A Theory of Organizations
To Supersede the Theory of the Firm
Yoram Barzel
Introduction
The theory of organizations that I formulate in this paper has three central themes: 1.The relationship between enforcement methods and organization. 2. The problems that the costliness of measuring inputs and outputs bring about and the usefulness of the measurement framework for organizing transaction costs models of organization. 3. The claim that the firm is not a useful concept and should be replaced with a set of organizations according to their ownership structure. The theory of organizations offered here opens the way for predicting which activities will fall under each organizational form.
Agreements within and across organizations must be enforced, and enforcement issues suffuse agreement formation and execution. I focus on the difference between state enforcement and enforcement by long-term relations. State enforcement is at the interface of organizations, whereas enforcement by long-term relations characterizes, among others, within-organization interactions.
Costly information, and its operational counterpart, costly measurement, is a basic ingredient in the analysis of institutions. Adopted here is a measurement costs framework that unifies and augments the analysis of the leading transaction costs models of the firm. I contend, however, that in spite of its popularity, the firm itself is not a useful concept, and that we should rather focus on organizations.
In Barzel (2002) I argue that the state, with control over violence, is the efficient enforcer of easy to-delineate (i. e., to measure) transactions. These, governed by contract, are classified as being “in the market.” I also argue that except for caveat emptor transactions, contracts are used to enforce only parts of transactions, and that long-term relations are used to enforce other parts of the same transactions. The comparative advantage of the latter is in enforcing components that are less explicitly specified. This difference is significant, among other things, for the study of vertical integration and for explaining the radical change in the character of the pre industrial revolution business enterprise to the one that emerged subsequent to it. More generally, I attempt to explain which enforcement method will be utilized in particular cases or, in different words, which activities will be undertaken in the market and which (mostly) within organizations.
The definitions of organization and its scope that I adopt are as follows:
Definition: An organization is a nexus of the agreements and parts of agreements guaranteed by centralized equity capital and enforced without the state’s assistance.
Definition: The scope of the organization is the ratio of its guaranteeing capital to (some measure) of its expected guarantee payments.
The measurement cost framework enables us to consolidate the existing transaction costs models of the firm (now understood as organizations). I show that this framework can unify the models by Alchian and Demsetz (1972), Williamson (1975), Klein, Crawford and Alchian (1978), and Barzel (1982). It also encompasses the role of guarantee capital (Barzel and Suen (1995) and Barzel (1997)). The need for such capital helps determine the size and boundaries of organizations. An implication that carries through all these models is that as the cost of measuring commodity attributes declines, more activities will be carried out in the market and fewer within the firm (or within organizations).
Although this paper is in the tradition of the transaction costs theory of the firm, the firm is a too diffuse a target for explanation. Some writers on the firm, especially in the law and economics literature, equate firms to for-profit corporations, subject to corporate law. This is sensible in that it is clear what falls under the definition. More importantly, the rather common assumption that the sole objective of the firm is profit maximization seems reasonable here (though it is easy to think of real life caveats). Students of the firm, however, tend to hold a more expansive view of what constitute firms. Viewing profits as owners’ sole objective ceases to be useful in the analysis of ownership forms such as partnerships, coops, or condominiums. In residential coops, for instance, an apartment that is offered for sale does not necessarily go to the highest bidder. Moreover, by invoking this assumption we forego the opportunity to ask why are these organizational forms needed; why aren’t they all organized as for-profit corporations? When owners maximize with respect to two or more objectives, we need to know what these are and to form our models correspondingly.
In any case, what is the firm? Economists seem to clearly know what firms are. This “knowledge,” however, evaporates once one is required to define or describe them. To broaden the coverage of the term beyond the corporation, we need to know what falls under it, but no useful stopping rule suggests itself for doing so. An operational theory of the firm requires dividing organizations into two mutually exclusive sets, one consisting of firms, and the other of non-firms. At the present state of knowledge such separation is unlikely to emerge; the existing transaction-cost based theories of the firm are not amenable for such separation. For instance, we observe employers employing teams of workers, paying them wages for their effort, thus seemingly conforming to the Alchian and Demsetz model. Such a production mode, however, is not unique to for-profit corporations. Team production also takes place, among others, in not for profit organizations and even in government. The notion of team production may help determine what takes place within organizations and what takes place outside them, but as it is not unique to any narrow concept of firms, it is not helpful in delineating them.
I propose to focus instead on explaining organizations. The rationales underlying the existing models of the firm may contribute to the explanation of organizations. I argue that it is useful to group organizations according to their ownership structure. The basic hypothesis here is that each ownership structure has a comparative advantage in reducing capture costs in an identifiable class of activities. Our task then is to predict which activities belong to which ownership structures. Such a theory should allow us to determine which activities will take place within which organization and when would each type expand or shrink.
Enforcement
Introduction
I make a distinction between economic rights and legal rights. The distinction underscores the notion that people gain (and thus acquire economic rights) not only from agreements that the state enforces but also from agreements that are enforced by other third party enforcers as well as from are self enforced ones. The agreements the state enforces are by contract and are carried out in the market. The others are carried out, in part, within organizations. In this section I attempt to determine which exchanges will be transacted by contract and carried, and which ones will be carried out within organizations.
The state imposes restrictions such as quid pro quo, and, in some instances, the use of official units of measurement that contracts must conform to. Aside from these, the transactors are free to stipulate in their contract whatever they wish. We expect contractors to stipulate those attributes that are most readily measureable relative to their values. As a rule, the state does not participate in forming contracts. Therefore, the enforcer’s information about what the parties actually agreed to is entirely based on the contracts themselves. To be effective, then, contracts must clearly stipulate the agreed upon terms. For this reason, transactors tend to use explicit, and where possible standardized measures of the attributes they stipulate in their contracts. In turn, contracts tend to be impersonal. One immediate implication is that the more valuable is exchange among strangers (provided they are subjects of, and reside in the same state) the more effective is the exchange by contract.
The importance of this observation is revealed, for instance, when we consider corporate share holdings. Consider common-stock corporations that impose no restrictions on share transfers. Share holders, then, need not know each other, and thus, as a rule, are not subject to long-term relations. Such organizations, therefore, cannot be governed by long-term relations. These organizations can be erected only by contract, subject to state jurisdiction. Where contract enforcement cannot be relied on, we expect the range of ownership forms to be rather limited. This, for instance, may explain why organizations in much of Southeast Asia are family-dominated. Moreover, where owners are not free to transfer their shares, it is unlikely that organizations with a large number of owners could function effectively. This applies not only to conventional for-profit firms, but also to organizations such as guilds or labor unions. Any of these organizations can be maintained if all owners belong to tight knit groups, or alternatively, if the organizers themselves can resort to of violence, as is the case in some labor unions. But then such organizations themselves acquire some of the characteristics of the state.
Non-contractual agreements are less formal and less clearly spelled out than are contractual ones. The comparative advantage of enforcement by long-term relations is in transactions that are less explicitly specified and that rely more heavily on the transactors’ and the enforcers’ reputations. For example, we observe that in the marriage relationship, the state enforces the distribution of marital property in divorce, but not love and affection, and not even the quality of cooking. Agreements that are enforced by a third party enforcers using long term relations tend to lack formal documentation. The parties must apprise the enforcer of their intentions when they make the agreement; otherwise the enforcer would not know well enough what to enforce. It is seen that the ease of measurement and thus the ease of delineation determines, in part, whether transactions will be in the market, enforced by the state, or within organization, enforced by other means.
Prediction: As attributes become easier to measure, their enforcement will shift from long-term relations to the state. This will happen, for example, when new commodity standards are formed. A movement between forward contracts (enforced by long-term relations) and futures contracts (enforced by the state) could test the proposition.
In the rest of this section I first clarify some terms. I then discuss the enforcement power of the state vis a vis that of long term relations. Next I indicate what is the nature of the activities over which each of these enforcement forms has a comparative advantage in enforcing. I conclude by bringing out some new notions about vertical integration.
Terms and definitions
Enforcement within organizations (or firms), as is well recognized, differs from enforcement in the market. Not much attention, however, has been given to the nature of the two enforcement forms. I compare enforcement by the use, or threat of use of violence, which is the realm of the state with that resulting from long-term relations that organizations use.
To proceed, certain terms have to be clarified and transaction costs defined. I make a distinction between agreements and contracts. Agreements encompass entire relationships. Contracts are agreements or part of agreements that the state enforces. In addition, by my definition, all agreements governed by contract are ‘in the market.’ I also distinguish between legal (property) rights and economic (property) rights. Legal rights are individuals’ rights that the state helps enforce. Economic rights are what people can do with their commodities or assets. This view of what economic rights are conforms to Alchian’s (1965, 1987) and Cheung’s (1970) notion of property rights.
Allen (1991) provides a most attractive definition of transaction costs (while, like Alchian and Cheung, using the term “property rights” for what is called here “economic rights”):
Definition: “Transaction costs are the resources used to establish and maintain property rights” [boldface in original]. Elaborating, he states that “They include the resources used to protect and capture (appropriate without permission) property rights, plus any deadweight costs that result from any potential or real protecting and capturing.”
As defined here, then, transaction costs are the “dual” to property rights.
Some aspects of state enforcement
The violence using enforcer, i.e., the state, does not rely on long-term relations for enforcement power, and is more likely to confiscate than are those using long term relations. Seemingly to abet the threat of confiscation, subjects of rule-of-law states impose various restrictions on the state such as the use of objective and explicit criteria for adjudicating contract-disputes (Barzel 2002). The more successful they are, the higher the quality of the safeguards, and the more attractive is state enforcement. We expect contractors to write contracts that are relatively clearly delineated in order to take advantage of state enforcement. Two additional factors affect the attractiveness of state-enforcement: 1. How well delineated the law is (e. g., more so for traditional transactions than for new types). 2. The probity of judges and of other components of the judicial system.
Consider now caveat emptor transactions. These are consummated at the time of the agreement, and lack built-in continuation. They can accommodate trade among anonymous transactors, then, only under state enforcement. What the state enforces here consists of the prevention of fraud. In spite of its minimal level, such enforcement tends to alleviate the last period problem, and of course plain theft, which is an extreme form of the last period problem.
Transactors’ attempt to determine the levels of the attributes of the commodities exchanged in caveat emptor transactions must occur before they are consummated. Certain transactions would not take place under such terms because adequate determination (measurement) is too expensive. The parties, however, may use the state to secure and enforce the levels of various attributes of the transaction. Alternatively, as elaborated on later, the parties may use long term relations for that purpose. Now, as shown in Barzel (2002), the state enforces by itself only caveat emptor transactions. The state never enforces the entirety of all other transactions; it enforces transactions at most only in part. Long-term relations are employed to enforce some of the other attributes of these transactions. Transactors, then, must form long-term relations to be able to engage in non caveat emptor exchange.
Enforcement by long-term relations
Seemingly, the literature does not fully recognize, and definitely does not much explore transactors’ need to use long-term relations for the enforcement of all of their non caveat emptor interactions. The two well-known conditions for transactions to be self-enforced are: 1. Each transactor expects the relationship to extend into the indefinite future. 2. Each perceives that at any point in the life of the relationship the other’s gain from reneging is less than the present value of the loss in opportunities for future exchange.
These conditions imply that in the absence of state enforcement, strangers cannot effect exchange even if they would have agreed on the terms of exchange of goods each has and the other desires. The reason is that neither expects the other to perform and thus neither of them views the exchange as worthwhile. Rather, if they actually meet, the stronger of the two will simply walk away with the other’s good. It may appear that breaking down a lumpy, one shot transaction, into many small ones to be traded successively may create conditions for exchange, since a person not paying for a batch would be deprived from additional units. As long as the number of batches is fixed, however, the exchange will not be effected. Whoever runs out of his commodity first will not get paid for the last batch because he cannot penalize the other for non-payment. But then he would not offer his last batch, and similarly for the succession of all earlier batches. Thus, as is well known, the transaction will unravel. Among transactions with positive joint net value that do not meet the indefinite life condition necessary for self-enforcement are lending transactions and sporadic exchanges. The latter may include the intermittent exchange of assets that are not economical to turn into the indefinite continuing exchange of the assets’ services.
Reputation is a substitute for the indefinite life of relationships. If each of two would-be transactors has some reputation, when one wishes to get what the other has he will attempt to demonstrate that his losses from reneging will be large. The more difficult it is to determine the value of what one receives, the more difficult it is to figure out if trade is worth while, and thus to reach agreement about it. In addition, the more the execution of the exchange puts one is in a position to take or to steal, the less desired he is as an exchange partner. Recall that there is no state here to deter theft. The main deterrent is the loss of reputation that one suffers when revealed to be a thief.
The higher is one’s reputation, the higher the likelihood that others will seek him as an exchange partner. The level or the quality of the reputation a person has depends on his investment in it. In a multi-installment exchange, for instance, as exchanges are carried out, each party is likely to encounter reasons to renege. Each time one gives the other the benefit of the doubt by deciding not to renege he engages in a reputation enhancing investment. The value of one’s reputation also depends on others’ perceptions; among other things, the larger the number of potential trading partners who become aware of such behavior, the higher the return form it. The value of the reputation also increases with its durability. Thus, for instance, a given amount of reputational capital tends to be more valuable to younger people than to older ones.
State enforcement may allow exchanges to take place where reputation is insufficient or where it is absent. Much of the time, however, it suffices to assign to the state a rather limited role. In the example of the transaction that is broken down into a number of successive steps, it suffices that the state enforces the last installment for making the entire transaction viable. We expect, among other things, that the easy-to-measure attributes of transactions will be enforced by the state and that other attributes will be enforced by long-term relations.
The enforcement of the employment relationship
The employment relationship, though usually called the “employment contract,” is governed only partly by contract. The contract typically stipulates attributes such as the wage rate and the length of paid vacation. It does not, however, spell out with any degree of completeness what the employer can demand of his employees, nor what the employer is expected to do for them. Indeed, under a contract that would have spelled out all such attributes, the employer would have had no leeway for making any master-servant type demands on employees. Thus the relationship would not become what is usually considered to be an employment relationship.
As the state does not enforce entire employment relationships, long-term relations must be deployed to enforce additional within-organization attributes. What is the nature of the enforcement power embedded in the long-term relations here and how potent is it? Note right away that the relationship is reciprocal. If either party has no reputation, they would be unable to form an employment relationship between them.
I proceed by first discussing a two person relationship, assuming that one will be employer and one employee, and then turn to briefly suggest and discuss a general reason why the employment relationship would be used. Consider an employer and an employee. Each expect the other to provide some valued services not stipulated in their contract, and each would require the other to possess a sufficient brand name to back his (implicit) promise. Suppose they already possess brand names, say acquired from their high school acquaintance. Each has some prior expectations of the other’s (as well as of his own) performance. As they interact, they presumably encounter opportunities to operate at the initially expected level, at a higher level, or at a lower level. If the employer acts at least as well as expected, the employee is likely to continue the relationship. On the other hand, if the employee perceives underperformance, he is more likely to quit. When others observe a “quit,” as when the separation occurs during a high employment period, they will lower the level of reputation they assign to the employer. However, when others observe a “dismissal,” as when the separation occurs during a low employment period, they will lower the level of reputation they assign to the employee. Periods without separation indicate that both are acting at or above their initial reputation level. Others, then, will assign both of them a higher level of reputation. One implication here is that women that temporarily leave the labor force to bear children and then return to their old employers lose not only labor force experience but also the opportunity to enhance their brand name to the degree that the longer employment duration would have generated. The loss in brand name would be still greater if (absent regulation to the contrary) the old employers do not invite them back.
How is the employer-employee relationship enforced? Casual observation may suggest that the value of the brand names that different individuals have is often low compared with the enforcement needed to secure the desired behavior. Transactors, however, have a choice of the form of the agreement they adopt. Changing a market relationship into an employment relationship induces a radical change of incentives. This change seems to be the main force for getting the parties to act in conformity with each other’s desires, and correspondingly affects the guarantees that others look for.
Consider the ‘make or buy’ choice. A transactor may acquire a product, the ‘buy’ decision, thus essentially operate in the market. Or he may acquire inputs, the ‘make’ decision, thus essentially operate within an organization. In both the ‘buy’ and the ‘make’ agreements, the contract governing the transaction is “incomplete.” Left out of the agreement are those aspects that are too costly to measure, document, and enforce. Forming and enforcing agreements is costly, and their incompleteness opens the door for capture activities. The services of employed workers is one of the inputs in the ‘make’ decision. Among the left out terms in the contract for these services are the exact tasks workers are expected to perform, their promotion as well as various aspects of working conditions which must be self-enforced (Rock and Wachter).
The transactors presumably will adopt the form they expect to generate the higher net gain (i. e., the one subject to lower transaction cost). We have to determine, then, conditions under which one form of agreement will be preferred over the other. An agreement regarding inputs may be preferred over one regarding the output when, for instance, evaluating the inputs is cheap relative to evaluating the product itself. This difference may seem innocuous, but it is actually of great significance.
Suppose the commodity to be exchanged is automobile breaks. When breaks fail, the resulting damage may be rather large. If determining the quality of the breaks at the time of purchase is difficult the buyer may require a guarantee. The reputation of an independent producer selling his devices, however, is unlikely to suffice for such purpose. The producer might have offered here a pecuniary guarantee, perhaps to be enforced by the state. But it is unlikely that his wealth would have sufficed for such a guarantee. We expect a seller who cannot be fully penalized for careless behavior to be more careless than is “efficient” and to produce a product that will cause “excessive” damage. He is not likely to be “careful enough” in seeking to maximize the value of his output. One may attempt to revert to a third party to provide the guarantee. This party, however, may charge an unacceptable high premium unless the seller agrees to be constrained; if not constrained, he has no incentive to avoid morally hazardous behavior. In some cases modest constrains on the seller would not suffice. One radical constraining method is to turn the seller into the guarantor’s employee.
An employee obviously has to be induced to perform; he has to have a reputation to lose in case the employer lays him off because of low performance level. Now turning a person from operating independently to becoming an employee does not make the product less prone to cause damage. The change is in the terms of the agreement governing the two types of operations. Consider again the potential for faulty breaks. Under the employment agreement the need to guarantee them falls on the employer. When an independent producer becomes an employee, however, his incentives are radically altered. Being rewarded by a wage, and given the incentives that the wage contract brings about, he is unlikely to extend himself as he gains little from working hard. But because of its indirect effect on product quality, this is the behavior that the employer wishes to induce. An employee who does not work hard is less likely to be careless and produce faulty products that can harm their users than is an independent producer. By the same token, a truck driver transporting delicate china who is paid by the hour is inclined to shirk by “taking it easy,” which in this case actually plays into his employer’s hand. His shirking is harnessed towards the desired outcome.
In terms of reputation, what the employee has to guaranty is diligent and careful work. For that a modest amount of reputation seems adequate. Thus by turning the worker from operating independently to becoming employee, the amount of reputation needed to guarantee performance is radically reduced.
On his part the worker has to believe that the employer is motivated to perform as implicitly promised. The employer’s reputation must be adequate to guarantee such things as good ambience in the work place. The work place may also pose the danger of injury. If the work involved exposes the worker to serious injury, he may view the employer’s reputation as inadequate to induce the appropriate behavior. There is no reason to expect the reputation of an individual employer to be adequate to guarantee his own action more than the worker’s guaranteeing his. The employer’s problem can be alleviated, however, if the need for injury compensation can be documented, as that part of the agreement can be included in the employment contract. The employer then will have to demonstrate that he is in command of sufficient pecuniary capital. Here he can take advantage of the scale economies to insurance by insuring many workers whose potential injuries are not highly correlated. He also insures the product. Obviously, he has to amass the capital necessary for that purpose.
Additional enforcement features of the employment relationship
In his critique of Alchian and Demsetz (1972), Holmstrom (1999, p. 80) states: “Any contract between the owner-monitor and the workers could just as well be carried out in the market as within a firm.” I agree, and this, indeed, seems to apply to any relationship that might take place either in the market or within a firm., however, Now we expect the costs of the two forms of organizing relationships to differ. Although Holmstrom acknowledges (footnote 20) that the “argument is weaker if we think of the contract as implicitly enforced, say through repeated interaction,” he does not follow through on this observation.
What is the nature of the difference between independent and employed workers, or between production that is carried out “in the market” and production that is carried out “within an organization?” As Holmstrom implies, the effort a worker exerts is equally observable to the would-be market buyers of his product as it is to employers. For three reasons the cost associated with observing workers and using the information within firms is less than the corresponding cost when they are observed in the market.
1. Market transactions are enforced by the coercive power of the state whereas firms enforce transactions that take pace within them. Consider the enforcement of the level of effort. Effort has to be explicitly documented if is to be adjudicated by the courts. Operating within the firm is different. An employer may rely on his informal judgement of the effort. He may reward a worker for “intangibles,” a term often used in team sports, which I interpret as observed behavior that is difficult to quantify and to document. The “employment at will doctrine” is the other side of this coin. Under this doctrine, an employer need not be able to demonstrate to the courts why he chose to terminate an employee. The greater the cost advantage of an implicit measurement relative to an explicit one, the higher the chance the transaction will be conducted within the firm.
2. An employer who pays wages must observe employees’ effort. When employing several workers he may place them next to each other to observe them together at the same time. Market buyers are less likely to operate that way; it seems “unmarket like” to constrain several sellers to operate side by side. The costs of the two forms would be the same only if the “market” arrangement would be subjected to the same restrictions (not state enforced) that are imposed on the employment relationship. But then such “market transactions” are turned, as the IRS would agree, into employment relationships.
3. Workers interact when one transmits a product to the other, share machines, or move in the same space. They function best as independent workers if it is easy to explicitly delineate their rights. Suppose, however, that the quality of the specimens that one worker transmits to the other varies, or that they intermittently use the same machines or move in the same space. If they operate independently, each is likely to attempt to gain at the other’s expense. Alternatively, a single employer may employ both individuals. The employer delineates the relationship between the two.
The instructions he gives and the rules he imposes constitute a redelineation of rights and he becomes a third party enforcer of these. The employer is expected to impose such rules and give such instructions so as to reduce the capture, and more importantly, reduce the incentive for capture. For instance, he may restrict the amount of time available for picking and choosing, and reduce the reward for choosing only high quality specimens. Or he may provide mutually exclusive territories to salespeople. The employer here induces a cooperative equilibrium. Independent transactors find such arrangements difficult to make, and Nash equilibrium is the more likely outcome here. I hypothesize that organizations are formed for the express purpose of creating rights that are more economically enforced by non-state means than by state means
These considerations yield the two following predictions:
1. When the state’s comparative advantage in enforcement has increased, (declined) more (fewer) stipulations will be agreed upon contractually.
2. When the state’s comparative advantage in enforcement has increased (declined), the scope of the state will expand (shrink).
Enforcement and Vertical integration.
Definition: Vertical integration is a state wherein a residual claimant of a firm, or of some other organization, has taken on himself the combined variability of two or more vertical operations.
The distinction between state enforcement and other enforcement forms is at the heart of vertical integration as it is defined here. Although the definition is idiosyncratic, it will be seen to be useful. Understanding the nature of within- organization enforcement sheds new light on the function of vertical integration. In general, some within-organization interactions are enforced within them but some, though seemingly falling within the operations of organizations, are enforced by the state, and thus are actually in the market.
What does it mean that the state enforces agreements regarding activities that take place within an organization? Consider, say, a battery producer selling batteries to Ford Motor Company and contractually guaranteeing them to car buyers. The battery producer, then, bears the variability in battery quality. It is true that the carmaker bought these batteries and installed them in his cars. But by and large, he acts here as his customers’ agent and does not bear the effect of variability in the quality of the batteries. The batteries, therefore, are not a part of his vertical structure even while they passed through his plant.
The merger of two vertical firms creates a single vertical firm out of them. I argue, however, that merger may actually reduce the level of vertical integration as it is defined here. Consider a production process with four vertical links B, C, D, and E. Suppose for instance, that B is a wholesale buyer of fresh vegetables, and C and D provide two successive links in transporting the vegetables to E, a supermarket. To proceed I make the following assumptions: 1. Freshness is a valued attribute. 2. Freshness is easy to inspect when the wholesaler B prepares the vegetables for shipment by C as well as when they are delivered to the supermarket E. 3. Under the most economical packaging method, freshness is expensive to inspect when vegetables are in transit. 4. Shippers’ costs are lower when skimping on maintaining freshness.
Suppose that while shippers C and D operate independently of each other. If B employs the low-cost packaging method and guarantees freshness to E, C and D can gain by skimping on maintaining freshness; each can blame the other for the loss of freshness. Under the guarantee, then, C and D gain from spending too little on maintaining freshness. Therefore, B’s guarantee of freshness to E will not survive for long.
As long as C and D continue to operate independently, freshness would be properly maintained only if B does not guarantee freshness to E. In the absence of the guarantee, E will pay D an amount commensurate with the freshness of the vegetables, and similarly D will pay C only if he is satisfied with what he receives. This arrangement, however, requires the use of the more transparent and more expensive packaging method that accommodates low-cost inspection in transit.
B would be willing to guarantee freshness to E provided that C and D merge to become C+D and the merged firm agrees to see to it that freshness would not be harmed by improper shipping methods. C+D and its transacting partners are likely to make such an agreement here since, given the assumptions, the cost is low of ascertaining whether or not the agreement was kept. If C+D agrees to maintain freshness, it presumably will simply instruct its employees to take the necessary steps to do so. The merger of C and D, then, allows B to leapfrog both and contractually guarantee freshness to E. The level of integration is actually lowered then. Indeed, this may well be the purpose of the merger.
As defined here, vertical integration is a matter of enforcement. Consider now the effects of changes in the cost of legal delineation on the level of integration. As discussed in the next section, the cost of legal delineation will fall when the cost of measuring commodity attributes falls, or when new commodity standards are formed. The ease of legally enforcing different stipulations varies; it depends, among other things, on what there is to enforce. I predict that a fall in the cost of legal delineation will induce firms to lower the degree of vertical integration, and to increase the rate at which they spin off some of their operations. This is a sharper prediction than the prediction that Williamson’s (1975) and Klein, Crawford, and Alchian’s (1978) model might yields. To them vertical integration is binary; it is either present or absent. By that model, two firms will merge, and thus integrate presumably when the quasi rents (which, to say the least, is not readily measureable) in their interface is “significant.”
Consider another question: Are a franchisor and his franchisees integrated? Echoing Cheung (1983), this question cannot be satisfactorily answered if we just try to determine whether or not they "belong" in the same firm or whether they constitute separate firms. The question can be answered by using my definition of vertical integration; it is designed to handle the notion that the level of integration is not constant and depends on the fraction of the exchanges between the franchiser and the franchisee that are enforced within the operation. The existing notion of vertical integration cannot cope with marginal changes of this kind.
Measurement.
Why are firms, or more generally, why are organizations needed; why don’t markets suffice to organize all interactions? The need arises, as seems to be generally agreed, from the fact that information is not free. Rather then focusing on “information,” however, I focus on the more operational concept of “measurement.” The switch is productive as measurement constitutes the quantification of information. Had the costs of measurement (and thus of information) been zero, measurement would be perfectly accurate, contracts would be complete, and the Coase theorem would have held. To underscore the empirical significance of measuring, consider what might be involved in measuring the sweetness or tartness of a supermarket apple, of the reliability of a worker, or of the appropriate price of a used car.
Every transaction requires the measurement of what each party agrees to yield to the other. I assume, as seems reasonable, that measurement is costly and that perfectly accurate measurement is prohibitively costly. Measurement errors are inevitable, then, and they result in economic rights that are not well delineated. Individuals are expected to spend resources to capture these (Barzel, 1982).
Had a producer’s contribution to output been easy to measure, he would have operated as an independent contractor, receiving the net value of his output. When measuring the contributions is difficult, each worker may attempt to gain more of the output value by imputing to himself output produced by others. Indeed, they can be successful in doing so when operating independently, and would then spend resources to induce the income transfer. As discussed below, organizing production within firms may, at a cost, avert such capture expenditures.
What I call the “measurement problem” most closely corresponds to Holmstrom and Milgrom’s (1991, 1994) multi-tasking approach. Their analytical framework seems very similar to the one I adopt for analyzing the exchange of multi-attribute, non-homogenous commodities. As is to be expected, Holmstrom and Milgrom’s analysis is much more formal than mine is. It seems to me, however, that by adopting the principal-agent framework, they unnecessarily limited the scope of their results. First, they seem to have overlooked the fact that their model can be easily adopted to analyze the exchange of non-homogenous commodities. Second, by adopting the principal-agent framework they implicitly assume that hierarchy and organizations already exist, and it deters one from asking what skills are needed to be a principal and what are the costs associated with that position. The principal-agent relationship is often asserted to be between one and many. Although this is usually true in practice, it is not always so, and the relationship should be explained rather then posited. Note that some law firms have many general partners (presumably the ‘principals’) and might have fewer employees than partners and it had been common for two physicians to employ one receptionist. In contrast, in applying the measurement notions, my starting point is that the relationships in any interaction between transactors are symmetric unless shown to be otherwise.
The cost of measurement framework can serve to unify the existing transaction-cost models of the firm as well as a tool for making the model more complete. I argue that all capture problems that this paper considers can be translated into measurement error problems. Indeed, as will be seen below, certain implications derived from the measurement error framework are common to all the models of the firm considered here. It is relatively easy to determine when measuring an activity or an attribute has become cheaper or more expensive, and thus to form refutable implications regarding the phenomenon at hand. I offer a number of implications in the following pages.
Alchian and Demsetz (1972) are the first authors subsequent to Coase (1937) to introduce transaction costs considerations to explain the firm. Firms, besides providing home to teams also accommodate vertically integrated operations. Two factors are asserted to lead to vertical integration. One which I have introduced (Barzel, 1982) is the costly measurement of intermediate inputs. The other is the capturability of quasi-rent by Williamson (1975) and by Klein, Crawford and Alchian (1978). Market transactions within both categories are asserted to be prone to costly wealth transfers that can be averted by turning the operations into within-firm ones. I first discuss Alchian and Demsetz’ contribution, albeit very briefly as they explicitly bring measurement costs into their model. I proceed with the measurement of intermediate products and then turn to the capture of quasi rent.
Alchian and Demsetz’ (1972) model operations that benefit from team production. They recognize that in team production, the contribution to output of individual team members is difficult to measure. They state that the entrepreneur can observe effort that is a proxy for output. He employs workers for wages, and induces them to make the appropriate effort. His reward consists of the difference between output value and the payments he makes to the members of the team. Being the residual claimant, he is induced to maximize the net output value. Note that as a rule, team-members’ wage contract governs only part of the employment relationship. The rest falls under the “master-servant” relationship, and it must be self-enforced. A measurement based prediction implied by the team production model is that as measuring the output of members of team production becomes cheaper, more team members will become independent operators.
The Exchange of Intermediate Products
Consider an upstream producer selling his output to a downstream one. In the case of some, if not all products, measuring their specimens and sorting them into strictly homogenous groups is prohibitively costly. The non-homogenous specimens, then, sell at a fixed unit price (or more generally, at a price that is not fully adjusted for all valued attributes). The asymmetry of information about the discrepancy in value between the equally priced specimens opens the door to resource consuming capture. For instance, the seller may attempt capture by producing and passing to the buyer low quality specimens that are made to appear like high quality ones, and the buyer by accepting or selecting only high quality ones. Between them they are expected to spend on producing and measuring (and sorting) the specimens of the commodity an amount that exceeds the jointly maximizing one (Barzel, 1982). This is a case where (economic) rights are not well delineated, and people compete to capture them.
The excessive measurement is largely avoided if an entrepreneur hires the two producers as wage employees thus forming a vertically integrated operation. Paid by the hour, the upstream producer would gain little from producing low quality specimens, and he need not fear a loss if the variability of his product exceeds the variability of his product as an independent producer. The specimens he is expected to produce, then, are of higher average quality, but subject to higher variability then what he would produce as an independent operator. That suits well the employer who is concerned more with measuring the mean, which is relatively cheap, then with measuring the variance, which is relatively expensive. In his turn, the downstream producer, also an employee, would gain from actions such as picking and choosing less were he an independent worker.
The use of labor services paid for by the hour incurs costs that arise from shirking, but it avoids pitfalls due to manipulations that the exchange in the market of expensive-to-measure specimens would have induced. Viewed from a different angle, when operating independently the two impose externalities on each other. Employing them within a firm as part of a nexus of contracts internalizes the externality. As with team production, here too enforcement within firms replaces enforcement by the state.
A central but not well recognized point is that the choice of the attributes to measure and the ones not to measure is also the choice of an incentive structure. In market exchange, measurement is primarily of the attributes of the exchanged commodities. In the employment relationship where the same commodities are produced, the main attribute being measured is the workers’ effort. The costs of operating the two organizational forms differ, as do the incentive structures underlying them. One of the expected results of these differences is that the attribute levels and the valuations of the two sets of commodities will differ from one another. If such is the case, then, the comparative advantage the firm has lies in activities different from those carried out in the market. This argument yields the following predictions:
1. As workers become cheaper to supervise, the within organization degree of specializing will increase. For instance, we expect that the movement from the putting out system to the factory system had been accompanied by a higher level of specializing; i. e., by a finer breakdown of operations into the constituent ones.
2. As workers become cheaper to supervise, the degree of vertical integration will rise. The introduction of the production line dramatically illustrates the lowering of the cost of measuring workers’ contribution to output. Where the production line is found to be useful, requiring individual workers to keep pace with the speed set for the line greatly lowers the cost of measuring their contribution. It is predicted that firm size would have increased with the introduction of the production line.
Consider further the cottage industry or putting out system that was widespread before the industrial revolution. In some cases, merchants who bought the cottage industries’ products found it difficult to determine the quality of the inputs that producers used. The producers of such products, then, could have gained by using low quality inputs. The merchants could have provided their suppliers with the desired inputs. Unless supervised, however, the suppliers could have sold these high quality inputs, replacing them with lower quality ones and pocketed the difference. Employing a number of workers on the same premises afforded low cost supervision. Indeed, a factory system on a modest scale emerged before the industrial revolution. These factories were not in response to new technology or to the need for a central power source. I hypothesize that the system emerged in order to reduce the costs of input supervision. Provided this was the main problem of the interaction, this obviated the need to measure the quality of inputs the workers used.
Prediction: The remuneration of pre-industrial revolution factory workers was primarily by the piece even though they worked in a factory setting.
When measuring intermediate goods or operations becomes cheaper, firm may then choose to reduce their level of integration. Firms have discretion over the precise point (or line) of separation between what they divest and what they retain. What they look for are easy to delineate boundaries between two sets of activities. Figuratively speaking, owners seek lines of perforations that will create a clean break between the parts. Suppose that a furniture-producing firm with its own retail outlets decides to divest them and sell the furniture to them. Should it polish the furniture, or sell it unpolished? If polished furniture is more uniform than unpolished furniture, we expect it to perform the polishing in house. If the furniture is more uniform before being polished, we expect it to sell the furniture before they are polished and let the buyers take care of the polishing.
Tracking stocks present another divesting issue. By issuing tracking stocks a mother firm creates a market for certain assets separate from the rest of its holdings. In order for such a stock to perform well, the separation must be clear. This requires the relatively easy measurement of the assets of the new entity. But if these assets can be cheaply and accurately measured then vertical disintegration becomes attractive. By the model here the expectation is that the mother firm will spin off the daughter firm within a short time period.
Captureable Quasi Rents
Not only can the analysis of the quasi-rent capture fits cleanly within the measurement framework, the latter also provides a superior understanding of the capture problem. The quasi rent of specialized assets seems prone to capture. Williamson (1975) and Klein, Crawford and Alchian (1978) assert that the existence of specialized assets is a sufficient condition for their capture as the courts are unlikely to protect the contractual arrangements involving these assets. This is a very cavalier view of the legal system. Why would anybody use it if bypassing it is so easy?
In general, enforcement problems arise when contracts are incomplete, and thus rights are not well delineated. The degree of incompleteness depends on two factors. One is the cost of measuring the assets’ (and the transactions’) attributes. The other is the nature of state enforcement. The lower the cost of measuring assets’ attributes, the clearer their delineation is and the easier the enforcement of their ownership. Thus as the costs of measurement decline, the chance that contracts will be enforced increases. The limiting case is that of costless measurement resulting in “complete contracts” which means clear delineation and full enforcement. The ease of measurement is seen to be a determinant of whether or not specialized assets would be captureable. Thus it also determines whether they will be exchanged by contract or protected through vertical integration.
State enforcement is never perfect. Transactors may view the enforcement of certain contract stipulations to be sufficiently inadequate to seek other means for enforcing them. The inadequacy is not constant, however. When enforcement is expected to improve, as would be the case, for example, when judges are deemed less likely to take bribes, or where precedents have become clearer, contracts will become more complete.
Consider a slightly different angle from which to view the capture problem. Suppose that two individuals own complementary assets. They might have interacted in the market in order to exploit the complementarity. However, the potential gain from the cooperation is not well defined and is thus subject to capture. To avert the capture the two may pool their capital. This they must do prior to committing it; only then can they proceed with the project without fear of capture. If done at the right time, the rent capture is averted. However, two new sources of dissipation arise with the new arrangement. First, each is of the two is only a partial owner of the integrated enterprise and thus neither would make the effort that a single owner would have made. The second source of dissipation regards managers’ behavior. Presumably, the combined enterprise requires two managers. Each would be remunerated according to the performance of his unit. Given the difficulty of measuring the interface between the two operations, which is the original source of the problem, each manager may gain by claiming as his part of the contribution of the other. This constitutes an income transfer at a resource cost. However, remuneration is only partially based on measured performance, and this portion of the dissipation is expected to be less than in the case where the attempt to capture the quasi rent is direct.
The measurement concept alerts us to the existence of solutions to the capture of quasi rent problem besides vertical integration. Consider the relationship between the small town owners of the printing press and the newspaper, which illustrates Klein, Crawford and Alchian’s capture problem. They claim that vertical integration is necessary to prevent capture. The two asset owners may, however, possess another method to reduce the dissipation which is to become horizontally integrated. A horizontally integrated owner will gain from acquiring and maintaining a reputation for fair dealings and is thus expected to avoid taking advantage of the leeway the legal system may afford. The horizontal integration here may restore vertical contracting and is a substitute to vertical integration.
Neither Williamson, nor Klein, Crawford and Alchian offer a general rule to predict when integration will occur. The measurement notion does offer such a rule:
As the delineation of the rights to the quasi rent becomes easier, writing contracts that the state will enforce also becomes easier and the incentive to integrate is reduced. Conversely, when such delineation becomes more difficult, contracts become less useful and we expect vertical integration to become more common.
The above discussion demonstrates that the existence of specific assets is neither a necessary nor a sufficient condition for vertical integration. First, the costly measurement of intermediate inputs is also a reason for such integration. Second, as the ability to measure specific assets is enhanced, transactors may interact by contract rather than by vertically integrating. Under certain conditions transactors may expect the courts to protect their assets at a lower cost than vertical integration does. Thus, they would be willing to make the appropriate contractual arrangements and forsake the prospect of integration.
Common to the three factors discussed thus far, i. e., team production, measuring intermediate inputs and captureable quasi rent is the argument that they determine which activities will be carried out within firms. As just shown, the measurement framework encompasses all these factors. That this is a useful framework is evidenced by reiterating a conclusion that holds for all these models, i. e., that the higher the cost of measuring commodities or activities other than workers’ productiveness, the greater the likelihood these will be conducted within a firm. Regarding workers’ productiveness, the lower the cost of measuring it, the more activities will be conducted within the firm.
None of the models discussed thus far sets bounds on the firm. For instance, it does not seem possible to explain the size and scope of Ford Motor Company by any combination of the three factors. We know neither why Ford makes engines but not tires, nor, why does it (seemingly) contain many distinct team production operations. Guarantee capital, another major ingredient of the measurement framework, provides such boundaries.
Guarantee Capital
Each of the firm models discussed above offers a rationale for performing certain activities within firms rather than across them. None of them, however, can explain where the boundaries of firms lie; whether a firm would consist of only one such activity or of a number of them. In this section I argue that the need for capital to guarantee one’s actions combined with the scale economies to such capital determine the size of firms and, to a degree, their composition. The scale economies to the assembled guarantee capital contribute to the interdependence among transactions, giving added contents to the notion of a nexus of contracts.
In productive endeavors the acquisition of inputs, the dispensation of output and the productive process itself are all subject to variability. By design or by default, each of the effects of the variability must be borne by somebody. The presence of variability opens the door for wealth capture; individuals may be able to claim gains they did not fully produce and divert to others losses for which they are at least partly responsible. Any agreement constitutes, among other thing, the allocation of the associated variability among the parties to the agreement.
The variability that organizations assume is a function of the contents of their agreements with the three sets of parties. For example, in the purchase of a raw material an organization may agree to pay the spot market price, or alternatively, a fixed long-term price. It obviously takes on the variability that price fluctuations generate in the former case but not in the latter case. Its contracting partners take on whatever variability remains. Similarly, if the firm sells its product under guarantee, it assumes the variability in quality, which it does not if the sale is governed by caveat emptor. Another example of an input contract is the purchase of fire insurance, which shifts the variability due to fire from the buyer to the insurer.
Perhaps the least obvious component of variability is the one associated with the interaction among workers, whether they are employees or operate as independent contractors. Workers interact with each other directly when sharing the workspace and equipment in it. In that case they expose each other to injuries and their equipment to damage. They also interact in delivering output from one to another. Typically the quality of the output is difficult to measure and tends to vary across specimens.
Wealth maximization or efficiency requires that a person who adds to the variability of outcome (for instance, by operation an innovative piece of euipment) will become the one who bears its effect. Bearing adverse outcomes, however, requires command over capital. A person who is otherwise the efficient bearer of variability but who does not command the capital to bear it can transfer wealth to himself by increasing the variability which induces adverse selection or moral hazard.
Between them, transactors stand to gain by reducing the amount of resources that are used to transfer wealth. Here it may take the form of cooperation between a person who lacks sufficient capital and one who does. Such cooperation will take place in the market, however, only if the capital provider could readily identify what his client actually does. He can then impose the appropriate restrictions such as requiring the use of high quality raw materials, or, in the case of a physician, prohibiting him from certain risky treatments whose expected value is low.
Restrictions are costly to monitor and to enforce. Another form of cooperation can reduce these costs: turning the client into employee. Employees must be supervised to ensure that they perform. When the sum of the costs of supervising employees and the costs associated with the shirking that remains is less than the cost associated with imposing restrictions (including the capture cost that remains) the employment relation will prevail. Applying here as well is the earlier prediction that the employment relation would become more common as the cost of supervising employees falls.
I now turn to the scale of the guarantee capital. The stock of guaranteeing capital is built up as the explicit or implicit guarantee premiums accumulate. As adverse events occur, the guarantor’s payoffs deplete this stock. Although the premiums are set to at least equal the expected losses, the net outflow may be negative at any time interval. A net outflow reduces liquidity, and is becoming more expensive as the imbalance increases. Ultimately, when the cumulative net outflow exceeds the available amount of guarantee capital, the assets backing this capital have to be fully liquidated. Such occurrence may be avoided by commencing with a large stock of guaranteeing capital, which is subject to economies of scale, basically obeying the law of large numbers.
As the operations guaranteed by the guarantee capital expand the divergence between ownership and control tends to increase, and with that, the cost of controlling them increases as well. The balance between the economies and these diseconomies determines the size of the operation covered by the guarantee capital. Guarantee needs vary with the variability in the value of different activities. The greater the variability, the larger the need for guarantee capital and the greater the gain from the scale economy to which it is subject. The appropriate measure of variability here extends beyond the second moment. A very small chance of a very large (negative) outcome calls for more guarantee capital than does a higher probability for more moderate outcomes even if the two produce the same variance.
Prediction: The greater the variability (and spikes in it) that the value of an interaction is subject to, the larger the size of the organization that will emerge to handle it.
Turning to boundaries, the boundaries of an organization or of a firm are demarcated by the activities that are guaranteed by its guaranteeing, or equity capital. This notion of boundary allows us to respond to questions such as: are franchised Ford dealers part of Ford? By inspecting the contracts between Ford and its dealers it can be determined what the portion of the variability in the relationship is that Ford assumes. The more of it they assume, the less they are part of Ford.
What is a firm?
In professional economic discourse as well as in common usage one may get the impression that the meaning of the term “firm” is clear. Such is not the case, however. I am confident that any economist, indeed, any contributor to the theory of the firm, who would try to come up with a definition will realize that no single, clear cut and restrictive meaning of the term exists.
As discussed in previous sections, between them the various authors suggested four factors that characterize and determine the size and the composition of firms. These are team production, the exchange of intermediate products, the capture of quasi rent, and the need for guarantee capital. Seemingly, however, each of these factors can affect any organization. For instance, team production is advantageous not only in shareholder corporations but also in cooperatives, in non-profit organizations in firms wholly owned by government and even in government itself. One example of team production that does not conform to Alchian and Demsetz notion of the firm is the symphony orchestra. Such orchestras are clear-cut teams where the conductor monitors the performance of team members. The conductor, however, is usually an employee, and often performing for a brief period as a guest rather than being the residual claimant of that team. By themselves, then, the above factors cannot constitute theories of the firm. It seems arbitrary to choose a subset of the organizations that fit the definitions and call these “firms.” Are law partnership firms? Are nonprofit universities firms? How about condominiums; professional team-sports leagues (an umbrella organization for for-profit members); local governments? This is not to imply that the above models should be discarded. If correct, they contribute to the theory of organization as they predict the activities that fall within organizations and as well as when will the organizations expand or shrink.
Whereas we cannot uniquely identify the firm, and thus cannot have a theory of the firm, we can make useful statements about the differences between classes of organizations. One factor that distinguishes between classes of organizations is the nature of the ownership of their guarantee-capital. This provides a criterion by which to classify the organizations. At this early stage in the investigation I simply point out that what should be done is to analyze what kind of behavior do different ownership forms induce, and note that the potential exists for predicting which kind of activities will fall into each of the categories. In the rest of this section I preliminarily indicate the nature of some of these predictions.
In an earlier section I discussed enforcement by altering employees’ incentives. The notion of altered incentives is relevant to the problem here too. Every form of organization and especially, every form of allocating rights to the guarantee capital generates its own distinct set of incentives. Every particular form, then, may be used to control different aspects of capture activities. To illustrate I briefly discuss how each of several ownership forms serves to alter incentives. In each case I predict the character of the resultant set of activities each will specialize in.
Consider the services that three sets of non-profit institutions--museums, universities and hospitals--produce. These services are characterized by the difficulty of assessing their quality. Perhaps not accidentally, these also tend to attract donors. Inducing suppliers to produce services of the appropriate quality as well as persuading donors to donate is problematic. Suppose that the producers would have been organized as profit seeking corporations, and their managers’ reward was tied to the value of the corporate shares. I argue that had this organizational form been used, it would have adversely affected both output quality and the level of donations.
Regarding output quality, it costs less to produce low quality output than high quality output. Because of the difficulty customers encounter in assessing quality, they are not expected to evaluate accurately the quality level of the service they get. Being profit maximizers, and abstracting from reputation, we expect producers to produce low quality output. We also expect the level of donations to be low because donors have reason to fear that some of their contributions will end up in the shareholders’ pockets.
When such operations are run by non-profit organizations; i. e., organizations that are not set to maximize profits, however, we expect these problems to diminish considerably. The reason is that in non-profit institutions the controllers of the equity capital do not gain much from producing low quality output, or from diverting donations away from their intended use, and so we expect such actions to be greatly reduced. In these dimensions, then, the operations of non-profit organizations would be more efficient than those of for profit corporations. The magnitudes of the costs and benefits from such ownership forms vary across case types. In each case, then, the costs have to be weighed against the benefits. Non profit organizations are expected to survive where the cost-benefit ratios are low.
A change in incentives is a tool that is used to induce the desired behavior; here the desired change is in the behavior of the controllers of capital. The altered action is attained at a cost; the organizations managers’ lesser incentives to produce output at least cost. Similarly, because of the managers’ reduced incentives resulting from the greater cost of monitoring them, managers will work less hard and will reward themselves with more difficult-to-measure perks than will those working for for-profit organizations. Given their lesser effort and higher perks, and given that salaries are easy to measure, we expect the salaries of managers of non-profit organizations to be lower than those of corporate managers. Indeed, because of the ease of measurement, this prediction seems relatively easy to test.
I conclude by sketching a number of additional illustrations where the ownership form induces particular pattern of activities associated with the particular incentive structures. Consider lawyers’ choice between forming partnerships and forming share holding corporations. The former organizational form better motivates the individual partners to contribute to their firm. The latter seems superior in raising capital. I predict that as law firms tend to themselves become plaintiffs in malpractice suits, as well as when they move into fields where they are more prone to such suit, that they would shift away from partnerships and into corporate form.
Second, consider liability in partnerships. This can be limited, or unlimited. In the latter case, all partners’ personal wealth backs the partnership in case of bankruptcy. Sharing in partnerships tends to induce shirking. In some cases there is the added problem that partners may be able to transfer to themselves some of the partnership’s assets. To the extent a partner gets enriched by such theft, in case of unlimited liability, he stands to lose his loot in when bankruptcy occurs. Theft seems to be relatively easy when a partner works out of his home and where the partnerships handles commodities that are easy to appropriate such as direct consumption goods. I predict that in such cases, provided wealth is difficult to conceal, the unlimited liability would be more common than when the partnership engages in other activities, say supplying legal services.
Finally consider consumer cooperatives. In certain situations consumers may anticipate that private entrepreneur that might serve them would be able to charge monopoly prices. Single price monopoly pricing results in a welfare loss. This loss is not averted even if the monopolist must pay a franchise fee equal to the monopoly profit. However, if the consumers organize as coops and serve themselves, they presumably will charge prices equal to marginal costs, and thus avoid the welfare loss due to monopoly pricing. They incur an extra cost in that they are not, in general, the most efficient producers of such services. I expect that consumer coops would be relatively common in isolated communities, and that their role will decline as communities become less isolated, either because population density increases, or because transportation costs decline.
Conclusions
This paper proposes a theory of organizations. It contains three major ingredients: 1. The importance of the enforcement of agreements in general, and the distinction between enforcement by the state and enforcement by long-term relations. 2. The adoption of a measurement cost framework to unify and augment various strands in the literature on the firm. 3. The claim that, on the one hand, the attempt to formulate a theory of the firm is futile as no useful definition of the firm seems to be available and thus far pinning it down has not been successful. On the other hand, though on a very preliminary basis, I demonstrate that it is possible to formulate an operational theory of organizations.
To underscore the first two ingredients consider another aspect of the formation of charitable organizations. A multitude of charitable organizations exist in the United States today, and donors can choose from these the ones to donate to. Two basic general features of the United States today that are relevant to such organizations are the high standard of accounting practices, and the accountability of fiduciary managers; they facilitate the choice for donors in general and for small donor in particular. When applied to charitable institutions, and especially charitable corporations, the high level of accounting practices allows making the performance of charitable organizations transparent, and thus make it easy to measure. The accountability of managers means that the level of state enforcement of managers’ duties is high. Therefore, people feel relatively secure that the money they donate will go to the charities they choose. In the absence of these two ingredients, people would be reluctant to donate to charitable corporations. Indeed, in most places and in most eras such corporations have not been widely observed. Instead, organizations such as churches, ethnic groups and worker associations managed these tasks. Large would-be donors often shunned charitable organizations altogether and instead personally directed their charitable activities.
The unifying aspect of the measurement framework is reflected in that it generates implications that apply generally, independently of particular firm models. The core dual implications are: 1. As the cost of measuring products declines (increases), we expect the scope of organizations to fall (increase). 2. As the cost of measuring labor inputs declines (increases), we expect the scope of organizations to increase (decline).
The focus on enforcement, and on the difference between state enforcement and enforcement by long-term relations proves to be a major ingredient in determining what will fall within organizations and what will fall across them. For instance, as the common law becomes more entrenched, we predict that a larger fraction of all interactions will be between organizations. A fall in the probity of judges will pull in the opposite direction.
Finally, whereas we cannot pin the firm down (is the Red Cross a firm?) the ownership structures of different organizations allows us to predict the lines of activities in which they have a comparative advantage, and such predictions may be readily tested.
References
Alchian, Armen A., “Some Economics of Property Rights,” Il Politico, (1965), Vol. 30, 816-29.
Alchian, Armen A. and Harold Demsetz, "Production, Information Costs, and Economic Organization," American Economic Review 62, (1972), No. 5: 777-95.
Alchian, Armen A, and Susan Woodward, “The Firm is Dead; Long Live the Firm,” Journal of Economic Literature 26, (1988), No.1: 65-79.
Allen, Douglas W., “What are Transaction Costs”, Research in Law and Economics, Vol. 14, 1991, 1-18.
Allen, Douglas W., “Transaction Costs,” in Encyclopedia of law and Economics, Bouckaert, Boudewijn and De Geest, Gerrit (eds.), The Encyclopedia of Law and Economics, vol. I, 893-926, Chelthenham, Edward Elgar (2000).
Alsberg, Carl L., "Protein Content: A neglected Factor in Wheat Grades," 2 Wheat Studies of the Food Research Institute, 163-76 (1926).
Bajari, Patrick, and Steven Tadelis, “Incentives versus Transaction Costs: A Theory of Procurement Contracts,” mimeo, October 2000.
Barzel, Yoram, "Measurement Cost and the Organization of Markets," JLE, April 1982.
Barzel, Yoram, “Transaction Costs: Are they Just Costs?” Journal of Institutional and Theoretical Economics, Vol. 141, 4-16, (1985).
Barzel, Yoram, Economic Analysis of Property Rights, Cambridge University Press, First Edition, 1989, Second Edition, 1997.
Barzel, Yoram, 2002, A Theory of the State: Economic Rights, Legal Rights, and the Scope of the State, forthcoming, Cambridge University Press.
Barzel, Yoram and Wing Suen, “Equity as a Guarantee: A Contribution to the Theory of the Firm,” mimeo, University of Washington, 1995.
Cheung, Steven N.S., “The Structure of the Contract and the Theory of Non-Exclusive Resources,” Journal of Law and Economics, Vol. 13, pp. 49-70 (1970).
Cheung, Steven N. S., 1983 “The Contractual nature of the Firm,” Journal of Law and Economics, April, Vol. 26, pp. 1-22.
Coase, Ronald H. (1937) “The Nature of the Firm,” Economica 4, no. 3: 386-405.
Coase, Ronald H. (1960), "The Problem of Social Cost," Journal of Law and Economics (October).
Coase, Ronald H. (2000). ”The acquisition of the Fisher body by general Motors,” Journal of Law and Economics (April) 43: 15-32.
Demsetz, Harold, 1988. “A Framework for the Study of Ownership,” in Ownership, Control, and the Firm. Oxford: Basil Blackwell.
Demsetz, Harold, 1998. “Review: Oliver Hart, ‘Firms, Contracts, and Financial Structure’,” Journal of Political Economy 106: 446-452.
Dittmer, Timothy
Grossman, Sanford, and Oliver Hart, 1986. “The Costs and Benefits of Ownership: A Theory of Lateral and Vertical Integration,” Journal of Political Economy 94: 691-719.f
Holmstrom, Bengt, 1999. “ The Firm as a Subeconomy,” Journal of Law, Economics & Organization 15: 74-102.
Holmstrom Bengt, Paul Milgrom (1991) “Multitask Principal-Agent Analysis: incentives Contracts, asset ownership and job design”, Journal of Law, Economics and Organisation, 7, 24-52.
Holmstrom, Bengt, and Paul Milgrom, 1994. “The Firm as an Incentive System,” American Economic Review, 84: No. 4, 972-991.
Jensen, Michael C. and William H. Meckling, 1976, “Theory of the Firm Managerial Behavior, Agency Costs and Ownership Structure,” Journal of Financial Economics 3, No. 3: 305-60.
Klein, Benjamin, Robert C. Crawford and Armen A. Alchian, 1978 “Vertical Integration, Appropriable Rents, and the Competitive Contracting Process,” Journal of Law and Economics, October: 21, 297-326.
Klein, Benjamin, (2000). “Fisher-General Motors and the Nature of the Firm,” Journal of Law and Economics, (April) 43: 105-142.
Masten, Scott E. “Contractual Choice,” in Encyclopedia of Law and Economics, B. Boukaert &G. De Geest (eds.), Edward Elgar Publishing and the University of Ghent, 1999.
McManus, John. 1975. “The Cost of Alternative Economic Organization,” The Canadian Journal of Economics: 8, 334-350.
Rock, Edward, and Michael Wachter. 2001. “Islands of Conscious power: law, Norms and Self Governing corporations,” Forthcoming in: Symposium, Norms and corporate Law, University of Pennsylvania Law Review, Spring 2001.
Williamson, Oliver E. 1975. Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press.
Notes.
1. Allan Schwartz: Rational Contracts, JLS 92 (271)
2. R&W p. 32 compare partnership with corporation
3. Milgrom & Roberts 92 book Org. &
4. When individuals produce services that are difficult to measure and difficult for outsiders to observe, sharing is an effective form of agreement between pairs of individuals. Such sharing agreements are self-enforced because each party directly gains from its own performance (shared ownership and operations of taxicabs; marriage).
5. How about incentive structure changes in Bernstein’s associations?
6. By what unit should size be measured—value of capital; of transactions guaranteed? Market value of firm? How can these be made operational?
7. It seems that (some) home-owners’ associations perform functions that the state would not. Their internal rule are enforced differently than are those of the state. Same applies to Bernstein’s associations.
