European Management Journal Vol. 15, No. 6, pp. 603-611, 1997
~)
Pergamon PII: 80263-2373(97)00044-3
© 1997 Elsevier Science Ltd All rights reserved. Printed in Great Britain 0263-2373/97 $17.00 + 0.00
What is the Optimum Amount of Organizational Slack? A Study of the Relationship between Slack and Innovation in 1 Multinational Firms NITIN NOHRIA, Harvard Business School RANJAY GULATI, Kellogg Graduate School of Management, Northwestern University
The relationship between organizational slack and innovation has remained an unanswered empirical question for decades and theorists continue to argue over the basic issue of whether slack facilitates or inhibits innovation. Opponents of slack claim that slack relaxes incentives to innovate and encourages wasteful investment in R&D activities, while its proponents counter that slack resources allow individuals and departments to experiment with projects that might lead to important innovations. In this article, Nitin Nohria and Ranjay Gulati attempt to reconcile the theoretical debate by postulating that slack is neither inherently destructive to an organization, nor is it a fail-safe cure. By discouraging any form of experimentation whose success is uncertain, too little slack inhibits innovation. Similarly, an abundance of slack inhibits innovation by fostering complacency and lax controls. These two extremes suggest the notion that an intermediate level of slack is optimal for innovation in any organizational setting. Multivariate analyses of survey data from 264 functional departments of two multinational corporations support the authors' proposition that both too much European ManagementJournalVo115 No 6 December 1997
and too little slack are detrimental for innovation. Thus, the authors argue that rather than focusing on whether slack has a uniformly positive or negative effect on innovation, theorists and managers should instead ask the question, 'What is the optimal amount of slack?' © 1997 Elsevier Science Ltd Innovation and slack are concepts that have received much attention from organization theorists in recent years. Amidst the rapid changes of the marketplace, innovation has proven to be an essential component of organizational adaptation and renewal; at least one expert has argued that organizations must innovate or die (Peters, 1990). At the same time, theorists have used organizational slack to understand a variety of organizational processes and outcomes, such as search behavior, goal conflict, political behavior, effectiveness, and innovation. Slack resources have been included as a variable in empirical investigations of numerous organizational phenomena (e.g., Singh, 1983; 1986; Bromiley, 1991). Although organizational theorists have been quick to accept slack as a rich and exciting construct, their 603
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONALFIRMS
conceptualization of slack has varied widely (see Bourgeois, 1981; Lant, I985 for comprehensive reviews). Some researchers define slack as an asset - excess resources that allow an organization to facilitate cooperation among competing coalitions and buffer itself from unforeseen circumstances (March, I976; 1981). Others view slack negatively, as an inefficiency resulting from an organization's failure to use resources optimally that results in performance that does not meet the full potential of the available resources (Leibenstein, 1969; Williamson, I963; 1964). While proponents of slack argue that it is essential for innovation, organizations confronting growing global competition feel pressured to eliminate all forms of slack. This paradox demonstrates the dynamic and previously undefined relationship between innovation and slack: if slack is both a form of inefficiency and also an essential ingredient of innovation, organizations that cut slack from their departments may find that their capacity to innovate is hindered. Attempting to avoid positive or negative value judgments, we define slack as the pool of resources in an organization that is in excess of the minimum necessary to produce a given level of organizational output. These slack resources include excess inputs such as surplus employees, idle capacity, and capital expenditures. They also include overlooked or unexploited opportunities to increase outputs, such the margins and revenues to be obtained from customers and innovations with links to the technology frontier. We echo Sharfman et al. (1988) and focus on slack resources that are recoverable, or employable in the future, because such resources should be more readily available to support innovative activity than absorbed, or long-term, slack (Bourgeois and Singh, 1983; Sharfrnan et al., 1988). We consider short-term slack to include resources of any kind that can be recovered and employed to influence performance over a typical temporal cycle relevant to managers' activities. Specifically, we limit short-term slack to excess resources that can be recovered within a year, the usual length of budgets and employee performance reviews. Our definition of slack applies across organizational levels, capturing the extent to which any organizational unit (be it an individual, a department, a function, a division, or a firm as a whole) can quickly deploy excess resources to meet internal or external contingencies.
The Case for Slack Our definition leads to the question, What are some benefits of slack? Cyert and March (1963) provided the seminal answer. They started with the premise that all organizations can be considered political coalitions consisting of subgroups with competing goals. Though these goals overlap to some extent, they are never perfectly aligned. They argued that slack plays a crucial and vital role in resolving this latent goal conflict and preventing the organization from breaking apart (Cyert and March, 1963). Other theorists have built on this idea 604
and argued that: slack prevents unhealthy conflict (cf. Bourgeois and Singh, 1983); it creates buffers that reduce information-processing and coordination costs across subunits (cf. Thompson, 1967; Galbraith, 1973); it allows for intertemporal adjustments to demand or supply fluctuations in various factor and product markets (e.g. use of excess labor to offset changes in labor markets); it creates an unexploited pool of refinements with which to meet changes in competitive intensity (cf. Levinthal and March, 1981); and it represents capital and cash reserves with which to meet changing pressures from capital markets. Researchers have also seen slack facilitating another kind of adaptive response: innovation. Slack causes relaxation of controls and represents a source of funds whose use may be approved even in the face of uncertainty. Slack allows innovative projects to be pursued because it buffers organizations from the uncertain success of those projects, fostering a culture of experimentation (Bourgeois, 1981). Slack resources enable firms to more safely experiment with new strategies such as introducing new products and entering emerging markets (Hambrick and Snow, 1977). In lean times, such projects might otherwise be killed. Furthermore, slack facilitates the pursuit of risky projects that sometimes yield positive results that can be of unexpected benefit to a firm (Nelson and Winter, 1982); innovation research is replete with stories of chance discoveries that resulted from slack search. A prime example is the much lauded invention of Post-it notes at 3M, an organization that encourages its employees to use slack time to pursue innovative ideas. As a result of the arguments in support of slack, most empirical studies on the organizational determinants of innovation include slack as a variable, and in some cases it has been found to have a positive effect (cf. Majumdar and Venkataraman, 1993; Zajac et al., 199I).
The Case against Slack Other theorists, particularly organizational economists such as Leibenstein (1969) and WiUiamson (1963, 1964), have adopted a more negative opinion of slack: they consider it to be synonymous with waste, and a sign of managerial self-interest, incompetence, and sloth. Like slack proponents Cyert and March, slack opponents begin by characterizing firms as coalitions of competing interests. However, they contend that these competing interests should be viewed as a system of nested principal-agent relationships (Jensen and Meckling, 1976). A firm's top management can be seen as a group of agents acting on behalf of its shareholders, or principals; divisional managers are agents acting on behalf of the top management, and so on down the line of management hierarchy. In this model, conflict arises because agents do not always have the incentive to act in the best interests of the principals. Accordingly, top managers don't always European ManagementJournalVo115 No 6 December 1997
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONALFIRMS
act to maximize shareholder value. Similarly, division managers don't always act to maximize corporate performance. This suboptimization arises because principals must delegate decision rights to the agents so that they can act, but the principals typically don't have all the information necessary to monitor their agents' performance accurately. Agents can take advantage of this 'information asymmetry' and act in their own interests. Indeed, Williamson (1963, 1964) argued that, left to their own devices, agents are primarily motivated to build empires for themselves. While advocates of slack may consider a loss of efficiency to be a necessary sacrifice for its additive benefits, its detractors find no redeeming value in slack. Indeed, a number of economists have argued that innovation, and hence adaptation, may actually be hurt by excess slack. Leibenstein (1969) coined the felicitous term X-inefficiency to highlight the discrepancy that slack creates between actual output and maximum output for a given set of inputs. Jensen (1986) argued that firms that have a high amount of slack often invest it in dubious projects, such as 'pet' R&D projects or unrelated acquisitions. Drawing on evidence from the petroleum industry in the late 1970s and early 1980s, he showed that most companies only gained about 60 cents for every dollar invested in exploration and development. In an even more graphic illustration of the problems of slack, he showed that if General Motors, Kodak, and IBM had simply invested the money they spent on R&D and capital expenditures in the 1980s in risk-free government securities, the value of the investment would have been enough for each of them to have bought out its principal competitor (Jensen, 1993).
An Argument for Optimal Slack Both the advocates and opponents of slack agree that slack promotes experimentation and the pursuit of new projects. Before innovation can occur, organizations must accept the uncertainty associated with innovative projects (Mansfield, 1963). This intrinsic uncertainty makes it difficult to gauge ex ante the net present value (NPV) of such projects. Not only can persistence and 'patient money' foster innovation, but they also provide the flexibility necessary to alter resource allocation levels as projects progress over time. As a pool of resources, slack eases adaptation to the ebbs and flows of the innovation process (Grossman and Shapiro, 1987), and frees managerial attention (Cyert and March, 1963). In tight organizations with little slack, managerial attention is likely to be consumed by short-term performance issues rather than by long-term, innovative projects. These findings support the assertion that the number of new initiatives undertaken by an organization will increase as slack increases. The relationship may not be linear over the entire range of slack; as the possibilities for innovation diminish, we expect diminishing returns from experimentation even as slack increases. The positive relationship between slack and experimentation European Management JournalVo115 No 6 December 1997
thus determines, in part, the relationship between slack and innovation, as shown in Figure 1. A related but opposing dynamic is the diminishing control placed on increased experimentation as slack increases. As slack increases, the amount of discipline that is exercised in the selection, ongoing support, and termination of projects becomes lax (e.g. Leibenstein, 1969; Jensen, 1993). In worst case scenarios, projects with high risk and negative NPV may be funded simply because agents realize they can afford to indulge in their pet projects. Once a bad project has been initiated, continual or 'escalating' commitment to this project will occur because the continued existence of slack makes it difficult to justify termination of someone's pet project (Staw et al., I981). In times of high slack levels, negotiations are more relaxed, and managers tend to be less insistent that projects meet their forecasted milestones (Cyert and March, 1963). Lax discipline not only increases the risk that poor projects will be blindly pursued in the face of negative information; it also adds to the risk that promising projects will be abandoned simply because someone ran out of energy, got bored, or ran into a tough problem. Thus, an excess of slack can lead to both Type I errors (selecting projects that should not have been funded) and Type II errors (stopping projects that should have been continued). Although too little discipline can cause inefficiencies, too much discipline can also be problematic because it can curtail legitimate innovative projects. To some extent, innovative projects are always uncertain, and positive NPV projects cannot always be easily distinguished from negative NPV projects. One thus has to be wary of the damaging effect of overly stringent discipline that may result from too little slack. Tight controls and very low levels of slack can give organizational members a diminished sense of self-efficacy and reduce their
~Discipl
imentation
/ Organizational Slack ]
Figure I The Predicted Relationship b e t w e e n Slack and Innovation 60.5
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONAL FIRMS
expectation that they can successfully propose and pursue innovative projects (Bandura, 1986). In the extreme, low levels of slack may engender a sense of futility regarding the prospect of pursuing any innovative project at all (Gecas, 1989). Rather than weigh in on one side of the debate for or against the innovation-enhancing benefits of slack, we propose a reconciliation of these perspectives. In sum, we expect the relationship between slack and discipline to be negative and the consequent relationship between discipline and innovation to be curvilinear, generally positive but falling off beyond a certain level. This prediction suggests that there is an intermediate level of slack in any given organizational setting that is optimal for innovation. Thus: Proposition I: The relationship between organizational slack and innovation is inverse U-shaped.
Two additional variables shown in Figure 2 mediate the stated relationship, one directly influencing degree of experimentation, and the other affecting degree of internal control. The first is the environmental context in which an organization is embedded. There is a substantial literature on the environmental conditions that stimulate innovation (see Kamien and Schwartz, 1982; Adler, 1989 for reviews). Implicit in most of these accounts is the notion that the degree of competition faced by an organization influences its degree of organizational experimentation. Since Schumpeter's (I926) classic treatise, competition has been seen as a vital spur to experimentation and thus innovation, and there is considerable empirical support for this positive relationship between competitive intensity and innovation (Majumdar and Venkataraman, 1993; Zajac et al., 1991). Another environmental factor that is known to stimulate E n v i r o n m e n t a l Context
Degree of Internal Control ]
Notes: + indicates positive effects -- indicates negative effects ~ indicates curvilinear effects [----I indicates observed variables C ) indicates u n o b s e r v e d variables
Figure 2 A M o d e l of t h e R e l a t i o n s h i p b e t w e e n Slack and Innovation
606
experimentation and innovation is the technological dynamism of the environment in which a firm is embedded (Lawrence and Lorsch, 1968; Porter, I990). Organizations embedded in dynamic technological environments recognize the importance of innovation for their success and are hence more likely to invest in innovative experiments. Given that economic action is embedded in social and institutional contexts, organizations in technologically dynamic environments are also more likely to innovate because they are in the midst of active networks of information and people flows and will experience inducements to mimic the behavior of other organizations in their institutional field (Granovetter, I985; Powell and DiMaggio, 1991). As a result: Proposition 2: The degree of competition and technological dynamism in the environment in which an organization is embedded will have a positive effect on its innovativeness,
In addition to the degree of slack present, an organization's internal control is the second factor that affects discipline in innovation. The importance of strong control systems that ensure that R&D as well as other capital expenditures lead to real value-added innovations was most vividly demonstrated by Jensen (1993), who provided compelling evidence that the internal control systems of most large organizations routinely fail to adequately control their resource allocation process. An effective internal control system optimizes the distribution of 'decision rights' in an organization and aligns the interests of individuals and the organization (Jensen and Meckling, 1991; Eisenhardt, 1985). The optimal distribution of decision rights takes into account an informational trade-off inherent in most organizations. It is important, on the one hand, for allocators to recognize that the individuals closest to the action, such as sub-unit managers, must have the discretion to employ their unique and valuable specific knowledge in making important decisions. It is equally important, on the other hand, to recognize that these individuals do not have a broad enough vision of the organization to make the trade-offs necessary for optimal resource allocation. A tight intemal control system can thus increase the amount of discipline exercised and the general knowledge applied to the selection of new projects. But if the controls are too tight and employees are given too little discretion over such decisions, the organization loses the benefits of their specific ability to introduce innovations. Thus: Proposition 3: The strength of an organization's internal control system will have a positive, but curvilinear effect on its innovativeness.
Research Method To test our propositions, we compare data obtained via a self-report questionnaire mailed to department managers at the national subsidiaries of two major multinational European ManagementJournaIVo115 No 6 December 1997
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONALFIRMS
corporations (MNCs), one European and the other
process, product or market opportunity that is perceived to be new by the manager of the innovating unit. We organization of MNCs, these data are described in detail deliberately adopted a broad definition of innovation elsewhere (Bartlett and Ghoshal, 1989). While both firms because our goal was to capture the extent to which are among the largest and most diversified MNCs in the each department was responsible for generating any world, we focused our study on the consumer electronics form of new knowledge that could benefit the MNC. We business in which the two companies competed directly considered the tangible economic benefits of innovations worldwide. In this particular business, the firms were by asking respondents to describe and estimate the total comparable in terms of size, geographic scope, and economic impact (the yearly savings and or additional competitive position. Each subsidiary had approximately revenues generated in millions of dollars) of their three the same number of departments, including most significant innovations made in the last year. This manufacturing, marketing, R&D, finance, and other allowed us the flexibility to compare the innovativeness administrative functions. Given that one of the major of a finance department that created a new hedging sources of variat!on in the instrument to protect its firm from characteristics of the There is an intermediate exchange rate fluctuations with the departments in our sample innovativeness of an R&D level of slack in a given was the subsidiary to which department that created a process they belonged, we chose a that improved organizational setting that innovation sample representative of the manufacturing yields by 3 per cent is optimal for full range of each MNC's (both real examples from our data). subsidiaries in consultation innovation with three corporate Additionally, we asked each managers from each firm responsible for the global respondent to report also the number of distinct consumer electronics business. In addition, these innovations accomplished within his or her managers completed a survey designed to ensure that department in the previous year. The robustness of our sample included departments located in small and our claims could thus be tested using more large subsidiaries, in advanced and developing nations, conventional measures of innovation, by estimating and in challenging and placid environments. each of our models against a dependent variable that measures the total number of innovations reported by each department in the previous year. Targeted departments represented functional subunits that had budgets specifying both their performance objectives and available resources. From a theoretical standpoint, the managers in charge of the departments Measuring Slack could be considered agents who had to satisfy the performance demands of the senior management Supported by case research by Schiff and Lewin (1970) responsible for the overall performance of the subsidiary that shows that managers routinely create slack by and firm and thus acted as principals. The setting was overestimating costs and underestimating revenues, we ripe with classic competing coalition and agency measured the degree of slack within each department problems (Nohria and Ghoshal, 1994). Department by asking the departmental managers the following managers' ability to garner slack resources was expected two questions: (1) 'Assume that due to some sudden to vary depending on their past performance, the development, 10 per cent of the time of all people negotiating leverage they enjoy, the primary functional working in your department has to be spent on work task of their subunit, and the critical contingencies it totally unconnected with the tasks and responsibilities controls (Pfeffer and Salancik, 1978). Thus, considerable of your department. How seriously will your output be variance in the amount of slack was likely across affected over the next year?' (2) 'Assume that due to a departments. similar development, your department's annual operating budget is reduced by I0 per cent. How We received the responses of 178 departmental significantly will your work be affected over the next managers from 14 national subsidiaries in the Japanese year?' The subunits whose managers anticipated that firm and 78 departmental managers from eight national output would be unaffected even with a 10 per cent subsidiaries in the European firm to be complete and increase in responsibilities or a 10 per cent decline in usable. The response rate was 87 per cent in the budget were considered to have higher levels of slack, European firm and 93 per cent in the Japanese firm. In no and those whose managers anticipated a decline in subsidiary was there a response rate of less than 83 per output of 20 per cent or more (disproportionate to the cent. suggested change) were judged to have low levels of slack. Using these transformations, we created a slack measure corresponding to each question with values Measuring Innovation ranging from 0 to 30. Japanese. Collected as part of a larger study on the
We developed a very broad definition of innovative accomplishments: any policy, structure, method or European Management Journal Vol 15 No 6 December 1997
We reasoned that departments whose output would be significantly affected by the speculative sudden 60 7
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONALFIRMS
changes would have less slack than those that would be less affected. Because the two measures were highly correlated, we added the two responses, constructing a composite measure of slack with values from 0 to 60.
Other Measures Three senior headquarters managers responsible for the overall global consumer electronics business in each MNC completed a survey that included measures of the degree of competition and technological dynamism confronted by each of the national subsidiaries in the sample. We found a high degree of convergence across the three respondents in both firms. Moreover, the above measures of competition and technological intensity were highly correlated. Because of these results, in our final analysis we included a composite measure of environmental forces for innovation, environment, which comprised the sum of the average measures of the degree of competition and technological dynamism reported by our respondents for each subsidiary. Because environment does not vary across the departments in a subsidiary, each department in a given subsidiary received the same overall subsidiary score. We included a number of measures that indicated the extent of control placed over a department's decisions. The first set of measures, developed by De Bodinat (1975), indicated the degree to which key decision areas were centralized. Respondents within each functional unit were asked to estimate the influence they enjoyed in making five types of decisions: (1) the modification of an existing product, (2) the modification of a production process, (3) the restructuring of the subsidiary organization involving the creation or abolition of departments, (4) the recruitment and promotion of managers to positions below that of subsidiary general manager, and (5) the career development plans for department managers. An additive scale of these five indicators was used as a measure called centralization. A second set of measures assessed the degree to which key areas of decision making were subject to formal controls. Each respondent was asked to indicate the extent to which the following five conditions applied: (1) for most tasks there are well developed rules and policies, (2) their decisions are closely monitored to ensure that rules and policies are followed, (3) for most situations, there are manuals that define the course of action to be taken, (4) for most jobs, there are written job descriptions, and (5) everyone has a well-defined and specific job to do. These questions followed the measures proposed by Pugh et al. (1968) to assess 'the degree of employee behavior that is defined by specialist jobs, routines, procedures, and formal written records'. Accordingly, formalization, an additive scale of all five indicators, was used to measure the organization's degree of formal control over decision making. 608
We also induded a number of control variables known or expected to affect innovation. The dummy variable 'company' for company-level effects; "subsidiary' measured the controlled relative resource levels of the subsidiary to which the department belonged; and three dummy variables, 'R&D', 'manufacturing', and 'marketing', identified functional areas. Finally, we constructed a measure of individual social capital, which included variables such as experience and length of employment. The degree of innovation was modeled using a standard ordinary-least-squares model available in the statistical package SAS. Since the dependent variable is continuous (the dollar value of innovations) and the data are crosssectional, such a model appeared to be adequate. Positive coefficients of variables indicate a positive contribution to the innovativeness of a subunit, and negative coefficients indicate otherwise. The number of innovations was modeled using an ordinal logit model available in SAS. The results obtained were later checked against those from Poisson and negative binomial regression models (Ramaswamy et al., 1994). We observed no differences in the directionality or significance of the results with these two specifications of the dependent variable.
Empirical Results The most important results of the multivariate analysis are those concerning slack. These strong results are consistent with the predicted non-monotonic effects of slack on innovation and are statistically significant. As hypothesized, slack has a significant, positive first-order effect and a significant, negative second-order effect on innovation. When the effect of a subtmit's environmental context and the degree of control exercised over it are isolated, slack effects remain strong. Further evidence of
45-
0 ~l 0
3
6
9
!11!1 III
12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60
Slack
Figure 3 T h e O b s e r v e d Relationship b e t w e e n S l a c k and i n n o v a t i o n
EuropeanManagementJournalVo115N o
6 December
1997
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONAL FIRMS
the role of slack in explaining innovation was gathered
by comparing the variance explained by the models. Introducing slack and its squared term lead to a statistically significant increase in the R-squared term, suggesting a better-specified model.
Summary and Conclusions Our results provide strong support for the inverse Ushaped relationship we hypothesized between slack and innovation. There was little evidence, however, that either the degree of competition and technological dynamism in an environment or the degree of control exercised over an organization influenced innovation.
Figure 3 plots the relationship between slack and innovation using the estimates in our multivariate analysis. In accordance with our earlier postulates, we observe here that this relationship has the shape of an Our arguments and results help resolve the debate inverted U within the between those who argue that observed range of innovation slack encourages innovation and Although many and slack. The point of those who suggest that slack may inflection at which innovation in fact inhibit innovation. The organizations could afford starts to diminish with middle ground we advocate to reduce a certain amount that slack has an inverse Uincreasing slack occurs at a slack score ranging from 32 to shaped effect on innovation of slack, they should be provides resolution to this 34 (on a scale of 0 to 60). This pattern also suggests that aware that drastic cuts may intractable debate. We propose two underlying mechanisms to the degree of reported sacrifice the very innovation explain this relationship: (1) the innovativeness at this point is effect of slack on the process of four times larger than it is at they are attempting to experimentation, and (2) the slack levels of 0. When slack promote effect of slack on the discipline equals the maximum possible exercised over experiments. in our survey, the level of Because it discourages any kind of experimentation innovation is once again a fraction of the maximum whose success is uncertain, too little slack is inimical to reached at intermediate levels. In other words, as slack innovation. Similarly, too much slack is also inimical to rises from the intermediate range to the high range, innovation because it fosters complacency and a laxness innovativeness drops dramatically. In fact, that lead to more bad projects being pursued than good innovativeness at the observed maximum slack is projects. about the same as when slack is at the minimum. The environmental context of a subsidiary had no influence on the innovative capacity of the functional units within it, suggesting no support for Proposition 2. No support emerged for Proposition 3 either. The directionality of the coefficients for internal controls was as expected but remained insignificant. To test the robustness of our claims concerning the effect of slack on innovation, we estimated unrestricted models for each company (these results are not reported here for the sake of brevity). The signs of the coefficients indicated that the postulated relationship between slack and innovation observed in the pooled sample hold true in both companies. We tried this procedure with dummy variables for each subsidiary and found no differences in the results. We also estimated the models separately for each functional area and again found consistent results. Lastly, we compared the reported results against those obtained using an alternative measure of innovation: the number of innovations accomplished by a sub-unit in the year prior to the survey. We found that the main results for slack remain the same; slack continues to show the predicted nonmonotonic effects, even when the other expected predictors of innovation are controlled. Thus, we can say with some confidence that for this sample, the inverse U-shaped relationship between innovation and slack is very robust. European ManagementJournalVo115 No 6 December 1997
In sum, we maintain that it is appropriate to think about the relationship between slack and innovation as inverse U-shaped. Thus, rather than focus on whether slack is uniformly good or bad for innovation, we should ask the question, 'What is the optimal amount of slack?' In our study, the optimum slack is a self-reported estimate for a five per cent reduction in yearly output if inputs such as budgets or time available are reduced by 10 per cent. This suggests that the optimal level of short-term recoverable slack is about five per cent of a unit's budget. It is interesting to compare our findings with reports of the amount of slack in highly innovative firms such as 3M, where employees are ostensibly allowed to devote as much as 20 per cent of their time to slack search, or the pursuit of innovation. Clearly, the optimal level of slack in our study is more conservative than that reported for 3M. These differences suggest that the optimal amount of organizational slack may depend on a series of situations that we have not explicitly explored in this paper. For instance, one might argue that an optimal amount of organizational slack is more likely to be found in a growing industry than in a declining industry because there will probably be more positive-NPV projects in the former than in the latter (Jensen, 1993). Similarly, there may be firm-level factors that determine the optimal amount of slack, such as a firm's culture and the nature of its internal control systems. One might, for instance, 609
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONAL FIRMS
expect the optimal level of slack to be greater in firms with strong cultures or strong internal control systems than in other firms because the additional slack is unlikely to be abused or frittered away. Subunit and individual level factors may also influence optimal slack. For instance, subunits that are responsible for critical contingencies may need more slack than others because they need a larger reservoir of slack resources to respond to these contingencies (Pfeffer, I992). Similarly, strong leaders may be able to manage slack resources better and can be trusted with more slack than weaker leaders. Relevant to the question of how much slack is optimal in a given situation are questions regarding the antecedents of slack (Sharfman et al., I988) and how the amount of slack in an organization can be changed. Although we are aware that good performance increases slack and bad performance decreases it, we continue to examine what managers can do proactively to change amounts of slack. In recent years, organizations have used initiatives such as LBOs, MBOs, leveraged recapitalizations, and business process reengineering to try to reduce slack. But there is not enough work on how internal control systems and incentive structures can be modified to reduce slack. Such research is important because, even though we have argued that some amount of slack may be necessary for an organization to be adaptive and innovative, slack is an economic liability and thus, efforts should always be made to reduce its optimal amount. Recognizing the inverse U-shaped effect of slack on innovation is not only of theoretical importance, but is also of great practical significance. As firms are increasingly forced to juggle simultaneous demands to be innovative and efficient (Bartlett and Ghoshal, I989; Tracey and Wiersema, 1994), it can be a challenge to maintain the degree of slack that is necessary to stimulate innovation. Firms that participated in costcutting programs of the 1980s, such as lean production, downsizing, and business process reengineering, are now feeling the negative effects of failing to invest in the future (Hamel and Prahalad, 1994). To many managers, slack represents nothing but wastefulness. We hope our results provide further warning against such a limited perspective. Although many organizations could afford to reduce a certain amount of slack, they should be aware that drastic cuts may sacrifice the very innovation they are attempting to promote.
Note 1.
This article builds upon a previously published article by the authors in the Academy of Management Journal (1996), 39: 5, 1245-1264, 'Is Slack Good or Bad for Innovation?' We thank the following for helpful comments and suggestions; Chris Bartlett and Sumantra Ghoshal for providing us with the data used in this study.
References Adler, P. (I989) Technology strategy: a guide to the literature. Research in Technological Innovation, Management and Policy 4, 25-151.
610
Bandura, A. (1986) Social Foundations of Thought and Action: A Social Cognitive Theory. Prentice-Hall, Englewood Cliffs, NJ. Bartlett, C.A. and Ghoshal, S. (1989) Managing Across Borders: The Transnational Solution. Harvard Business School Press, Boston, MA. Bourgeois, L.J. (198I) On the measurement of organizational slack. Academy of Management Review 6, 29-39. Bourgeois, LJ. and Singh, J.V. (1983) Organizational slack and political behavior within top management teams. Academy of Management Proceedings 43-47. Bromiley, P. (I99I) Testing a causal model of corporate risktaking and performance. Academy of Management Journal 34(1), 37-59. Cyert, R.M. and March, J.G. (1963) A Behavioral Theory of the Firm. Prentice-Hall, Englewood Cliffs, NJ. De Bodinat, H. (1975) Influence in the multinational corporation: The case of manufacturing. Unpublished DBA dissertation. Harvard Business School, Boston. Eisenhardt, K.M. (1985) Control: organizational and economic approaches. Management Science 31(2), 134-149. Galbraith, J.R. (I973) Designing Complex Organizations. AddisonWesley, Reading, MA. Gecas, V. (1989) The social psychology of self-efficacy.Annual Review of Sociology 15, 291-316. Granovetter, M. (1985) Economic action and social structure: a theory of embeddedness. American Journal of Sociology 91(3), 481-510. Grossman, G. and Shapiro, C. (1987) Dynamic R&D competition. The Economic]ournal, 97(June): 372-387. Hambrick, D., and Snow, C. (I977) A contextual model of strategic decision making in organizations. In Academy of Management Proceedingseds. R.L.Taylor, J.J. O'Connell, R.A. Zawacki, and D.D. Warrick. pp. I09-112. Hamel, G. and Prahalad, C.K. (I994) Competing for the Future. Harvard Business School Press, Boston, MA. Jensen, M.C. (1986) Agency costs of flee cash flow, corporate finance, and takeovers. American Economic Review 76(2), 323-329. Jensen, M.C. (1993) The modem industrial revolution, exit, and the failure of internal control systems. The Journal of Finance 48(3), 831-880. Jensen, M.C. and Me&ling, W.H. (I976) Theory of the firm: managerial behavior, agency cost, and ownership structure. Journal of Financial Economics 3, 305-360. Jensen, M and Me&ling, W.H. (199I) Specific and general knowledge, and organizational structure. Paper presented at Nobel symposium no. 77 on Contracts: Determinants, Properties, and Implications, Stockholm, August I8-20, 1990. Kamien, M. and Schwartz, N. (I982) Market Structure and Innovation. Cambridge University Press, Cambridge, UK. Lant, T. (1985) Modeling Organizational Slack: An Empirical Investigation. Stanford University Research Paper #856. Lawrence, P.R. and Lorsch, J.W. (1968) Organization and Environment. Harvard Business School, Division of Research, Boston. Leibenstein, H. (1969) Organizational or frictional equilibria, Xefficiency, and the rate of innovation. Quarterly Journal of Economics 600-623. Levinthal, D. and March, I.G. (1981) A model of adaptive organizational search. Journal of Economic Behavior and Organization 2, 307-333. Majumdar, S.K. and Venkataraman, S. (1993) New technology adoption in US telecommunications: The role of competitive pressures and firm-level inducements. Research Policy 22, 521-536. Mansfield, E. (1963) Size of firm, market structure, and innovation. Journal of Political Economy 41, 556-576. March, J.G. (1976) The technology of foolishness. In Ambiguity and Choice in Organizations eds. G. March and P. Olsen, pp. 69-81. Universitetsforlaget, Bergen. March, J.G. (1981) Footnotes to organizational change. Administrative Science Quarterly 26, 563-577. EuropeanManagementJournalVo115No 6 December 1997
A STUDY OF THE RELATIONSHIP BETWEEN SLACK AND INNOVATION IN MULTINATIONAL FIRMS
Nelson, R., and Winter, S. (I982) An Evolutionary Theory of Economic Change. Harvard University Press, Cambridge, MA. Nohria, N. and Ghoshal, S. (I994) Differentiated fit and shared values: alternatives for managing headquarters-subsidiary relations. Strategic ManagementJournal 15(6), 491-502. Peters, T. (1990) Get innovative or get dead, California Management Review 33(2), 9-26. Pfeffer, J. (I992) Managing with PowerHarvard Business School Press, Boston, MA. Pfeffer, J. and Salancik, G. (1978) The External Control of Organizations. Harper and Row, New York. Porter, M. (1990) The CompetitiveAdvantage of Nations. Free Press, New York. Powell, W. and DiMaggio, P. (eds.) (I99I) The New lnstitutionalism in Organizational Analysis. University of Chicago Press, Chicago. Pugh, D.S., Hickson, D.J., Hinings, C.R. and Turner, C. (1968) Dimensions of organization structure. Administrative Science Quarterly 13, 65-105. Ramaswamy, V., Anderson, E. and DeSarbo, W. (1994) A disaggregate negative binomial regression procedure for count data analysis. Management Science40, 405-417. Schiff, M. and Lewin, A.Y. (1970) BehavioralAspects in Accounting Prentice-Hall, Englewood Cliffs. Schumpeter, J. (I926) The Theory of Economic Development. Harvard University Press. Cambridge, MA.
Sharfrnan, M.P., Wolf, G., Chase, R.B. and Tansik, D.A. (1988) Antecedents of organizational slack. Academy of Management Review 13(4), 601--614. Singh, J.V. (1983) Performance, slack and risk-taking in organizational decision making: test of a structural equation model. Unpublished doctoral dissertation, Graduate School of Business, Stanford University, Palo Alto, CA. Singh, J.V. (I986) Performance, slack and risk-taking in organizational decision making. Academy of Management Journal 29(3), 562-585. Staw, B.M., Sandelands, L.E., and Dutton, J.E. (1981) Threatrigidity effects in organizational behavior: a multi-level analysis. Administrative ScienceQuarterly 26, 501-524. Thompson, J.D. (1967) Organizations in Action. McGraw-Hill, New York. Tracey, M. and Wiersema, F. (1994) The Discipline of Market Leaders. Harper & Row, New York. Williamson, O.E. (I963) A model of rational managerial behavior. In A Behavioral Theory of the Firm, eds. R.M. Cyert and G. March, Prentice-Hall, Englewood Cliffs, NJ. Wi[liamson, O.E. (1964) The Economicsof Discretionary Behavior: Managerial Objectivesin a Theory of the Firm. Prentice-Hall, Englewood Cliffs, NJ. Zajac, E.J., Golden, B.R., and Shortell, S.M. (I99I) New organizational forms for enhancing innovation: The case of internal corporate joint ventures. Management Science 37(2), 170-184.
NITIN NOHRIA, Harvard Business School, Soldiers Field Road, Boston, Massachusetts 02 I63, USA.
Nitin Nohria is an Associate Professor of Business Administration at the Harvard Business School, and in I997 a Visiting Professor at London Business School. He holds a Ph.D. in management from the Sloan School of Management, MIT. Currently, he is investigating the dynamics of organizational change through a series of projects that include studies of corporate downsizing, the spread of strategic alliances, and the impact of total quality management and reengineering programs
European Management Journal Vol 15 No 6 December 1997
R A N J A Y G U L A T I , J.L.
Kellogg Graduate School of Management, Northwestern University, 2001 Sheridan Road, Evanston, Illinois 60208-200I, USA. ~
Ranjay Gulati is an Associate Professor of Organization Behavior at the Kellogg School, Northwestern University. He has researched and published in the fields of interfirm strategic alliances, managing innovation, and the formation of supply chain partnerships. He received his Ph.D. from Harvard University, and is currently an editorial board member of the Administrative Science Quarterly, and the Academy of Management Review.
6 11