Measurement Costs and Governance Perspectives: two views about the same subject.
Decio Zylbersztajn1 /email@example.com
School of Economics & Business at the University of São Paulo
National Council of Scientific Research (CNPq)
Abstract: The evolution of the theory of the firm departed from the production function paradigm towards a more general scope dealing with coordination between firms as well as the internal organization of the firm. The transaction cost perspective motivated a large volume of empirical studies based on the relation between risk of appropriation of quasi-rents and investments in specific assets. Another theoretical approach to the firm is the measurement cost theory. Also rooted in the transaction cost perspective, it offers an empirically promising structure yet did not show the same empirical impact. Both theories share similar grounds but differ in internal logic, explicit assumptions, and key measurable variables leading to methodological implications. The aim of this paper is to contrast the two theories discussing the theoretical and empirical boundaries.
It is organized in five parts. In the second part the theoretical constructs of the measurement cost and the transaction cost economizing perspectives are presented. The third part discusses the testable hypothesis and predictions offered by both models. Part four discusses a case and possible contradictions that similar observed phenomena can bring by using both lenses. Finally, part five concludes with the delimitation of the theoretical scope, and particularly the discussion on possible accommodation of eventual frictions present in both constructs.
Key words: measurement costs, transaction costs, theory of the firm.
September 2005 – preliminary version
Measurement Costs and Governance Perspectives: two views about the same subject. Plainly an integrated treatment of governance and measurement is ultimately needed (Williamson, 1996). The measurement cost approach is applicable to vertical integration. Moreover it is more general and more operational than that of specific assets. (Barzel, 2003).
The present paper aims to compare the Governance Perspective (GP) and Measurement Cost Theory (MCT) considering three basic elements. First, the meaning of the firm, second the theoretical main cases, and third the derived testable hypothesis.
The theory of the firm has evolved based on contributions of Coasian rooted authors such as Harold Demsetz, Oliver Williamson, Yoram Barzel and Oliver Hart, among others. The set of theories contributed to depart from the scope of the production function paradigm towards more general coordination problems, including vertical and horizontal coordination as well as the internal organization of the firm. Particularly the transaction cost perspective motivated a large volume of empirical studies, and refutable hypothesis tests. The governance perspective (GP) has paved the road to make transaction cost theory a case of success, as Williamson stated. Another vein based on transaction cost perspective is Barzel´s measurement cost theory (MCT). This theory offers a more general perspective since it embraces horizontal coordination and the internal structure of the firms however it did not show the same empirical impact. Both theories share similar grounds but differ in internal logic, explicit assumptions, and key measurable variables leading to methodological implications, which might explain the differences in observed empirical impact. The aim of this paper is to contrast the two theories suggesting an explicit connection between its basic constructs, namely the measurement cost and asset specificity. Based on the proposition the paper discusses empirical limitations and presents a case of puzzling contradiction.
The most frequent proposition presented by the family of theories of the firm based on the transaction perspective is that all of them depart from, and offer criticisms to, the Walrasian model and to the production function paradigm. The criticism is essentially placed on the zero transaction cost assumption and the answer is built based on the Coasian analysis of the firm as a nexus of contracts, where positive cost of transaction applies. Once the introductory remarks are made, then diverse roads are chosen to explain the existence, scope, size and internal organization of firms, leaving the suspicion that some work still remains to be developed in order to clear the theoretical interfaces and advance towards a future unified theory of the firm, both useful to explain its existence as well as the inter-firms relations and their inner organization. The present study represents an incremental effort towards this objective, being limited in the sense that it is both narrow and focused only in two of the existing transaction cost based theories, namely the measurement cost (Barzel,1997,2001) and the efficient governance perspectives (Williamson,1989; Klein,Krawford,Alchian,1978).
The motivation for this particular study is originated from the suggestive theoretical constructs offered by theories, its clean logic and actual as well as potential use for empirical analysis. The theoretical evolution allowed the advance of the study of the modern firm, mainly within the paradigm of vertical coordination, however leaving the users, in many cases, puzzled with respect to the complementary or competing nature of the theoretical constructs. As stated by Williamson (1996, p.65) in discussing the measurement branch;”….plainly, an integrated treatment of governance and measurement is ultimately needed.” Also the same author states that both approaches are not independent. “The difference in emphasis is nevertheless real and needs to be highlighted” (Williamson, 1975, p. 81).
Likewise, Barzel considers that t is difficult to distinguish among the several theories of the firm, since its assumptions are most of the times, implicit. Also Barzel (2003) makes explicit comments on the governance approach. He concludes that the notion of measurement cost is more general than that of specific assets, being also more operational. Moreover he stated that the GP perspective deals with difficult to measure variables and the empirical tests are of fragile nature, while the MCT is more readily testable.
Formalization intentions absent, this introductory study aims to contribute to the analysis of the modern theory of the firm, placing concepts on a comparative means. It runs in the same stream as other authors, as Gibbons (2004) moving towards the direction of comparative analysis, maybe reflecting the momentum of the status of the theory, and the concerns of its users.
The present paper is organized in five parts. In the second part the theoretical construct of the measurement cost model and the transaction cost economizing perspective are presented. The third part discusses the testable hypothesis and predictions offered and part four presents some of the dynamic aspects present in both models. Considering that alternative theories can be discriminated by its empirical results, part four introduces a theoretical connection linking asset specificity (k) and measurement costs, giving room for comparative empirical tests. The same chapter discusses an empirical example and possible contradictions raised when the same phenomenon is observed adopting both lenses. Finally, part five concludes with the comparative delimitation of the theoretical scope, empirical perspectives and particularly the discussion on possible accommodation of eventual frictions present in both constructs.
2. Theoretical Construct:
Definitions matter, in the sense that they offer guidelines to the users of the theory. Contrasting how a firm is defined by GP and MCT is an adequate starting point. Under the governance economizing perspective Williamson sees the firm as a governance structure, defined as the institutional framework in which the integrity of a transaction, or related set of transactions, is decided (Williamson, 1996 p.11). This definition stresses organization rather than technological features to define the firm. Barzel (1997, p.81) defines the firm as “..a set of contracts whose variability is contractually guaranteed by common equity capital. The firm, then, is a nexus of outcomes guarantees”. The same author offers another definition (Barzel, 2001, p.20): “The firm is a nexus of the agreements and parts of agreements guaranteed by centralized equity capital and enforced without the state’s assistance. The firm’s owners are the individuals who are the residual claimants to the value guarantee capital. The scope of the firm is the ratio of its guaranteeing capital to that of (some sort of) its expected guaranteed payments.”
Williamson (1985,p.24) presents a cognitive map of contract, where he places both approaches, governance and measurement, in the same branch of literature, both are rooted in the efficiency perspective and both share a transaction cost perspective. The author chooses the governance approach however makes the interdependence between both approaches explicit.
The transaction cost argument proposed by Williamson is based on the existence of different transaction dimensions, namely; frequency, uncertainty and asset specificity which, in presence of opportunism are expected to indicate the choice among alternative feasible governance modes, in a discriminatory way, moved by transaction cost economizing perspective. Transaction is the observable unit of analysis and individuals follow a decision rule marked by the efficient alignment between the transaction characteristics and governance choice, given the institutional environment and the behavioral assumptions of bounded rationality and opportunism. Individuals decide to place the transaction within the boundaries of the firm, in cases where the risk of ex-post expropriation of valuable specific assets is high. Therefore, strict control is associated with vertical integration; while market transactions are related to those where low asset specificity prevails, leaving no room for expropriation. The theory deals with the need to coordinate independent agents to promote cooperation in production and recognizes the existence of contract breaches and the possibility of external and unpredictable shocks, added to the demand of specific investments in non-deployable assets, framing the basic structure of the model.
The key argument is borrowed from Klein, Krawford and Alchian (1978), of expropriable quasi-rents associated to specific investments. The basic alignment hypothesis is represented in figure 1, where institutional environment, transaction characteristics and behavioral assumptions interact to define the efficient solution of governance.
Changes in the institutional environment have an impact on the transaction cost minimizing solution, acting as shift factors inducing the realignment. Therefore institutions are taken as exogenous constructs of the TCE theory2.
The measurement cost model departs from the proposal that property rights matters, which is not a unique proposition in the family of theories of the firm. However it adds that there is a difference between economic and legal rights, being the first guaranteed by firms and the second by the state. The main case states that what precludes a Pareto optimal contractual solution is the cost to measure the attributes being transacted. The easy to measure dimensions, are contracted while the difficult to measure attributes open room for capture of value and are expected to remain within the firm. Measurement costs are associated to capture of value, therefore one can state that if measurement is needed in order to avoid capture, there is an implicit assumption of opportunism, yet it seems that MCT does not consider necessary to make an explicit argument for this particular behavioral element. The purpose of efficiency is present in the sense that the logic of property allocation follows a value maximization perspective. Three building blocks are relevant. The first is the contractual nature of the firm where property rights are transacted to achieve production3. The protection of rights is in the root of the model, sharing therefore the transaction cost perspective. Second, both product and inputs suffer from imperfect measurement, making it difficult to properly allocate incentives. Costly measurement makes contracts more difficult to make, precluding the advantages of specialization and allowing some transaction dimensions to become of public domain. Organizations differ in the way they build mechanisms to protect rights associated with given dimensions of transactions. Under Barzels´ lenses property rights might be of double nature, legal or economic rights, differing in the way enforcement is achieved. Both, weakly defined property rights and difficult to measure dimensions of transactions, are associated with positive transaction costs. The third building block of the model relates to the allocation of property based on the value maximization criteria. Property is efficiently allocated with the party that affects product variability and is able to offer guarantees, becoming therefore a residual claimant. As Barzel (1997,p.78) stated, “…as the effect a party exerts on the value of the outcome increases, rights will be better defined if that party assumes a larger share of the variability of outcome.”, becoming more a residual claimant.
Barzel proposes that the value of economic rights is affected by legal ownership, including the enforcement quality offered by the state. In the case of contractual dispute, the state might offer a solution at a lower cost than private arrangements. Otherwise, agents prefer to build alternative private solutions to protect its economic rights.
Both GP and MCT consider that institutions matter. In both cases institutions are taken as given, being therefore a shifter of the expected governance alignment. For MCT institutions matter since legal ownership is based on the state’s capacity to offer guarantees to the transaction based on a contract (see figure 2).
Both difficult to measure attributes and weakly defined property rights affect transaction in similar ways, i.e., they increase the costs of transaction and the likelihood of vertical integration. The existence of firm relies on its capacity to offer guarantees related to the variability in production, or in any significant dimension being transacted4. If variability cannot be anticipated, it cannot be contracted out.
Contrast 1: While GP deals with different effects of unanticipated shocks, where highly sequential shocks, in the presence of asset specific investments, are aligned with internal transactions whereas transactions characterized by inconsequential shocks are placed in the market, under MCT lenses the assurance of guarantee, either by the state or by the residual claimant ability to offer guarantees defines the firm, therefore the ownership is expected to be allocated to whomever offers equity capital.
Contrast 2: Both theories handle variability. The GP focus on the existence of external shocks and its effects classified as sequential, non-sequential and strictly sequential. The alignment hypothesis is rooted in the compatibility between the consequence of external shocks and the need of coordination. Likewise, the measurement cost perspective deals with variability, both of the product as well as of the factors of production. Therefore the theory recognizes that factors are not homogeneous, and product is not stable. Agents have to cooperate in production aiming to maximize the value of the outcome subject to positive measurement costs, and positive costs to write contracts, therefore opportunities for capture of value. In cases of high measurement costs the transacting agents are more exposed to value dissipation. In such cases the internal organization is an efficient response to the coordination problem, namely, one that places incentives that maximize the value of production avoiding losses of capture.
Table 1 explores some comparative features for both theories. The construct of the GP is condensed in the books Economic Institutions of Capitalism and The Mechanisms of Governance. Barzel´s contributions are first presented in the book Economic Analysis of Property Rights (1997), A Theory of State (2002) and several articles, especially “A Measurement Cost Based Theory of the Firm” (2001, 2003).
Contrast 3: Both theories consider the relevance of post-contractual hazards, however coming from different origins. The GP suggests that in the absence of reputation mechanisms, agents are expected to break contracts if benefits exceed the costs to keep the promise. Likewise MCT proposes that information is costly to produce, motivating errors in the allocation of incentives and therefore ex-post disputes for margins. As stated by Barzel (2003), “ …errors open the door to the expenditure of resources to capture wealth, because in their presence, among other things, people can pass off their poor performance as a result of random error”. Therefore the opportunistic behavior is present in both models in a more or less explicit way.
Both theories at a first glance, offer alternative explanations to the contractual phenomena and its empirical application follows from the possible establishing of parameters of its basic constructs and the possibility to propose refutable hypothesis.
Contrast 4: On behavior assumptions: The GP offers clear statement for the relevance of behavioral assumptions, basically opportunism and bounded rationality. The relevance is that contracts are incomplete and selective or distorted use of information can lead to ex-post expropriation. Barzel does not make explicit assumption related to opportunism, but signals that maximizing behavior is sufficient to explain the incentives for transactors to spend resources to capture unprotected margins.
In summary, the building blocks of both theories can be contrasted in terms of the unit of analysis, the behavioral assumptions, namely rationality and opportunism, how the firm is defined, rationality, relevance of institutions. Each one will have empirical implications. Both aspects are discussed bellow.
3. Testable Hypothesis:
Non-Walrasian theories of the firm, as any other theories, can be evaluated based on their capacity to offer refutable hypothesis and the adherence to the real world of organizations. The more clearly stated the main case and the better defined the building of theoretical blocks, the easier it is to the users to derive testable hypothesis from the theory and perform tests. The clear identification of the theoretical blocks actually facilitates the adoption of empirical tools.
The GP offers as its fundamental prediction the following: “the higher the level of asset specificity, the more vertical integration is expected to be observed as a way to economize in transaction costs.” Additional elements can be derived and follow from the theoretical perspective. First the alignment perspective permits to compare other observed transaction characteristics (not only asset specificity) with the observed governance mechanisms. Uncertainty plays a role and offers another empirical dimension far less explored in the literature, maybe due to the difficulty of measurement. Uncertainty and asset specificity suggest the benefits of post-contractual coordination mechanisms to deal with the effect of the uncertain and consequential impacts. It permits to choose the efficient governance mechanism based on more or less strict coordination devices, ranging from vertical integration to markets, including hybrid governance modes.
Cooperation is necessary and incentives for mutual adaptation are related to the performance of any given institutional arrangement. Frequency of transactions open room to introduce reputation mechanisms dealing with actors involved in a given transaction, being the same or not. It seems that reputation mechanisms associated repetitive of transactions are commonly studied, since it can be extended to network effects, in cases where the relevant information is transmitted to other players.
By and large the most relevant explanatory variable offered by GP is the level of asset specificity, where the empirical literature is mostly based. The larger the level of asset specificity, the more uncertain the transaction outcome, the more coordination adds value to the transaction. In such cases the governance mode more likely observed is vertical integration. Recursive transactions offer a mechanism of transaction stability, in the sense that it increases the hold up costs, suggesting a large self - enforcing contractual range acting as a protection to the value of specific investments, therefore less vertical integration solutions are expected.
The empirical use has been predominantly applied to the study of vertical integration and the governance choice between internal or market transactions. The theory evolved towards the study of hybrid forms, as discussed by Williamson (1991) and Menard (2004), however offering less empirical evidences to shed light on the organization of hybrid contractual formats. Both authors stress that the intermediary formats of organization are what we expect to find in the real world.
The measurement cost theory offers an alternative vein for empirical analysis, especially to deal with the hybrid forms. Barzels´ basic prediction and testable hypothesis is that: “Ownership of assets attributes is expected to gravitate into the hands of who are the most inclined to affect the income flow.” The scope of the firm is defined by the set of contracts that affect the product variability. Therefore the firm is seen as the nexus of outcome guarantees, or a set of contracts guaranteed by the equity capital and one can predict that the more variability, more liability, more residual claimancy. Therefore the theory applies to the question of which services should be owned or contracted. This part of the theory complements the GP in the sense that both deals with the size and scope of the firm, using somewhat similar lenses. To MCT the scope of the firm is defined by the nexus of agreements that affect the product variability, or the variability of a given dimension.
The role of institutions in MCT connects with the property rights guarantees. If the state offers good quality of guarantee of property rights then agents are more inclined to contract instead of owning the asset. Contractual hazards are controlled if disputes are properly handled by courts; therefore the value of the firm is expected to increase by an external contract. The easier to measure the attribute characteristics, either related to factors or to products, the more efficient will it be to place the transaction outside the firm, since measurability is easier to be handled and courts will have less difficulty to handle disputes. The firm is expected to design contracts and agreements in such a way that minimizes dissipation. Both the costs to organize and the benefits of decreasing value dissipation are contrasted, which means that the theory recognizes that cooperation is costly due to the risk of capture of value.
One can say that asset ownership is associated to the right over the income flow generated by the asset. If attributes vary, the flow becomes more difficult to measure therefore a non-owner might have opportunity to capture margins. More variability in production is associated to smaller predictability of the income flow, therefore more difficult to contract.
Relevant Theoretical Constructs Dimensions: The GP states clearly which characteristics of transactions are relevant dimensions to be considered, being asset specificity the most relevant (Williamson, 1996. p.45). Moreover the literature offers alternative measures for “k”, classified as location or site specificity, time, physic and human capital specificity. Different authors have found evidences of the relevance of different measures of k to explain vertical integration, as for instance Joskow (1993), Shelanski, Klein (1995) and Zylbersztajn, Lazzarini (2005) developing or reporting empirical tests for TCE based on the transaction cost minimizing alignment hypothesis.
The contrasts between the two approaches have been limited to the statements made by the main authors. Williamson (1985, p.81) offers a criticism to the MCT by stating that “…the relevant dimensions for asserting where the measurement difficulties reside remain somewhat obscure.” On the other hand, Barzel (2003) argues that the notion of measurement costs is more general than that of asset specificity. His argument stresses other mechanisms to protect quasi-rents, particularly considering the relevance of standardization. Moreover, Barzel opens room for a relevant vein yet to be explored, relating the cost to generate homogeneity, which will be addressed in the next section.
The key variables in MCT are less precise than in GP, however indicating the importance of the causes of transaction variability decoupled in its valuable dimensions. The unit of analysis can be considered as being the transaction, being general enough to handle transactions of inputs or products, as well as horizontal or vertical transactions. Under the MCT lenses there are at three aspects to be considered, two of them discussed in the literature and the third introduced here. The first related to the transaction characteristics, the second related to the transactors characteristics and the third introduces shifters in the efficient arrangement suggested by the conceptual model. Related to the first, any transaction can be decoupled in its different margins, which differ in terms of variability and are subject to capture. Also each margin can be measured at a different cost affected by the dispersion or the attribute, the predictability, the existing measurement technology. Related to the second, the relevant characteristic is the capacity to offer guarantee of capital, which determines ownership. Related to the third point, the question is about the external elements that can alter the efficient decision. I suggest three variables; the first the measurement technology which can make a given measure easier. The second are the investments associated with the generation of homogeneity (standards) and the third is the competence of courts (state) to obtain relevant information to handle disputes as well as the private mechanisms to deal with the non-contractible transaction dimensions.
The measurement technology can change during the process of transaction, changing the measurement costs. If costs are reduced, then a new efficient solution might become possible. Questions on the incentives to develop technology of measurement are related to the specific interests of transactors, unless government decides to intervene. The origination of private or public standards can also change the equilibrium, since given the measurement technology available, the enhanced homogeneity reduces the measurement costs. Standards, being public or private, have an effect in the technology of production, introducing a selective effect. Finally, the competence of courts to handle disputes is a variable to be considered, since the measurement technology might be available but a high cost to be handled by the state5. The expected-observed agreement is represented by the set of margins that are part of the whole transaction. If one consider the legal environment as given, it determines the predictability of the measurement, therefore affects the decision to transform the agreement in a contract (external transaction) or offers private guarantees (internal transaction). The competence of the state to handle disputes is one relevant issue for reform of legal systems.
Complex contracts: While GP points to the existence of complex contracts or hybrid forms, its main case is the make or buy paradigm in the vertical coordination domain. On the other hand the measurement cost perspective handles intermediary cases, horizontal integration and joint asset ownership. For instance, it can be predicted that two contracting parts, A and B, do not necessarily own an entire asset, but only some of its dimensions. Therefore if guarantees are too large to be handled either by A or B, then joint ownership can be predicted in a value maximizing way or an external agent might offer guarantees6. Recognizing that both GP and MCT are close theories, a possible interpretation of its boundaries, or main cases, is that the empirical history of the first deals predominantly with polar governance modes while the MCT can potentially deal with the observed complex contractual formats. However it must be said that the GP has a much more vigorous empirical literature than MCT.
Dynamics: Both theories point to the importance of changes over time, and equilibrium shifters, however many points remain somewhat obscure.
About the dynamics of k: Handling changes in asset specificity is a relevant aspect. If we allow for long term effects of transactions how will k change? Williamson considers that prices (p), safeguards (s) and asset specificity (k) are jointly determined. Can a more general model embrace the endogeneity of k? Likewise, the way measurement costs change over time and if the costs are given or not or the incentives for changes in technology of measurement can they be incorporated in MCT model? Similar considerations can be made about the treatment given to changes in institutions by both theories7.
In some sense GP has evolved sufficiently to motivate criticisms, as expressed in Langlois (1992,1998), Langlois and Foss (1999) relating to the dynamic elements and the negligence to incorporate routines and capabilities, and Ghoshal (1996) looking to organizations as having other operational advantages not considered by Williamson. The evolution of GP also permitted the existence of comparative empirical studies as expressed by Poppo and Zenger (2000), offering support to the GP arguments. It is my perception that the lack of criticisms on the MCT has to do with the lack of more empirical work, the ultimate ground to compare theoretical performance.
4. Linking MCT and GP
Both theories share common objectives, namely to explain the size, scope and structure of firms as well as patterns of coordination of production. Assuming that theories can be contrasted either by the realism of their assumptions as well as by their predictive power; the aim of this section is to offer a possible vein to empirically contrast both theories. Considering that governance costs and measurement costs are moved by value maximization purposes, both being equally acceptable, the distinctive element relies on the empirical strength of each theory to offer explanatory motives and testable hypothesis to discriminate among alternative institutional arrangements, based on efficiency principles. Before discussing an empirical example, it is necessary to explore a possible “missing” linkage between both theories in order to facilitate empirical contrasts.
4.1. Missing Linkage: The issue of the limits of the firm is discussed by Barzel (2001,2003) by means of a comparative analysis of GP and MCT which came to the following results:
There is a distinction based on the origin of incentive to integrate, which is the protection of quasi-rents (based on an ex-post perspective) and the alternative of saving in measurement costs. As stated by Barzel measurement concepts apply at any time while the quasi-rents rational actually represent an ex-ante perspective. Barzel states that the solutions offered by MCT cover a wider range of possibilities, including horizontal integration.
It can be stated that GP also offers support to explain hybrid contractual forms in general, however the empirical production has been, by and large, linked to the vertical integration paradigm.
Barzel affirms that MCT offers a more general rule: “As the delineation of the rights to the quasi-rents become easier, writing contracts becomes easier thus the incentive to integrate is reduced”. This rule is promptly able to generate testable hypothesis, since one can admit that a better delineation of rights might be associated with changes in measurement technology, improvement on the efficiency of courts, and changes in horizontal coordination tools as exemplified by the development and adoption of standards.
The GP also offers a general rule, stated as: “The higher the level of asset specificity, the larger the quasi-rents generated, the more likely to observe vertical integration as a form to protect non-redeployable investments. The general statement also suggests testable hypothesis, departing from measurement of asset specificity and uncertainty, being the first the most common support found in the empirical literature.
I maintain that whether the rule suggested by the Williamsonian GP is more or less general than the one offered by the Barzelian MCT, it is a matter of empirical analysis, since asset specificity and measurement costs are different and independent constructs and show similarly independent genesis. In order to leave the circularity of the reasons offered by both authors, two points deserve to be discussed; first, what is missing in the theoretical construct in order to make them comparable, and second, what the empirical analysis suggests as a response to the theoretical comparability. Let’s explore both aspects.
First, what is missing to make the connection between theories feasible? The answer must touch the keystones of each theory, namely the variability of relevant transaction attributes and asset specific investments. Considering that both are independent variables, and focusing a given transaction as unit of analysis, four possibilities can be observed (table 2). The case where k = 0 and measurement costs (MC) are irrelevant, which applies for perfectly homogeneous dimensions and easy to measure attributes, in which both theories predict that external transaction is likely to occur. Quasi-rents and risks of value capture are absent. GP predicts that transaction will be placed in the market and MCT predicts that external contracts are easy to be handled and courts are expected to solve any post-contractual dispute. No private guarantees are necessary.
The case where k = ∞ and MC = ∞ lead to the same result of internal solution. It is more efficient to vertically integrate to economize in post contractual transaction costs or, otherwise, to avoid capture of value related to difficult to measure attributes. Contracts are difficult to be crafted and courts will face enormous problems to solve disputes over property rights.
The other cases pictured in table 2 present the zone of dispute between both theories. The case of k = ∞ and MC = 0 indicates contradictory results. The GP suggests that vertical integration is the expected efficient solution while the MCT suggests that external contracts are expected to handle the transaction efficiently. Private guarantees are not necessary since courts will have no problems to solve post contractual disputes being easy to write contracts and govern the transaction out of the limits of the firm. However, the high level of asset specificity points to the other solution, namely the vertical integration to save in ex-post transaction hazards. Similarly the case of k = 0 and MC = ∞ indicates diverse efficient solutions.
The GP handles intermediary levels of asset specificity evolving in the direction of complex or hybrid contractual forms. The governance of contractual relations under the GP might evolve in complex patterns of cooperation, including network mechanisms based on reputation and relational aspects, crafting mechanisms to solve disputes and fill contractual breaches. Similarly the MCT allows the decoupling of several dimensions of a single transaction, allowing the immersion in the reality of complex transactions. Formal contracts and informal agreements are expected to govern particular dimensions of a given transaction, as pictured in figure 3.
Therefore, both theories handle complexity, but in different perspectives, leading to the possible complementarity of explanatory tools. However, in both cases, the empirical tests of hypothesis related to the choice of complex forms are so far rare and mostly based on descriptive case studies.
4.2 The Example: Consider the typical situation of vertical coordination among technologically separable activities, as pictured by Barzel (2003). The firm A supplies a product to a second firm B that delivers to the final consumer, C. Consumers are willing to pay a premium for a reduction in variability in one or more dimensions (Di) of the product.
< Insert figure 4>
If A and B can supply lower variability of Di with no specific investments, then they will be able to capture value since C is willing to pay for lower level of variability, being a trivial solution. Two different situations can be considered, one if A, B offer a measurable reduction in the variability of the valued attribute or if they have to join efforts to achieve the desired level of reduction. Certainly there are other situations where A, B or A+B will produce a reduction in variability, leading to arguments of efficient allocation of rights, generating testable hypothesis about who captures the value and who offers guarantees.
Now, suppose a simple case where A supplies B and makes a highly specific investment associated to the production of homogeneity. Consider that A must make a non re-deployable investment (highly specific asset) in order to produce a lower variability in Di. In that case, GP suggests that A and B must be vertically integrated in order to reduce the potential ex-post loss due to the hold-up risk. However MCT suggests that less vertical integration is associated with the reduction in variability of Di, with a contract being possible, since the state will be equipped to offer guaranties for that measurable dimension. It seems that we face a theoretical puzzle here.
Adding horizontal coordination: A common case can be observed in the agro-industrial markets, where a given attribute of quality is generated at the farm level and must be carried through the processing stage, to reach the final consumer. It is a typical situation observed in a large number of cases in the food industry (figure 5).
Suppose that A is segmented in many independent production units, each offering Ai to B, who intermediates the transaction with the final consumer C, as represented in figure 5. A new source of variability is introduced, since each firm Ai has different skills, different workers even if the same production technology is assumed to exist. Random variability applies. If the final consumer values a given level of variability in the final product, which is related to a maximum variability in A, say VAi, then it might introduce incentives for horizontal cooperation to produce Ai.
In this example, the production or maintenance of a given level of variability might be associated to alternative assumptions relating the presence of asset specific investments, as well as different incentives for horizontal coordination that include network externalities or cost sub-addictivity. Assuming, for simplicity that no network or cost incentives are present in A, basically one can foresee four situations:
a) No specific assets are demanded to produce or maintain the desired level of variability.
b) Only Ai must make asset specific investments in order to produce VAi.
c) Only B must make specific investments to reach the same goal.
d) Both, A and B must make asset specific investments to reach the desired level of production of attribute variability.
The first situation shows that some horizontal coordination is sufficient and no specific investment is necessary to produce VAi. The expected governance mechanism suggested by GP is market, matching with the MCT solution.
Case b represents a situation in which the production of the desired level of variability is associated with asset specific investments by Ai. Two aspects are relevant. First, a given mechanism of exclusion at the horizontal coordination level might be relevant, in order to preclude free riding, especially if the investments are made individually. The observability of the investments or the observability of the attribute plays a relevant role to guarantee the horizontal coordination.
The lower variability resultant from the horizontal coordination implies that contracts can be drafted to coordinate the transaction with B. On the other hand the specific investments made in A suggest that, under GP, vertical integration applies in order to economize in transaction costs.
Case c indicates a situation where variability is controlled by B, who makes specific investments. For Ai, no asset specific investments have been made, and the high level of variability is maintained. The result is that, under MCT, the high level of variability precludes the draft of contracts with B, and B does not run the risk of expropriation, since he sells to C the desired level of attribute, generating standards. Both theories agree in the coordination mechanisms adopted at transaction between A and B.
Case d presents a situation of credible commitment where both parties make specific investments whose value depends on the accomplishment of the transaction. It shows a situation of presence of bi-lateral specific investments associated with the reduction in variability. GP suggests that a contract might evolve since variability is under control; so does the MCT since specific investments have been made by both parts in the transaction. Therefore the same prediction is reached but for different reasons.
This chapter suggests that the choice of transactions governance is affected by the degree of asset specificity and by measurement cost reasons. In some cases, both outcomes are the same, however due to different reasons. In other case, namely, one where low variability is reached through specific assets investments, the expected governance mechanisms is opposite, suggesting that more empirical comparative work is necessary to be carried.
The analysis presented suggests some new veins to be explored in the interface of two theories of the firms and organization. First, the non Walrasian theories of the firm are still in a stage of development of formalization that hinder the comparative analysis. Second, there is a clear difference in the empirical effort found in the literature related to the GP and MCT. The contribution offered by Williamson has proven to be able to generate testable hypothesis treated in empirical studies carried in different countries and in almost all areas of economic interest, ranging from auto-industry to agriculture, and also finance, labor, and international trade, among others. A different stage is observed in the MCT, with a limited number of empirical studies and no surveys to collect the existing literature in a well organized format.
Third, the details matter. One can observe the movement of the theory out from the polar results of markets or hierarchies, or make or buy discussion, towards the study of hybrid forms. It suggests that MCT has a promising vein to be explored, in terms of the study of the details of complex mechanisms of cooperation. Basically the possibility to treat a given transaction in its decoupled dimensions opens room for very detailed studies and suggests the evolution of empirical studies based on the predictions offered by Barzel. The degree of detail is different from the one offered by Williamson, that also decouples the transaction in a different way, based on its characteristics of asset specificity, uncertainty and recursivity.
Fourth, institutions are present in both constructs and suggest the possibility to study comparatively the effects of inter-temporal changes, international contrasts of institutional arrangements, effects of reputation mechanisms and technological change.
Fifth, there are some puzzles yet to be treated. One is, if variability shows a predictable pattern, then it can be contracted. Governance structures can be designed to deal known patterns of variability. Two is the polar cases pictured in the previous chapter, where both theories point to different predictions. This is a good opportunity for comparative empirical analysis.
It seems that there are good reasons to believe that both theories have complementarities to be explored empirically. However the sharp interfaces must be lapidated in order to make less obscure the apparent frictions offered at the actual stage of knowledge. I do believe that the job to create new approaches to interpret the complex organizations was and still is being carried with elegant modes, enriching the availability of tools to be tested empirically.
Side by side with theoretical efforts to shape the theories, more comparative empirical tests might shed light in this particular matter. Considering MCT, its theoretical evolution invites the definition of standard aspects to be contrasted between different transactions, as well as measures of performance. Without improving the theoretical frame the tendency of the research based on MCT will be to follow an ad-hoc approach.
It can be stated that GP has shown to be more user friendly than MCT, in the sense that it has offered to the empirical research a wide range of possibilities to adopt the usual empirical tools in economics, based on better defined transaction dimensions. A further effort to evolve the empirical analysis based on MCT is ultimately needed to clarify the limits and the strains of the theory. I consider this a real opportunity for empirical work on firms structure, agreements and contracts.
Barzel,Y.1997. Economic Analysis of Property Rights. Cambridge University Press.161 p.
Barzel,Y.2001. A Measurement Cost Based Theory of the Firm. Working Paper. Unpublished.
Barzel,Y. 2002. A Theory of the State: Economic Rights, Legal Rights, and the Scope of the State. Cambridge University Press.289 p.
Barzel,Y.2003. Organizational Forms and Measurement Costs. Working Paper. Unpublished.
Gibbons,R.2004. Four Formal(izable) Theories of the Firm? Working Paper 04-34. Massachusetts Institute of Technology. SSRN collection. http://ssrn.com/abstract=596864.
Goshal,S., and Moran,P.1996. Bad for Practice: A Critique of the Trnsaction Cost Theory. Academy of Management Review, vol 21, n. 1. p.13-47.
Joskow,P.L. 1993. Asset Specificity and the Structure of Vertical Relationships: Empirical Evidence. In Williamson,O and Winter,S.G. The Nature of the Firm> Oxford University Press.
Klein,B.,Crawrord,R.A., Alchian,A.A.1978. Vertical Integration, Appropriable Quasi-Rents, and the Competitive Contracting Process. Journal of Law and Economics 21 (October):297-326.
Langlois,R.N.1992. Transaction-cost Economics in Real Time. Industrial and Corporate Change, vol 1, n.1:99-127.
Langlois,R.N.1998. Capabilities and the Thory of the irm, in: Economic Organization, Capabilities and Co-ordination, Foss,N and Loasby, eds. Routledge.
Langlois,R.N. and Foss,N.J.1999. Capabilities and Governance: The Rebirth of Production in the Theory of Economic Organization. Kyklos, vol. 52, n.2:201-218.
Lueck,D.2005. Yoram Barzel and the New Institutional Economics. University of Arizona. Unpublished paper based on the address given at the Plenary Session of the 8. meeting of the International Society for the New Institutional Economics.
Menard,C.2004.The Economics of Hybrid Orgnizations. Journal of Institutional and Theoretical Economics, 160:345-376.
Poppo,L and Zenger,T.2000. Substitutes or Compkements? Exploring the relationship between formal Contracts and Relational Governance. SSRN series, abstract_id=223518.
Shelanski,H.A. and Klein,P.G. 1995. Empirical Research in Transaction Cost Economics: A Review and Assessment. Journal of Law Economics and Organization, v. 11, n. 2.:335-361.
Williamson,O.E.1989. Transaction Cost Economics in Richard Schmalensee and Robert Willig eds. Handbook of Industrial Organization, vol 1.
Williamson.O.E. 1975. The Economic Institutions of Capitalism.: Firms, Markets and Relational Contracts. The Free Press.449 p.
Williamson,O.E.1996. Transaction Cost Economics. In: The Mechanisms of Governance. Oxford University Press.
Zylbersztajn,D and Lazzarini,S.G.2005. On the Survival of Contracts: assessing the stability of technology licensing agreements in the Brazilian seed industry. Journal of Economics and Business Organization. Vol 56:103-120.
1 The author acknowledges the comments and suggestions offered by Douglass Allen and the participants on the research group of the University of São Paulo and the Center for Studies o Law, Economics and Organization. The author is solely responsible for the contents.
2 Williamson discusses institutional change, but the assumption is that in the time horizon of interest, they are stable. The endogeneity of institutions is treated more adequately by Douglas North.
3 Barzel distinguishes two kinds of agreements. Self -enforced agreements are kept within the firm and corresponds to difficult to measure attributes. Otherwise they are guaranteed by third parties, namely the state, and are defined as contracts.
4 That is why Barzel insists in the importance to define what is being really transacted.
5 For example, in some countries, courts have limited access to DNA tests to be used as evidence of paternity, due to the high costs, limiting the use to produce evidences.
6 Barzel suggests that guarantees show scale economies based in risk sharing arguments. He does not discuss the possibility of scope economies, based on the concept of portfolio handling.
7 Institutions and governance are endogenous to the GP, however for practical purposes in dealing with governance mechanisms institutions can be considered as exogenous.
8 It would be relevant to clarify whether MCT relies on maximization behavior or opportunism.
9 As stated by Williamson (1975, p. 81) governance problems will vanish if either bounds on rationality or opportunism are presumed to be absent.