1. INTRODUCTION

This study investigates the governance of high-tech alliances. The governance of high-tech alliances, through legal, private, and relational ordering, is a challenge as it needs to balance between realizing benefits and safeguarding risks (Dodgson et al. 2008; Dodgson 2000; Nooteboom 2004). Our focus is on the role of legal ordering, or formal contracts, in this process. Formal contracts are written, legally binding agreements between two or more parties (Lyons & Metha 1997). They are important instruments for the governance of exchange relations between economic actors because they represent promises or obligations to perform particular actions in the future (Mayer & Argyres 2004). However, empirical research on interfirm contracts is sparse because they are often subject to confidentiality and therefore rarely published. This hampers the understanding of the content and role of alliance contracts.

In this article, we report an in-depth study of 391 contracts of Dutch high-tech alliances. By doing so, we intend to fill the aforementioned research gap.

Traditionally, contract studies have considered a contract as a static, legal document and have therefore paid little or no attention to the active role contracts may play in interorganizational alliances (Lyons 1996). Transaction cost theory (Williamson 1985) has contributed greatly to the study of interorganizational exchange because it specifies in detail the nature and extent of risk in transactions and provides indications that allow the construction of schemes for 'governing' transactions in such a way that risks are reduced (Brousseau & Glachant 2002). Various alliance scholars (e.g. Arino & Reuer 2004; Parkhe 1993; Crocker & Reynolds 1993) provide evidence that complex formal contracts or contracts with many clauses that are strictly specified allow mitigating the risk of opportunistic behaviour. Nevertheless, there are also empirical contradictions (David & Han 2004) and theoretical limitations to transaction cost economics (Dodgson 1993; Nooteboom 1996).

Most importantly, the available evidence suggests that the level of detail of alliance contracts varies greatly. Very simple contracts seem to be able to regulate very complex collaborations. Empirical research of economists on contract clauses in innovation alliances show that in stead of entering into details about future activities, contracts tend to focus on a few core issues (Grandori 2006). This brings many questions. Which provisions do high-tech alliance partners specify in their contract? Does the number of clauses in formal contracts of high-tech collaboration vary and if so, why? What is the role or function of a contract? The key objective of this paper is to answer these questions.

Our research concerning the antecedents of contract details aligns with transaction cost economics but complements this with insights that derive from social exchange theory. We argue that the behavioural assumption of opportunism is one of the theoretical limitations in transaction cost economics that may explain the empirical anomalies (cf. Nooteboom 2002). More in particular, we suggest that fear of opportunistic behaviour by a potential or actual partner and a willingness to trust and reciprocate may be mutually considered by those designing and implementing contracts to manage interfirm alliances.

The study of contracts seems to ignore the human aspects of economic transactions, in particular the development of trust (Dodgson 1992, 1993). Hence, we argue that for a thorough understanding of the content and role of contracts in high-tech alliances we need to account for the development of trust as well as for the safeguarding of positions.

The cooperative nature of the interfirm relationships under study presents an interesting arena for the study of contracts. Contract studies often focus on vertical relationships such as procurement relationships where prices, quantities and qualities can be established and agreed upon (Crocker & Reynolds 2006; Anderson & Dekker 2005).

There have been few studies of contracts in high-tech cooperative relationships, where parties have no hierarchical relationship and outcomes cannot be predetermined (notable exceptions are Robinson & Stuart 2004; Ryall & Sampson 2006). Our sample is interesting as the relationships focus on the development of new knowledge (intangible assets); prices and budgets might be difficult to set ex ante and the verifiability of tasks and performance are likely to be low (e.g. man hours are specified but the result is still unknown). These differences in relationship characteristics and context, as compared with other contract studies, can encourage new insights into the study of contracts and interfirm relationships.

The outline of this paper is as follows. In the next section, we will further explain the theoretical foundations of our study. We draw upon transaction cost economics and social exchange theory to identify key theoretical drivers of the choices that firms make when they design their interfirm agreement. A discussion of the research methodology and our empirical results will follow. We conclude with implications for managers and suggest avenues for future research.

2. THEORETICAL FOUNDATIONS

2.1 Dedicated assets

Dedicated assets is our first antecedent for the level of contractual detail. Dedicated assets are crucially important for the governance of strategic alliances (Poppo & Zenger 2002; Reuer & Arino 2007; Joskow 1988). Transaction cost economics argues that particularly in conditions of asset specificity, alliance parties should safeguard to the maximum, but at the same time it acknowledges that this might be impossible due to bounded rationality and uncertainty (Williamson 1985). Dedicated assets are the result of dedicated investments and are required to support the focal alliance. They have, in contrast to general purpose assets, little or no salvage value outside the relationship. That is, when asset specificity is low, resources can be easy deployed to other relationships and continuity of the alliance is therefore not important.

However, when a firm makes transaction-specific investments, it creates dedicated assets, which increase its switching costs and makes the focal firm more dependent. The more dependent the focal firm, the more difficult it is to replace the alliance partner and the more vulnerable it will be to opportunistic behaviour. An alliance manager will find it beneficial to negotiate a more detailed contract when the potential value of loss due to hold-up behaviour exceeds the costs of negotiating safeguards. Therefore we suggest:

Hypothesis 1: Asset specificity will result in a more detailed alliance contract.

2.2 Spill-over risks

Our second antecedent concerns the risk of spillover (Inkpen 2000). Knowledge is a key asset for high-tech companies for which technology development is a core activity. The purpose of a business relationship is to benefit from this firm-specific, path-dependent competencies and resources as it complements the firm's own specialist knowledge and know-how (Nooteboom 2004). The exchange of specialist knowledge is a prerequisite for the development of new knowledge. However, specialist knowledge is often highly confidential because it is part of the core competence of the firm and therefore offers sustainable competitive advantages.

Spill-over is not the same as the loss of a resource (like the risk of dedicated assets). Under spill-over the company still owns the knowledge but it is no longer exclusive. In the setting of high-tech alliances contracts are designed to govern spill-over risks associated with the knowledge exchange essential to innovation. As this specialist knowledge is often the basis of future competitive advantages, firms have a strong incentive to manage risks of spill-over, particularly when the existing or potential partner firm is or could be a competitor. This gives

Hypothesis 2: Spill-over risks will result in a more detailed alliance contract.

2.3 Trust

By bringing trust into the equation of contracting behaviour, we align our research with the ongoing discussion concerning formal and relational governance (Lane & Bachman 1998; Nooteboom 2002). Transaction cost economics denies the importance of trust as a meaningful governance mechanism (Williamson 1993), but this is in conflict with empirical evidence showing that in interfirm alliances trust exists and has value (Nooteboom 1996): it facilitates joint action, reduces the need for hierarchical control, and is a key condition for the development of new knowledge within and between organizations.

In this study we focus on interorganizational trust (Dyer & Chu 2003; De Jong & Klein Woolthuis 2008) defined as a positive perception of the partner's behaviour, that is, the perception by the respondent of the focal firm that a partner organization will not engage in opportunistic behaviour even in the face of opportunities and incentives to do so. We can expect this confidence or perception (trust) where the partner firm: a) shows forbearance from opportunism, and b) acts with care and concern, and c) the focal firm hence shows a lack of monitoring. Zaheer et al. (1998) show that there is a strong orrelation between interpersonal and interorganizational trust and that, although conceptually different, it is the latter in particular that improves interfirm performance.

Trust and contracts are generally viewed as substitutes (Poppo & Zenger 2002; Reuer et al. 2003). There are two important arguments why trust and formal contracts can be considered as substitutes. First, trust economizes on costly contracts. Alliance managers who trust each other have less inclinations and need to impose control on others. Trust reduces the need to negotiate and specify e.g. tasks, investments, responsibilities, planned outcomes and accountability. Second, it has been argued that detailed contracts may destroy trust, for example by creating suspicion. Fehr and Schmidt (2002), for instance, argue that contracts should remain less detailed because this is closer to the implicit and explicit norms and values of human interaction, represented by their behavioural preferences. We therefore expect the following: