marketing channel. J. Mark. Res. 21: 278–289.

Johnston, J. 1984. Econometric Methodsed. McGraw-Hill, New York, NY.

Kasulis, J. J., R. E. Spekman. 1980. A Framework for the use of power. Eur. J. Mark. 14(4): 180–191.

Kelman, H. C. 1958. Compliance, identification, and internaliza- tion: Three processes of attitude change. J. Conflict Resolut. 2: 51–60.

Kennedy, P. 2003. A Guide to Econometricsed. MIT Press, Cam- bridge, MA.

Kessler, J. B., S. Leider. 2012. Norms and contracting. Management Sci. 58(1): 62–77.

Ko, D.-G., L. J. Kirsch, W. R. King. 2005. Antecedents of knowl- edge transfer from consultants to clients in enterprise system implementations. MIS Q. 29: 59–85.

Kumar, N., L. Scheer, S. Jan-Benedict. 1998. Interdependence, punitive capability, and the reciprocation of punitive actions in channel relationships. J. Mark. Res. 35: 225–235.

Lewis, M. A., A. D. Brown. 2012. How different is professional service operations management? J. Oper. Manag. 30(1): 1–11.

MacCallum, R. C., M. W. Browne. 1993. The use of causal indica- tors in covariance structure models: Some practical issues. Psychol. Bull. 114: 533–541.

Maloni, M., W. C. Benton. 2000. Power influences in the supply chain. J. Bus. Log. 21: 49–73.

Masten, S. E., J. W. Meehan, E. A. Snyder. 1991. The costs of orga- nization. J. Law Econ. Organ. 7(1): 1–25.

Mayer, K. J., J. A. Nickerson. 2005. Antecedents and performance implications of contracting for knowledge workers: Evi- dence from information technology services. Organ. Sci. 16(3): 225–242.

Mudambi, R., S. Helper. 1998. The “close but adversarial” model of supplier relations in the U.S. auto industry. Strateg. Manag. J. 19: 775–792.

Parker, D. B., G. A. Zsidisin, G. L. Ragatz. 2008. Timing and extent of supplier integration in new product development: A contingency approach. J. Supply Chain Manag. 44: 71–83.

Parkhe, A. 1993. Strategic alliance structuring: A game theoretic and transaction cost examination of interfirm cooperation. Acad. Manag. J. 36: 794–829.

Pfeffer, J., G. R. Salancik. 1978. The External Control of Organizations: A Resource Dependence Perspectiveed. Harper & Row, New York.

Poi, B. P. 2006. Jackknife instrumental variables estimation in Stata. Stata J. 6(3): 364–376.

Poppo, L., K. Z. Zhou, S. Ryu. 2008. Alternative origins to interorga- nizational trust: An interdependence perspective on the shadow of the past and the shadow of the future. Organ. Sci. 19(1): 39–55.

PriceWaterhouseCoopers. 2009. Outsourcing comes of age: The rise of collaborative partnering. Available at https://www. pwc.com/gx/en/operations-consulting-services/pdf/outsourc ingcomesofage.pdf (accessed date July 13, 2015).

Provan, K. G., J. B. Gassenheimer. 1994. Supplier commitment in relational contract exchanges with buyers: A study of interor- ganizational dependence and exercised power. J. Manage. Stud. 31(1): 55–68.

Pugh, D. S., D. J. Hickson, C. R. Hinings, C. Turner. 1969. The context of organization structures. Adm. Sci. Q. 14(1): 91–114.

Roh, J. A., J. M. Whipple, K. K. Boyer. 2013. The effect of single rater bias in multi-stakeholder research: A methodological evaluation of buyer–supplier relationships. Prod. Oper. Manag. 22(3): 711–725.

Sako, M. 2006. Outsourcing and offshoring: Implications for produc- tivity of business services. Oxf. Rev. Econ. Policy 22(4): 499–512.

Schilke, O., K. S. Cook. 2015. Sources of alliance partner trustwor- thiness: Integrating calculative and relational perspectives. Strateg. Manag. J. 36(2): 276–297.

Schmenner, R. W. 1986. How can service businesses survive and prosper. Sloan Manage. Rev. 27(3): 21–32.

Schoenherr, T., V. A. Mabert. 2011. A comparison of online and offline procurement in B2B markets: Results from a large- scale survey. Int. J. Prod. Res. 49(3): 827–846.

Siemsen, E., A. Roth, P. Oliveira. 2010. Common method bias in regression models with linear, quadratic, and interaction effects. Organ. Res. Methods 13(3): 456–476.

Skinner, S. J., J. H. Jr Donnelly, J. M. Ivancevich. 1987. Effects of transactional form on environmental linkages and power- dependence relations. Acad. Manag. J. 30(3): 577–588.

Skinner, S. J., J. B. Gassenheimer, S. W. Kelley. 1992. Cooperation in supplier–dealer relations. J. Retail. 68(2): 174–194.

Spekman, R. E. 1979. Influence and information: An exploratory investigation of the boundary role person’s basis of power. Acad. Manag. J. 22(1): 104–117.

Stock, J. H., J. H. Wright, M. Yogo. 2002. A survey of weak instru- ments and weak identification in generalized method of moments. J. Bus. Econ. Stat. 20(4): 518–529.

Stump, R. L., J. B. Heide. 1996. Controlling supplier opportunism in industrial relationships. J. Mark. Res. 33: 431–441.

Svensson, G. 2004. Interactive vulnerability in buyer–seller rela- tionships: A dyadic approach. Int. J. Phys. Distrib. Log. Manag. 34(8): 662–682.

Unsal, H. I., J. E. Taylor. 2010. Modeling interfirm dependency: Game theoretic simulation to examine the holdup problem in project networks. J. Constr. Eng. Manag. 137(4): 284–293.

Von Nordenflycht, A. 2010. What is a professional service firm? Toward a theory and taxonomy of knowledge-intensive firms. Acad. Manag. Rev. 35(1): 155–174.

https://www.pwc.com/gx/en/operations-consulting-services/pdf/outsourcingcomesofage.pdf

Wathne, K. H., J. B. Heide. 2000. Opportunism in interfirm rela- tionships: Forms, outcomes, and solutions. J. Market. 64: 36– 51.

Wilkinson, I. F. 1979. Power and satisfaction in channels of distri- bution. J. Retail. 55: 79–94.

Williamson, O. E. 1975. Markets and Hierarchies: Analysis and Anti- trust Implicationsed. Free Press, New York, NY.

Williamson, O. E. 1985. The Economic Institutions of Capitalismed. Free Press, New York, NY.

Williamson, O. E. 1991. Comparative economic organization: The analysis of discrete structural alternatives. Adm. Sci. Q. 36: 269–296.

Wooldridge, J. 2012. Introductory Econometrics: A Modern Approach. Cengage Learning, Mason, OH.

Copyright of Production & Operations Management is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.