* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download sustaining networks - IESE Business School
Survey
Document related concepts
Transcript
Working Paper WP-115 October, 1986 SUSTAINING NETWORKS J. Carlos Jarillo Joan E. Ricart IESE Business School – University of Navarra Av. Pearson, 21 – 08034 Barcelona, Spain. Phone: (+34) 93 253 42 00 Fax: (+34) 93 253 43 43 Camino del Cerro del Águila, 3 (Ctra. de Castilla, km 5,180) – 28023 Madrid, Spain. Phone: (+34) 91 357 08 09 Fax: (+34) 91 357 29 13 Copyright © 1986 IESE Business School. IESE Business School-University of Navarra - 1 SUSTAINING NETWORKS J. Carlos Jarillo1 Joan E. Ricart2 Abstract The notion of networking is gaining widespread acceptance as a strategic tool for obtaining the efficiency and flexibility that are so badly needed in today's economic environment. It is also seen as the preferred way for means-constrained entrepreneurs to overcome their limitations and launch successful enterprises. The success of many business forms that can be conceptualized as networks, from franchising to the “Benetton system” to the web of subcontractors behind well-known Japanese firms, reinforces the perception of networking as an important phenomenon. Unfortunately, networks are not easy to deal with at a conceptual level. The very essence of a networking arrangement entails cooperation, and the mental frameworks of business policy and microeconomics are much better prepared to address competition. Of necessity, we must postulate a profit-maximizing behavior by business firms, leave open the possibility for opportunism, and recognize the fact that contracts are, in many cases, not enforceable at a reasonable cost. It is precisely for this reason that economists have traditionally seen large organizations as the only alternative to open, competitive markets. However, a network is something clearly in-between: independent, profit-oriented firms that cooperate over the long term. Strategy scholars have traditionally been equally wary of cooperative arrangements: relationships between firms are normally analyzed in terms of “power relationships” or “competitive positions”. The need to understand why cooperation is indeed feasible, at a theoretical level, is patent: without such understanding, all analysis of the phenomenon, let alone any prescriptive statement for practitioners, will be condemned to superficiality or irrelevance. It is revealing that most research on networks has been carried out on not-for-profit firms, where the cooperation-competition dichotomy is, at least, less acute. This paper presents some basic theoretical ideas on why cooperation is indeed possible for profit-maximizing business enterprises. It covers both fronts: first, why cooperation may be an efficient arrangement; and second, how cooperation can be sustained over the long run. Drawing heavily from games and agency theory, some basic models are established that shed some light on real-life cooperative arrangements. In the final analysis, long-term outlook and the ability to elicit personal trust are seen as key ingredients for a successful network. 1 Assistant Professor, Strategic Management, IESE 2 Assistant Professor, Economics, IESE IESE Business School-University of Navarra SUSTAINING NETWORKS 1. Introduction “Networking” is a fashionable topic. It is receiving increasing interest in popular management publications, as well as specialized academic journals. After many years of study of the topic from the field of organizational theory, strategy scholars are now discovering it, particularly in the United States (Miles and Snow, 1984; Thorelli, 1986). In Europe, a longer tradition exists, based around British and Swedish scholars (Farmer and McMillan, 1976; Macmillan and Farmer, 1979; Johanson and Mattsson, forthcoming; Mattsson, 1986i Lorenzoni, 1982; forthcoming). Although increasingly more phenomena that fall under the general heading of networking are drawing the attention of strategy scholars, the overall phenomenon is not well integrated within the general framework of current strategy thinking. The main reason is that all networking arrangements rest upon some sort of cooperative behavior among independent business units. This is hard to fit in a neat conceptual way with the overall competitive paradigm. This paper addresses this problem by showing how cooperative behavior is indeed consistent with profit maximizing behavior, given certain conditions. We will concentrate on cooperation between pairs of firms, as this is the simplest possible case. This analysis of dyadic cooperative relationships will provide a building block for understanding complex cooperative networks. The paper first tries to define carefully the concept of strategic network; section 3 analyzes cooperation as the theoretical stumbling block, and offers some well-known insights from game theory. Section 4 analyzes the dyadic relationship from the point of view of agency theory. Finally, section 5 summarizes the conclusions and opens up some avenues for further thinking. 2 - IESE Business School-University of Navarra 2. Strategic Networks A strategic network is a form of organization that it is somewhere in between “hierarchies” and “markets”, to use Williamson’s terminology (1975)1. It can be defined as a web of “long-term, purposeful arrangements among distinct but related for-profit organizations that allow those firms in them to gain or sustain competitive advantage vis-à-vis their competitors outside the network” (Jarillo, 1986a)2. The different firms in a network are independent in most dimensions, which is what differentiates a network with respect to a vertically integrated or quasi-integrated organization (Blois, 1972). It is a mode of organization that is not based on the price mechanism, nor on a hierarchical relationship, but on coordination through adaptation (Johanson and Mattson, forthcoming). It is a long-term relationship based on implicit contracts, without specific legal ties. We can analyze now in more detail how a network is “between” markets and hierarchies. The different “modes” of organizing complex economic activities can be categorized by representing them in a matrix determined by two variables: the “legal” organization and the “kind of relationship” (see Figure 1.) This representation gives us four “prototypes” or pure forms of organization (See Jarillo, 1986a, for a more detailed explanation): Figure 1 Approach to the Relationship Legal Form Hierarchy Market Zero-Sum Game Non Zero-Sum Game Classic Market Strategic Market Bureucracy Clan “Classic Market” is an arrangement where many players interact on a spot basis. “Bureaucracy” is a hierarchical organization with many characteristics of an open market (basically, the pursuit of different interests by the interacting agents). It can be exemplified by the antagonistic labor-management relationship. 1 Williamson’s well-known work states that “markets” are the “normal” mode of organization. However, transactions between independent players in those markets do have a cost. Sometimes, the transaction cost is so high, that it is more efficient to group the two players in the transaction under the same roof, in fact eliminating the market transaction and originating what Williamson calls a “hierarchy”, which is the equivalent to a “firm” or an “organization”. Thus “markets” and “hierarchies” are two different modes of organizing economic activity. The most efficient one for a given kind of repeated transaction will eventually prevail. 2 This second section of the paper draws heavily from Jarillo (1986a), where the argument is developed in full. IESE Business School-University of Navarra - 3 “Clans” (Ouchi, 1980) are long-term relationships, carried out through non-specified contracts within the formal environment of an organization. The unity of purpose within the organization is strong, which lowers the cost of internal supervision.3 “Strategic Networks” are the focus of this paper. A firm has a special relationship with the other members of the network, which are independent organizations with no point of contact in many other dimensions. However, this relationship also has many characteristics of a “hierarchical” relationship: relatively unstructured tasks, long-term point of view, relatively unspecified contracts. Once we have classified the forms of organizations, the next natural question should be to try to understand how a network can be more efficient than any other mode of organization since, by definition, if it weren’t more efficient, it would not survive. We can view a firm as a collection of activities that add value between suppliers and customers (Porter, 1985). Each of these activities can be performed internally at a given internal cost IC, or be subcontracted at a given external cost EC. In the latter case, we will also incur in a transaction cost TC. Therefore, a particular activity will be integrated or not depending on the relation between IC and EC+TC. Thus, a first technological constraint for a network to exist is that EC<IC.4 Given this, it is still necessary that the cooperating firms be able to reduce the transaction cost to the point where EC+TC<IC. Thus, the difference between a strategic network and the market is that, in the case of the market, TC’s are low enough for any player in the market, out of their own nature. In the case of networks, there is a first high cost that the firms taking part in the network are able to lower through conscious efforts. Not every firm, then, is in a situation to perform those functions externally (because they cannot get TC’s that are low enough). The firms in the network, therefore, enjoy a competitive efficiency out of being in the network (all other things being equal, of course). A network can be sustained only if it is both effective and efficient (Barnard, 1968, first introduced those concepts applied to organizations, but they can be used fruitfully here). It is effective if it can achieve its end at a lower total cost than alternative modes of organization. This effectiveness depends on two factors: technological considerations and transaction costs, as we have just seen. The network is efficient if it offers the firms taking part in it more than it demands from them. This also depends on two factors: first, the participants realize there is a larger pie to share (which implies both effectiveness and the perception of it); and second, there is a fair mechanism for sharing the pie, i.e., each individual participant thinks it will gain more by being part of the network. 3. Cooperation as the stumbling block Williamson (1975) argues, and we have followed him, that markets fail and hierarchies emerge because of costs that make an arrangement based on market transactions inefficient. Those “transactions costs” stem from four reasons: man's “bounded rationality” (Simon, 1976), uncertainty about the future, the presence of a “small number” of players for a given kind of 3 Working within Williamson’s framework, Ouchi realized that many “hierarchies” do not, in fact, eliminate “transactions costs” between their internal constituents. He then distinguished between “bureaucracies” and “clans”. 4 A typical reason why external costs may be lower than internal costs is the possibility of capturing external economies of scale, i.e., buying from a specialized, high-volume supplier. Stiegler (1968) develops the argument in full. 4 - IESE Business School-University of Navarra transaction, and the possibility of “opportunistic behavior” on the part of (at least) some of the players. Those are the hurdles an entrepreneur has to overcome in order to establish a strategic network (or gain his or her way into an already existing one). There is only one way to do this and that is to generate trust. As we are about to see, trust is the great transactions-cost-lowerer (see Jarillo, 1986b). Leaving out such important points as personal congruence in objectives, friendship and other related psychological and social factors, we shall concentrate now on analyzing the problem of trust, from a formal point of view, by drawing some insights from the theory of repeated games and related economic areas. A well known problem in game theory is the prisoner’s dilemma that we will use as a starting point of our analysis. The general form of the game, as well as a particular version of it, is given in Figure 2: Figure 2 Cooperation Cooperation No Cooperation Cooperation R,R S,T Cooperation No Cooperation T,S P,P No Cooperation No Cooperation 15,15 5,21 21,5 12,12 In the static version of the game, each player (row or column) must choose, simultaneously to and independently of the other player, one of the two strategies available, cooperate or not cooperate. If both cooperate, they receive the reward R. If none cooperates, they receive the punishment P. The problem arises because each player has a temptation value T that is greater than the reward R. Furthermore, if one player cooperates while the other does not, he will receive the lowest possible payoff, called the “sucker's payoff” S.5 For the general situation to be a prisoner's dilemma, we must have: T > R > P > S6 With these relations, as shown in the example, it is better for both players not to cooperate, regardless of what the other player does. In the game theory jargon, not to cooperate is a “dominant strategy” for both players, so that the expected outcome of the game is not to cooperate for both players, resulting in the payoff pair P, P. Since this outcome is “socially dominated” by the cooperative outcome R, R, the result is highly negative. In our context, given that many situations that arise in a business relationship can be represented in this way, we 5 The “prisoner’s dilemma”, the name given to this general model, comes from its first formulation. It runs as follows. Two (guilty) criminals have been arrested by the police. The police has some evidence that would cause the criminals to receive a short sentence, but they lack substantial evidence about the more serious crimes. Both criminals (separately) are offered freedom if they give testimony against the other. If one of them does, the other is hanged. So, the reward for their cooperation (not talking) is to obtain a light sentence. If one talks, he goes free and the other is hanged. But if both talk, both get heavy sentences (for there is incriminating evidence against both). The outcome of the game, without absolute trust in each other, is that both get heavy sentences. 6 Strictly speaking, (T + S) < 2*R is also a necessary condition, especially if we allow for repetition of the game. IESE Business School-University of Navarra - 5 encounter a big difficulty for sustaining networks.7 Still more negatively, if we repeat this game any finite number of times, one can readily realize by backward induction that not cooperating is the only equilibrium of this game for both players. The prisoner’s dilemma is a good formalization of the basic problem of lack of trust. In spite of the obvious benefit of cooperation, there is a strong temptation to default and take a larger benefit. But, even more importantly, our own fears of being fooled (and get S) induce us to act first and not to cooperate. Since both parties, if they are rational, would think along the same lines, we end up at the inefficient point, i.e., the relationship is not possible. How can an entrepreneur deal with such a negative situation? One first obvious direction would be to try to change the game they are playing. Any effort in this direction is worthwhile and can help avoid the dilemma by eliminating it. As a static game, however, it is not always easy to change the game so that the temptation is avoided. One trivial way to do this is to sign a contract or similar, so that both players (one is not enough) are committed to cooperating. But contracts are often not enforceable, at least at a reasonable cost: this is precisely the essence of transaction costs. Intuition tells us that repetition should foster cooperation. Theory, however, prescribes that no cooperation is the only equilibrium in the finitely repeated game. An experiment contradicts this point. In his book “The Evolution of Cooperation” (1984), Robert Axelrod describes a computer game with characteristics similar to those of the prisoner's dilemma. A large number of professional game theorists sent subroutines ready to play the role of one of the players against all the other submitted routines. The game was repeated 200 times and everybody knew this information. The winner was a very simple routine submitted by Prof. Anatol Rapoport from the University of Toronto. His strategy is now known as TIT FOR TAT. The strategy specifies the following rule: start by cooperating; at any other stage, do exactly what your opponent did in the previous move. That's all. This strategy proved to be robust even against players that knew that this strategy was the winner in the first round. Furthermore, it was also robust in an evolutive simulation where the winning strategies reproduced themselves and changed the environment for the next round. Therefore, there seems to be hope for cooperation to survive and end up dominating, even in a world of mistrust and no cooperation. Analyzing the traits of the winning strategy, the author finds three relevant characteristics: first, the strategy is nice, since it never starts a non-cooperative move by itself; second, it is provocable, i.e., it “gets mad” quickly at defectors and retaliates; third, it is forgiving, since retaliation is proportional to the length of defections. These three “good personality traits” of being nice, provocable, and forgiving, seem to be the essence of both the strategy’s robustness and its survival in an aggressive environment. Without trying to push too much in this direction, we may think that these are personality traits we should find in an entrepreneur able to create and sustain a network. It also gives us hope that cooperation can survive even in a non-cooperative environment and end up being the prevalent “way of doing business”. We also realize that the strategy’s robustness indicates 7 One example is joint R&D. If several firms decide to take on a given project in a cooperative way, it is to the best advantage of each to assign a second-rate scientist to the project, while expecting that the others will assign their best. No firm wants to send their best scientist, fearing that the others will send their seconds. So everybody sends seconds, with the result that the project fails. 6 - IESE Business School-University of Navarra that it is beneficial to have a reputation for being a “TIT FOR TAT”, since it beats more aggressive behavior in the long run, even if this behavior is designed against it. Finally, note that this strategy never really beats its opponent; at most, they end up equal; the benefit comes from doing well on average against any opponent. Thus, taking a long-term approach seems to bring an effective improvement to the problems of cooperation. Let us, then, try to analyze in more detail what including the long-term outlook contributes to the problem. To be more realistic, the game has to include an “uncertainty factor”, ∂. This means that the game will be repeated with a probability ∂, depending (presumably) on the satisfactory results of previous games, among other things. Then, the probability that the relationship will still survive in period t is ∂t. The incentive to cheat in a particular game is tempered by the loss of future potential gains. Obviously, the game’s results depend on the value of ∂, i.e., how much the future is valued. It can be seen that, as ∂ increases, so does the incentive for cooperation. Following the case in Figure 2, the incentive for not cooperating is 21-15 = 6. But, if cooperation is lost forever, we must take into account the 15-12 = 3 loss that comes from it, adjusted by the probability of having more chances to play. In other words, the opportunistic gain is sure, while the future loss depends on the probability of having more games to play. We can calculate a value for ∂ that corresponds to the cut-off point between cooperation and noncooperation. Thus, the future value of cooperation is 15 12 1 The opportunistic gain is 6. Therefore, there will be cooperation if 6 15 12 1 The cut-off value can be found by solving 6(1-∂) = 3∂. The result is ∂=2/3, which means that if the parties think that there is at least a chance of 2/3 for long-term repetition of the game, they will cooperate.8 Given our interpretation of the game as a strategic network relationship between two parties, it seems that an infinite game with discounting that incorporates a probability of continuation is an acceptable representation of the situation. Note that if ∂<1 the game will finish in a finite time with probability 1, but the final time is a random variable. We are, in essence, saying that the threat to severing the relationship becomes the main deterrent to non-cooperation.9 8 In general, 9 For this to hold, there must be a “subgame perfect equilibrium”, i.e., the threat must be credible. ( T – R ) (1 – ∂ ) = ( R – P ) ∂ ; T–R=∂(T–P) ∂=(T–R)/(T–P) ∂ is always less than 1, for R is greater than P, by assumption ∂ can also be interpreted as a discount rate. A cash flow A obtained in period t will have a present value of A∂t. IESE Business School-University of Navarra - 7 The lessons to be gained by these results is that cooperation may be sustainable, even allowing for opportunistic behavior. There are three points that should be highlighted: Repetition permits cooperation. Threats must be credible. The discount factor matters. Impatience makes the agreement more difficult.10 Both the theoretical solutions and the empirical analysis show the same characteristics in the optimal strategy for supporting cooperation. This seems to be telling us something about the way one should act to address cooperation in the long run. In summary, be nice, provocable and forgiving. Furthermore, any particular situation can be represented as a game like the one in our example. If we are able to modify the game, we can make cooperation somewhat easier. We can try to reduce the temptation level, we can include stronger punishments and make them “cheap” to implement, or we can try to modify the discount factor by increasing the (perceived) probability of continuing the relationship. Building a reputation becomes an essential consideration, as it clearly introduces the long-term factor. 4. The supplier relationship So far, we have presented some fairly general points on cooperation. Let's now turn to some aspects more directly related to actual business practices. We will start by examining contributions from the field of agency theory. Agency theory is the branch of economic theory that, for the last 10 years, has been formally studying the relationship between a “boss” or “owner” and an “employee”. We will try to gain some insights from this work (for an up-todate introduction, see Pratt and Zeckhauser, 1985). Although it formally deals with “inside” relationships (i.e., within the organization) we will apply it to the external relationship between a larger firm and a subcontractor, for networking relationships are, again, in between external and internal transactions. It should not be surprising, then, that an agency relationship can be shown to have a good fit with real-life relationships between firms in a network. We say that an agency relationship is present when we have a party, called the “agent”, that has to act on behalf of another party, whom we call the “principal”. The a priori problem is to find an arrangement that solves the incentive problem. There is an incentive problem if the following three elements are present in the relationship. First, there should be some kind of uncertainty, so that given a particular outcome, it is not possible to know with certainty if the agent chose the right actions or not. Otherwise, an authoritarian relationship would solve all the problems. Second, we must have some discrepancy in the agent’s and principal’s objectives. Otherwise, the only problem is sharing the joint profits, but not how they are obtained. Third, the agent should be risk-averse. This last condition is necessary since otherwise the problem is trivial: give all the risk to the agent in exchange for a constant value. Given a relation that complies with the above conditions, we have a problem of finding a tradeoff between sharing risk and giving enough incentives to the agent to choose the “right” action that is best for the relationship. The “optimal” arrangement is the one that achieves equilibrium 10 An interesting twist is that ∂ can be lowered and the cooperation equilibrium will still hold if we include the possibility of penalties. 8 - IESE Business School-University of Navarra between the two opposing goals. By way of illustration, suppose that the principal is riskneutral. Then, an arrangement that gives a flat wage to the agent is the best for risk-sharing but does not give the agent any incentive to work hard. On the other hand, an arrangement that gives a constant payment to the principal is the best incentive for the agent to excel, but it is a disaster in a risk-sharing sense. Along these lines, Holmstrom and Milgrom (1985) present an interesting model where the agent controls the drift of the outcome stochastic process, i.e., the result of the agent's acts is determined by his effort and some randomness. With some assumptions about the utility function and the distributions, they prove that the optimal incentive scheme is a linear function of the aggregate outcome. The choice of the parameter of the linear function determines the trade-off between risk-sharing and incentives. Kawasaki and McMillan (1986) use this model for an empirical test of Japanese subcontracting, a well-publicized case of successful networking practices. Subcontracting is a very extended practice in Japan. There are four times as many wholesale transactions as retail transactions in Japan; in the United States, Britain and West Germany, in contrast, the ratio of wholesale to retail transactions is between 1.6 and 1.9. In the Japanese automobile industry, an average of 75% of a car's value is provided by outside suppliers, and only 25% is produced within the firm; in the U.S. automobile industry, 55% is provided by outside suppliers. This overall trend is increasing over time. The Ministry of International Trade and Industry (MITI) has stated that the “Japanese manufacturing industry owes its competitive advantage and strength to its subcontracting structure” (the preceding evidence is provided and referenced by Kawasaki and McMillan, 1986, p.1). It has been said that subcontracting practices are simply a mechanism by which large firms export some of their business risk to smaller, defenseless subcontractors. The larger firms would, for instance, keep production in-house when there is a slack in demand, thus avoiding lay-offs. The agency relationship would then be based on exporting all of the risk to the subcontractors. We are about to show that this does not seem to be the case. For a principal to take on some of the risk of the relationship, there must be an agreement that shifts (at least) part of the variance in the subcontractor's costs to the principal. Let's say there is an agreed price b. If there is a cost overrun (or underrun), the price may be maintained, and the subcontractor is forced to bear all the risk. Or, conversely, there may be a factor , that represents the share of variance taken up by the principal. Thus, if = 0, then the subcontractor bears all the risk (it is a purely fixed price contract); if = 1, the principal bears all the risk (it is a cost-plus contract). The problem with = 0 is that there is no real network relationship (it would be the case of buying something at a given, fixed price, with no further involvement between the firms). The problem with = 1 is that there is no incentive whatsoever for the subcontractor to be efficient. Our prediction, from what has been said on strategic networks (and the success of the Japanese system), is that there would be an > 0. Kawasaki and McMillan examine a large sample of subcontracting arrangements, and find the following: first , the subcontractors are indeed risk-averse (as may be expected from their small size, compared to the principals); second, the contracts have the principal absorbing some of the risk on behalf of the subcontractor ( > 0); third, grows , among other things, with the subcontractor’s degree of risk aversion and the size of the cost fluctuations; finally, the average is 0.69, with many of them being above 0.75. This means that the contracts are closer to the cost plus end of the spectrum. IESE Business School-University of Navarra - 9 Although an in-depth analysis of Japanese subcontracting practices is clearly beyond the scope of this paper, and would have to include considerations from many different fields, the previous points allow us to realize that an agency relationship offers a good representation of the actual arrangements of extremely successful industrial networks. Thus, the risk-sharing agreement is basic to the relationship’s long-term success, and the principal has to be willing to take it. The supposed “exploitation” is certainly nowhere to be found. The arrangement gives flexibility to the large firm while the subcontractor is better off because of the risk absorbed by the big firm, which is presumably more risk-neutral. Furthermore, the contract seems to lean more toward the risk-sharing feature than to the incentive function. The authors try to explain this feature by the fact that periodical recontracting, taking into account competitive offers, is by itself a strong enough incentive to perform adequately. 5. Conclusions The fields of game theory and agency theory can shed more light on modeling basic networking mechanisms. For instance, a firm can specialize in designing the game that other firms play. It is a fact that cooperation is often impossible because of information asymmetries or some other problem intrinsic to the situation. But a third party -who can also be involved at a different level- can act as mediator, regulator, arbitrator, etc. The inventory of conceptual models for representing activities in a network is thus enlarged well beyond the mere cooperation to include network building and proactive sustaining activities. In any case, the simple models reviewed here provide the following conclusions for understanding why and how cooperative relationships are possible. First, there is certainly room for “in-between” situations. Relationships of the kind studied here are neither “arms-length” nor purely internal (“hierarchical”). It is important for managers to realize this, particularly when the decision at hand is to integrate a function (i.e., performing it “in-house”) because of the need to control it, when it would be more efficient to subcontract it. An entrepreneur who is trying to establish his or her firm must also realize this point, and act accordingly. Trust is of the essence. This comes as no surprise, but we have seen how it is at the root of the problem for cooperation. But we haven't only seen that, we have also analyzed why, following the simple “prisoner's dilemma”. This analysis has shown the two general avenues for overcoming the problem: develop long-term relationships (or, for an entrepreneur, do all you can to show you are in it for the long run, that your reputation is at stake), and try to modify the game, acting on the four variables: cooperation reward (increase); opportunistic gain (reduce); punishment (increase); “sucker’s payoff” (reduce); and, above all, the importance of the future. The agency model reviewed shows how risk is indeed an important consideration and how the large firm can trade risk and incentive with the smaller subcontractor. Again, in the case of the entrepreneur, it is up to him or her to realize the situation and implement terms that will be beneficial to both. These are still fairly general ideas, reduced to dyadic relationships within a network. But they can help in conceptualizing how cooperation is possible between firms, and which are the key variables that entrepreneurs and large firms alike must act on to improve overall efficiency. 10 - IESE Business School-University of Navarra References Axelrod, R. (1984), “The Evolution of Cooperation,” New York: Basic Books. Barnard, C. I. (1968), “The Functions of the Executive,” Cambridge, Harvard University Press. Blois, K. J. (1972), “Vertical Quasi-Integration,” The Journal of Industrial Economics, Vol. 20. Farmer, D. H. and K. MacMillan (1976), “Voluntary Collaboration vs. “Disloyalty” to Suppliers,” Journal of Purchasing in Materials Management, Vol. 12, No. 4, pp. 3-8. Jarillo Mossi, J. C. (1986), “On Strategic Networks,” Working Paper No. 112, IESE, Barcelona. Johanson, J. and L. G. Mattson, “Interorganizational Relations in Industrial Systems: A Network Approach Compared with the Transaction Cost Approach”, International Journal of Management and Organization, forthcoming. Kawasaki, S. and J. McMillan (1986), “The Design of Contracts: Evidence from Japanese Subcontracting,” Mimeo, University of Western Ontario. Lorenzoni, G. (1982), “From Vertical Integration to Vertical Disintegration,” Paper presented at the Strategic Management Society Conference, Montreal. Lorenzoni, G., “Venturing without Investing,” Journal of Business Venturing, forthcoming. MacMillan, K. and D. H. Farmer (1979), “Redefining the Boundaries of the Firm,” The Journal of Industrial Economics, Vol. XXVII, No 3, pp. 277-285. Mattson, L. G. (1986), “Management of Strategic Change in a ‘Markets-as-Networks’ Perspective,” Working Paper prepared for the “Management of Strategic Change” Seminar at the University of Warwick. Miles, R. E. and C. C. Snow (1984), “Fit, Failure and the Hall of Fame,” California Management Review, Vol. XXVI, No 3, pp. 10-28. Ouchi, W. G. (1980), “Markets, Bureaucracies and Clans,” Administrative Science Quarterly, Vol. 25, pp. 129-142. Porter, M. (1985), Competitive Advantage, New York: The Free Press. Pratt, J. W. and R. J. Zeckhauser, eds. (1985), “Principals and Agents: The Structure of Business,” Harvard Business School Press. Simon, H. A. (1976), “Administrative Behavior: A Study of Decision-Making Processes in Administrative Organizations,” The Free Press. Thorelli, H. B. (1986), “Networks: Between Markets and Hierarchies,” Strategic Management Journal, Vol. 7. pp. 37-51. Williamson, O. (1975), “Markets and Hierarchies,” The Free Press. IESE Business School-University of Navarra - 11