Transaction cost economics: the main case

The benchmark against which TCE has developed is the neoclassical framework of complete contingent contracting under which initial ownership rights and organizational structures do not matter for economic efficiency (see Debreu, Reference Debreu1959). Hence, a major contribution of TCE is arguably the introduction of behavioral and economic conditions that prevent complete contingent contracting, thus making property-right allocations and governance structure relevant. Specifically, the choice of discrete governance structures (such as market, hybrid contracts, and hierarchies) matters for economic efficiency because transactions occur between bounded rational and opportunistic economic agents that are subject to bilateral dependency in the form of asset specificity. Since boundedly rational actors are unable to completely foresee all future contingencies that affect the execution and fulfillment of contracts, these contracts are inevitably incomplete and the contractual parties are, as a result, at the risk of being subject to opportunism. Economic agents in TCE are assumed to be farsighted, so that they are able to predict and calculate the economic impact of the hazards that the contractual incompleteness gives rise to and, as a result, produce a discriminating match between transactions (that differ in the attributes of asset specificity, frequency, and uncertainty) and governance structures (that differ in their capacity to protect the integrity of the contractual relation) that is comparatively efficient (Williamson, Reference Williamson1985).

TCE rests on the behavioral assumptions of bounded rationality and opportunism.Footnote ² Williamson (Reference Williamson1985: 44) introduces these assumptions in contrast to neoclassical maximizing rationality and simple self-interested behavior with an appeal to assumptional realism and an aim to characterize key aspects of human nature ‘as we know it’. Bounded rationality refers to the fact that agents are assumed to be ‘intendedly rational, but only limitedly so’ (Simon, Reference Simon1945: xxiv). The first part of Simon's (Reference Simon1945) definition of bounded rationality (‘intendedly rational’) refers to a central component of TCE, namely that economic agents are farsighted, adaptive, and planning. This component is central to the TCE in terms of eliciting economizing behavior. Without this belief in the ability of agents to engage in farsighted planning and accurately forecast the impact of potential contractual hazards, there is no reason to believe that the contractual parties in any given transaction will choose the economizing governance structure. The second part of the quote (‘only limitedly so’) fulfills a quite different purpose than the first, namely, providing a behavioral justification for contractual incompleteness and the particular institutional responses to this incompleteness that the theory claims will result. Hence, economic agents’ limited ability to predict future contingencies that may affect the integrity of the contract leads to incomplete contracts, which prompt decision-makers who aim to minimize the negative effects of contractual hazards to adopt particular governance structures, such as hierarchy (unified ownership) and long-term contracts that safeguard against contractual incompleteness.

While bounded rationality in conjunction with opportunism may be sufficient to create contractual incompleteness, they do not necessarily lead to significant ex post contractual hazard of the kind suggested by TCE if the parties to the transaction can renegotiate contracts at a low cost. Hence, the magnitude of contractual hazards also depends heavily on the degree of bilateral dependence, or asset specificity, between the parties. Together, the concepts of bounded rationality, opportunism, and asset specificity constitute the core of TCE's explanatory apparatus. However, the assumption of bounded rationality, and the associated informational implications (e.g., uncertainty) have not been consistently applied in TCE (Foss, Reference Foss2003). This has important consequences for how firms’ governance choices are explained.

Explanatory mechanisms in transaction cost economics

The partial introduction of uncertainty into contractual relationships (e.g., contractual relations have known payoffs but are subject to unknown contingencies) should be understood against the historical and disciplinary background of TCE as part of the broader field of new institutional economics (Williamson, Reference Williamson1985), which was originally developed as a reaction to particular perceived shortcomings in the conventional neoclassical theory of production and exchange (e.g., Arrow, Reference Arrow1974; Arrow and Debreu, Reference Arrow and Debreu1954; Debreu, Reference Debreu1959). Basically, new institutional economics suggested a series of modifications of the conventional microeconomic framework that were not primarily aimed at fundamentally rejecting or recasting the basic premises of this framework, but rather to build upon it to cast new light on institutional phenomena (Furubotn, Reference Furubotn2001). For example, the role of individual decision-making was reevaluated to acknowledge that, while decision-making can be conceived of as maximizing, it is not the firm that maximizes its profit, but rather the individual agents that maximize their utility given the constraints established by property rights and organization (Furubotn and Pejovich, Reference Furubotn and Pejovich1972). This allowed theorists to consider a broader range of possible ownership and organization patterns, which in turn allows for the analysis of the complex relationship between institutional and property rights arrangements and economic behavior.

Hence, the basic equilibrium-oriented marginalism and maximization mode of conventional economic analysis was not fundamentally challenged, but rather, partially modified and extended to apply to individual decision-makers’ utility function, and thereby also to new domains such as contracting and organization. The result of this approach was ‘hybrid models’, such as TCE, which ‘change only a few elements of the basic neoclassical structure while leaving most of its other features intact’ (Furubotn, Reference Furubotn2001: 135).Footnote ³

The hybrid character of TCE is perhaps most clearly illustrated by the application of bounded rationality. Hence, rather than modeling actors according to a full-fledged constraint on rationality that affects economic behavior in all relevant domains, bounded rationality instead ‘briefly enters, leaving only one essential trace, contractual incompleteness, in an otherwise largely orthodox theoretical structure, but is essentially suppressed for the remaining part of the story’ (Foss, Reference Foss2001: 405). For example, while Williamson's (Reference Williamson1985) behavioral justification of contractual incompleteness in terms of actors’ inability to imagine future contingencies, and therefore being subject to potential surprise in the contract-writing domain, is close to Shackle's (Reference Shackle1961) work on imagination and choice, TCE instead builds on a highly restrictive application of Simon's (Reference Simon1945) notion of bounded rationality. This may simply be a pragmatic assumption aimed at making the TCE more compatible with an orthodox agenda of farsighted (equilibrium-oriented) contracting, but it nonetheless raises important questions about the nature of cognitive processes and how these are described in TCE. For example, Simon's (Reference Simon1945) notion of bounded rationality as decisions made within a given closed system that is discovered via limited search and where, consequently, limitations to imagination and the potential of surprise are both absent, is arguable much more in line with a comparative-static and equilibrium-oriented agenda than a full application of the alternative notion of open-ended situations where decision-makers have to rely on their imagination to articulate the problem to be solved (see Shackle, Reference Shackle1961: 100). Hence, a partial or limited use of bounded rationality may enable TCE to avoid some of the more radical implications of uncertainty (see Earl and Kay, Reference Earl and Kay1985; Lachmann, Reference Lachmann1976). However, the different conceptualizations of cognitive processes in TCE may also turn out to have broader research implications beyond tractability, prediction, and the justification of incomplete contracts. As will be discussed in detail, more extensive forms of uncertainty allow for addressing systematic biases in governance choices, the effects of current organization on decision-making, and situations where organizational decisions are reversed or significantly revised over time as decision-makers understand or learn about circumstances that were previously unknown.Footnote ⁴

The approach in TCE described above may be related to the fact that there are many technical challenges involved in modeling extensive uncertainty and bounded rationality in a tractable way (Kreps, Reference Kreps1996). However, there may also be more subtle barriers involved related to the training of organizational economists and a preference for the economists’ traditional toolkit. Hence, a somewhat rationalized reconstruction of the problem indicates that while organizational economists do welcome the opportunity to study institutional phenomena that can only be derived from assumptions of uncertainty and behavioral limitations, they are simply not willing to generalize these assumptions to the extent that it may threaten to change the basic economic modus operandi, irrespectively of whether this leads to an apparent logical inconsistency in how assumptions are applied (Hodgson, Reference Hodgson2011). These practice points toward a core problem in hybrid models such as TCE: the lack of symmetry with which behavioral assumptions and causal mechanisms are applied (Foss and Foss, Reference Foss and Foss2000). Rather than being mere heuristic devices for selectively achieving modeling aims, the consistent application of a theory's core assumptions, such as uncertainty and bounded rationality in TCE, plays an important role in providing a credible causal story (Foss and Hallberg, Reference Foss and Hallberg2014).

Comparative efficiency in transaction cost economics

The introduction of transaction costs is likely to have more general effects on the possibility of economizing behavior, equilibrium analysis, and the possibility of optimal organizational outcomes than what is currently acknowledged in TCE (Furubotn and Richter, Reference Furubotn and Richter1997). Hence, a fundamental problem in TCE is thus that while it partly rejects core principles in microeconomics, it has yet to identify a fully consistent alternative explanatory mechanism. According to Langlois (Reference Langlois, Jussawalla and Ebenfield1984), this largely stems from a failure to fully acknowledge the explanatory role of extensive uncertainty and develop a set of new explanatory mechanisms that are consistent with this assumption. In a sense, this reflects a tendency to not acknowledge the full set of logical consequences of uncertainty. According to Langlois (Reference Langlois, Jussawalla and Ebenfield1984: 1), a complete acknowledgement of this core analytical concept of TCE, in turn, ‘points towards a reinterpretation of “efficiency” explanations of internal organization in terms of an evolutionary logic of explanation’. Specifically, there are three possible forms of causal grounding of the comparative-efficiency explanation that may be examined to pinpoint the current role of uncertainty in TCE and the possible effects of adopting a more consistent application of this notion: calculative intentionality, organizational learning, and competitive market pressure.

Explanation and prediction in TCE build on an emphasis on the comparative efficiency of existing governance structures and the assumption that organizational and contractual decisions are made in a farsighted, adaptive, and economizing manner. Consequently, the most straightforward interpretation of the causal story underlying TCE is that governance structures are consciously chosen by the parties to a transaction after carefully calculating and comparing the transaction costs associated with every possible governance structure. Building on Hayek's (Reference Hayek1988) original distinction, Smith (Reference Smith2008: 2) refers to this as constructivist rationality (‘the deliberate use of reason to analyze and prescribe actions judged to be better than alternative feasible actions that might be chosen’) as distinguished from ecological rationality (‘emergent order in the form of the practices, norms, and evolving institutional rules governing action by individuals that are part of our cultural and biological heritage and are created by human interactions, but not by conscious human design’). In the context of governance and contracting, there are at least three different problems associated with this constructivist justification that are related to logical consistency, governance inseparability, and appropriability.

First, there are problems of logical consistency related to TCE's use of bounded rationality. Hence, if boundedly rational decision-makers are expected to not be able to include future contingencies that may affect their relationship in the original contracts, they will hardly be able to in advance effectively identify the specific governance structures that are most efficient for dealing with these unknown contingencies. In other words, if ‘transaction costs incorporate the value of adaptations to future circumstances [. . .] then the parties cannot know (or agree upon) the objective transaction costs they face. Such knowledge would imply cognitive abilities sufficient for comprehensive contracting, which has already been ruled out as infeasible’ (Dow, Reference Dow1987: 27). Thus, while actors’ calculative intentionality is restricted by their inability to imagine all possible future contractual contingencies and incorporate these into the contract, these same actors are fully able to accomplish the arguable more difficult task of comparatively evaluating the differential governance structures that may best protect the integrity of the contract against the unknown contingencies.

Second, there are also indications that economizing on transaction costs might necessitate taking longer time periods into account than economizing decisions that are made for single transactions. Hence, in cases where the governance of a certain transaction is linked in an inseparable way to the governance of other transactions (e.g., by prior contractual commitments), predicting the choice of governance structure may not simply be a matter of transaction-specific comparative analysis, but rather require an analysis of a larger set of temporally separated but contractually inseparable transactions. Argyres and Liebeskind (Reference Argyres and Liebeskind1999) argue that such governance inseparability plays an important role in explaining why similar transactions may be governed in different, and sometimes comparatively inefficient, ways. For example, the exclusive franchising agreements with independent bottling companies that Coca-Cola entered in the past have in some cases prevented forward integration through internal expansion and forced Coca-Cola to buy out these independent franchisees at a significant price premium in the market for corporate control (Argyres and Liebeskind, Reference Argyres and Liebeskind1999). Since governance inseparability increases the scope and complexity of comparative institutional analysis by introducing path-dependence, widespread occurrence of this phenomenon indicates that more extensive calculative intentionality is required to identify the best governance structure. This makes calculative rationality a less likely causal mechanism for linking governance structures and transactions.

Third, questions of appropriability are likely to enter decision-makers’ calculation of payoffs from a certain combination of transaction attributes and governance structure. Hence, in line with its emphasis on comparative efficiency and economizing on aggregate transaction costs, TCE does not directly account for the potential differential appropriability of transaction cost savings (i.e., payoffs) under different governance structures (Dow, Reference Dow1987). If in fact transaction cost savings are differentially internalized by the contractual parties, this may affect the parties’ preference across different governance structures and even give rise to situations where asymmetric appropriability/payoffs prevent desirable contractual solutions. The above conditions may thus create social dilemmas where private appropriation threatens joint value creation (Miller, Reference Miller1992). Contrary to, for example, principal-agent models (e.g., Arrow, Reference Arrow, Pratt and Zeckhauser1985; Grossman and Hart, Reference Grossman and Hart1983; Holmström Reference Holmström1979; Jensen and Meckling, Reference Jensen and Meckling1976), where incentive conflict is typically resolved (at a cost) by means of behavior- and outcome-based contracts (Eisenhardt, Reference Eisenhardt1989), the more restrictive informational assumptions in TCE make such contractual solutions less likely than in the principal-agent model where principals are only selectively unable to observe or verify some parameter, such as the agent's effort, while being otherwise fully informed (see Foss and Foss, Reference Foss and Foss2000). The impact of appropriation possibilities on governance choice would of course also be less significant if the parties to the transaction, at the time of the governance choice had access to, and complete knowledge about, a wide range of alternative contracting partners. This form of symmetrical governance (where actors have equal or no bargaining power) is, however, less likely to occur under conditions of uncertainty and bilateral dependence. This further increases the complexity of governance decisions and reduces the likelihood that boundedly rational agents are consistently and homogeneously able to reach a comparatively efficient solution.

Together, the three problems outlined above introduce complexities into uncertain governance choices that make it less likely that boundedly rational agents will in a homogeneous fashion be able to reach comparatively efficient solutions based solely on calculative intentionality. Instead, other causal mechanisms, such as environmental selection, competitive market pressures, and organizational learning, may be invoked to defend a comparative-efficiency mode of explanation (i.e., ecological rationality; see Smith, Reference Smith2008: 2). However, while these evolutionary mechanisms may be strong in the long run and on an aggregate level, they do not explain the relatively frictionless and homogeneous adoption of governance structures that result in the TCE model (see Hodgson (Reference Hodgson2013) for a discussion of selection and adaptability in organizational evolution). For example, competitive market selection involves some form of initial variation upon which market forces may work over time to select forms that are competitively and environmentally fit (Nelson and Winter, Reference Nelson and Winter1982). Similarly, change through organizational-learning mechanisms operates based on a notion of organizational variation and selection mechanism that due to myopia and organizational rigidity take substantial time to reach efficient outcomes (Levitt and March, Reference Levitt and March1988). Hence, as discussed in the next section, there are important insights to be gained from a more realistic representation of the process by which governance choices are made.

Evolutionary and learning explanations of governance

The problems associated with calculative intentionality as a causal mechanism for explaining the governance of incomplete contracts have led scholars to search for alternative mechanisms in the realm of market selection, economic evolution, and organizational learning. While TCE has not yet been integrated with these perspectives (Williamson, Reference Williamson1996), it has been argued that such a reorientation of TCE is a necessary next step for advancing our understanding of economic organization (e.g., Argyres, Reference Argyres2011). In line with this argument, prior studies have highlighted: the effects of economic change on governance (Langlois, Reference Langlois1988), dynamic governance costs (Langlois, Reference Langlois1992; Langlois and Robertson, Reference Langlois and Robertson1995), productive capabilities (Argyres, Reference Argyres1996; Langlois and Foss, Reference Langlois and Foss1999), bounded rationality and organizational inertia (Kreps, Reference Kreps1996), governance inseparability (Argyres and Liebeskind, Reference Argyres and Liebeskind1999), learning in contract design (Argyres, Reference Argyres2011; Argyres et al., Reference Argyres, Bercovitz and Mayer2007; Argyres et al., Reference Argyres, Felin, Foss and Zenger2012; Mayer and Argyres, Reference Mayer and Argyres2004), and governance as a determinant of learning and entrepreneurship (Foss and Klein, Reference Foss and Klein2012).

A common theme in the above-mentioned papers and books is that frictions, such as transaction costs, are positive across a broader range of economic domains than assumed in TCE. In other words, while TCE has traditionally isolated a rather narrow set of explanantia (linked to contractual incompleteness, opportunism, and bilateral dependence) and explananda (linked to firm boundaries, alliances, and decentralization), a common denominator in the above-mentioned literature is that it seeks to expand both these sets by including new forms of frictions and economic outcomes.Footnote ⁵ This reorientation leads to an increased emphasis on the process and learning aspects of contracting and governance, key aspects that are necessarily studied in disequilibrium as opposed to the type of equilibrium predictions that are commonplace in TCE.

As argued above, empirical research on contracts has, due to its focus on equilibrium outcomes, not extensively examined evolutionary and learning patterns in contract formation (Argyres et al. Reference Argyres, Bercovitz and Mayer2007). Contracts between exchange partners may however, in addition to their governance functions, also function as repositories of knowledge that incrementally improves as the exchange relationship matures (Mayer and Argyres, Reference Mayer and Argyres2004). This indicates that important aspects of contractual relationships are better described along the lines of evolutionary and behavioral notions of local and incremental learning (e.g., Cyert and March, Reference Cyert and March1963; Nelson and Winter, Reference Nelson and Winter1982) than with Williamson's (Reference Williamson1996) notion of farsighted (equilibrium-oriented) contracting. The lack of attention paid to this aspect of contracting is in part due to a too narrow view on the suitable application domains of incomplete contract theory where this theory is seen as explaining firm boundaries and asset ownership rather than providing broader accounts of how organizational primitives such as incentives, information, and authority enable an efficient organization of contracts (Argyres, Reference Argyres2011).

In a case study of a computer software firm's contracting with one of its major customers, Mayer and Argyres (Reference Mayer and Argyres2004) found that the contracting parties were relatively slow in identifying important contractual contingencies and including these into contracts. Consistent with TCE predictions, the original contract between the two parties was found to be highly incomplete. However, as the relationship unfolded and contractual problems arose, the parties made incremental modifications to the contract in order to enhance interfirm communication, clarify responsibilities and expectations, plan for contingencies, and modify the contract format. Mayer and Argyres (Reference Mayer and Argyres2004) describe this incremental process of learning to contract as building (heterogeneous) contract design capabilities. Similarly, in a larger sample of contracts for IT services, Argyres et al. (Reference Argyres, Bercovitz and Mayer2007) find evidence that firms learn over time how to discover and exploit combinations of different types of transactions and different contractual provisions, and that this may produce learning spillovers and complementary effects between different types of contractual provisions such as task description and contingency planning. Contractual contingency planning was found to become more prevalent over time in the studied contracts, which indicates that firms were becoming better at using contracts to safeguard future contingencies. Further support for that the performance of exchange relationships is dependent on the contract design capabilities of the involved parties is provided by Argyres and Mayer (Reference Argyres and Mayer2007). Specifically, they argue that firms may develop competitive advantage based on their contract design capabilities, especially if they are able to successfully define roles and responsibilities in the contract design process, provide for communication between the parties, and secure contingency planning.

The developments in TCE discussed above illustrate how the traditional more restrictive application of uncertainty in exchange relationships has limited the scope of institutional analysis to the distribution of activities and transactions across alternative governance structures. To the extent that organizational capability and learning have entered the discussion of governance (e.g., Argyres and Mayer, Reference Argyres and Mayer2007; Argyres et al., Reference Argyres, Bercovitz and Mayer2007; Mayer and Argyres, Reference Mayer and Argyres2004), it has primarily been through a ‘capabilities first’ logic where capabilities affect governance but not the other way around (Argyres et al., Reference Argyres, Felin, Foss and Zenger2012). Overcoming this one-sidedness and exploiting the full potential of TCE for understanding organizational phenomena involves a change in TCE's basic assumptions related to the application of bounded rationality and the extent to which knowledge and technology are assumed to be perfect, or rather fixed, outside the contract-writing domain. Such a change would not only result in a more plausible and consistent causal story but also allow the theory to address new explananda such as learning, organizational novelty, entrepreneurship, and competitive heterogeneity. For example, what may be termed a ‘governance first’ or a ‘govern to learn’ logic would directly examine how contractual parties ‘differ in the skill, agility, and competence with which they govern’ (governance capability) and the ‘entrepreneurial process of assembling unique resource configurations and selecting the associated form of governance’ (governance of judgment) (Argyres et al., Reference Argyres, Felin, Foss and Zenger2012: 1217).

Knight, Coase, and the role of judgment in the theory of the firm

The modern theory of the firm has been hugely influenced by Frank Knight (Reference Knight1921) and Ronald Coase (Reference Coase1937). These seminal contributions to organizational economics provide two related, but yet different, alternatives for theorizing the firm. Coase's (Reference Coase1937) basic argument was that there are costs associated with using the price mechanism in markets, which makes it cheaper, under some conditions, to organize transactions by means of administrative direction (i.e., firms). Firms in the Coasian framework may thus be viewed as primarily (transaction) cost-reducing entities, whose boundaries are established by the marginal analysis of the costs and benefits of organizing transactions inside firms versus organizing them on markets. The role of the manager/entrepreneur in establishing and governing a firm in this setting is one of ‘minimizing the cost of producing a given output with given inputs’ (Boudreaux and Holcombe, Reference Boudreaux and Holcombe1989: 148). In Coase's (Reference Coase1937) original analysis, these costs arise primarily as a result of incomplete and dispersed knowledge (e.g., marketing costs), while in Williamson's (Reference Williamson1985) later version of TCE, they arise from incompatible incentives and having to safeguard the risk of opportunism in exchange relationships (Loasby, Reference Loasby1976). Hence, while sharing the view of the firm as a primarily transaction cost-reducing entity (as emphasized by Boudreaux and Holcombe, Reference Boudreaux and Holcombe1989), there is a subtle but important difference between Coase and Williamson regarding how they relate to uncertainty in contractual relationships (Foss, Reference Foss1997). Whereas Williamson (Reference Williamson1985) emphasizes managerial farsightedness in the comparative analysis of differential governance structures, where a large number of contingencies prevent boundedly rational managers from drawing up contracts that completely safeguard opportunism, Coase's (Reference Coase1937) analysis is more concerned with the comparative benefits of internal organization in terms of its superior flexibility (due to unspecified employment contracts) when confronted with unknown contingencies (Loasby, Reference Loasby1999: 79).

Despite the above-mentioned differences between Coase (Reference Coase1937) and Williamson (Reference Williamson1985), where Coase's (Reference Coase1937) analysis arguably builds less on farsighted planning, Boudreaux and Holcombe (Reference Boudreaux and Holcombe1989) argue that the transaction cost view of the firm is altogether one of general equilibrium since the actual goods that are produced, and the methods by which these goods are produced, can be taken as given in the comparative analysis of transaction costs (see also Earl, Reference Earl and Pressman1996: 47). Hence, while the transaction cost approach has been successful in terms of providing tractable and empirically operationalized answers to many important questions in economic organization (e.g., Grossman and Hart, Reference Grossman and Hart1986; Hart and Moore, Reference Hart and Moore1990; Williamson, Reference Williamson1985), it nonetheless begs the question why transaction costs, which are the basic reason for the existence of firms in Coase's and Williamson's framework, are supposed to arise in the first place if markets are in general equilibrium, a condition that is associated with frictionless competitive markets (see Loasby, Reference Loasby1999: 80–81).

Knight (Reference Knight1921) approached the existence of firms from a different, while potentially complementary (see Foss, Reference Foss1993), perspective than Coase (Reference Coase1937) by explicitly building his theory in a disequilibrium setting subject to extensive uncertainty that does not allow entrepreneurs to costlessly and accurately map the relationship between inputs, outputs, and prices (i.e., the givens in the TCE framework). Hence, while institutional analysis in terms of comparing bureaucracy and market costs on the margin may still be important for firms’ organizational decisions, Knight (Reference Knight1921) views decision-making under true uncertainty about which productive activities to engage in as the crucial role of the entrepreneur and the fundamental reason for the existence of firms.

Economic organization under true uncertainty

An important contribution of Knight (Reference Knight1921) was the distinction between risk and true uncertainty, and the linking of the latter to the existence of profits through entrepreneurs’ willingness and ability to accept accountability for this specific form of risk by exercising judgment, a form of intuitive estimation in business situations that resist logical and statistical probability assessment (Loasby, Reference Loasby2007; O’Driscoll and Rizzo, Reference O’Driscoll and Rizzo1985).Footnote ⁶ Since the marginal product of judgment cannot be assessed because the ‘decisive factors in the case are so largely on the inside of the person making the decision that the “instances” are not amendable to objective description and external control’ (Knight, Reference Knight1921: 311), judgment cannot be paid a wage, traded, or contracted upon. As a result, the entrepreneur is required to start a firm in order to take economic actions based on his/her judgment. The Knightian entrepreneur is thus to be characterized as a person who exercises judgment under true uncertainty, owns the productive assets necessary for realization of the business idea that result from exercising judgment, and that due to the ownership of the necessary assets is the residual claimant.

There have been several attempts to integrate the Knightian disequilibrium-based theory of the firm with the Coasian theory of the firm that today dominates organizational economic (e.g., Boudreaux and Holcombe, Reference Boudreaux and Holcombe1989; Foss, Reference Foss1993; Foss and Foss, Reference Foss and Foss2000; Foss et al., Reference Foss, Foss and Klein2007; Foss, Foss, Klein, and Klein, Reference Foss, Foss, Klein and Klein2007). An important critique in this relatively new stream of literature is that mainstream theory introduces uncertainty into models in a too piecemeal fashion (Foss and Foss, Reference Foss and Foss2000) and that the typical comparative-static equilibrium analysis that is used to explain boundary-, governance-, and ownership choices, limits the theory's applicability for explaining organizational phenomena related to organizational learning, novelty, entrepreneurship, and competitive heterogeneity (Foss and Klein, Reference Foss and Klein2012; Furubotn, Reference Furubotn2001).

Organizing judgment

Rather than building the analysis of firm boundaries and asset ownership on well-specified optimization problems for exogenously given transactions, Foss and Klein (Reference Foss and Klein2012) suggest that boundary and ownership questions may instead be explained in terms of the focal entrepreneur's incentives to secure ownership and control over assets that are complementary to his/her judgment-based business idea (see also Teece, Reference Teece1986). Boundary and governance decisions may thus be viewed as search activities aimed at identifying and securing new asset combinations that are complementary, not only to the entrepreneurial business idea, but also to the firm's current stock of assets. Navigating in such an unknown space of possible complementary asset combinations and directing search activities requires that entrepreneurs form specialized mental models as part of their judgment that are conducive to realizing their business idea. According to Foss and Klein (Reference Foss and Klein2012: 179), entrepreneurs form these heterogeneous ‘cognitive representations about which assets they need to secure the services from, the major contractual hazards associated with such procurement, and the most effective ways of protecting against such hazards’.

An important advantage of the approach described by Foss and Klein (Reference Foss and Klein2012) is that it allows for the analysis of heterogeneous organizational outcomes. Even when operating under similar circumstances, cognitive representations are likely to differ across entrepreneurs that form different judgments about the conditions they are operating in. Hence, one is to expect that similar transactions may be governed in different ways due to the differential cognitive representations formed by the entrepreneurs and the managers these may put in place (Foss and Klein, Reference Foss and Klein2012). The heterogeneity in outcomes that differential cognitive representations give rise to provides a space for addressing evolutionary and learning processes that have been raised as a source of concern in relation to the static treatment of boundary and ownership decisions in TCE (e.g., Langlois, Reference Langlois1988; Mayer and Argyres, Reference Mayer and Argyres2004) and reduces the emphasis placed on more extreme forms of calculative intentionality as a causal mechanism linking transaction attributes with governance structures. Furthermore, the approach relieves the problem in traditional TCE analysis related to logical consistency: decision-makers are modeled as severely limited by information and calculative constraints in all relevant domains.

A key question is how the more extensive informational and cognitive constraints suggested by Foss and Klein (Reference Foss and Klein2012) may manifest themselves in governance choices? As discussed earlier, Williamson's (Reference Williamson1985) rather superficial treatment of bounded rationality (mainly referencing the early work of Herbert Simon (Reference Simon1945)) provides little guidance on this question. However, bounded rationality and the particular heuristics and decision-biases that it gives rise to have been subject to detailed examination in social-cognitive psychology and experimental economics (e.g., Gigerenzer, Reference Gigerenzer1991; Gigerenzer and Hoffrage, Reference Gigerenzer and Hoffrage1995; Gigerenzer and Selten, Reference Gigerenzer and Selten2001; Goldstein and Gigerenzer, Reference Goldstein and Gigerenzer2002; Kahneman, Reference Kahneman2003; Smith, Reference Smith1991, Reference Smith2003, Reference Smith2008; Tversky and Kahneman, Reference Tversky and Kahneman1974, Reference Tversky and Kahneman1981). I suggest that this research provides important insights concerning how boundedly rational agents select governance structures under uncertainty, as well as the relationship between uncertain governance choices, current organization, and asymmetries in actors’ access to decision-supporting systems. Specifically, while acknowledging that some of the results from these studies are contested (see Gigerenzer, Reference Gigerenzer1991; Gigerenzer and Hoffrage, Reference Gigerenzer and Hoffrage1995, etc.), I primarily build on what has been known as the ‘cognitive biases’ stream (e.g., Tversky and Kahneman, Reference Tversky and Kahneman1974, Reference Tversky and Kahneman1981, etc.) in order to highlight how asymmetries in human decision-making under true uncertainty (i.e., cognitive biases) may translate into asymmetries in governance and economic outcomes.

Studies of judgment under uncertainty in social-cognitive psychology

Experimental studies of human judgment under uncertainty have shown that decision-making is guided by particular heuristic rules that may give rise to errors (see Kahneman (Reference Kahneman2003) for an overview). In their seminal paper on judgment under uncertainty, Tversky and Kahneman (Reference Tversky and Kahneman1974) show that judgment suffers from errors originating with the common use of three types of heuristics: representativeness (probabilities are evaluated by the degree to which A is representative of or resembles B), availability (probabilities are evaluated based on the ease with which instances or occurrences can be brought to mind), and anchoring/adjustment (probabilities are evaluated by starting from an initial value that is later adjusted to yield the final answer).

The types of biases described above are related to the specific ‘decision frame’ that an individual adopts for a certain decision or judgment. Tversky and Kahneman (Reference Tversky and Kahneman1981: 453) define ‘decision frame’ as ‘the decision-maker's conception of the acts, outcomes, and contingencies associated with a particular choice’. Decision frames are thus specific mental models or cognitive representations that result partly from the formulation of the problem at hand and partly from the particular norms, habits, and other individual characteristics of the decision-maker (cf. Foss and Klein, Reference Foss and Klein2012). According to Tversky and Kahneman (Reference Tversky and Kahneman1981), there is experimental evidence that decision frames significantly affect human decision-making in ways that are not consistent with the rational choice principles of consistency and coherence. In other words, changes in the formulation of the same decision problem produce shifts in preferences.

The experimental findings discussed above have been used to develop a more general critique of expected utility and rational choice theory on the basis that assumptions, such as invariance and dominance, under some conditions appear to be invalid descriptions of actual choice behavior (Tversky and Kahneman, Reference Tversky and Kahneman1986). For example, the validity of expected utility theory has been studied in the context of subjects’ judgment as to their willingness to pay (WTP) and willingness to accept (WTA) under market-like experimental conditions (e.g., Knetsch and Sinden, Reference Knetsch and Sinden1984; etc.). Several of these studies have found a disparity between WTP and WTA that can be attributed to how these decisions are differently framed as either giving up something perceived as an endowment or not so perceived (Knez et al., Reference Knez, Smith and Williams1985). While these observations of WTP and WTA divergence are initially significant, there is also evidence that values converge over time as subjects learn to become ‘more rational’ from market experience (Coursey et al., Reference Coursey, Hovis and Schulze1987; Knez et al., Reference Knez, Smith and Williams1985) and develop institutional arrangements that are conducive for mitigating initial biases stemming from bounded rationality (Ostrom, Reference Ostrom1998).

The ‘anomalies’ in exchange decisions that are discussed above have been identified and described in several studies. For example, Samuelson and Zeckhauser (Reference Samuelson and Zeckhauser1988) report several experiments that show support for what they term a status quo bias in decision-making. Similarly, Kahneman et al. (Reference Kahneman, Knetsch and Thaler1990) report support for what they term an endowment effect in trades. As pointed out by Kahneman et al. (Reference Kahneman, Knetsch and Thaler1991), these effects are important because they indicate that economic behavior (and thus also organization) may be much less symmetrical and reversible than predicted by standard equilibrium models.Footnote ⁷ This may in turn significantly affect the bargaining positions and incentives of exchange parties. Specifically, I argue that these asymmetries in economic judgment produce corresponding asymmetries in governance and payoffs.

While many studies of judgment under uncertainty have focused on more immediate and intuitive forms of judgment, decisions about exchange and its organization are in many (if not most) cases made in organized and thereby at least partly rule-based forms. Hence, while the above discussion suggests that humans are subject to quite substantial biases in their judgments about uncertain situations, organizational, and other rule-based formats that allow decision-makers to take on an ‘outside view’ (as opposed to an ‘inside view’) have proved to be able to correct many of these potential misjudgments by individual agents (Kahneman and Lovallo, Reference Kahneman and Lovallo1993; Smith, Reference Smith2008). For example, Stanovich and West (Reference Stanovich and West2000) distinguish between what they term System 1 and System 2 operations in decision-making: System 1 processes are described as ‘typically fast, automatic, effortless, associative, implicit [. . .], and often emotionally charged; they are also governed by habit and are therefore difficult to control or modify’; while System 2 processes are characterized as ‘slower, serial, effortful, more likely to be consciously monitored and deliberately controlled; they are also relatively flexible and potentially rule governed’ (Kahneman, Reference Kahneman2003: 698). The interplay between System 1 and System 2 is particularly important for judgments made in organizations. In this setting, TCE predicts that significant resources will be spent on designing organizational systems that are adaptive to, and safeguard against, the risk of opportunism (Williamson, Reference Williamson1985). However, as is discussed in the next section, it is also reasonable, given the prevalence of decision-biases, to conjecture that these organizational systems will be designed to counter or lessen the potentially negative economic impact of known decision-biases. In other words, boundedly rational agents acting under uncertainty will not only seek to protect themselves against unknown contractual contingences but also against the risk of making ‘mistakes’ or simply miscalculating the benefits and costs associated with a certain combination of transaction and governance structure.

The judgment corrective role of current governance structures

As argued by Argyres and Liebeskind (Reference Argyres and Liebeskind1999), governance choices may be history-dependent to the extent that future contractual relations are contingent on prior contractual commitments and organization. However, this path-dependence may not be solely born out of the fact that historical commitments affect the value of future contractual commitments, but also from the effect that current governance structures have on actors’ judgment of future exchange relationships and the preferred form of governance for these exchange relationships. Hence, the distinction between System 1 and System 2 operations (see Stanovich and West, Reference Stanovich and West2000) indicates that the detection of biases in the contracting parties’ judgment of what constitutes an efficient governance structure for a given transaction may be highly dependent on the type of decision-supporting systems that are in place prior to the decision (see Kay, Reference Kay2000). For example, due to status quo bias, there might be an initial reluctance among decision-makers to adjust governance structures for a given transaction as transaction attributes change. Consider, for example, Kodak's reluctance to adjust firm activities in response to the emergence of the new digital photography technology, which made traditional elements of Kodak's business model outdated (see Hiltzik, Reference Hiltzik2011). The probability that these types of biases in governance choice are detected and corrected is likely higher if decision-makers operate within an organizational framework that provides slower, serial, effortful, consciously monitored, deliberately controlled, and rule-governed procedures for how such decisions are to be made (i.e., System 2 operations). In other words, the particular organizational framework in which governance and boundary decisions are made may significantly affect the outcome of such decisions (e.g., the organization of Kodak may have affected the decision to not switch focus from film photography).

Decision-biases might also be related to cognitive frames, representativeness, and availability. For example, because governance choices are typically framed as decisions of firm boundaries for intrafirm decision-makers, the particular framing of this situation may in some situations give rise to what may be termed a ‘hierarchy bias’ in governance choices. The detection and correction of this type of bias is dependent on the particular organizational framework within which the decision is made: a particular form of internal organization might be chosen because the transaction in question resembles other known stereotypically internal transactions (representativeness), a particular form of internal organization might be chosen because of its higher retrievability by decision-makers that operate from inside a firm and who are thus surrounded by internally organized transactions (availability), etc. Hence, under conditions of uncertainty, it is reasonable to conjecture that current governance structures do not only perform the function of protecting the integrity of contracts from the risk of opportunistic behavior but also play an important role in supporting the identification of judgment errors and enabling an effective framing of future governance decisions made within the realm of that governance structure.

For example, it has been shown that the use of common value auctions in the oil and construction industries leads to asymmetrically large payoffs for the auctioneer due to judgmental failures in the competitive bidding process known as the ‘winner's curse’ (Capen et al., Reference Capen, Clapp and Campbell1971; Dyer and Kagel, Reference Dyer and Kagel1996; Thaler, Reference Thaler1988). However, studies also show that these judgmental failures are mitigated by the adoption of particular institutional arrangements, such as dealing with many projects, so that gains and losses may even out over time and specialization in certain types of projects (Dyer and Kagel, Reference Dyer and Kagel1996). More generally, this illustrates how particular governance structures may result in asymmetrically distributed payoffs among cognitively biased actors with differential access to decision-supporting systems. Specifically, I argue that governance structures are heterogeneous with regard to the extent that they allow for consolidation (dealing with groups of similar cases rather than with individual ones) and specialization (specializing an individual's judgment to dealing with a particular type of uncertain situation).Footnote ⁸

Contractual positions within a particular type of governance structure that are associated with high contractual intensity of a certain type carry comparative benefits. In other words, there is in many cases an asymmetry between contractual parties in terms of the extent to which they can use the current governance structure or organization to consolidate and specialize in the particular type of contractual decision. An example in the area of sales force integration is the contractual relationship between entrepreneurs (or managers with delegated authority) and sales representatives (see Anderson, Reference Anderson1985; Anderson and Schmittlein, Reference Anderson and Schmittlein1984). The entrepreneur may, for example, prefer employment contracts to the alternative of outsourcing sales activity to an independent market firm, not because it is the most efficient solution in terms of reducing the cost of ex post opportunism, but because given the parties’ differential access to decision-supporting systems, it is the solution with the most favorable payoff for the entrepreneur: the entrepreneur may enter hundreds of contracts with sales representatives in a year and have extensive possibilities for consolidation and specialization, while the sales representatives may only enter one such contract and have no possibilities for consolidation and specialization in area of contracting for sales services. Hence, the sales representative may not, due to lack of possibilities for consolidation and specialization, be able to properly judge the situation, and agree to unfavorable terms that would not be entered if the decision were made rationally with complete information.

Governance asymmetry and expected biases

The internalization of the benefits and costs associated with a particular form of governance may be asymmetric across contracting parties. This is particularly true if, as indicted above, agents enter contractual relationships with decision biases that prevent them from considering the full economic consequences of the relationship. Without full appropriation of benefits and costs, the governance choice becomes decoupled from the comparative efficiency criterion that has traditionally guided TCE. This asymmetry in the distribution of payoffs may be further intensified if the future payoffs from the relationship are uncertain and the parties’ estimates of this payoff are subject to asymmetrically distributed cognitive biases. Consider, for example, the choice of a particular internal organizational form in the context of employer–employee exchange relationships (e.g., partnerships with an ‘up or out setup’ in the context of professional service firms). In this context, the choice of organizational form may not only be chosen based on its transaction cost economizing properties compared to other organizational forms but also based on the extent to which it supports a favorable distribution of surplus for either party based on the known biases of the involved actors and the available resources for correcting these biases. For example, to the extent that an employee is expected to overestimate future total payoff/compensation (e.g., due to framing effects related to a high availability of ‘success cases’ of high earners or status quo biases that over time decrease mobility and bargaining power), employers may select organizational forms that make these biases even more severe (e.g., by offering a large share of employee compensation as future conditional payment as in a partnership with an up-or-out structure), and thereby maximize their private payoff from the relationship. Hence, under conditions of uncertainty, it is reasonable to conjecture that governance structures provide contractual parties with differential access to decision-supporting systems to detect and correct biases, which lead to systematic deviations in the correctness of different actors’ judgments and the benefits they derive from the relationships. This, in turn, implies that actors’ expectations of other parties’ particular judgment biases are likely to affect their preferences over differential governance structures.

For example, consider an entrepreneur's decision of whether to work with employed vertically integrated sales representatives. Contrasting the situation of the entrepreneur who has an established hierarchy setup to govern sales activities, and the situation of the individual sales representatives, points to several areas where the entrepreneur because of the attributes of the current hierarchy is likely to be better equipped to make economically sound judgments about the future contractual organization of sales activity. The entrepreneur might, due to the current managerial hierarchy, at a low cost have access to aggregated sales/salary data from repeated product and factor market transactions (consolidation) and specialized human capital for market forecasting (specialization). Such attributes of the current governance structure affect uncertain governance choices, and thus also the specific future governance structures that are chosen. In other words, the informational advantage of the entrepreneur is likely to lead to future governance choices that further reinforce this advantageous position of the entrepreneur. Hence, as the sales organization increases in scale and contractual intensity, the judgmental and informational benefits that the entrepreneur gains from consolidation and specialization increase, thus further increasing the incentive to remain vertically integrated into the activity.

While the above example concerns employer–employee relationships between an entrepreneur and individual sales representatives, the economic effects of governance asymmetry and expected biases should not be viewed as limited to this specific type of exchange relationship. The economic effects of governance asymmetry and expected biases can be generalized to any exchange relationship, which is characterized by an asymmetry in the exchange parties’ access to decision-supporting systems. Generally, different types of governance structures, ranging from hierarchy (e.g., partnerships with ‘up or out setup’) to market (e.g., common value auctions subject to winner's curse), may satisfy the requirement of granting asymmetric access to decision-supporting systems. However, as indicated by the example of sales force integration, the typical employment contract exhibit specific properties (e.g., exclusivity, flexibility, etc.) that may be particularly important for how governance asymmetry and decision-biases affect firm boundaries and vertical integration (see Loasby, Reference Loasby1999). Specifically, based on the above analysis, I find that employment contracts that grant a focal entrepreneur asymmetric access to decision-supporting systems may in general provide entrepreneurs with greater incentive to vertically integrate than alternative governance structures where exchange partners are less exclusively tied to the entrepreneur. Hence, in deriving preliminary predictive implications from the theoretical argument presented above and relating these implication to TCE, one may argue that by overstating the power of constructivist rationality in uncertain governance choices and thereby disregarding the potential effects of decision-biases and the role of decision-supporting systems, the standard TCE model is likely to underestimate the degree of vertical integration in actual firms.