Unilateral Commitments and the Importance of Process in Alliances

The rapid proliferation of strategic alliances during the past decade has captured the interest of both the business press and the academic literature. A number of empirical studies have documented the unprecedented growth of joint ventures in a variety of industries, and much has been made of the beginning of a new era of cooperation in which firms seek partnerships in several facets of their operations. Many see the growth of alliances as a key to sustained competitive advantage for U.S. industry.

Yet, despite the popularity and presumed strategic importance of alliances, the fact that they often fail suggests that our understanding of the appropriate ways for managing different kinds of alliances is quite limited. There are numerous cases of firms that have become disenchanted with alliances. Several studies have reported failure rates as high as 80 percent, failure that usually leads to their dissolution or acquisition by one of the partners.1 Prominent failures in the automobile industry alone include, for example, alliances between General Motors and Daewoo Corporation, General Motors and Isuzu Motors, Chrysler and Mitsubishi Motors, Chrysler and Maserati, Fiat and Nissan.

We believe that part of the reason contemporary alliances fail so often is because many partners view them, in game theory terms, as prisoner’s dilemma situations. Each partner fears that the other will get the larger payoff by acting opportunistically while it cooperates in good faith. The result is that both partners choose not to cooperate and are worse off than if they had cooperated. Conceptualizing alliances in this way, while appropriate in some instances, is restrictive, leading to the false belief that one management style fits all alliances and all stages of a particular alliance. In fact, as we will show in this paper, alliances vary considerably and, depending on their structure and how they evolve, they may need to be managed very differently from the recommendations that result from seeing them in a prisoner’s dilemma context.

One shortcoming of the prisoner’s dilemma framework is that it underestimates the importance of each partner acting unilaterally to enhance the possibility that all partners will cooperate. Our fieldwork has shown how vital unilateral commitments can be to the success of alliances. In this paper, we present a view of alliances that helps us understand why unilateral commitments can be so helpful.

We also argue that process is important. When unexpected changes (technological, political, regulatory, or industry changes that are beyond the alliance partners’ control) alter the economics and underlying incentives of the participants, different formulations and, hence, different management styles become appropriate. The management process for the day-to-day running of the alliance matters because institutionalizing methods to ensure that all parties appreciate such changes is important to the alliance’s continued success. That managers and academics underappreciate this is fairly clear. A Coopers & Lybrand study suggests that, while executives spent 23 percent of their time on developing alliance plans and 19 percent on drafting legal documents, they spent only 8 percent actually managing the alliance.2 Even if different stages of the alliance do require different time commitments by managers, the little time spent on managing its day-to-day affairs is startling.

For this study, we interviewed 143 managers in seventeen organizations in the United States and abroad that are engaged in alliances. The interviews were open-ended, with a focus on understanding some of the factors critical to the alliances’ success. The analytical framework we present here is derived from our observations during these field interviews. We base our discussion on the following observations and illustrate it with examples from our fieldwork. We observed firms that made unilateral commitments with the professed intent of somehow trying to influence the outcome of the alliance. These commitments, which appeared in a variety of guises, are difficult to rationalize if, indeed, the prisoner’s dilemma formulation is appropriate for all alliances at all times. We also observed that unexpected disruptions, completely outside the control of any of the alliance participants (individually or collectively), were commonplace in virtually every alliance we studied. Further, the way managers dealt with these disruptions differed widely across organizations, suggesting the importance of process issues in managing alliances.

We first review one approach to analyzing alliances —the prisoner’s dilemma — that has received the greatest attention. We then offer an alternative way of thinking about alliances, also couched in the language of game theory, and suggest the major differences.

The Limitations of the Prisoner’s Dilemma

The payoff that a firm gets from participating in an alliance depends not only on its own actions, but also on its partner’s. The tools of simple game theory are ideally suited to suggest conceptual frameworks that capture this interdependence. For our purposes, game theory also provides a way to explain the counterintuitive findings from our research and gives us a solid basis for our normative recommendations.

The most frequently used game, the prisoner’s dilemma in its simplest form, involves two players, A and B, who can either cooperate or not cooperate with each other. Table 1 shows their payoffs, depending on their choice of action. If prisoners A and B cooperate with each other, they have a greater payoff than if neither cooperate. However, if one does not cooperate, while the other does, then the noncooperative prisoner receives the highest possible payoff, while the cooperative prisoner loses out. The only equilibrium is if each prisoner does not cooperate.3 Thus, despite the existence of an alternative that is better for each prisoner (the 7,7 payoff that is obtained if both parties cooperate), this formulation for alliances suggests that the only thing that each prisoner will do is not cooperate. On the face of it, the much discussed failure of alliances is not surprising (if failure is equated with disappointment that the best outcome is not realized).

Of course, if alliances based on the prisoner’s dilemma theory are bound to fail, then each participant should rationally anticipate this and not enter the alliance in the first place. The reason alliances are formed at all is that most participants expect to get better payoffs than they would without the alliance. Thus, even if neither cooperate, they still expect to get more than if they did nothing. A manager at a firm with many alliances commented:

The benefits we obtain from our alliances are manifold. There are, of course, the direct benefits from the partnership itself. But above and beyond that, even if there are no positive payoffs from the alliance, there are other benefits that make it worth our while. Some alliances allow us to signal our interest in a particular area that might deter entry by our competitors, and others serve as important stimuli for our own internal engineering staff to redouble efforts in particular areas. Generally, the more alliances we do, the better we get at doing them.

To understand what we mean by cooperating or not, let us assume that any alliance requires something from each partner. Typically this investment is the sum of some tangible investment plus some unobservable effort. Noncooperation can arise if either partner reneges on its tangible investment commitments or shirks on its unobservable investments. In either case, in the payoff structure of the game, we would expect one partner to benefit and the other to lose.

As a conceptual framework for thinking about alliances, the prisoner’s dilemma is limited to particular types of situations. It assumes that each company’s best response to its partner’s choice of action is the same, regardless of what its partner chooses. Thus, it always pays for me not to cooperate with my partner, even if my partner is cooperating. In this framework, it is this dominance of one alternative (“not cooperate”) over the other (“cooperate”) that results in only one equilibrium outcome.

This conceptual framework gets more complicated if firms engage in a series of alliances over time. In repeated situations, an equilibrium may arise if firms are able to sustain mutual cooperation by threatening to punish the rival by not cooperating in the future if the latter does not cooperate now. Experimental game theory confirms that such “tit-for-tat” strategies outperform the strategy of not cooperating in every episode of a prisoner’s dilemma game. In recent years, management scholars have argued that alliances can be seen as having the same structure as repeated prisoner’s dilemma games and have recommended that firms adopt a tit-for-tat strategy in their ongoing governance of alliances.4

The literature on alliances suggests several additional modifications of this repeated prisoner’s dilemma game. A mechanism such as building a reputation might be used so that both players cooperate in each iteration of the game. Hamel, Doz, and Prahalad view the benefits of not cooperating in repeated games as increasing over time, and argue that this is especially true for the firm that learns more quickly from the alliance during the period of cooperation.5 Their recommendation, therefore, is that firms view alliances as learning races: the key is to learn fastest so that one can later back out of the alliance at an advantage.

Our fieldwork also suggests that many firms use a similar framework for their alliances and adopt tit-for-tat strategies. But we observed that such strategies were often unsuccessful. For instance, in an alliance between a U.S.-based office automation equipment manufacturer and an Indian engineering firm to manufacture and distribute office automation equipment in India, each partner agreed to provide twenty employees to the alliance. At the outset, the Indian firm had an internal crisis, compounded by a slow hiring and training process; as a result, it sent only fifteen employees to the alliance, promising that five more would follow. The U.S. partner immediately withdrew five of its own employees and insisted that they would return only when the Indian partner provided its share of employees. Shortly thereafter, the alliance was terminated. Further interviews with managers in the U.S. firm suggested that they had not had much success with alliances. Their behavior in this example was consistent with their behavior in others; they followed reactive tit-for-tat strategies. One manager remarked, “We are very careful in our alliances and monitor our partner’s behavior very carefully. Our experience suggests that we must look for early indicators of the integrity of our alliance partner. . . . It doesn’t take much for our warning bells to go off.”

This example illustrates how blind adherence to the recommendations of the prisoner’s dilemma concept may not always be the best approach. This is particularly true for alliances formed in the past decade, since their character has become quite different. Alliances are increasingly strategic, moving closer to the core of a firm’s activities than to its periphery. Today, alliances encompass traditionally protected areas such as R&D, and there is a greater parity between the partners’ relative size and contribution to the alliance.6 In contrast, alliances in the past usually had one dominant partner that contributed much of the technology and manufacturing prowess, and another that made smaller contributions such as providing personnel, financial resources, or specific market knowledge.

Because of their changing scope, contemporary alliances tend to be more interdependent, requiring crucial inputs from all partners involved to be successful. In such alliances, cooperating and not cooperating are abstractions for, say, “invest properly” and “withhold quality resources from the alliance,” respectively. The prisoner’s dilemma framework is unlikely to be the right one for such alliances since it is difficult to see how a partner can gain the most by withholding its crucial input. It is precisely in alliances where each partner is crucial to its success that it is advisable for each to make unilateral commitments to ensure the venture’s success. Thus, it is important to broaden our understanding of different kinds of alliances and the appropriate managerial strategies. We now turn to some alternative game theory views of alliances that coincide more closely with our field research observations.

The Importance of Unilateral Commitments

Other simple models highlight ways in which alliance participants can attain the best possible payoffs. Here we present one alternative, akin to many alliances in our research, and detail how the behavior patterns it suggests are considerably different from those of the prisoner’s dilemma formulation.

The feature that distinguishes the game shown in Table 2 from the prisoner’s dilemma game is that there are two possible equilibriums. No longer is a company’s best response to its rival’s strategy the same, regardless of what the rival does. Thus, Company A’s cooperation is best responded to by cooperation from Company B; Company A’s failure to cooperate is best answered by a similar failure to cooperate by Company B. The outcome under which both companies get a higher payoff (9,9) is now a possible equilibrium. The other possible equilibrium, however, is one in which both players do not cooperate. The question remains: How can we influence which equilibrium will arise in practice? Making unilateral commitments emerges as one way to influence the outcome.

Commitment to Sacrifice

There are two kinds of unilateral commitments. The first refers to actions that a particular company can take to alter only its own payoffs under particular outcomes. What is such a unilateral commitment? It could be something like signing a long-term contract with a supplier (not the partner in the alliance, but a third-party supplier) for material needed for the alliance. Then, if the company signing the contract chose not to cooperate, it would still be saddled with the contract’s implications and could lose heavily as a result. Or it could be emphasizing the importance of the alliance and committing the firm’s reputation to its success. Then failure sends a bad message to the market about the firm’s ability to implement its strategy.

Examples from our fieldwork illustrate such unilateral commitments more concretely. One of the goals of an alliance between a U.S.-based software firm and a hardware manufacturer was to combine the latter’s superfast chip technology with the former’s new PC operating system. To demonstrate its commitment to the alliance, the hardware manufacturer decided to disband unilaterally its own internal PC software development group. By amputating a crucial part of the overall strategy’s success, the hardware manufacturer set itself up to be a big loser if it did not cooperate with its software development partner by providing it with information to take advantage of the chip’s capabilities. In making this unilateral commitment, the hardware manufacturer indicated that it had every intention of cooperating with its partner, hoping that the partner would also cooperate.

If the software firm lived up to its side of the bargain, the hardware manufacturer felt that it had a good chance of making its chip the leader in the PC market. But the software firm made no such commitment in turn; it could continue to develop its new operating system, so it did not rely as heavily on the advantages of the hardware manufacturer’s new chip. But since the new operating system could be superior if designed to take advantage of the new chip’s capabilities, the hardware manufacturer believed that, by making a unilateral commitment, it had created a situation in which it was more likely that both firms would cooperate fully.

In another example, a startup U.S.-based firm that designed computer disk drives and a large Japanese disk drive manufacturer agreed to an alliance in which the Japanese firm would manufacture computer disk drives for the U.S. firm in exchange for the rights to market them in Asia. The Japanese firm agreed up front to pay any costs that arose from design changes. Just after the Japanese firm had spent $6 million tooling up for production, a technology shift forced the U.S. firm to modify its designs. This, in turn, required the Japanese firm to spend an additional $600,000 adjusting its manufacturing processes. The Japanese firm kept its commitment and chose to absorb the costs and respond promptly to the changes. The Japanese manufacturer’s unilateral decision to pay any subsequent additional costs arising from technology shifts was an important signal to the U.S. startup of the Japanese manufacturer’s commitment to the partnership.

In a game theory context, the unilateral commitment mechanism does not work according to the prisoner’s dilemma scenario. In that situation, nothing that, say, prisoner A does to alter his own situation changes the fact that it is better for prisoner B to choose not to cooperate, regardless of prisoner A’s choice. The dominant strategy for each is always to choose not to cooperate, regardless of what the partner does.

In the situation illustrated by Table 2, Company A is best off choosing not to cooperate if Company B does the same. Now suppose that Company A undertook a unilateral commitment beforehand so that, rather than getting 5 when both companies chose not to cooperate, it got something less than 4, say 4-e. Then Company B will realize that if it chooses not to cooperate, Company A will no longer respond by not cooperating. In this new situation, both partners not cooperating is no longer an equilibrium. Thus, it is possible for one company to act first to change the payoff structure and cause the more desirable outcome (see Table 3).

Unilateral commitments can be thought of as analogous to posting a bond: Company A goes to its lawyer and says it will pay $1 million to charity if the alliance fails. By doing so, it convinces its partner, Company B, that it really does not want the alliance to fail. Convinced of Company A’s intentions, Company B also cooperates. The result is that Company A does not lose the $1 million and both firms derive the full benefits of cooperation. Thus, the commitment to sacrifice something in the event that a less desirable outcome emerges can be sufficient to ensure that this outcome does not emerge in the first place.

Sequential, Irreversible Commitments

A second kind of unilateral commitment is conceptually distinct from the one discussed above. Rather than acting to alter its payoffs, a firm irreversibly commits to one of the choices already available to it. Then the other company takes this into account in making its decision. Sequential commitments can take a variety of forms. As the following examples illustrate, one form is to commit resources unilaterally. For instance, in one alliance we studied, one partner gave the other the résumés of all fifty of its development engineers and let the partner choose ten engineers that it considered best suited to the alliance. The same firm also let its partner decide how to deploy the engineers and who would supervise their work. By making this commitment, this firm clearly demonstrated its intention to cooperate fully toward the success of the alliance, removing the partner’s doubt about sharing its technology. The partner responded with equal cooperation, and the alliance led to the development of a successful telecommunications switch that exceeded both companies’ initial expectations.

Sequential actions can also take the form of unilateral technology commitments. For instance, in a joint venture between a Japanese automobile manufacturer and an Indian firm, the Japanese firm made its car designs available well in advance of the Indian firm’s investments. Thus, the Japanese firm signaled to its partner that it had every intention of transferring auto manufacturing technologies to the Indian firm and was not simply taking advantage of the low labor costs in India. Impressed with its counterpart’s gestures, the Indian partner did everything it could to make the venture succeed, ensuring the joint venture’s runaway success.

To further explain sequential commitments in terms of game theory, let us reconsider the game in Table 2. Suppose Company A is able to make its choice first (rather than both companies moving simultaneously). Company A knows that if it chooses to cooperate, B will respond by cooperating; conversely, a choice of noncooperation will result in Company B choosing not to cooperate. But Company A prefers the payoff resulting when both cooperate, so it will choose to cooperate, and so will Company B.

If an alliance is structured so that choices are made sequentially rather than simultaneously, a better outcome can result. What does it mean in practice for moves to occur simultaneously or sequentially? Sequential moves imply that the first company’s choice is already known when the second makes its move. Then the second company simply chooses the best response to the first company’s move. The first company considers this fact when initially making its choice.

In contrast, in moving simultaneously, neither company can simply respond to its rival’s move because the latter’s actions are contingent on its own choices that have not yet been made. Thus simultaneous moves are an abstraction for saying that one does not know the other player’s choice, in the sense that one’s own choice will actively influence the other’s choice in real time. So an alliance that is structured so that one company moves first and is able to demonstrate its choice to the other company should be analyzed as a game with sequential rather than simultaneous moves.

Another form of sequential unilateral commitments is the promise of exclusivity. In our research, there were many instances in which one company in an alliance made an offer to work exclusively with the partner, even when it could have chosen others with comparable capabilities. For instance, some firms agreed to offer an exclusive license to their partners, when they could easily have kept open the option of licensing their technology to others. In another case, a firm committed to tendering all bids for local telecommunications contracts in collaboration with its foreign partner, even though that was not a stipulation in their agreement. In all these cases, the managers explained their actions as an attempt to demonstrate to the partner that they were fully committed to the alliance and hoped for a reciprocal commitment. We found no examples in which the partner took advantage of this unilateral gesture; in all cases, the partner responded by cooperating.

It appears as though the game theoretic framework suggested here is valid in a number of situations. In each of these examples, the firm making the unilateral commitment was always able to credibly convey the choice it had made. Thus the firm that made its engineers available to its partner first relieved them of their existing responsibilities. Similarly, the Japanese company’s release of automobile design plans was an irreversible move. However, such credible commitments may not always be feasible. For example, if by cooperating a company invests heavily in research and if by not cooperating a company does not invest in research, it may be hard to convincingly demonstrate to one’s partner that one has in fact chosen to cooperate.

There is something counterintuitive in one company allowing its rival to move first and thus influence the final outcome. Yet this results in a payoff that is better for both parties. Certainly, not all formulations will result in this outcome. The crucial feature is the coincidence of both companies’ interests; Company B is best off exactly when Company A is best off. To clarify this, consider an example in which there is a coordination problem similar to that in Table 2, but where the coincidence of interests does not arise (see Table 4).

Sequentially, if Company A moves first, cooperation by both will result, since Company A will choose the action giving the highest payoff. However, Company B would have preferred that neither cooperate, because it would get more in that situation. This example illustrates that structuring the alliance so that sequential moves provide the appropriate analytical measure need not result in a choice that both partners favor.

The prisoner’s dilemma formulation completely misses how important the distinction is between the sequential and the simultaneous dynamic. In the prisoner’s dilemma context, regardless of the nature of the moves, the only possible outcome is for both partners to not cooperate. One partner’s unilateral commitments are of no consequence. Again, this points to the restrictive nature of the prisoner’s dilemma formulation.

The Importance of Process

Managers must understand the economics of an alliance. Their recommendations for appropriate actions depend crucially on the nature of the alliance. An alliance in which the economics are better represented by one game versus another can lead to different avenues to ensure the participants’ cooperation. For example, unilateral commitments play a much greater role in the formulation shown in Table 2 and play no role in the prisoner’s dilemma formulation.

If we think of the two actions available to each party as “cooperate” and “don’t cooperate” (which are abstractions for different degrees of cooperation by the parties), then the first question that a manager should ask is: What does cooperation mean in the context of my particular alliance? The next issue to ponder is the size of the pie that the alliance hopes to get. The pie could exist for two broadly different reasons: the alliance could be pursuing a new market opportunity that has become feasible due to, say, new technologies developed in the alliance, or the alliance could be targeted at product or process improvements that result in delivering a higher quality per unit cost. It is important to ask if the alliance is of any value if neither party cooperates. Is there some benefit to be had simply by being exposed to the partner’s environment? These are indicative of the kinds of questions needed to get a sense of the payoffs in an alliance. Such questions also provide guidance in the event that there are environmental changes and the payoff structures need to be reassessed. Though the exact payoffs are important to properly value the alliance as one of several investment options, it is the structure of the payoffs that determines whether the alliance is viable in the first instance, and how it should be managed to get the most value out of it.

In addition to understanding the payoff structure of any alliance, managers must also remember that the payoffs may change over time due to a whole host of unexpected factors. For example, when the alliance between Digital Equipment Corporation and Tandy was first formed, Tandy was to be the sole source of low-cost PCs for DEC’s entry into this market, whereas DEC’s installed customer base would allow Tandy to increase its production volume and achieve greater scale economies. The alliance worked well while low-cost production was the dominant strategy in the industry. However, the benefits of the alliance changed when Compaq forced the market toward a performance, rather than price, orientation. Increasingly, DEC felt it had to offer higher-performance PCs to remain competitive in the industry, and the value of its alliance with Tandy suddenly diminished as DEC now needed a partner with more sophisticated design and engineering capabilities than Tandy had to offer.

DEC sought an alliance with Intel to develop its high-performance PCs, reducing the scope of its venture with Tandy. But, in the rapidly changing PC industry, DEC was shortly forced to reassess its situation because the success of Dell changed the competitive dynamics of the industry yet again. Price again became the competitive imperative, as Dell was able to match Compaq on performance and yet offer much lower prices due to its innovative use of a direct marketing and sales strategy. To respond to this new competitive challenge, DEC created its own direct sales and mail-order organization, terminated its alliance with Tandy, started manufacturing its own low-cost PCs in Taiwan, and even moved away from its alliance with Intel to develop its own high-performance machine based on a new alpha chip.

Moreover, after Robert Palmer replaced Kenneth Olsen as DEC’s CEO, there was a marked shift in DEC’s strategy in the PC market. Under Olsen, the PC business had always been the orphan child; he had repeatedly called PCs mere toys. Soon after his appointment, Palmer declared his intention to make DEC a major competitor in the PC market, bringing in outside leadership. This strategic change further altered the way DEC viewed the payoffs from its alliances.

While the DEC story illustrates how unforeseen events lead to reduced expectations of payoffs from alliances, there are times when events can increase the payoffs. In the case of the joint venture between the Japanese and Indian automobile firms described earlier, changes in the Indian government’s economic policies more favorable to foreign investors and international trade, coupled with a stronger Japanese yen, led to significant changes in the benefits that both partners thought they could derive from the alliance. Their reassessment led to changes in the scope of the venture; the Japanese partner increased its investment and is now planning to manufacture cars for export to the European market instead of only for domestic consumption in India. For the Japanese firm, the venture provides a low-cost platform, whereas, for the Indian partner, the venture is now a way to earn much-needed foreign exchange.

The general lesson of these examples is that a series of circumstances, technological and otherwise, can cause payoffs to be altered. Coupled with our earlier discussion of how different payoff situations need to be managed differently, this leads us to recommend that alliances be managed quite actively.7 It is extremely important for an alliance and its parents to adapt to changes in roles and relationships as circumstances demand. Even more so than typical organizations, an alliance is particularly likely to change significantly over time. Not only is it subject to changes in its own environment, which is usually highly risky and volatile, it is also subject to derivative changes that come from shifts in one or more of its parent firms. In a game theory context, this translates to the fact that the payoffs associated with the partnership may alter over time, thus altering the incentive structure for each participant.

Operating in such a dynamic context, alliances’ very survival may hinge on adapting rapidly to any internal or external changes. Cooperation of the parent firms in such a relationship may actually be contingent on their ability to modify the alliance to meet their changing needs, since an alliance that was once important can soon become obsolete. In such situations, the alliance must be flexible enough to adapt itself to changing demands. A McKinsey & Company study suggests that such flexibility is a significant element of many successful alliances.8

Many of the successful alliances we studied had contractual clauses whereby the relationship was assessed and renegotiated every few years. Such opportunities were not mere legal formalities but were used at distinct junctures to reformulate the relationship and further the understanding between the parent firms. CFM International, a fifty-fifty joint venture between General Electric and Snecma, the French jet engine manufacturer, is an example of a successfully open relationship that may be terminated once it has outlived its usefulness to its partners. The seventeen-year-old alliance has taken orders for 10,300 engines totaling $38 billion. Despite this obvious success, the alliance is neither permanent nor monogamous. Each partner exercises its right to enter into alliances with other competitors and reserves the right to exit from the alliance when its usefulness runs out.

Parent firms can ensure their alliances’ adaptability by conducting periodic reassessments with alliance managers. Frequent meetings between the parents’ top managers ascertain the alliance’s proper functioning, further mutual understanding between the partners, and realign and reorganize the alliance in response to changing parental needs.

An important element in managing the process dynamics of alliances is ensuring the open and free flow of information between the partners and the alliance. Many firms rely extensively on their assigned liaison managers or teams to manage the information interface between the firm and the alliance. In addition to gathering information, such boundary-spanning teams or individuals are also responsible for overseeing the inflow and outflow of resources, as well as managing any conflicts that may arise. These individuals must possess considerable diplomatic skills to keep the alliance together. DEC has designated “relationship managers” who are responsible for all aspects of the firm’s relationship with specific partners. Ford relies heavily on its top management to establish concrete communication channels with the partner’s top management. For instance, in Ford’s successful alliance with Mazda, the top managers of both companies meet with the alliance managers three to four times a year to discuss the general and specific developments in the relationship. Such individual connections can build close personal ties and trust across organizations and provide a useful hands-on assessment of the alliance.

In addition to informal or personal links among parent firms and their alliance, partnering firms can establish formal systems to provide useful performance information, such as detailed reports and frequent presentations by alliance managers to parent firm representatives. The information must be broader than simple performance reports and should encompass the alliance’s strategic decisions, current activities, and future concerns and agenda.

Information flows between the parent firms and the alliance provide the partners with an early warning system to indicate if the partnership is going askew. The alliance can provide partners with estimates of performance; simple accounting results are usually not appropriate, and the parent firms may need to rely on more creative measures. Unfortunately, our research confirmed that many firms use standardized measures of output to assess the performance of their alliances. In most cases, we found that at least one of the parent firms had standard accounting and measurement systems for evaluating the alliance. Furthermore, many parent firms made no distinctions between the alliance and their other wholly owned divisions. In such instances, the alliance managers expressed frustration at being unable to convince their parent firms of the alliance’s effectiveness. Not only were such standard measures misaligned with the alliance’s goals, they led parent firms to become so discouraged that they imposed stringent financial controls or, even worse, withdrew prematurely from the alliance.

There are many alternative measures of performance more appropriate for alliances. One solution that some firms in our research used was the subjective assessment of their partners, at least in the early stages of the venture. Another, less frequent, solution was to rely more on input measures such as morale, learning, and level of harmony, as opposed to simplistic output measures. In seeking appropriate measures for alliance performance, successful firms also took into account the evolving character of alliances and modified the measures accordingly. Some even used straightforward output measures once the alliance had matured into a stable business. On this issue, managers within many alliances have taken the initiative and developed appropriate measures that coincide with the parent company’s needs. The crucial point here is that, for the alliance to be successful, parent firms need a regular, comprehensible stream of information about its progress.

Conclusion

Each partner’s understanding of the alliance’s economics is crucial for understanding the incentives to cooperate and for realizing the possible ways each can unilaterally influence the alliance’s outcome. Indeed, one of our main points in this paper is that the payoff structure matters.

We have used two contrasting payoff structures to show their fundamentally different implications for how alliances should be managed. Our framework emphasizes how various unilateral commitments, observed in the field, played an important role in ensuring alliances’ success. The popular prisoner’s dilemma framework that is traditionally used to conceptualize alliances does not recognize a role for these commitments.

Further, we have drawn attention to the fact that the payoff structure in any alliance changes over time. Circumstances, technological or otherwise, can cause the payoff structure to be altered. Since different payoff situations have different managerial implications, we suggest that companies structure and manage alliances to recognize changes in the payoff structure and build in explicit mechanisms to acquire an appropriate managerial style.

Most important, we have provided a rich framework for thinking and talking about alliances. In the past decade, alliances have become a ubiquitous aspect of the corporate landscape. Yet many have failed to live up to the partnering firms’ expectations. While some of their disappointing results may be due to the partners’ unrealistic expectations, we think some of the failures may have been the result of the very restrictive models and frameworks used for thinking about alliances. By using different frameworks for securing cooperation between partners, managers will get more out of their alliances.