Reinvention Labs

Reinvention Laboratories were created in the United States as part of the National Performance Review and represented a novel approach to strategic change in the federal government. In assigning an important role to middle management in the design and implementation of new work technologies, the lab program contrasted with more conventional approaches to top-down change in which management was dominant.

This study examines the viability of the generative model that the lab program represents as a means of inducing agencies to employ post bureaucratic organizing principles. A conclusion is that although a variety of factors will affect the degree of success achieved, the model holds promise as a means of achieving strategic change in public organizations with a bottom-up approach to change

A distinguishing feature of the National Performance Review (NPR) is the degree of attention directed to what Arnold (1998) calls the "micro-context of reform," referring to changes to work processes and relationships within agencies. Consistent with a central theme of modern management theory, the sponsors of NPR endorse a shift away from organizing practices associated with the bureaucratic-scientific management model whereby workers are assigned highly specialized tasks, extensive written rules govern work behaviors, and hierarchical reporting and communication prevail. The new post bureaucratic (Barzelay, 1992) model presumes greater autonomy for frontline workers and is accompanied by a structure that facilitates the use of that autonomy to encourage high quality service, self- management, and innovation.

Extensive theoretical justification has been provided as to the suitability of this new model for modern conditions. A central problem for organizations seeking to adopt this model, however, is that of transitioning from the new to the old, of getting "from here to there" (Kanter, Stein, & Jick, 1992). The issue is presented in the management literature as one of strategic change (Tichy, 1983). Insights into the problem of strategic change can be gained from a variety of theoretical perspectives. A key issue relates to the relative roles of top and lower level management in the design and implementation of alternative structures and work processes. Within the innovation literature, for example, Marcus (1988) contrasts rule-bound and autonomous approaches to innovation implementation. The rule-bound approach presumes a "central authority who wants to see subordinates follow rules to carry out his or her policies," whereas the autonomous approach "accepts that people in the lowest echelons of an organization exhibit autonomy by redefining policies during the course of implementation" (p. 237). Marcus concludes that "implementation is likely to be more effective when policy implementors [i.e., lower level managers] are free to design and determine the specifics" (p. 251).

Within the organizational change literature, Kanter et al. ( 1992) note the scarcity of "practical examples of organizations . . . that have fully transformed themselves to attain this ideal [i.e., the post bureaucratic model]" (p. 5) and observe that "grassroots innovation-often referred to as bottom-up change-is often preferred to large-scale top- down change as a source of enduring results" (p. 7). Similarly, Beer, Eisenstat, and Spector (1990) note the widespread failure of top-down programmatic approaches to change such as quality circles or attempts to change organizational culture. They endorse a model in which lower level managers take the lead in devising alternative means of processing work.

The reinvention lab program, which is an NPR initiative launched by Vice President Al Gore and his staff as a means of embedding post bureaucratic organizing practices within the federal executive establishment, presents an opportunity to examine the viability of change strategies in which lower level managers are allowed a central role. In the discussion below, this generic approach is identified as generative, in contrast with a command or top-down approach to change. This research is exploratory in nature; no attempt is made to test the relative efficacy of the two approaches. Nevertheless, because the generative model represents a departure from conventional approaches to change, a review of its viability is important. A second purpose is to examine factors that account for the relative effectiveness of the generative model in different organizational contexts.

Strategy-Making Process Models
Hart (1992) identifies 11 separate typologies of strategy-making processes, which he organizes into a master typology or integrative framework. Strategy, in this context, is a rather amorphous concept. Shrivastava and Grant ( 1985), referencing Chandler (1962), define strategic decisions as "ones that involve a commitment of a large amount of organizational resources for the fulfillment of organizational goals and purposes through appropriate means" (p. 98). For the purpose of this discussion, the change from bureaucratic to post bureaucratic structures represents a specific strategy suited for analysis within Hart's (1992) framework.

The modal approach to strategy making is what Hart (1992) calls the command mode, which presumes a highly centralized structure in which lower level participants abide by directives from the top. In this mode:-

Strategy formulation is exclusively the function of top management,
Top management relies primarily on formal means (rules, hierarchical direction, staffing) of inducing compliance by other organizational participants
Implementation is separate from and subsequent to formulation.

Each of these three features represents a dimension of the strategy-making process according to which other models can be categorized. Hence, one dimension is the degree of domination by top management of the formulation process (top-levels dominant vs. shared), a second is whether primary reliance is placed on formal or informal levers or means of inducing behavioral change, and a third is whether formulation and implementation occur iteratively or sequentially.

Multiple strategy-making models (or modes) identified by Hart (1992), Mintzberg and Waters (1985), and Bourgeois and Brodwin (1984) arrayed according to these three dimensions. Both Mintzberg and Waters and Bourgeois and Brodwin include in their typologies models that share the three essential features of Hart's command mode. All three sets of authors also include an approach or approaches that represent the polar opposite of the command mode. Although different terms are employed, for the purpose of this discussion this generic approach will be labeled generative, consistent with Hart's terminology. In the generative model, responsibility for strategy formulation is shared by managers at all levels, informal rather than formal means serve to gain compliance, and formulation and implementation occur in an iterative rather than sequential manners

Hart (1992) describes the generative mode as one in which "strategy is made via intrapreneurship-new product ideas emerge upward, and employee initiative shapes the firm's strategic direction" (p. 338). Similarly, in the "crescive" model identified by Bourgeois and Brodwin (1984), strategy comes upward from the firing line, rather than downward from the top. The role of the CEO has moved from designer to that of premise-setter and judge.... Here, the strategic problem revolves around the CEO's ability to define organization purposes (i.e. set decision premises) broadly enough to encourage innovation, and to select judiciously from among those projects or strategy alternatives that reach his attention. (p. 254)

Mintzberg and Waters (1985) describe the "umbrella strategy" as one in which "leaders who have only partial control over other actors in an organization . . . set general guidelines for behavior-define the boundaries- and then let other actors manoeuvre within them"; the general guidelines may be established by means of a "vision emanating from the central leadership" (p. 263).

These theorists offer ideas on the circumstances for which these approaches to strategic change are most suitable. Mintzberg and Waters (1985) posit that their umbrella strategy is suited to circumstances in which the leader has "only partial control over other actors" and an environment that is "complex, and perhaps somewhat uncontrollable and unpredictable as well" (p. 263). The planned strategy (analogous to the command mode), in contrast, is suitable to environments that are, "if not benign or controllable, then at least rather predictable" (p. 259), because separation between the formulation and implementation phases requires that the formulator be able to anticipate circumstances that will prevail at the time of implementation. Where formulation and implementation occur in an iterative manner, as in the umbrella mode, there is an opportunity to adjust to changing circumstances-an approach that is conducive to strategic learning.

Bourgeois and Brodwin (1984) regard their crescive model as particularly suitable for "the large divisionalized firm" (p. 259). They question the validity of many of the assumptions that underlie the command mode, observing that, in such firms, "the chief executive cannot monitor all significant opportunities and threats" and that "information collected at lower levels often becomes stale before it can be assimilated" (p. 257). Both considerations mitigate in favor of a shared approach to strategy formulation in which information accessible to those at lower levels can be applied to the solution of strategic problems. They further observe that strategies formulated in a participative manner are more likely to be successfully implemented.

Applying the Generative Model
Is such an approach to strategic change suitable to the public sector generally and to the federal government in particular? It is possible to identify features of the federal environment that appear well suited to the generative approach. The federal executive establishment broadly resembles the divisionalized firm with substantial authority residing at the department and bureau levels. The environment confronting federal managers, although less dynamic than that confronting most private-sector firms, is nevertheless increasingly turbulent and- with multiple stakeholders, fragmented sources of external authority, and a proliferation of administrative controls- highly complex (Rainey, 1991).

The reinvention lab program qualifies as an example of the generative approach to strategy making and strategic change. The intent of the program's designers was that managers of units designated as labs would be allowed sufficient autonomy to identify and test new approaches to processing work and that those deemed successful would be institutionalized across other units. Elements of the program that are similar to those of the generative model include the following:

Lab managers and their staffs would be allowed and in fact encouraged to participate in the formulation of specific strategies for their unit and/or organization that fit the broad guidelines set forth by the vice president and his staff in the report of the NPR (U.S. NPR, 1993).
Vice President Gore and his staff relied more on informal than on formal means of inducing compliance with their vision. For example, the reinvention lab designation carries with it no formal authorities; rather, it serves to convey to other stakeholders the legitimacy the change efforts so designated.
In most cases, the units designated laboratories were sufficiently small to allow an iterative process of formulation and implementation such that the design could be continually adjusted and refined. Consistent with the lab rubric, the spirit was intended to be one of experimentation and learning.

The lab program offers an opportunity to examine the suitability of the generic model to this particular environment. In the following section, data collected on the lab program serves as a basis for analyzing the efficacy of the generative model as a means of inducing strategic change in the federal government.

The Reinvention Lab Experience
The reinvention lab program was formally inaugurated in the spring of 1993 when Vice President Gore sent a letter to the heads of all departments and of the larger independent agencies asking each to designate two or three programs or units as "laboratories for reinventing government" (Gore, 1993). As with other elements of NPR, implementation was left largely up to the departments (or agencies). According to U.S. NPR (1998), reinvention labs are designated to lead the way and set the pace of change. They are asked to experiment with new processes and new ways of doing business. They are called upon to be creative and innovative, and radically improve service and performance.

In 1994, when the research for this article began, there were approximately 140 reinvention labs distributed among 28 departments and agencies. Thirty-five randomly selected labs provide the primary data set for this study; 176 interviews have been conducted with 102 different individuals at the 35 labs. The interviews occurred in three phases: The first phase was from late 1994 to mid1995, the second phase was from mid-1995 to early-1996, and the third phase occurred in early 1998. The initial contact was with the person identified as lab manager by NPR.

Subsequent interviews were conducted with other individuals who had knowledge of the project and its history and with those, generally from the same agency, who could offer a perspective different from the direct participants. A semi structured interview protocol was employed; archival materials were also used as source of data.

The primary unit of analysis in this study is the reinvention lab program as a whole, consistent with a focus on the applicability of the change model the lab program represents to the federal government. Embedded units of analysis (Yin, 1994) include the individual labs and specific changes or proposed changes labeled innovations. One reinvention lab might have multiple innovations. For example, the debt collection service in the Department of Education has three separate innovations: the reorganization of workers into teams, the institution of a power-sharing arrangement with the employee union, and an attempt by the unit to gain a degree of autonomy from departmental budget processes. As listed in Table 2, there were 48 separate innovations at the 35 laboratories.

Innovation Processes
The reinvention lab program is analyzed within a framework devised by innovation theorists. The three-phase model of innovation employed represents a hybrid of those proposed by Zaltman, Duncan, and Holbek (1973) and Wilson (1973). The phases identified are

Initiation/design
Implementation
Institutionalization.

Initiation/design represents that phase when the innovation is first conceptualized and its component elements identified; implementation occurs when the innovation is put in place at the local (unit) level. Implementation implicitly incorporates a design element as testing occurs and adjustments to the original concept are made.' Finally, institutionalization occurs when the innovation is replicated by other units within the organization.

The intent of the lab program is to induce organizational innovation, defined as "the successful utilization of processes, programs, or products which are new to an organization and which are introduced as a result of decisions made within that organization" (Rowe & Boise, 1973, p. 6). The emphasis in the lab program is less on new products or programs than on changes to work processes consistent with the goals identified above.

The reinvention lab program is premised on certain assumptions regarding the behaviors of midlevel federal managers; these assumptions correspond to the different phases of the innovation process and are congruent with the tenets of the generative approach to change cited above, as follows:

Initiation/design: The program presumes that midlevel and lower level managers will participate in strategy formulation by putting forth ideas for improving the performance of their organization consistent with reinvention precepts.
Implementation: The lab designation is intended to legitimize innovative activity by lower level managers and to facilitate implementation by deterring interference from other units. Implicit in the lab designation is that these units will be granted some degree of autonomy to test innovative approaches to service delivery.
Institutionalization: Although this approach to change can have a systemic impact only if individual innovations are successfully replicated, the program sponsors did not specify how this would be accomplished.

The following discussion evaluates the extent to which the assumptions listed under initiation/design and implementation have been borne out and the extent to which institutionalization has occurred despite the lack of an identifiable strategy toward that end. This permits a preliminary examination of the suitability of the generative model to the federal environment. Thus, the model can be regarded as viable to the extent that lower level managers have responded by proposing changes, lab status facilitated implementation, and successful innovations were institutionalized. Table 2 lists the 35 labs included in this study and the innovations associated with each as well as summary information for each phase of the innovation process. Brief case histories of several of the labs are presented below. The cases were selected specifically because they illustrate how the model was intended to work; they constitute best practices and are not representative of the broader sample in this respect.