business paper

Journal of Asia Business Studies SPRING 200832 Journal of Asia Business Studies SPRING 2008 33

INTRODUCTION The determinants of high-performance in joint ventures

(JVs) have been extensively studied. Many scholars have of- fered insights into factors that affect economic and other measures of JV success or failure. Recently, part of the lit- erature has converged on particular specific important cat- egories of factors, unified by the concept of fit. Based on the literature, we argue and test for the primacy of three particu- lar fit-related factors’ effects on JV performance. Our work rigorously explores the effects of all three of these factors on JV performance. Using detailed data comprised of over 80 recent China-U.S. joint ventures operating in China, this pri- marily empirically paper scrutinizes the impact of strategic, organizational and cultural fit on JV performance. The arena of China-U.S. equity-based joint ventures is rich and diverse, providing an appropriate venue for a study of the three speci- fied categories of partner-fit.

When considering the extant “literature of fit,” we are struck by the preponderance of studies theorizing about or empirically focused on one rather than all three factors. Some scholars look at many factors (e.g., Douma et al, 2000), but treat only one or two of the three factors we assert the litera- ture identifies as critical. Part of the reason for this may be a lack of agreement in the literature as to precisely which ob-

servable phenomena and managerial attitudes comprise valid measurements of fit. In the first sentence of the conclusion to their article, Hennart and Zeng (2005:113) assess the situation well: “The literature on the determinants of alliance perfor- mance is increasingly diverse, with no clear consensus as to what the dependent and independent variables ought to be.” Our study partly addresses this issue by acknowledging the diversity in definitions of fit and suggesting two very different definitions for each of three types of fit--strategic, organiza- tional and cultural. Therein lies a major part of our contribu- tion: we test a reasonably complete model of the determinants of JV performance using a well-controlled-for, diverse set of reliable measures of fit. Additionally, we construct and model four independent measures of JV performance, our dependent variable: perceived satisfaction with JV performance, per- ceived prior financial performance trends, perceived overall competitiveness of the JV, and perceived efficacy of JV deci- sion making structure. We use diverse JV performance indica- tors in recognition of the myriad ways in which performance may be measured. Our approaches to measuring our indepen- dent variables as well as performance, while admittedly im- perfect, avoid some issues encountered by other scholars.

The academic literature concerning international joint ventures has its roots in the 1980s, and is admittedly vast. In this paper, for simplicity, we confine our attention to papers

STRATEGIC, ORGANIZATIONAL, AND CULTURAL FIT: EFFECTS ON PERFORMANCE IN CHINA-US JOINT VENTURES Bruce A. Heiman San Francisco State University

Weining Li South China University of Technology

George Chan University of San Francisco

Salvador D. Aceves University of San Francisco

ABSTRACT

We explore the effects of three categories of fit on US-China joint-venture performance using four performance measures. Many studies prescribe strong fit across multiple categories as necessary for high performance, but little rigorous analysis supports this. Three important threads of existing “fit” research resonate in the literature: strategic, cultural and organizational fit. We analyze an original survey dataset of over 80 US-China JVs, and test for effects of fit-categories using two measures for each thread. Additionally, multiple control factors give a compelling look at a complete model of fit’s effects on JV per- formance. Objective congruence (strategic fit) among JV partner-firms, impacts two performance-measures. Efficacy of managerial communications (cultural fit) also matters, as does harmony regarding hiring decisions (organizational fit). Our findings are a step forward empirically, and partly resolve persistent questions about partner-fit in JVs and performance.

Keywords: strategic, fit, China, joint-venture, performance, empirical.

explicitly concerned with fit between equity-based joint ven- ture partner-firms. By focusing our discussion of the litera- ture on papers discussing fit and JV performance, we usefully narrow the universe of our citations. A number of notable, foundational works looking at JVs are not, however, explicitly concerned with fit, and bear brief mention. One core work, Anderson and Gatignon (1986) looks at the determinants of foreign entry mode. Teece (1981) examines far-sighted ef- ficiency of governance safeguards under a transaction cost lens when the activity is sharing knowledge between inter- national partner-firms. Both articles look at the determinants of governance choice (the dependent variable) and assume performance is high. This is a common feature of early analy- ses; their intent was to shed light on the bases of governance choice. Hennart (1988; 1991) advances thinking about JVs by micro-analytically looking at two bases for market failure that JVs serve to remedy: scale and link JVs both solve differ- ent market failure scenarios. Kogut and Zander (1992) assert a “social community” effect over Hennart’s more Spartan (more farsighted) logic: social network-based knowledge recombi- nation within the firm competes with the need to collaborate between JV partner-firms—the market for collaboration with outsiders is considered thin. Despite drawbacks, Kogut and Zander’s analysis of important knowledge-based contingen- cies is valuable and sheds some light on how successful JVs operate. More recently and more in the spirit of our work, Hen- nart and Zeng (2002) showed that cross-cultural differences affect JV longevity (where longevity proxies performance). Dhanaraj and Beamish (2004) look at the effects of different equity ownership structures, also on longevity. Longevity is a good measure of performance, but it is not the only one. Al- though focused on our independent variables of interest, the pioneering antecedent literature noted above generally ne- glects to measure performance in a simple, meaningful cross sectional sense, e.g., financial outcomes after multiple years of JV operations, an issue addressed by our work.

This paper proceeds as follows: first we discuss the litera- ture on “fit” in four parts, starting with (a) studies that treat multiple aspects of fit (works with main thrusts close to our own). We then briefly (b) discuss a particularly relevant study regarding joint ventures, and (c) offer our own definitions of fit, partly based on the literature. We also (d) argue for six fit- savvy testable hypotheses--we touch on the specific literature for each of the two components of fit derived from the three main fit-categories. We subsequently describe our methodol- ogy and the sample data as well as variable construction. We then present the results of a data analysis using ordinary least squares regression. Finally, we discuss the implications of the results, issues, and future directions for this line of research.

LITERATURE OF FIT-RELATED FACTORS’ EFFFECTS ON PERFORMANCE

In an article that discusses managing the dynamics of fit, Douma et al (2000:582) present a “generic fit framework” and conclude that alliance success requires good fit in no fewer

than five areas: strategic fit; organizational fit, cultural fit, op- erational fit, and human fit. These five types of fit are all pur- ported to be indispensable for alliance success. Their work fo- cuses primarily on strategic and organizational fit, and takes a (useful) dynamic rather than static view of fit whereby the degree of fit may be improved over time. The authors main- tain that the more partner-firms complement one another, the greater the probability that an alliance will be success- ful (resource complementarity). Other drivers for strategic fit include the compatibility of strategies (similar to this paper’s congruence of objectives) and strategic importance.

Regarding organizational fit, Douma et al (2000) focus on ex ante alliance design/building issues, such as alliance potential and feasibility) and maintain that a good design should address organizational differences, provide flexibil- ity, and enable effective management control for both part- ners. Their approach avers that fit may be “designed into” al- liances. Although dynamic, Douma et al’s focus on alliance design does not offer insight into important organizational fit issues that may arise after the JV is established and oper- ating—farsightedness with respect to performance is lacking. Alliance design, given the impossibility of complete con- tracting (Williamson 1975), can never articulate a complete set of contractual clauses that precisely mandate the scope of knowledge-sharing, or the precise delineation of scope and division of work prior to a JV building some operating ex- perience. Contracts are necessarily incomplete, and besides measuring efficacy of the alliance design activity, this line of research is of limited utility in predicting successfully oper- ating JVs. Our work acknowledges that transaction cost fac- tors affect a JV’s performance.

Although they offer meaningful insights, Douma et al base their extensive conclusions on a case study, potentially limit- ing the generality of their findings. Their chosen case study JV excelled at all five of their specified fit parameters. This seems to us to represent an onerous standard for a successful alliance. We suggest that it may be possible for a JV to experi- ence high performance under less than “perfect” conditions, i.e., when not all factors are optimal.

Faure (2000) discusses the impact of strategic, organiza- tional and cultural issues on performance in the setting of negotiations prior to entering into a JV. Significant difficul- ties encountered by both parties during negotiations are ex- plored and divided into three categories: strategic, cultural, and organizational cause. These areas correspond strongly to our conception of strategic, cultural, and organizational fit, but, like Douma et al, since Faure applies his framework sole- ly to the negotiation process, the results are of limited use for us—they ignore the nature and extent of performance during a JV’s lifetime. Faure’s strategic cause corresponds to one part of strategic fit for us: congruence of objectives between part- ners is a primary factor determining the success of negotiat- ing outcomes (in our study, JV performance outcomes).

Regarding cultural causes (what we call cultural fit), at the beginning of negotiations two parties agree to cooper- ate based on their respective knowledge, procedures and

Journal of Asia Business Studies SPRING 200834 Journal of Asia Business Studies SPRING 2008 35

resources--initial characteristics that are frequently very dif- ferent for each partner. Mutual partner-ignorance of the de- tails of firm-specific cultural nuances ensues—partner-firms are uncertain of the endowments and practices of their part- ners. These cognitive gaps are dangerous because they are not necessarily perceived directly, and may introduce periods of costly mutual misunderstanding (Heiman and Nickerson 2004). Commonality of partners’ cultures is positively asso- ciated with performance. For Faure, performance comprises the successful outcome of the negotiating process, a worthy, but incomplete goal which neglects the potential richness of in-stream operational performance data. Operational per- formance data as well as managerial attitudes toward per- formance (after operating over time) are critical elements of JV performance. Does a successful negotiation portend a successful JV? Perhaps, but Faure does not shed light on this question. Our approach to the logic of cultural fit is, however, broadly compatible with Faure’s.

Luo et al (2001) explore the parent-JV relationship as it affects performance. They discuss the relationship between control and performance and investigate the potentially moderating influences of objective congruence, culture, par- ent competency and equity stake on that relationship. In a more expansive empirical effort, Luo and Park (2004), via an- alyzing 250 China-foreigner JVs, show a strongly dominant empirical link between high levels of partner-cooperation and performance in JVs. They find that perceived high-qual- ity cooperation among both partners tends to occur only in the presence of partner-goal congruence. We interpret their work as using rough controls for cultural and organizational fit factors while focusing on strategic fit (in the form of goal congruence) and the implications for performance. Their re- sults for cultural and organizational fit factors vary depend- ing on which partner is assessing the JV, suggesting some in- conclusiveness regarding findings. Their work, in some ways, most closely echoes our approach, but with some important differences. For example, our dataset is composed solely of China-US JVs. Although they investigate strategic, cultural and organizational determinants of performance, the two above-noted works focus on the impact of these three factors on the relationship between control and performance, and the relationship between cooperation and performance. Luo and Park’s major contribution, from our perspective, intro- duces a logic wherein all fit factors may not be necessary for a JV to achieve high performance. We agree with this approach and specifically test for its veracity. In this paper, however, we discuss the direct impact of the above fit-related factors on JV performance, rather than their role as moderators between control and performance.

Buchel (2002:199), in a longitudinal case study of one joint venture, considers the effects of “driving forces” on JV “diver- gence and convergence in group relations.” In the formation stage, the strategic fit factor of goal congruence matters most. In the subsequent adjustment stage, the cultural fit element of managerial communication is the prime factor determin- ing performance. During the evaluation stage, renegotiation

may result remediations to existing agreements regarding ownership and distribution of rents—this reflects action in the realm of organizational fit. Buchel usefully identifies stages where certain types of fit matter most depending on the stage. Our work looks at “mature” stage JVs, all of which have operated for over two years. In Buchel’s terms, we scruti- nize either the adjustment or evaluation phase. We question Buchel’s implicit view that the primacy of partners’ goal con- gruence is expected to fade over time in high performing JVs.

JV performance. The above section has mentioned numer- ous studies involving fit-related factors, each of which mea- sures JV performance differently. How to evaluate perfor- mance remains a controversial topic (Calantone and Zhao 2001). The measurement of joint venture performance can be usefully divided into four categories (Probst and Buchel 1997): economic, strategic, behavioral, and learning. The eco- nomic perspective focuses on how to add value to the parent company. For example, some researchers use financial per- formance ratios such as ROI, ROE, or profits as indicators of performance (Calantone and Zhao 2001; Li et al 2001). The strategic perspective emphasizes non-financial, but nonethe- less important outcomes. In contrast to an economic view, it recognizes long-term interests, mainly focusing on the company’s size, product, market share, and competition. We view JV longevity or probability of dissolution, respectively per Barkema et al (1996) and Park and Ungson (1997; 2001), as also related to the strategic view. Park and Ungson (1997; 2001) treat the premature dissolution of JVs as an indicator of poor performance. They note the deleterious effects of oppor- tunistic threats, cultural differences, and competitive rivalry in determining the premature dissolution of JVs. Barkema et al (1996) treat JV longevity as a key performance indicator. Additionally, the strategic view concerns itself with a ven- ture’s core competence(s), and the possibility of strategic malfeasance (wherein the contractual arrangement between partners is breached by guileful, self-interested, opportunis- tic actors—see Williamson, 1985; Oxley, 1997). The behavioral perspective concentrates on the likely actions of partners and prefers to scrutinize antecedents and processes in JVs rather than outcomes. Behavioral-view oriented indicators of joint venture performance include trust between partners, transparency of the joint venture, division of responsibility, the ability to deal with crisis, participant satisfaction, and the ability to hedge risk. The learning perspective is substantially different from the other three, more conventional, viewpoints. The learning view focuses on outcomes, but also pays atten- tion to the process elements of activities. Typical empirical measurements for this perspective include the achievement of mastery over technology, marketing, partnerships, and management skills. Probst and Buchel’s (1997) taxonomy of JV performance is useful because it allows us to think about performance outcomes in a systematic manner.

Other indicators of JV performance. Another popular indi- cator of JV pe rformance is parent firms’ overall satisfaction with the joint venture and the likelihood of employee reten- tion in the JV (Isobe et al 2000; Luo and Park 2004; Lasserre

1999). Other researchers evaluate JV performance by looking at the effectiveness with which both partner-firms achieve their alliance objectives (Sirmon and Lane 2004). Our work, in Probst and Buchel’s (1997) terms utilizes one financial, one economic and two behavioral measures of performance in or- der to create opportunities to measure effects of our fit-related factors on a variety of reasonable performance indicators.

This section has recounted several notable works wherein all three of the fit factors in which we are interested are treat- ed. We have also touched on the multi-faceted nature of JV performance. Generally, existing work is of laudable quality, but suffers from a preponderance of case studies, and analy- ses of negotiation-stage JVs. The above-cited approaches, al- though rich in dynamic detail, may lack generality of find- ings; our analysis, while admittedly of primary relevance for China-U.S. joint ventures, employs inferential statistics on a geographically diverse sample and, uses original, reliable survey-based measures while addressing issues of under- specified models or anecdotal empirical approaches.

A PARTICULARLy RELEVANT PRIOR STUDy ON ASIA-WEST JVS

Lasserre (1999) identifies several factors associated with the perception of satisfactory joint venture performance from western managers’ views. Lasserre’s study solely treats western manager satisfaction across JVs in multiple Asian countries as a measure of performance (a behavioral mea- sure). Western partners show satisfaction if joint ventures are instrumental in achieving the partner’s objectives and if costs of cooperation are minimized. The author also explores the impact of strategic and cultural fit on satisfaction—both factors are positively associated with satisfaction. Strategic fit exists when the western partner perceives that both part- ners share a common view about the objectives of a venture. A perceived short-term, extractive orientation by an Asian partner has a very strong negative influence on the expressed satisfaction of Western managers.

Western partner satisfaction is chosen as a measure of joint venture success because the author states that the “…percep- tion of satisfaction or dissatisfaction of the people involved in JV management reflects an important dimension of joint venture performance independent of whether the joint ven- ture is financially or commercially successful” (Lasserre 1999:3). We find this thinking cogent, and share Lasserre’s view of perceived performance, though we question whether the performance of any JV is necessarily well-measured by the perceptions of only one partner. Much research, includ- ing ours, suffers to some degree from similar drawbacks. Ad- mittedly, the article states that it’s goal is to measure western partner satisfaction without claiming to measure overall JV performance. Our approach improves on Lasserre’s by intro- ducing more diverse measures of performance and by con- trolling, as well as the data permit, for the difference between participants originating from either Chinese or American partner-firms versus participants hired independently by the JV management team. Note that our sample includes

some “Chinese” managers who possess multiple years of ex- perience in U.S. firms, prior to joining the U.S.-Chinese JV un- der study. Some of these managers are U.S. citizens, and have lived in, worked in, and acculturated to the U.S. for years. This renders the line between “Chinese” and “American” managers substantially less clear. Owing to a growing, fluid job market between the U.S. and China, there is increasing lack of clarity regarding the characteristics of a modern “Chinese” manager as contrasted to an “American” manager. This may engender diminished or no detection of differences in perceived perfor- mance owing to partner-firm’s “national” origin. We mention this issue again, briefly, in Section 3, Method and Data.

CONCEPTUALIZATION AND DEFINITIONS OF FIT Having described the treatment of fit in the literature, we

now offer brief definitions of the three dimensions of fit. The underlying concept behind our approach to defining fit is to recognize and embrace the diversity of fit. We allow for the existence of multiple, legitimate types of fit. Our defini- tions have the benefits of (1) being broadly compatible with the literature’s views of fit, (2) possessing intrinsic empirical tractability, and (3) recognizing multiple perspectives on fit. One potential drawback to our diversity approach is the pos- sibility of a lack of orthogonality between measures when looking at multiple types of fit in one model. We address this issue in the empirical analysis. To recognize the diversity of definitions of fit and retain simplicity, we choose to bifurcate each dimension of fit into two intendedly orthogonal parts. Our pragmatic approach to defining fit recognizes that fit is in the eye of the beholder, and allows us to construct the first complete model incorporating the effects of multiple con- ceptions of fit on JV performance.

Strategic fit. Fit may reflect perceived degree of agreement or commonality between partner-firms, as in the case of ob- jective congruence between JV partners. For example, when partners both intend, over time, to profitably manufacture and sell automobile parts in export markets (outside China) we interpret this as a high degree of objective congruence. We also recognize, however, that in addition to agreement, har- mony, or other “sameness-based” definitions, fit may entail “efficient differences” or complementarities between partner- firms, as in the case of resource complementarity. Under dy- adic resource complementarity, one partner-firm, Partner1, is strong in, say, Resource1 (one of two critical resources needed for a product to be manufactured and sold), while Partner2 is weak in Resource1. Additionally and necessarily, Partner1 is weak in Resource2, in which Partner2 is strong. For example, if one partner provides low-cost manufacturing resources while the other provides marketing and distribution savvy, and each is weak in the arena of the other’s strength, we in- terpret this as resource complementarity: a degree of fit exists between partners—fit based on differences between partners. Our two dimensions of strategic fit are based on sameness (objective congruence) and differences (resource comple- mentarity) among partners.

Cultural fit. We split cultural fit into two categories, one

Journal of Asia Business Studies SPRING 200836 Journal of Asia Business Studies SPRING 2008 37

conventional, the other unconventional. The first, based on Hofstede’s (1980) work reflects a prevailing view of cultur- al fit: cultural similarity facilitates inter-partner relations and minimizes communication errors, leading to increased performance. The cultural similarity approach implied by Hofstede’s work represents the conventional view of culture in firms—it has the advantages of being both comparative and measurement-oriented.

Our alternative approach to cultural fit is less convention- al: the nature and quality of interactions between managers. Frequency of interactions, use of high bandwidth commu- nications and investment in co-specialized language assets (building shared meaning) are signs of high-commitment, high-quality interactions between managers. Increasing quality of interaction processes between partner-firms’ man- agers impacts performance in JVs.

Organizational Fit. The formal division of ownership be- tween partner-firms impacts JV performance because the allocation of ultimate control in a JV is determined by share- holding structure. The presence or absence of a dominant partner is expected to have a different impact on performance than a structure where partner-firms possess equal-shares of JV stock. Formal decision power, based on partner-firm shareholding, in addition to determining the dominant part- ner-firm, reflects JV design decisions regarding the nature of decision making in the JV. The direction of the effect of this type of organizational fit on performance is uncertain—a dominant partner might make mostly good or bad decisions, or might choose not to assert control at all. Both the nature of decisions and the nature of structure-induced decision pro- cesses (share ownership) affect performance. The structura- tion of ownership may dominate the game of control: equally shared ownership may stymie good decisions or any action at all, owing to the possibility of deadlock.

The second aspect to organizational fit is the extent of harmony among members of a JV’s management team—we see this as a form of “informal” organizational fit versus the formality of ownership structure. Managerial control within a JV, in contrast to formal ownership, often occurs without regard for ownership of shares, but rather for the purposes of maximizing performance and minimizing costs—the in- tention is to place people into positions in which they can perform at a high level. The resultant question that arises is “do these people work well as a management team in the JV?” The extent to which JV management personnel work well to- gether is expected to affect performance.

Defining JV performance. Having briefly defined multiple forms of fit, our independent variable of interest, we offer definitions of JV performance, the outcome of interest in our study. Each definition of performance answers a differ- ent question: (1) What is the degree of managers’ satisfaction with JV performance? (2) What do past financial perfor- mance measures tell us? (3) How does the JV compare to its competitors? and (4) Do managers generally make good deci- sions in the JV? The first and fourth performance definitions are largely subjective appraisals of JV performance, while the

second and third definitions are less so. Like the definitions of our causal factors above, our definitions of performance benefit from diversity, tractability and compatibility with the literature.

The literature-inspired definitions of fit-related factors af- fecting JV performance represent part of our conceptual con- tribution, which is enhanced by the recognition of a diverse set of factors via the bifurcation of each main effect into two intendedly orthogonal factors. Our conceptual setup allows for the testing of existing theory regarding fit in a uniquely complete manner. In particular, testing multiple factors in one fully specified, well-controlled model allows us to de- termine which, if any, of strategic, cultural or organizational fit is most critical to JV performance. Disagreement as to the importance of individual fit-dimensions exists. Also, claims of the necessity of concurrently achieving multiple types of fit are not well-supported by evidence. Our work takes a major step towards resolving these questions. We explore the veracity of previous incomplete work asserting the need for fit across many concurrent dimensions. The managerial contribution of the paper is to offer direction for managers as to which aspects of fit matter most for performance. Cer- tain dimensions may tolerate a lack of fit without adversely affecting performance. Knowledge of which dimensions are most closely associated with increased JV performance has the potential to shape managerial thinking regarding where to concentrate attention, a precious commodity. We provide the first and only completely-specified, well-controlled, reli- ably measured empirical model of the effects of JV partner-fit on performance. The next section offers hypotheses—two for each fit-category—asserting the effects of empirically tracta- ble proxies for the above-defined factors on JV performance.

HyPOTHESES REGARDING STRATEGIC, CULTURAL AND ORGANIZATIONAL FIT IN JVS

Strategic fit I: Congruence of partner-objectives. Many re- searchers maintain that congruence of objectives is one of the basic determinants of joint venture success (Inkpen and Cur- rall 1998; Tomlinson and Thompson 1977; Tung1984). Simply, congruent objectives mean that partners seek the same ends. As an example of lack of objective congruence, consider one author’s experience with a JV between a Dutch chemical firm and a Chinese chemical firm (Heiman and Nickerson 2004), wherein the Dutch partner sought to build operating exper- tise in China and to run a processing plant for profit (from ex- port initially, to be followed by sales to the domestic market later). The Chinese partner, on the other hand, was primarily interested in draining cash from the JV with little regard for profitability or the need to operate profitably in the future. i

The JV was ultimately designated a failure. Neither partner’s goals were met owing to the divergent nature of each part- ner-firm’s objectives.

Congruent objectives are seen as minimizing transaction costs and agency costs, while the incongruence of objectives contributes to conflict between managers and also affects the stability of collaborative relationships (Buchel 2002). With

different goals, misaligned incentives may arise, resulting in opportunistic behavior (Williamson 1983; 1991) and subse- quent venture failure. Personnel from each parent-firm may pursue their partner-firm’s interests instead of mutual part- ner-interests, leading to alliance failure (Park and Ungson 2001). As congruence of partner-firms’ objectives increases, joint venture performance increases.

H(1): Ceteris paribus, congruence of partner-firms’ objec- tives is positively associated with joint venture per- formance.

Strategic fit II: Complementary resources. Since it is compar- atively more costly to build relationships in foreign coun- tries than domestically, parent firms invest overseas when resources are unavailable at home (Chen et al 2004; Dunning 1980; 1988). ii Scholars maintain that resource complemen- tarities are important because jointly utilizing complemen- tary resources generates higher returns than using these re- sources separately (Chi 1994).

Researchers have also pointed out, however, that an alli- ance is less stable if an asymmetric contribution situation exists and creates an effectively unilateral dependence which leads to asymmetric bargaining power between part- ners (predicted by Williamson 1983; 1991; empirically sup- ported by Park and Ungson 1997; Hamel 1991). Our view of complementarities also recognizes the issue of asymmetric contributions by a partner. We acknowledge the role of mu- tual hostages/dependency (Williamson 1983), or the need for “balance” in partner-contribution to the ability to gener- ate rents in a JV (Teece 1977; 1980; 1981). The operative logic is economizing in its thrust (Williamson 1975)—resource access to foreign productive assets, when granted under properly safeguarded governance, economizes on both the costs of production and governance—mutual benefits en- sue (Williamson 1985; 1991). The presence of complemen- tary resources in a JV might include, for example, distinct unique bodies of knowledge possessed by each respective partner. The degree to which partner-specific knowledge- resources are intangible (difficult to share) is positively associated with increasing need to safeguard a JV against malfeasant behavior by a partner-firm, particularly if the intangibility is not present to a similar degree in partners’ knowledge-resources (Williamson 1985; 1991; Oxley 1997). Complementary resources may, of course be more tangible, and for example, also include one partner with manufactur- ing capabilities and another with distribution capabilities. Both tangible and intangible resource complementarity fo- ment increases in partner incentives to behave owing to the existence of mutual hostages. In this paper, we assert that owing to alignment of incentives to perform from mutual hostages, complementary partner-resources have a positive relationship with JV performance.

H(2): Ceteris paribus, increasing resource complementari- ties between partner-firms in a JV are positively relat- ed to the venture’s performance.

Culture Fit I: Culture similarity. A common belief articu- lated in the literature states that the greater the cultural distance between the parent firms, the more difference in their managerial and organizational practices. Low levels of cultural similarity is purported to increase coordination costs because of the cultural adjustment involved in a joint venture (Kogut and Singh 1988; Barkema et al 1996; Li et al 2001).

Some theorists (Sirmon and Lane 2004) divide culture into three categories: national, organizational, and profes- sional. They believe that the closer the domain of a social group is to the value-creating activities of an alliance, the more disruptive the culture difference between the part- ners’ members in that social group becomes. Differences in professional culture will be most disruptive, followed by or- ganizational culture differences and national culture differ- ences. Though we treat only the organizational and national culture fit components described by Sirmon and Lane, we recognize the value of their multi-level thinking.

Barkema, et al (1996), use an organizational learning ap- proach, treat JV-longevity as the performance measure of interest, and show that cultural distance plays an impor- tant role in determining JV-longevity. Their work covers 13 Dutch firms engaged in 225 JVs. Though we do not look at longevity as a dependent (performance) variable, Barkema et al’s thinking is largely compatible with ours. Hypothesis 3 expresses our assertion regarding the relationship between organization culture similarity (fit) between partners and the performance of a JV:

H(3): Ceteris paribus, increasing cultural similarity be- tween partner-firms in a JV is positively related to a joint venture’s performance.

Culture Fit II: Managerial communication between partners. We recognize that there are alternative ways to conceive of cultural fit. Specifically, managers in multi-cultural joint ventures frequently mention issues arising from poor com- munications owing to culture and language barriers. Lack of cultural or linguistic competency leads to a lack of under- standing of assignments and poor communications (Oxley 1997; Heiman and Nickerson 2004). These conditions incur costs for remediation. Each partner-firm tends to base its managerial interaction style on its unique past alliance/JV experiences as well as its prior unique internal interactions. In the worst-case, individuals in each partner-firm may in- correctly interpret certain events or statements by JV team- members from the other partner-firm; distrust or conflict between partner-firms may arise (Buchel 2002). The lack of efficacious communication between participants in a JV re- sults in increased coordination costs (e.g., the cost of creating a common vocabulary of terms for increased communica- tion efficacy), and may negatively impact JV performance.

H(4): Ceteris paribus, increasing efficacy of managerial communication between partners is positively associ- ated with a joint venture’s performance.

Journal of Asia Business Studies SPRING 200838 Journal of Asia Business Studies SPRING 2008 39

Organizational fit I: Equity structure. Frequently an exam- ple of fit by design, equity has been typically interpreted as a measure of control and dominance in a joint venture (Blodgett 1992). The division of control is thought to be a critical determinant of joint ventures’ stability (Geringer and Hebert 1989). Some contrary views regarding division of con- trol maintain that JVs are more stable when partners begin with a 50/50 shared control structure owing to high levels of trust required to make decisions mutually (a form of the above-mentioned mutual hostages argument). Unequal di- vision of ownership gives a majority shareholder the power to dictate terms. This power may manifest itself in the fre- quent opportunistic unilateral renegotiation of joint venture contracts/terms and subsequent dissatisfaction by the non- dominant partner-firm. Blodgett (1992) sees this syndrome as an indicator of instability in JVs. Predominant evidence from the literature, however, shows the opposite to be most likely true within limits. When a dominant partner exists in a JV, stability increases. When one partner has majority con- trol, decision-making is easier and less time consuming (Kill- ing 1983; Geringer and Hebert 1989) and a deadlocked vote by share ownership is not possible. Right or wrong, decision- processes are efficiently facilitated. When an equal share- control structure (50/50) is used, however, partners may experience coordination (decision making) problems and related increased costs, reducing the value captured by a JV (Geringer and Hebert 1989). There remain, nonetheless, oth- ers who failed to show a significant positive relationship be- tween imbalance of ownership control and JV performance (Steensma and Lyles 2000). Overall, prior studies shed little light on the relationship between equity structure and joint ventures’ performance. We side with the predominant per- spective, compelled by the “efficient facilitation-of-decisions” argument mentioned above. We define a dominant partner as having the ability to exert exclusive control over a JV.

H(5): Ceteris paribus, the existence of a dominant (equity- based) partner-firm in a JV is positively associated with performance of the JV.

Organizational fit II: Top management team decision making process. In addition to the importance of the formal division of control within a JV, we assert that a high degree of mutual- ly agreed-upon decision-making within a JV’s management team is likely to lead to high performance. When studying joint venture performance, some researchers have highlighted the impact of cultural distance or control structure, but lim- ited research has dealt with another pivotal factor: the extent of harmonious decision making in a JV. A JV’s management team typically consists of five to ten managers. Each parent firm designates a team, each of which often maintains strong allegiance to their respective parents (Li et al 1999). Each JV’s team has management responsibility for JV operations, but individuals from particular parent-firms also act as delegates of their respective parent-firms (Hambrick et al 2001). Each (self-interested) parent would like to exert more control over the JV than the other. In a given JV, since top managers in some critical positions have more control, parent companies often

prefer their own people to occupy these positions (Li et al 1999). For example, our data shows that some China-US Joint ventures create two similar positions executing largely the same function, with identical or slightly different titles. This is one possible manifestation of partner-firms’ desire for con- trol by their own managers. Role ambiguity may arise, which harms a joint venture’s performance by increasing coordina- tion costs. The reliability of communications (particularly managerial fiat) decreases—unnecessary inaccuracy and re- dundancy in communication gives rise to increased remedial costs. Some researchers also assert that increasing imbalance in management control leads to a higher level of parental conflict and a high likelihood of alliance failure (Steensma and Lyles 2000). We assert that mutual agreement on stra- tegic matters within top management is associated with increased performance, while one-sided control (authority- based decision making—see Nickerson and Zenger 2004) is associated with comparatively inferior performance:

H(6): Ceteris paribus, within a JV’s top management team, partner-firms’ representatives consistent mutual agree- ment in decision making is positively associated with JV performance.

Note that Hypothesis 6 deals with control of the actual, or informal decision making structures in the JV, while Hypoth- esis 5 explores the effects of the formal (legal) contractual dis- tribution of control (equity shares). iii

METHOD AND DATA Our empirical approach to understanding the role of three

distinct types of fit in determining JV performance employs a survey instrument, with which we gathered data from over 100 Chinese JVs during the period 2003-2005. Our data came primarily from four loosely designated regions, Beijing, Shanghai, Guangdong and Tianjin. In total, data was obtained from 112 China-US JVs distributed roughly evenly by region. A few JVs from outside the above regions were also included in the survey data. In each region, the first step was to obtain a list of registered joint ventures from local governments. We then filtered the lists to include only (1) China-US JVs with over 100 employees, and (2) JVs that have existed for at least two years--this ensured that managers would have had suf- ficient time to form reasonably accurate beliefs about a given JV’s performance. Subjecting the government-provided lists of JVs to our inclusion criteria resulted in an original dataset of about 2000 JVs being reduced by over 80% to about 400. Of these China-US JVs, 112 managers agreed to be interviewed by our team who personally administered the survey instru- ment in order to offer clarification as needed to respondents. This constitutes an overall 28% response rate. Our team’s in- terviewers are comprised of a Chinese university professor and several PhD students.

Extensive discussions were held within the China-based research team regarding the survey questions and how to consistently record responses (discussions after a small ini-

tial round of pilot interviews were particularly useful). Sub- sequent to pilot interviews, minor editorial revisions to the survey instrument were made in response to problems with question comprehension and wording uncovered by the pilot effort. All interviews and surveys were conducted in Chinese. The original survey instrument was subsequently translated into English for data analysis and reporting. Since the origi- nal survey was written in Chinese and the intention of the re- search team is to publish in English, back-translation to Eng- lish from Chinese (in order to correct the Chinese version) was performed as follows: (1) the original Chinese survey was translated to English by the China-based portion of the re- search team (the primary survey authors); (2) Team members in the U.S. (graduate students previously unacquainted with our project, literate in both Chinese and English) also trans- lated the original Chinese into English; (3) the two English versions were compared, differences discussed and resolved, and minor changes were made to the Chinese version. The only (minor) translation issue that arose regarding the col- lected data was the need to understand the names and loca- tions of cities with JV sites in order to classify each JV prop- erly by Chinese region.

The respondents were either hired independently by the JV (i.e., not from a parent-firm in the JV), or came from a par- ent-firm in the JV. In some cases, respondents were Chinese people with substantial experience at western (American) firms who came from these firms to work at the JV. Some of the Chinese participants in the JV had previously worked at American firms in the US, including at the American par- ent. The sample is diverse, and we caution the reader that owing to the effects globalization on labor mobility, the line between “Chinese” and “American” respondents is no lon- ger as clear at it once was. We interpret this development as possibly mitigating the issue of potential bias based on a respondent’s “origin.”

Using the dataset described above, we develop constructs for independent and dependent variables. We recognize, however, potential issues with common-methods bias in measuring our independent and dependent variables, but elect to treat this as a minor methodological disadvantage. Doty and Glick’s work (1998:374) supports this approach. Having empirically detected and the distinct effects of com- mon methods bias, they note that while common methods bias exists, “…[the] level of [detected] bias is cause for concern but does not invalidate many research findings.” The survey data is rich in that there are many items that can be usefully combined to create reliable measures of several of the ele- ments of fit and performance. In many instances, we perform useful combinations and transformations on multiple sur- vey items in order to create meaningful composite measures. Multiple composite measures are then combined into a use- able (or not) construct depending on the results of tests using Cronbach’s Alpha (Alpha ≥ 0.7 is our standard for construct reliability). The resulting reliable constructs are then used in an analysis employing ordinary least squares (multiple) re-

gression. Below, we discuss the construction of variables and provide summary descriptive statistics of our constructs as well as a correlation matrix.

CONSTRUCTION OF VARIABLES JV PERFORMANCE— MEASURING THE DEPENDENT VARIABLE

Our analysis recognizes the diversity of multiple measures of JV performance by using four distinct JV performance measures. We treat dependent variables as reflecting un- derlying continuous factors. Following Probst and Bucher’s (1997) taxonomy, one of our dependent variable constructs is a conventional economic measure (perceived financial performance trends). Two performance measures are slightly unconventional behavioral measures (perceived managerial satisfaction with the quality of inter-firm cooperation and perceived quality of JV decision-making. One performance measure is strategic in nature: competitiveness versus other industry players (not necessarily JVs).

For some measures (where feasible), we specify that perfor- mance is scored as high only when respondents indicate that participants from both partner-firms agree about perceived high performance. For example, the performance measure satisfactperform is comprised of the mean of six distinct items from the survey, some of which inquire regarding per- ceptions of U.S. partner satisfaction and some of which mea- sure perceived Chinese partner satisfaction. All constructs are comprised of non-standardized survey items. When a calculated Cronbach’s Alpha is deemed sufficiently high to suggest a reliable construct, the construct, unless otherwise noted, is calculated by averaging the relevant test items, but adjusting case-by-case as needed to account for missing data in the calculation of the mean. This prevents distortion of the construct’s value for cases with missing data while allowing use of as much of the available data as possible.

Dependent Variable I: Perceived satisfaction with JV perfor- mance. Returning to our first performance construct, sat- isfactperform , in two survey items, we ask respondents to indicate their perceptions of each partner-firm’s satisfaction with the cooperation between partner-firms in the JV. That is, in separate items, the respondent indicates his/her per- ception of the satisfaction level of the Chinese partner-firm and then indicates his/her perception of the satisfaction level of the American partner-firm. One respondent per JV was surveyed. The respondent originated from the Chinese or American partner-firm, or was hired independently by the JV. iv We create a composite variable that is set equal to 1 only if respondents perceive that both partners are satisfied with the cooperation that occurred in the JV, and is set equal to 0 otherwise. This comprises a conservative approach to recog- nizing high performance. Where relevant, and when the data permit, several other constructs discussed are built in a simi- lar fashion. The perception of a high degree of satisfaction on the part of both partner-firms regarding the achievement of high performance is required to code a variable as reflecting

Journal of Asia Business Studies SPRING 200840 Journal of Asia Business Studies SPRING 2008 41

perceived high performance in the JV. We construct a similar measure from two additional survey items to indicate part- ner-firm satisfaction with JV achievements to date. In addi- tion to these two composite measures, we use more mundane measures of satisfaction with performance in developing the construct. One question directly asks if respondents are sat- isfied with the cooperation in the JV to date. The response is measured on a 7-point Likert-type scale with a range of 1-7, where 4 is neutral, 1 is completely unsatisfied and 7 is com- pletely satisfied. Another question asks if respondents think the JV is successful, and is scored as for the previous ques- tion. Two more items inquire about respondents’ impressions of top managers’ satisfaction with sales and profits for the most recent year, and are also scored on a 7-point scale. The construct, satisfactperform, scored a Cronbach’s Alpha=0.82 when all six items were considered, considerably above our standard of 0.7 for inclusion in the analysis. Higher values of the construct satisfactperform reliably indicate increas- ing satisfaction with JV performance as perceived by the respondent.

Dependent Variable II: Perceived prior financial performance trends. The second performance measure looks at perceived past trends in the JV’s financial performance. The construct, trendperform, is comprised of six measures that inquire re- garding respondents’ perceptions of the most recent main trend in the JV’s sales, profits, ROI, ROA, and ROS as well as the JV’s overall achievements over the past three years.v In many (though not all) cases, respondents were able to look up or calculate the actual performance figures, rather than having to estimate, thereby increasing the accuracy of re- sponses. The construct, when Cronbach’s Alpha is calculat- ed, has an Alpha=0.97, and is deemed reliable. Higher values of the trendperform construct indicate a greater degree of perceived prior high financial performance trends.

Dependent Variable III: Perceived overall competitiveness. Our third performance measure, a construct labeled over- allcompetitiv measures the perceived overall competitive- ness of the JV. The construct combines seven measures of respondents’ perceived positions versus their competitors. The following aspects of the JV versus its competition are rated on a 7-point scale where 1 is much weaker, 4 is equal (neutral) and 7 is much stronger: management skill, produc- tion technology, production equipment quality, patent port- folio, product quality, distribution network, and quality of employees. When the items are combined into a construct, the Cronbach’s Alpha=0.88, indicating reliability—we are re- liably measuring the perceived overall competitiveness of a JV versus its competition. We note, however, the remarkable agreement of performance measures across various areas of competition is ostensibly somewhat anomalous. This might be explained by the relative homogeneity of JVs’ value chains associated with a particular constant form of governance: an equity JV. That is, successful JVs excel in a consistent set of areas versus their competition. We acknowledge, however, that a less monotonous competitive landscape may in fact exist—we may have simply failed to measure it.

Dependent Variable IV: Quality of the decision making structure. The fourth performance measure is a construct labeled deci- sionperform, which asks “what is the quality of the decision- making in the JV?” We ask about the perceived influence of the existing decision-making structure on JV performance using 11 survey questions. The measure is intended to capture the aggregate performance of the JV’s managerial decision structure by combining rated performance across different areas of decision making. The data makes available detailed information regarding the perceived efficacy of decision making in multiple areas in the firm. This original construct measures behavioral aspects of both partner-firms’ perfor- mance as perceived by respondents. It is somewhat similar in character to managers’ perceived satisfaction with a JV’s inter-firm cooperation and achievements (satisfactperform) which is also comprised of multiple measures that vary by area of JV function.

The construct decisionperform consists of eleven survey items pertaining to the influence of decisions by area on JV performance. The following areas of decision-making com- prise the construct: incentive/salary policy, appointment of middle managers, distribution of dividends, financial con- trol, marketing strategy, purchasing plans, corporate strategy, quality control, training, R&D investment, and production plans. Respondents were asked to estimate the contribution to performance of decision making in a particular area, where a score of 1 represents “decisively lowered JV performance,” 4 is neutral (no effect on performance), and 7 represents “de- cisively increased the performance of the JV.” The construct comprising all eleven items reliably reflects an overall mea- sure of perceived decision making performance in the JV (Alpha=0.93, implying a highly reliable measure). Overall decision making performance seems a reasonable construct for use as our fourth of four dependent variables.

In order to validate the performance measures, we consult- ed with several academic experts as well as several long-time practitioners with experience in U.S.-China JVs. Among our experts, there was general agreement about the validity of our proposed measures. This process suggests that our pro- posed measures of JV performance possess face validity.

INDEPENDENT VARIABLES Strategic Fit I: Objective Congruence. The first indepen-

dent variable, objectivecongruence, reflects the similarity of partners’ objectives. It is a construct comprised of six survey items, four of which are rendered into two composite mea- sures, and two of which are incorporated as individual items, in all, consisting of four total measures subject to Cronbach’s Alpha calculation. The two composite items are constructed in an analogous manner to some performance measures in that we require perceived partner-firm agreement across the items regarding objective congruence of partner-firms. The first composite measure combines two survey items that ask about the type of benefits each partner seeks: long-term ver- sus short-term. From these two items we construct a conser- vative composite measure that is set equal to one only if the

respondent indicates that both partners are not interested primarily in short term benefits—we interpret one partner- firm engaging in short-term thinking as not reflecting goal congruence, as well as both partners thinking short-term (which might, for example, precede untimely dissolution of the JV by one partner or other unilateral opportunistic be- havior). When both partners are interested in long-term ben- efits, the value of the measure is set to one, and zero other- wise. The second composite item asks whether each partner is only interested in quick profits and reluctant to reinvest in the joint venture. When both the Chinese and Ameri- can partners are perceived as not interested in quick profits and not reluctant to reinvest in the JV, this is coded as a one and zero otherwise. The remaining two survey items com- prising this construct are 7-point Likert-type “agreement” scales (where 1 is completely disagree 4 is neutral and 7 is completely agree). They make the following two statements to respondents: “The JV partners have a shared strategic vi- sion and goal for the JV,” and “The establishment of the JV is compatible with the investment and corporate strategies of both parties.” The Cronbach’s Alpha for this construct is 0.65, a number close to, but not quite, meeting our stated criterion for acceptance of 0.7. We nonetheless deem the level of this original construct’s Alpha to be acceptable (while admitting that this is partly owing to the fact that we have no suitable alternatives), and we use this construct as representative of objective congruence in our analysis. This original construct is used to test Hypothesis 1, regarding the increasing effects on performance of partners’ increasing objective congruence. A caveat regarding objective congruence is in order. When measuring the extent to which a respondent perceives two parties believe in the same goals, there is a risk of desirability bias, wherein the response reflects a desire to offer socially ac- ceptable responses (Fisher, 1993). An examination of the data shows objectivecongruence has a range of 1.25-6.0, a mean of almost exactly 3, and a standard deviation of approximately 0.7. These properties of the data suggest that desirability bias may not be an issue. There exists substantial variation in the data and a central tendency that is not skewed towards desir- ability (high degree of objective congruence). The data sug- gest this measure is most likely valid for the purposes of this paper.

Strategic Fit II: Resource Complementarity. The construct la- beled resourcecomplement measures the extent to which one partner’s lack of resources is compensated for by the other partner. The construct is comprised of nineteen composite measures, each of which is derived from two survey items. Unlike for objectivecongruence and similarly constructed measures, we are looking for complementarity: multiple ar- eas where one partner is high while the other is low indicate a high degree of resource complementarity. When this is the case the measure is assigned a value of 1, and 0 otherwise. The following areas of resource complementarity are measured in this construct: management skills, branding, logistics, distri- bution network, financial management, human resources management, production facility/land, public/government

relations, training (workers and managers--two items), un- specified distinctive resources, skill in managing resources, business acumen (i.e., one partner interested in learning about industrial processes versus the other in gaining man- agement know-how), expertise in exploration, production know-how, equipment contribution, R&D contribution, and marketing/sales contribution. When, for example, a respon- dent perceives that one partner is strong in financial man- agement, while the other is weak in the same area (based on two survey items), a composite measure is set equal to one, and set to zero otherwise. The resulting composite items are then analyzed for reliability using Cronbach’s Alpha.

The construction of the “business acumen” item above in- volves building one measure from two distinct survey items to look at the implicit resource-assumptions inherent in having a particular partner-firm-level objective.vi In this ap- proach, in one survey item the respondent indicates whether or not the objective of the American partner is to learn by gaining business experience in China—this assumes the Chi- nese partner has specific knowledge resources to share. In a separate item, if the respondent indicates that the objective of the Chinese partner is to acquire management know-how, the assumption being that the Chinese partner believes that the American partner has management know-how resources to share. This approach avers that partners perceive the ex- istence of complementary resources via an articulation of their objectives. That is, one partner-firm’s goals are set par- tially according to perceived valuable resources possessed by another partner-firm. We set the value of the measure to 1 if the partners have complementary implicit resource as- sumptions. When the Cronbach’s Alpha is calculated for the construct resourcecomplement, the value is 0.91, indicating a reliable construct measuring overall resource complementa- rity. This construct is used to test Hypothesis 2, wherein we expect to detect positive effects of degree of resource comple- mentarity on performance.

Culture Fit I: Culture similarity. We divide cultural similar- ity fit into two measurable categories, labeled nationalcultr and organizationcultr. The first measure, nationalcultr, is based on one survey item, which inquires directly about the clash of partner-firm’s national cultures. The second mea- sure, organizationcultr, is also comprised of one survey item which asks respondents about the clash of partner-firms’ organizational cultures. Each item inquires regarding the ex- tent to which differing national or corporate cultures are per- ceived to cause conflict in the JV. We reversed the coding of these items to create positive measures wherein the value of the measure increases with increasing cultural compatibil- ity. These factors are used to test Hypothesis 3, which avers that as cultural compatibility increases, JV performance in- creases.

Culture Fit II: Managerial communication effectiveness. Mana- gerialcommunic is a construct that measures the perceived effectiveness of communications between managers. It is comprised of 5 survey items. The items ask about the extent of understanding of and agreement regarding communica-

Journal of Asia Business Studies SPRING 200842 Journal of Asia Business Studies SPRING 2008 43

tion between managers. For example, using a scale of 1 (No understanding/cooperation) to 7 (complete understanding/ cooperation), two survey items ask about, respectively, under- standing of and intention to cooperate with JV resolutions (agreed by the JV’s Board of Directors). Increasing perceived cooperation with and comprehension of Board resolutions by managers is taken as positive evidence of effective mana- gerial communication. Another item inquires as to the extent to which the nature of managerial communications is formal (regular, year-end, and other routine meetings). We interpret increased use of routine meetings as reflecting comparatively more effective managerial communications. It is conceivable, however, to think of this measure as acting in the opposite sense to that proposed. That is, fewer meetings may reflect in- creasingly effective managerial communications in a JV. The effect on performance may be either positive or negative. For simplicity and consistency in this paper, we treat increased duration and numbers of scheduled, routine meetings as a sign that managers are communicating comparatively more effectively than under decreased duration and numbers of routine meetings (For a detailed justification of this ap- proach, see Heiman and Nickerson 2004)

The last two of five items comprising the construct mana- gerialcommunic are positively worded inquiries about com- munication effectiveness, and ask the respondent about the perceived effectiveness of problem solving for formal and in- formal communication (again using a 1-7 scale). The survey explains that formal communications include not only meet- ings as described above, but also written reports. Examples of informal communications are described as non-work related interactions between JV employees: coffee breaks, lunch, sports, extra-JV activities, socializing, and dining out. The five items described immediately above together yield a Cron- bach’s Alpha of 0.72. The construct managerialcommunic is deemed suitable for inclusion in our analysis as an indepen- dent variable reflecting the effectiveness of communication. Our managerial communications construct tests Hypothesis 4, which predicts positive effects on JV performance.

Organizational Fit I: Equity Structure. We aver that the pres- ence of a dominant partner (in terms of equity ownership) will increase performance of a JV because the decision mak- ing process will likely rarely, if ever, deadlock. Decisions will get made and we interpret this as better than the lack of forward progress associated with cycling in decision mak- ing processes. This measure, labeled dominantpartner, is set equal to 1 if either the Chinese or the American partner is identified as holding a majority (>50%) share of JV equity, and 0 otherwise. This construct is used to test Hypothesis 5, regarding the effects of a dominant partner (by equity owner- ship) on JV performance.

Organizational Fit II: Top Management Decision Making Effec- tiveness. We use a construct labeled agreehires to measure the extent to which top management agrees regarding the place- ment of top managers and board members into the JV from parent firms. Specifically, the extent of agreement regarding hiring the following mission-critical positions is measured:

Board Members, General Manager/CEO, CFO, VP Market- ing, VP Research & Development and VP Production. We use the extent of partner-agreement regarding top-management hiring decisions as evidence of high performance: a smooth- ly functioning, coordination cost-minimizing top admin- istration in the JV bodes well for performance. Harmonious decision-making is interpreted as effective decision making, admittedly an explicit assumption on our part. Given ex- tensive contention regarding placements of top managers, comparatively more managerial time, energy, and attention is consumed by the process of achieving agreement—oppor- tunity costs are incurred—managers are decreasingly able to properly fulfill their primary duties. Also, lingering resent- ment from prior disagreements might incur sub-optimal re- prisals in subsequent joint-decisions or, perhaps in the worse case involving voting for appointments, deadlock over suc- cessive ballots. We argue for the efficacy of mutually agreed- upon decisions (a positively-signed coefficient). A smoothly functioning decision making team incurs lower compara- tive costs versus a contention-driven decision making team. We interpret either a joint-parent-based appointment or an independent appointment by JV managers as evidence of harmonious decision-making in the affairs of a JV; decisions driven solely by one parent are viewed as potentially having comparatively greater negative consequences. Initially, for the ease of respondents, the survey coded distinct responses for the following situations: (a) the manager was hired by the American partner’s parent firm, (b) the manager was hired by the Chinese partner’s parent firm, and (c) the JV’s man- agers/Board were allowed to make the decision independent of parent involvement. Respondents were told to select any and all choice(s) that applied. Since we are interested in mea- suring the extent of agreement between partners, the survey items are recoded as follows: from the six possible response permutations agreehires is set equal to 1 only if both JV par- ents jointly decided on the person to hire, or the JV’ s manag- ers were given independent control over the person to hire. Otherwise (one partner-firm made the hire) the measure is set equal to 0. These six recoded survey items are then sub- jected to Cronbach’s Alpha analysis to determine the extent to which they reflect the level of overall agreement between partners regarding hiring in the JV. The Alpha value is 0.8, which suggests that the measures, as recoded, reliably reflect the underlying construct of agreement regarding hiring top managers. This construct is used to test Hypothesis 6, which asserts the positive effects of top management agreement in hiring decisions on JV performance.

CONTROL VARIABLES Age of the joint venture: Experience effects. Older JVs, ow-

ing to their greater operating experience and familiarity with their partner-firm, suppliers and customers, might be expected to outperform relatively younger, more” naïve” JVs. In order to control for this effect, we employ a measure com- prised of each JV’s age from inception. The youngest JV in our sample is 4 years old, while the oldest is 32 years old (at the

time of data gathering, the youngest JV in our sample was 2 years old). We label this control measure agejv.

Involvement of the Chinese State in a Dominant Partner. In order to control for possible positive effects of Chinese gov- ernment involvement and influence on China-US JV-perfor- mance, we construct the control variable, chinastateinvolvd, which is a dummy variable, set equal to 1 if the Chinese State directly owns shares in the Chinese JV and that partner is identified as the dominant partner in the JV. Otherwise, the variable is set to 0. We recognize state involvement as poten- tially occurring at any of local (city, town), regional (prov- ince) and national levels.

Balance of Control. This construct measures the extent to which the Chinese partner’s parent increasingly controls 6

areas of the JV as reflected by 6 survey items. The construct balanceofcontrol looks at respondents’ perceived (success- ful) assertion of authority by a partner over a particular area, regardless of equity stake held by one partner. The areas of the JV covered include each respondent’s perceptions of ac- tual control over formulation of quality control standards, dividend decisions, employee training processes, research & development investment, public relations, and proportion of mid-level managers hired by a particular parent. This factor measures perceived “social” control over decision making in multiple areas and controls for the effects of strong individual personalities or strong culture-specific assertions of control by participants in JVs, which is not part of fit as discussed herein. We assert the possibility of an effect, but not the sign of the effect. The Cronbach’s Alpha is high at 0.88, suggesting a reliable construct for subsequent analysis. This construct, in addition to being a control variable, may shed additional

Dependent Variables

satisfactperform Performance measure of the degree of satisfaction with the JV--combines six different measures of (perceived) satisfaction.

trendperform Assessment of JV’s past 3 years of financial performance via respondents estimation of six financial measures.

overallcompetitiv Respondent’s perceived overall competitive- ness of the venture vs. competition--combines seven different estimates of performance versus competition: management skill, production technology, production equipment quality, patent portfolio, product quality, distribution network, and quality of employees.

decisionperform Measures the extent to which the decision structure of top management is perceived to be efficient. Using eleven survey items it comprises an overall measure of perceived decision making performance in the JV.

Independent Variables

objectivecongruence Degree of the extent of the perception that a JV’s parent-firm-partners agree about the long-term goals and vision for the JV.

resourcecomplement Measures the number of areas in which one partner’s lack of resources is compensated for by the other partner.

organizationcultr Measures the extent to which the respondent perceives conflict or tension between partner-firms’ corporate cultures in the JV--comprised of one survey item.

nationalcultr Measures the extent to which the respondent perceives conflict or tension between partner-firms’ home-country cultures in the JV--comprised of one survey item.

managerialcommunic Measures perceived effectiveness of communication between managers--combines five different communication measures.

dominantpartner =1 when the Chinese or American partner owns more than 50% equity share (i.e., not equal shares of 50%)--there exists a dominant partner by equity ownership fraction.

agreehires measures the extent to which top manage- ment agrees regarding the placement of top managers and board members into the JV from parent firms

Table 1: Summary of construction of variables

Control Variables

agejv The age of the JV, in years, from its inception.

balanceofcontrol Reflects extent to which the Chinese partner’s parent increasingly controls 6 areas of the JV-- reflected by 6 survey items. Looks at respondents’ perceived (successful) assertion of authority by a partner by area, regardless of equity stake situation. Areas covered: formulation of quality control standard, dividend decisions, employee train- ing processes, research & development investment, public relations, and proportion of mid-level managers hired by a particular parent. More negative values indicate increasing perceived American control, while more positive numbers signal predominantly Chinese control.

balancecontrolsqrd Is an estimate of a parabolic effect, with an expected negatively signed coefficient by squaring balanceofcontrol. Controls for possibly diminishing returns to extremely one-sided balance of control in a JV.

numemplos Measures eight size categories of JVs from less than 50 employees to more than 1000 employees.

beijingregn =1 if the JV is located in the Beijing region.

guangdongregn =1 if the JV is located in the Guangdong region.

shanghairegn =1 if the JV is located in the Shanghai region.

tianjinregn =1 if the JV is located in the TianJin region.

jobposition The measure decreases from 3 to 1 as mana- gerial seniority of respondents increases.

yearsworked Total years worked at the JV.

education Increases with education level through 4 ordered categories: primary school, secondary school, bachelor’s degree and master’s degree or higher.

age Age of respondent in years.

Construction of Variables

Journal of Asia Business Studies SPRING 200844 Journal of Asia Business Studies SPRING 2008 45

light on the issues raised by Hypothesis 5 regarding the ef- fects of a dominant partner. We coded the composite measure to be increasingly negative for ascendant Chinese control and increasingly positive for ascendant American control. Zero or near-zero values indicate balanced control. If significant, when the sign of coefficient for this factor is more negative, ascendant American control positively affects performance, and as it becomes increasingly positive, increased Chinese control predicts higher performance. A near-zero, significant coefficient value might reflect the importance of balanced informal control—assertions of superior performance from unbalanced control would be cast into doubt.

Caveat: Diminishing Returns to highly asymmetric social con- trol. We are concerned that there may be noise from the ef- fects of diminishing returns to “over-balanced” social control by one partner. That is, one partner-firm or that firm’s JV team-members may feel so excessively excluded from deci- sion making processes to the extent that they elect to mini- mize their meaningful participation or even quit a JV—neg- atively affecting performance. To control for this possibility, we estimate a parabola with an expected negatively signed coefficient by introducing the square of the construct bal- anceofcontrol, a term we label balancecontrolsqrd. The two constructs are not problematically correlated at Pearson’s r = -0.29. We deem this “squared-term” for degree of asserted control to be suitable for inclusion as a control variable.

Industry Controls. From a framework for industry-level con- trols developed for strategic management research by Dess, et al (1990:21), our approach falls into the “Multiple Industry Controls” category. We recognize Dess, et al’s (1990) asser- tion of an important limitation to our approach: “[Subjective] judgment [is] involved in selection of ‘critical’ environmental dimensions (21).” In addition to the above-discussed controls, we examined a number of industry-level control variables, and none were significant in any models. We omit industry- level controls from our final models, presented below.vii

JV-level Control Variable: Number of Employees. We control for any effect on performance of increasing size of JVs with the variable numbremployees, which measures eight size cat- egories of JVs, coded 1 through 8, from less than 50 employees to more than 1000 employees.

Controls for regional effects. Does JV performance vary by region? The differential availability of resources (natural, labor, materials supply) as well as regional variation in po- litical and legal constraints within China suggest that con- trols for regional effects are appropriate. We employ dummy control variables for four regions. Controls are set equal to 1 if the JV is located in a particular region. The base condition (all dummies equal to 0) includes several JVs that lie outside the four regions which comprise the bulk of our data: Beijing, Shanghai, Guangdong and Tianjin.

Respondent Controls. The possibility exists that respon- dents’ individual characteristics might affect their perceived performance of JVs. We control for respondent-level char- acteristics via four measures. The first measure, jobposition, captures whether the respondent is a top manager, assistant

top-manager, or middle manager, and is an ordinal measure. The value of the measure decreases as managerial respon- sibility increases. The second respondent control is year- sworked, the number of years the respondent has worked for the JV. The third control for respondent attributes is educa- tion, which increases with education level through 4 ordered categories: primary school, secondary school, bachelor’s de- gree and master’s degree or higher. The fourth control is the age of the survey respondent (across 7 ordered categories), labeled age. A fifth respondent-level control, gender of the re- spondent, is omitted from the final models as there was never any hint of significance for this factor in any of the models we tested. Having described the construction of dependent, independent and control variables in some detail, the next section presents the summary statistics for our variables and discusses any associated issues. Table I summarizes the con- struction of all the variables discussed above.

DESCRIPTIVE STATISTICS Table II presents the summary statistics for our measures.

Table 2: Summary statistics for dependent, independent and control variables

Variable Obs Mean Std. Dev. Min Max

satisfactperform 112 3.714 0.670 1.333 5.000

trendperform 107 4.144 1.200 1.000 8.000

overalcompetitiv 110 5.198 0.864 2.857 7.000

decisionperform 112 5.189 0.827 1.636 7.000

objectivecongruence 112 3.020 0.734 1.250 6.000

resourcecomplement 110 0.220 0.276 -0.059 0.895

organizationcultr 107 5.981 1.421 2.000 7.000

nationalcultr 107 5.879 1.478 2.000 7.000

managerialcommunic 112 5.171 0.775 3.400 6.600

dominantpartner 112 0.857 0.351 0.000 1.000

agreehires 112 0.217 0.314 -0.333 0.667

agejv 109 10.963 5.126 4.000 32.000

chinastateinvolvd 93 0.419 0.496 0.000 1.000

balanceofcontrol 111 0.083 1.990 -6.000 5.000

balancecontrolsqrd 111 3.932 7.108 0.000 36.000

numbremployees 112 4.366 2.294 1.000 8.000

beijingregn 110 0.318 0.468 0.000 1.000

guangdongregn 110 0.282 0.452 0.000 1.000

shanghairegn 110 0.273 0.447 0.000 1.000

tianjinregn 110 0.091 0.289 0.000 1.000

jobposition 112 2.438 0.792 1.000 3.000

yearsworked 109 5.537 4.725 0.250 24.000

education 109 3.128 0.336 3.000 4.000

age 110 2.345 1.571 1.000 7.000

The table suggests that no problems exist regarding lack of adequate variation in the data. We note that the mean of the dummy variable dominantpartner seems to indicate that JVs in our sample are predominantly characterized by a domi- nant partner (86%). The remainder of the descriptive analy- sis does not suggest any likely sources of problems with the data. We deem the data to be suitable for further analysis.

Table III presents the correlation matrix (using Pearson’s r) for all the variables.viii The highest absolute magnitude of any correlation coefficient is 0.708, indicating an (unsurprising) strong association between corporate culture (organization- cultr) and national culture (nationalcultr). We deem this not problematic as it is below the threshold of an absolute mag- nitude of 0.8. The next largest correlation coefficient is 0.571 between respondents’ yearsworked and respondent age, an unsurprising relationship. The third highest correlation is 0.538, between two independent variables, managerialcom- munic and objectivecongruence. The next highest correlation is 0.505, between two dependent variables, overallcompetitiv, and decisionperform. We interpret these relationships as not problematic for our analysis. Only 4 of 276 relevant correla-

tion coefficients have an absolute value of 0.5 or higher--the regression analysis may proceed with no cause for worry re- garding overly correlated factors.

REGRESSION ANALYSIS RESULTS Table IV, below, presents the results of ordinary least squares

regression analysis using all four of our dependent variable performance measures in four distinct models. Owing to miss- ing data, the number of complete cases in the models varies slightly from n=77 (Model 2) to n=80 (Models 1 and 4). Model 1’s behavioral dependent variable is satisfactperform, a mea- sure of perceived satisfaction with JV performance. Model 2’s economic dependent variable, trendperform, reflects respon- dents’ perception of the JV’s financial performance over the past three years. Model 3’s strategic dependent variable, over- allcompetitiv, is a construct reflecting multiple different per- ceived aspects of the relative competitiveness of the JV versus other industry players. Model 4’s behaviorial dependent vari- able, decisionperform, measures the extent to which decision making in the JV is perceived as effective by respondents.

Dependent Variable:

Model 1 Model 2 Model 3 Model 4

satisfact- perform

trendperform overalcompeti-

tiv decision- perform

Number of obs 80 77 79 80

F(dfmodel, dfresid.)

2.02* (20, 59)

1.82*(20, 56)

2.08* (20, 58)

1.33 (20, 59)

R-squared 0.406 0.394 0.418 0.311

Adj. R-squared 0.201 0.178 0.218 0.077

coefficient (standard error)

objective- congruence

0.267 * 0.231 0.342 * 0.175

(0.149) (0.277) (0.189) (0.177)

resource- complement

-0.434 -0.723 0.083 -0.303

(0.331) (0.612) (0.409) (0.392)

organization- cultr

-0.098 -0.010 -0.113 -0.010

(0.084) (0.152) (0.103) (0.099)

nationalcultr 0.108 -0.059 0.068 -0.010

(0.076) (0.136) (0.094) (0.090)

managerial- communic

0.103 -0.766 *** -0.049 0.040

(0.152) (0.276) (0.189) (0.180)

dominant- partner

-0.178 -0.213 -0.236 -0.054

(0.229) (0.409) (0.284) (0.271)

agreehires -0.157 0.929 * -0.027 -0.486

(0.280) (0.511) (0.352) (0.331)

agejv 0.003 0.018 0.014 -0.021

(0.016) (0.030) (0.021) (0.019)

chinastate- involvd

0.159 0.595 * -0.113 0.147

(0.196) (0.358) (0.244) (0.231)

balanceofcontrol 0.045 0.153 * -0.088 † -0.115 *

(0.043) (0.080) (0.054) (0.051)

balance- controlsqrd

0.000 0.035 0.021 0.016

(0.014) (0.024) (0.017) (0.016)

numbr- employees

0.077 * 0.110 0.085 * 0.042

(0.038) (0.069) (0.047) (0.045)

beijingregn 0.591 2.212 * 0.577 -0.327

(0.703) (1.255) (0.868) (0.832)

guangdongregn 0.824 1.613 1.065 -0.315

(0.700) (1.246) (0.864) (0.828)

shanghairegn 0.885 2.086 * 0.753 -0.186

(0.685) (1.218) (0.846) (0.811)

tianjinregn 0.345 0.625 0.575 0.075

(0.736) (1.307) (0.908) (0.871)

jobposition -0.107 -0.056 0.151 -0.268 **

(0.108) (0.200) (0.138) (0.128)

yearsworked -0.030 -0.059 -0.061 * 0.001

(0.020) (0.038) (0.026) (0.024)

education -0.247 0.198 -0.426 0.226

(0.259) (0.489) (0.332) (0.306)

age 0.059 0.197 * 0.227 *** 0.078

(0.060) (0.109) (0.076) (0.071)

constant 2.475 * 4.392 4.409 *** 4.889 ***

(1.210) (2.226) (1.514) (1.432)

* p≤0.05 **p≤0.01 ***p≤0.005 † (near-significant) p≤0.0525

Table 4: Regression analysis results: Effects of fit-related factors on JV performance

Journal of Asia Business Studies SPRING 200846 Journal of Asia Business Studies SPRING 2008 47

Our first reflection on the analysis is that Model 4’s F-Sta- tistic, at a standard of p≤0.05, does not allow us to confidently reject the hypothesis that all coefficients are equal to zero. Our dependent variable in this case, decisionperform, is not well predicted by our independent variables. The R-Squared for Model 4 is 0.31, the lowest of all the models. The other three models (1, 2 and 3) have sufficiently large F-Statistics to confidently reject the hypotheses that all coefficients have a zero-value (p≤0.05). For Models 1, 2 and 3, respectively, the R- Squared figures are 0.41, 0.39, and 0.42. A substantial amount of variation in JV performance is apparently explained by our independent variables. Using the adjusted R-Squared sta- tistics shown in Table IV in place of the raw R-Square figures tells us that less variation in the data is explained in Models 1-4 than the R-Square statistic initially suggests. According to the adjusted statistic, Models 1-4 explain 20 %, 18%, 22% and 8% of the variation in the data, respectively. Models 1, 2, and 3 significantly out perform a model comprised of a constant and explain substantial variation in the data. These models are suitable for further discussion and interpretation, while Model 4, regrettably, is not.

For Model 1, which predicts perceived satisfaction with JV performance, the effect of objective congruence is signifi-

cant at a level of p≤0.05 and signed as predicted. Hypothesis 1 is the only test assertion supported by Model 1. Consider- ing the range of the variable, 1.25 to 6.0, the impact of the magnitude of the coefficient on performance is substantial. Among control variables in Model 1, the only significant fac- tor (p≤0.05) is numbremployees, a JV-size measure, but the effect is largely insubstantial, with a coefficient magnitude of 0.077 (the data values range from 1 to 8).

In Model 2, as the effectiveness of managerial communi- cation increases (managerialcommunic) managers’ percep- tions of financial performance over the last 3 years (trend- perform) decreases (p≤0.005), a significant finding that suggests the opposite of Hypothesis 4—lower duration and numbers of routine managerial meetings (formal commu- nications) seem to predict higher JV performance. The co- efficient magnitude is substantial (-0.766), suggesting the effect of this factor on performance is strong (across a data range of 3.4 to 6.6). Agreehires, which measures the extent of agreement among managers regarding key position hiring, is significant and positively signed, a finding that supports the veracity of Hypothesis 6 regarding the positive effects of harmonious decision making (coefficient magnitude of 0.929, acting on a data range of -0.333 to 0.667—a substantial

effect). Also noteworthy in Model 2 are several other signifi- cant control factors. The control variable, balanceofcontrol, is significant and signed so as to suggest increasing Chinese partner control predicts increased performance. This find- ing supports Hypothesis 5, regarding returns to control by a dominant partner. JVs from the Beijing and Shanghai re- gion also significantly predict higher performance relative to other regions. The age of respondents is a significant pre- dictor of perceived performance. Finally, Model 2 is the only model where involvement of the Chinese state with a domi- nant partner is significantly and positively associated with prior increased performance.

Turning to Model 3, objectivecongruence predicts per- ceived overall competitiveness of the JV (overallcompeti- tiv) at a level of p≤0.05, is signed as predicted, and boasts a moderately large coefficient (equal to 0.342, acting on a data range of 1.25-6). This factor has a moderate impact on per- formance as measured by perceived overall competitiveness of the JV. Among control variables in Model 3, the coefficient of balanceofcontrol is near-significant, but its magnitude is small, suggesting that perceived predominance of control by one partner (irrespective of equity stake) does not substan- tially impact performance as measured by perceived overall

competitiveness. Model 4, as noted above, does not predict decision making performance better than a constant (coef- ficients are not significantly different from 0) and is thus excluded from the discussion.

Scanning across the rows of coefficients in Table IV for significance and consistency of coefficient signs is some- what illuminating. Objectivecongruence is clearly the most consistently significant factor predicting 2 of 3 performance indicators (from Models 1, 2, and 3) at a level of p≤0.05. These two models have coefficients that are signed as predicted by Hypothesis 1. Among other factors, managerialcommunic apparently matters, and is a highly significant predictor only of perceived financial performance trends, but the sign of the coefficient is opposite that predicted. Agreehires, mea- suring degree of managerial harmony regarding whom to hire, is a also significant predictor of financial performance. Informal control increasingly slanted towards the Chinese partner (measured by balanceofcontrol) predicts financial performance, while the overall competitive performance of a JV versus its competitors improves slightly as the Ameri- can partner increasingly controls the JV via informal means. Further, informal control has an effect on performance while formal control (measured by dominantpartner), apparently,

(01) (02) (03) (04) (05) (06) (07) (08) (09) (10)

satisfactperform (01) 1

trendperform (02) 0.2516 1

overalcompetitiv (03) 0.263 -0.0191 1

decisionperform (04) 0.0981 -0.0743 0.5052 1

objectivecongruence (05) 0.3802 -0.0207 0.3091 0.1604 1

resourcecomplement (06) -0.2103 0.0701 0.0251 -0.0408 -0.022 1

organizationcultr (07) 0.1172 -0.0917 -0.1055 -0.0959 0.2644 -0.008 1

nationalcultr (08) 0.163 -0.1134 -0.0191 -0.0043 0.2143 0.1235 0.7083 1

managerialcommunic (09) 0.3261 -0.1785 0.1104 0.0776 0.5379 0.0119 0.4759 0.4034 1

dominantpartner (10) -0.1731 -0.1061 -0.2048 -0.041 -0.1743 0.0193 0.0453 0.0637 0.1176 1

chinastateinvolvd (11) 0.0713 0.047 -0.0962 -0.0117 0.0459 0.0951 0.3006 0.3359 0.4201 0.3173

agreehires (12) -0.1482 0.1385 -0.0987 -0.0062 0.1116 0.2031 0.2264 0.1526 0.2171 0.2163

agejv (13) 0.0738 0.1593 0.0746 -0.0008 0.1821 0.0293 0.0222 -0.0212 0.1454 -0.0775

balanceofcontrol (14) 0.1384 0.1644 -0.2276 -0.3418 -0.1711 0.0569 0.1197 0.0292 -0.0326 -0.022

balancecontrolsqrd (15) -0.1578 0.1242 0.1214 0.175 -0.2294 0.2646 -0.357 -0.1408 -0.1639 0.0961

numbremployees (16) 0.0756 0.1788 0.1751 0.1012 -0.0197 -0.0733 -0.2291 -0.2552 -0.1461 -0.1445

beijingregn (17) -0.107 0.2579 -0.0469 0.001 0.1596 0.2226 0.1359 0.131 0.1722 0

guangdongregn (18) 0.1608 -0.2369 0.1301 -0.0732 0.0945 -0.2808 0.0923 -0.0165 0.2898 0.0343

shanghairegn (19) 0.1203 0.1036 -0.0382 -0.0443 -0.0945 0.0149 -0.0662 -0.0325 -0.2085 -0.1132

tianjinregn (20) -0.2507 -0.1707 -0.0203 0.2051 -0.1655 0.1239 -0.2016 -0.0395 -0.3474 0.1187

jobposition (21) -0.0285 -0.1091 0.0563 -0.2375 -0.1607 -0.2301 -0.1569 -0.1478 -0.211 -0.0761

yearsworked (22) -0.0417 0.0514 -0.0484 0.0005 0.0302 -0.1221 0.088 0.107 0.1398 0.0223

education (23) -0.2206 0.018 -0.084 0.037 -0.015 0.2581 -0.1084 0.0046 -0.0027 -0.0825

age (24) 0.174 0.1126 0.199 0.1477 0.2539 0.0921 0.2276 0.2497 0.3778 -0.0253

Table 3: Correlation matrix

(11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23)

1

0.3042 1

0.2412 0.0021 1

-0.0419 -0.2303 -0.0343 1

-0.038 0.0435 0.0534 -0.2183 1

-0.1875 -0.2667 0.253 0.0489 0.1598 1

0.1258 0.2957 0.1486 -0.0836 0.1607 0.0446 1

0.2292 0.0291 0.0289 0.0232 -0.1992 -0.1821 -0.4121 1

-0.2527 -0.3179 -0.1113 0.1735 -0.102 0.0187 -0.4246 -0.4666 1

-0.126 0.0689 -0.0488 -0.2762 0.2175 0.1249 -0.178 -0.1956 -0.2016 1

-0.2207 -0.4453 -0.0573 0.1123 -0.1 0.1663 -0.2421 0.0789 0.2049 -0.1239 1

0.2001 0.0959 0.3616 0.0254 -0.0143 0.0759 0.1591 0.1229 -0.1343 -0.2033 0.0373 1

-0.1159 0.1071 0.2647 -0.0912 -0.0668 0.1655 0.0495 -0.1575 0.0067 0.1958 -0.076 -0.001 1

0.1748 0.1963 0.1453 0.0504 -0.1559 -0.0901 0.2247 0.0676 -0.1421 -0.1949 -0.2396 0.5706 0.0094

Journal of Asia Business Studies SPRING 200848 Journal of Asia Business Studies SPRING 2008 49

does not. The effect of asymmetric control by one partner- firm on performance is supported (Hypothesis 5), but not as proposed, via formal/legal means. Rather, informal controls have a significant effect on performance, but the sign of the effect is inconsistent across types of performance measure.

ISSUES, CONTRIBUTIONS, DIRECTIONS Using a sample of China-U.S. JVs, our study tests a model

that predicts JV performance as a function of three categories of fit: strategic, organizational and cultural. Using an empiri- cal approach that recognizes the diversity inherent in defini- tions of both performance and fit, we test a complete model of JV performance by scrutinizing six distinct varieties of fit, two from each of the three above-mentioned fit-categories. The model also takes into account multiple salient control factors in order to offer a reasonable, largely complete em- pirical picture of the main hypothesized effects.ix From our examination of six different types of strategic fit, we find that the congruence of objectives (a form of strategic fit) consistently matters for two types of performance measure: perceived satisfaction with JV performance, and perceived overall competitiveness of a JV—behaviorial and strategic performance measures, respectively. Resource complementa- rity, cultural factors, and the presence of a dominant equity- owning partner do not impact performance in significant ways. Effectiveness of managerial communication, and top management agreement regarding important hires mat- ter (regarding predicting recent financial performance), but managerial communications efficacy has a negative effect, suggesting that high performance is associated with fewer meetings and interactions between JV participants. Depend- ing on how performance is defined, different aspects of “fit” matter. Hypotheses 1 (objective congruence—strategic fit) and 6 (agreement regarding hiring—organizational fit) are most strongly supported, while there is evidence that Hypothesis 4 (the efficacy of managerial communication—cultural fit) affects performance, though not in accord with our predic- tion.

Given the attention that “fit” has received in the academic literature, our results offer opportunities for refined insight into the effects of fit on performance. We interpret the findings as evidence that JVs can and do succeed without necessarily meeting the demands of multiple stringent criteria. Notably, organizational and cultural fit seem to matter, but somewhat less than strategic fit, specifically objective congruence. We are puzzled that resource complementarity does not seem to matter—one possible explanation is that a high degree of complementarity (as we measure it, spanning many areas) may not be necessary, but complementarity in one or two key areas may drive this factor. The lack of detected significance for resource complementarity may reflect an overly stringent (and hence poor) measurement of resource complementarity. That is, we define a high degree of resource complementar- ity as a large number of detected complementarities versus

a small number or zero complementarities between partner- firms. We literally count the number of incidences of com- plementarity that the data manifest. Our measure assumes that more complementarity between partner-firms is better. It is conceivable, however, that resource complementarity in as few as one area of performance-critical partner comple- mentarity is a sufficient condition to trigger a positive effect on performance. This reality would introduce substantial noise into our measure. If the above assumption is incorrect, our measure for resource complementarity is suspect.

ISSUES Although our study has offered some interesting insights

into the fit-related determinants of JV performance, there remain several limitations to our study. Our findings may be limited in generality owing to (1) our sample comprising China-U.S. JVs, and (2) our including only JVs that have op- erated for at least two years. Although three of our four pro- posed models (1) predict outcomes significantly better than a model comprised solely of a constant, and (2) explain sub- stantial variation in the data, the veracity of Model 4, with a dependent variable designed to reflect decision making per- formance in JVs is highly doubtful. A more accurate method of measuring this construct is needed, perhaps through the use of additional or different measures. Additionally, our sample size is somewhat small, though sufficient for the pur- poses of our analysis. A larger sample might help us to detect additional significant effects. We recognize our models are affected by a trade-off between predictive power and our use of many controls. Given that our intention in this paper is to create the most complete model to date, we have elected to include the most complete available set of control factors in our model—we seek to explore a fully-specified empirically model. In the course of running preliminary models that were comprised of few or no control factors, as well as differ- ent combinations of controls, the results were not substan- tially different than those reported in Table IV—generally, across the different models we tried, the same control and test factors were consistently significant.

Finally, though we sampled JVs from four major regions of industrial activity in China, the situation there is far from static, and a future survey could constructively contribute to refining our results by sampling more regions in China. Our results may lack generality owing to the sample being com- prised solely of U.S.-China JVs. It is likely that our results are most relevant for U.S.-China JVs, and should not be taken as “general” findings reflecting the calculus of performance in all JVs.

CONTRIBUTIONS AND DIRECTIONS Generally, our study identifies which types of fit in JVs

play a significant role in determining four distinct aspects of performance. From a managerial perspective, we offer evi- dence that there can be flaws in a JV (i.e., imperfect fit of one JV partner to another) and it can still succeed. Some areas are

more sensitive to flaws than others--focusing on objective congruence among JV partners seems warranted by the re- sults. Of the other fit-areas we tested, substantial effects were detected on perceived financial performance of (a) cultural fit via managerial communications efficacy (though in an unexpected way), and (b) organizational fit via agreement regarding hiring.

From a scholarly view, previous work in the study of the effects of different types of “fit” on JV performance has been distinctly piecemeal in nature; rigorous statistical analysis seems the exception rather than the rule. Case-based studies, while useful for developing theory, do not offer substantial generality of findings. Existing work also seem over-focused on ex-ante negotiation process efficacy rather than operating performance over time, an omission addressed by our study. Our work adds value to the debate by utilizing a reasonably (geographically) diverse convenience sample of U.S-China JVs that have been operating for some time in China, and subjecting that sample to a fully-specified, controlled empiri- cal exploration.

Our study, via a survey of extant literature, identifies three types of fit--strategic, cultural and organizational—and em- pirically tests two distinct constructs for each fit-type. Via our method and models, we acknowledge the diversity of definitions of fit as well as the diversity of ways to measure performance. Our approach is to test for significant factors (ceteris paribus) with multiple relevant controls in order to determine which ones matter versus the others—in this we are successful. Our findings suggest that some of the litera- ture espousing the absolute necessity of many variables im- pacting performance of JVs may be overstating the number of key success factors (e.g., Douma et al 2000; Buchel 2002). We find that the most strongly confirmed performance-en- hancing factor is that of objective congruence, part of strate- gic fit. The effectiveness of managerial communication (an aspect of cultural fit) is significant, but the direction of this factor’s action on performance is counter to our arguments. We can briefly explain the unexpected negative impact of managerial communication efficacy by reviewing an aspect of its construction. One of the main components, frequency of meetings, stands out. We assumed in our construction of the measure that increasing numbers of meetings enhance transparency, a positive outcome that would be associated with increased performance. Another, more diminishing- returns-savvy interpretation is that increasing numbers of meetings reflect increasing need to meet in order to handle vexing communications problems between partner-firms. This would explain the opposite-than-predicted sign of the measure in our model. A simpler explanation might be that the decreased costs incurred by holding fewer meetings en- hances performance.

Agreement among JV participants from different partner- firms regarding hiring, an organizational fit-related factor, affects performance positively, as predicted. Dominant infor- mal control, as noted, also plays a role, though the direction of its effects leack clarity as the vary by performance measure.

Interpreted generously, our analysis identifies four impor- tant factors in all three of our fit categories: strategic fit (via objective congruence—strongly supported), organizational fit (via agreement regarding hiring—moderately supported, and informal dominance—weakly supported) and cultural fit (though the efficacy of managerial communications im- pacts performance, it acts in a surprising manner—increas- ingly Spartan interaction suggests increased performance). The impact of cultural similarity on performance is called into question by our findings, and future work should incor- porate a better measurement for resource complementarity. The work of scholars who espouse the necessity of multiple sources of fit for the achievement of high performance is con- firmed; we add specificity to the extant findings by identify- ing particular factors that affect performance in a series of fully-specified models. Further work in understanding the effects of fit-related factors seems warranted.

REFERENCES Anderson, E. and Gatignon, H. (1986), “Modes of foreign entry: A transac-

tion cost analysis and propositions”, Journal of International Business Studies, 17(3), 1-26.

Barkema, H.G., Bell, J.H.J. and Pennings, J.M. (1996), “Foreign entry, cultural barriers, and learning”, Strategic Management Journal, 17(2), 151-166.

Blodgett, L.L. (1992), “Factors in the instability of international joint ven- tures: An event history analysis”, Strategic Management Journal, 13(6), 475-481.

Buchel, B. (2002), “Joint venture development: Driving forces towards equi- librium”, Journal of World Business, 37(3), 199-207.

Burt, R.S. (1992), Structural Holes: The Social Structure of Competition, Cambridge:Harvard University Press.

Calantone, R.J. and Zhao, Y.S. (2001), “Joint ventures in China: A com- parative study of

Japanese, Korean, and U.S. partners”, Journal of International Marketing, 9(1), 1-23.

Chen, T.J., Chen, H. and Ku, Y.H. (2004), “Foreign direct investment and local linkages”, Journal of International Business Studies, 35(4), 320-332.

Chi, T. (1994), “Trading in strategic resources: Necessary conditions, trans- action cost problems and choice of exchange structure”, Strategic Man- agement Journal, 15(4), 271-290.

Dess, G.G., Ireland, R.D. and Hitt, M.A. (1990), “Industry effects and strate- gic management research”, Journal of Management, 16(1), 7-27.

Dhanaraj, C. and Beamish, P.W. (2004), “Effect of equity ownership on the survival of international joint ventures”, Strategic Management Journal, 25(3), 295-305.

Doty, D.H. and Glick, W.H. (1998) “Common methods bias: Does common methods variance really bias results?” Organizational Research Meth- ods, 1(4), 374-406.

Douma, M.U., Bilderbeek, J. Idenburg, P.J. and Looise, J.K. (2000), “Stra- tegic alliances– Managing the dynamics of fit”, Long Range Planning, 33(4), 579-598.

Dunning, J.H. (1980), “Toward an eclectic theory of international produc- tion: Some empirical tests”, Journal of International Business Studies, 11(1), 9-31.

Dunning, J. H. (1988), “Transnational corporations in a changing world en- vironment: Are new theoretical explanations required?” in Transnational Corporations and China’s Open Door Policy,

W. Teng and N. T. Wang(Eds.), 3-26, Lexington, MA:Lexington Books. Faure, G.O. (2000), “Negotiations to set up joint ventures in China”, Interna-

tional Negotiation, 5(1), 157-189. Fisher, R.J. (1993), “Social desirability bias and the validity of indirect ques-

tioning.,” Journal of Consumer Research, (20) September, 303-315.

Journal of Asia Business Studies SPRING 200850 Journal of Asia Business Studies SPRING 2008 51

Geringer, J.M. and Hebert, L. (1989), “Control and performance of inter- national joint ventures”, Journal of International Business Studies, 20(2), 235-254.

Hambrick, D.C., Li, J., Xin, K. and Tsui, A.S. (2001), “Compositional gaps and downward spirals in international joint venture management groups”, Strategic Management Journal, 22(11), 1033-1054.

Hamel, G. (1991), “Competition for competence and interpartner learning within international strategic alliances”, Strategic Management Journal, 12(Special Issue, Summer), 83-103.

Heiman, B.A. and Nickerson, J.A. (2004), “Empirical evidence regarding the tension between knowledge sharing and knowledge expropria- tion in collaborations”, Managerial and Decision Economics, 25(6/7), 401-420.

Hennart, J.-F. (1988), “A transaction costs theory of equity joint ventures”, Strategic

Management Journal, 9(4), 361-374. Hennart, J.-F. (1991), “The transaction costs theory of joint ventures: An

empirical study of Japanese subsidiaries in the United States”, Manage- ment Science, 37(4), 483-497.

Hennart, J.-F. and Zeng, M. (2002), “Cross-cultural differences and joint venture longevity,” Journal of International Business Studies, 33(4), 699-716.

Hennart, J.-F. and Zeng, M. (2005), “Structural determinants of joint venture performance” European Management Review, 2(2), 105-115.

Hofstede, G. (1980), “Culture’s consequences: International differences in work-related values.” Beverly Hills, CA: Sage.

Inkpen, A.C. and Currall, S.C. (1998), “The nature, antecedents and con- sequences of joint venture trust”, Journal of International Management, 4(1), 1-20.

Isobe, T., Makino, S. and Montgomery, D.B. (2000), “Resource commit- ment, entry timing, and market performance of foreign direct invest- ments in emerging economies: The case of Japanese international joint ventures in China”, Academy of Management Journal, 43(3), 468-485.

Killing, J.P. (1983), Strategies for Joint Venture Success, New York: Prae- ger.

Kogut, B. and Singh, H. (1988), “The effect of national culture on the choice of entry mode”, Journal of International Business Studies, 19(3), 411-432.

Kogut, B. and Zander, U. (1992), “Knowledge of the firm, combinative ca- pabilities and the replication of technology”, Organization Science, 3(3), 383-397.

Lasserre, P. (1999), “Joint venture satisfaction in Asia Pacific”, Asian Pacific Journal of Management, 16(1), 1-28.

Li, J.T., Lam, K. and Qian, G. (2001), “Does culture affect behavior and performance of firms? The case of joint ventures in China”, Journal of International Business Studies, 32(1), 115-132.

Li, J.T., Xin, K.R., Tsui, A.S., and Hambrick, D.C. (1999), “Building effective international joint venture leadership teams in China”, Journal of World Business, 34(1), 52-68.

Luo, Y., Shenkar, O. and Nyaw, M.K. (2001), “A dual parent perspective on control and

performance in international joint ventures: Lessons from a developing economy”, Journal of International Business Studies, 32(1), 41-58.

Luo, Y. and Park S.H. (2004), “Multiparty cooperation and performance in international equity joint ventures”, Journal of International Business Studies, 35(2), 142-160.

Nickerson, J.A. and Zenger, T.R. (2004). “A knowledge-based theory of the firm - The problem-solving perspective”, Organization Science, 15(6), 617-632.

Oxley, J.E. (1997), “Appropriability hazards and governance in strategic alliances: A transaction cost approach”, Journal of Law Economics and Organization, 13(2), 387-409.

Park, S.H. and Ungson, G.R. (1997), “The effect of national culture, orga- nizational, complementarity, and economic motivation on joint venture dissolution”, Academy of Management Journal, 40(2), 279-307.

Park, S.H. and Ungson, G.R. (2001), “Interfirm rivalry and managerial com- plexity: A conceptual framework of alliance failure”, Organization Sci- ence, 12(1), 37-53.

Probst, G. and Buchel, B. (1997), Organizational Learning: The Competi- tive Advantage of the Future, London; New York: Prentice Hall.

Sirmon, D.G. and Lane, P.J. (2004), “A model of cultural differences and international alliance performance”, Journal of International Business Studies, 35(4), 306-319.

Steensma, H.K. and Lyles, M.A. (2000), “Explaining IJV survival in a tran- sitional economy through social exchange and knowledge-based per- spectives”, Strategic Management Journal, 21(8), 831-852.

Teece, D.J. (1977), “Technology transfer by multinational firms: The re- source cost of transferring technological know-how”, Economic Jour- nal, 87(346), 242-261.

Teece, D.J. (1980), “Economies of scope and the scope of the enterprise”, Journal of Economic Behavior and Organization, 1(3), 223-247.

Teece, D.J. (1981), “The market for know-how and the efficient international transfer of technology”, The Annals of the American Academy of Politi- cal and Social Science, 458(1), 81-96.

Tomlinson, J.W.C. and Thompson, M. (1977), A Study of Canadian Joint Ventures in Mexico, Office of Science and Technology, Department of Industry, Trade and Commerce, Ottawa, Canada.

Tung, R.L. (1984), “Strategic management of human resource in the multi- national enterprise”, Human Resource Management, 23(2), 129-143.

Williamson, O.E. (1975), Markets and Hierarchies, Analysis and Antitrust Implications: A Study in the Economics of Internal Organization, New York: Free Press.

Williamson, O.E. (1983), “Credible commitments: Using hostages to sup- port exchange”, American Economic Review, 73(4), 519-540.

Williamson, O.E. (1985), The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting, New York; London: Free Press.

Williamson, O.E. (1991), “Strategizing, Economizing, and Economic Or- ganization”, Strategic Management Journal, 12(Special Issue, Winter), 75-94.

ENDNOTES i. Historically, the example JV under discussion reflects issues encoun-

tered mostly in the early days of Western-Chinese alliances—the au- thors have no desire to promote an inappropriate stereotype. This JV operated approximately from 1991-1993. Present-day alliances may not experience these specific issues to such a degree as indicated in our example. Improvements in enforcement of contract law and evolu- tion of common practice surrounding JVs in recent years have muted the severity of this issue. It stands as a clear example of incongruent objectives between partners.

ii. Contrary to Chen, Dunning (1980) sees this as resource-seeking be- havior, but only amid part of a larger tapestry of causality employing multiple theoretical perspectives. For China we expect that a large fraction of JVs are driven by resource-seeking—U.S.-Chinese JVs should be ripe with varying degrees of resource complementarity.

iii. Burt (1992), among others, has documented the influence of informal control structures.

iv. In a preliminary model, we created a dummy control variable set equal to 1 only if the respondent was hired by the JV management (and not by one or another partner-firm). This factor was never significant in any early models, so it was dropped from the analysis. The origin of a respondent has no measurable effect on his/her assessment of JV performance.

v. If the JV had been in existence for less than three years, we asked for the perceived trend over the period of its existence, which is never less than two years for any JV in our sample. We asked respondents to report regarding gross sales and net profits before taxes. The abbre- viations ROI, ROA and ROS mean, respectively, return on investment, return on assets, and return on sales. Again, respectively, our team ex- plained that these financial ratios were defined as fractions: net profits divided by capital invested, net profits divided by total assets, and net profits divided by sales (less commissions).

vi. “Business acumen” is intended as a convenient placeholder term for the particular resource complementarity described in the main text. The authors do not intend to assert a definitive meaning for the term.

vii. The dummy controls we tested for industry-effects included whether or not a JV was in an “old technology” industry (clothing, textiles or food), and whether or not a JV transitioned from an old technology industry to a new technology industry (anything except textile, cloth-

ing or food). Remnants of the old technology industry might have a persistent (negative?) effect on performance, e.g., unused productive assets still carried on the books or taking up space. Unsuccessful transitions from old to new technology industries by managers and workers, and/or lack of managerial expertise in the new technology arena might also impact JV performance. We also examined indi- vidual industry control dummies for eight distinct industries plus an “other industry” category. The industries tested included electronics/ communications, textiles, clothing, transportation, electrical machin- ery, food machinery, chemical/medical, and construction. None were significant in any models and have been excluded from our final mod- els, presented below.

viii. In addition to the correlation matrix shown in Table III, which utilizes all the available data to calculate Pearson’s r, we calculated the same matrix using pairwise deletion for cases with missing data—a more conservative approach to obtaining correlation coefficients. The re- sults are substantively similar to the matrix shown.

ix. For thoroughness, we modeled the data in several different ways, including allowing for the possibility of interaction effects between vari- ous factors. We created an interaction term combining JV age (age) with each of our test factors (thanks to Lihua Wang, San Francisco State University, for this suggestion). Increasing JV maturity might compound the effects of our test factors owing to effects from com- paratively greater operating experience for older JVs. No such distinct effects were detected. We further sought to identify important inter- actions, specifically, between the two measures of fit within each fit- category. For strategic fit, for example, we created an interaction term that multiplied objective congruence by resource complementarity to ask if a combination of fit factors is required for high performance. For the few interaction terms (across all three fit-categories) that are statistically significant, the magnitude of the estimated coefficient is very small relative to the range of the data, suggesting miniscule, if any, effects. The majority of interaction term coefficients are not statisti- cally significant.