INSEAD, Fontainebleau, France
Over the last ten years, an impressive number of studies have examined entrepreneurial behaviour within large organizations. In light of recent principal agent models (Holmstrom and Milgrom, 1991, 1994), it can be argued that entrepreneurial behaviour in large organizations is to a some extent a paradox. When some tasks are easy to measure and some are hard to measure, the principal must necessarily offer dampened incentives, combined with bureaucratic controls on agent behaviour to compensate for the lack of clear performance measures. Holmstrom (1989) thus concludes that large organizations have a relative disadvantage in innovative activities compared to smaller ones.
Even though theoretical reasoning is not that encouraging for entrepreneurial organizations, arguments can still be made how organizations could become more entrepreneurial. A straightforward way is to follow the logic in Holmstrom and Milgrom (1991) and group tasks according to measurability, e.g. hard to measure innovative tasks are allocated to one group of people (R & D department). If this functional solution is not satisfactory, for example because co-operation across functions is important, then a way has to be found to deal with the difficulty of measuring differing tasks, and resulting incentives and bureaucratization problems.
We propose a framework based on recent work by Crémer (1995), Aghion and Tirole (1996), and Koskinen (1997). In this framework, employees put up effort and if they are successful propose new projects. The role of management is to approve or reject the projects, not to monitor employees' activities. In these models, management's project evaluations and strict selection criteria for projects can have adverse effects on employees' effort, and therefore on the number of projects proposed. Hence, in an environment where employees have more autonomy, such that new projects are encouraged and mistakes tolerated, entrepreneurial behaviour is easier to sustain than in a traditional hierarchical organization.
We reviewed case research in five entrepreneurial or transforming organizations, including 3M, DEC, Rhone Poulenc, MHZ and Siemens Nixdorf. In each company, multiple interviews were conducted, through many layers of management, in many companies, and across functional groups. The research originally looked at a broad range of factors contributing to entrepreneurial behaviour; in this paper we used a new and different focus on the data.
Our review confirms the value of the theoretical argument which links monitoring with effort, and adds a rich managerial perspective. Management engaged in limited direct monitoring of people's activities, preferring instead to emphasis the actual outcomes of their work. This approach applied to all kinds of work, both the routine tasks and the more 'entrepreneurial' and vaguely defined activities. The outcomes of employees' work were always measured in terms of value added in the market, that is, customer benefit and/or improvements to the companies' bottom line, which are relatively objective standards for measurement. The rewards for the positive performance outcomes were allocated to individuals, according to their contribution as judged subjectively by management. Typically, the rewards had a long-term and/or corporate nature, although the general emphasis was on non-pecuniary rewards. Rather than cash bonuses, the organizations used a mixture of recognition (which turned into promotions with long-term salary increases, improved informal networks, and better access to resources for future projects) and stock options (which depended on the whole companies' performance, not just the individual successful project).
Theoretically, the incentives systems used in the organizations have some key risks which could unbalance the whole system and lead to demotivated employees doing inferior work. Management could worry that employees direct their efforts inappropriately, because the actual work processes were not being monitored. Employees could be concerned that rewards were not properly allocated. On the one hand, they needed to trust management or the system to make accurate evaluations of individual contributions, and not be fooled by free-riding colleagues. On the other hand, they could worry that management would steal recognition for their winning project, despite prior commitments to reward successes.
Management and employees' worries seem to be handled the much same way in each organisation. First, evaluation of performance was conducted regularly, by peers as well as management, and the evaluations were public knowledge within the company. Likewise, all objective performance measurements (product by product, unit by unit) were transparent to everyone in the company. Thus, everyone was judged by more than one source, and the quality of any evaluation could be compared to other evaluations. Second, anti-social behaviour, as defined in these companies, was punished with bad reputation, and people with bad reputations found they had limited access to the informal network. As the organizational work processes relied heavily on the informal network to access crucial information and support, this punishment directly impacted future performance. A person's reputation in the informal network also directly affected their promotion opportunities. Thus, a combination of social reputation, combined with transparency of multiple performance measurements, ensured that management honoured commitments to appropriately reward high performers.