The Role of External Involvement in the Creation of Management Innovations

External Involvement in the Management Innovation Process

We conceptualize the management innovation process as transpiring through multiple stages according to an evolutionary logic (Burgelman, 1991; Damanpour & Schneider, 2006; Zbaracki, 1998). These stages have been called motivation, invention, implementation, and theorization and labelling (Birkinshaw et al., 2008). The process involves internal change agents as driving forces and may also involve external change agents. The process takes place within an organizational and environmental context that constrains and enables these agents. While we do not wish to downplay the significance of that context here as it may, for instance, be of significant impact on organizational creativity (Amabile, 1997), it is not our research focus.

External involvement is the use of inputs in the innovation process that originate outside the boundaries of the focal organization. Internal change agents are assumed to use external involvement whenever they consider it to be beneficial. These benefits may be more tangible, for instance, when external involvement helps shape an innovation directly, or less tangible, for example, when external involvement is sought primarily to legitimize an innovation (Staw & Epstein, 2000). We acknowledge that the use of external involvement is costly, although some costs will have been incurred prior to an innovation effort, i.e. the marginal costs may be small.

We suggest it is useful to distinguish between three forms of external involvement, each of which potentially affects the process of management innovation through a different mechanism. The first is external change agents (Birkinshaw et al., 2008, p. 832), individuals who ‘are independent consultants, academics, and gurus who are proactive in creating interest in, influencing the development of, and legitimizing the effectiveness and retention of new management practices’. Birkinshaw et al. (2008, p. 832) contrast this with internal change agents, ‘who are the employees of the innovating company proactive in creating interest in, experimenting with, and validating the management innovation in question’. The second is external knowledge sourcing – the observation of related practices in other organizations and contexts that are transferred into the focal organization. External knowledge sourcing is a well-known means of innovating in the technology and product domains (Ahuja, 2000; Leiponen & Helfat, 2009) and has been theorized to affect management innovation as well, a view that is increasingly supported by empirical evidence (Ganter & Hecker, 2011; Mol & Birkinshaw, 2009). The third form is external experience (Bapuji & Crossan, 2004; Hoang & Rothaermel, 2010), which occurs when an internal change agent brings in relevant novel experience from the outside, either from industry or through some form of training. We next discuss these forms in more detail; Table 1 summarizes their key characteristics.

OPEN IN VIEWER

Table 1.

Key Characteristics of the Three Forms of External Involvement.

	External agents	External sources	External experience
Acting agent is…	external to organization	internal to organization	internal to organization
Transferred into the organization when…	external agent becomes involved in innovating	internal agent sources knowledge from elsewhere	internal agent starts to work for organization
Mechanism relies on…	providing legitimacy and expertise by lending credibility to inventions, acting as sounding boards or action researchers, and theorizing about/labelling an innovation	informal transfer (usually) of knowledge from other organizations and individuals about elements of related practices	possession of relevant formal training to provide a method or on-the-job industry experience to improve decision-making and legitimacy
Role in motivation phase	Identifying new threats or opportunities	Generating insights into novel problem	Contextualizing novelty of problem
Role in invention phase	Generating new ideas	Translating existing practices into hypothetical practices	Transforming elements of previous practices into hypothetical practices/methodical innovation
Role in implementation phase	Conducting in-vitro thought experiment	Applying lessons on successful implementations of new practices	Testing viability of in-vivo new practices
Role in theorization and labelling phase	Creating theorized practice (out of immediate context)	—	Packaging new practice

A Typology of Management Innovations

In this section we present a typology of management innovations, which allows us to identify what we see as primary dimensions along which management innovations vary, as a first step in theory building (Doty & Glick, 1994). Our typology uses the broader literature on the management of innovation. It is worth noting that this typology is concerned with the characteristics of the innovations as such, and not with questions of impact or effectiveness. This approach is in line with our focus on the genesis of management innovations: we are more interested in the role external involvement played in shaping the type of innovation that transpired, rather than its organizational impact or subsequent diffusion. Equally, note that the focus is not on organizational characteristics.

One versus many functions

The second dimension concerns the scope (or breadth) of the innovation’s impact on the focal organization. We characterize each innovation as either focusing on a discrete functional activity, e.g. the human resource function or the supply chain, or affecting multiple functions across the organization and the linkages between them. This is an important dimension because it influences both the degree of risk and complexity in implementing the innovation in the first place, and the difficulty that other organizations will have in adopting the innovation if it proves to be successful. As the resource-based view of the firm has shown, the greater the degree of interdependence in the firm’s knowledge assets (Winter, 1987), and the greater the degree of causal ambiguity in the relationship between inputs and outputs (Lippman & Rumelt, 1982), the more likely it is that a firm’s internal resources and capabilities will be a source of competitive advantage. More systemic management innovations may therefore offer greater potential for competitive advantage but equally are likely to be more difficult to implement effectively.

In Figure 1 we summarize this framework and include some examples for the purposes of exposition. For example, scenario planning, as developed at Shell (Wack, 1985), provided a tool to help those in the strategy function to think more effectively about the future. Return on investment, introduced by Donaldson Brown at Du Pont in 1912, was a radically new idea for calculating the prospects and performance of projects, but its focus was not on the system as a whole (Johnson & Kaplan, 1987), but on investment decision-making. The M-Form structure was a radical management innovation at the time of its introduction because it required executives at the top of the organization to give up power over operational decision-making, and it required many employees to work in business units rather than in central functions (Chandler, 1962), resulting in changes throughout General Motors. By contrast, management by objectives (MBO) was more incremental, as the pursuit of objectives had been around, but it was the manner in which they were defined and used that was novel (Greenwood, 1981), even though it too had implications for the entire General Electric (GE) organization. These examples refer to the original versions of these innovations, not the subsequent evolution of the innovations as they diffused (Ansari et al., 2010). For instance, brand management at P&G was initially focused on organizing the marketing function only, but brand managers subsequently extended their reach into such areas as production management (Dyer, Dalzell & Olegario, 2004).

OPEN IN VIEWER

Figure 1.

A Typology of Management Innovation.

We believe this typology provides a useful way of characterizing management innovations and has close parallels to the well-known typology for technological innovation introduced by Henderson and Clark (1990). We acknowledge the two dimensions may not be orthogonal: we might expect, for example, that radical management innovations are more likely to affect multiple parts of the system, while incremental innovations might be more likely to transpire within a single function. However, our examples suggest that the dimensions are not entirely overlapping either, and it is ultimately an empirical question what proportion of innovations we observe in each box.

External Involvement and Type of Management Innovation

We now bring the two aspects of our conceptual development together, by asking the question: How does the nature of external involvement vary depending on the type of management innovation? Prior research is of limited help to us here, because the typology developed above has not been widely applied in the management innovation literature and work on the genesis of innovations is very rare. We therefore refrain from developing explicit hypotheses. Rather, we consider the competing arguments that can be drawn from the extant literature, and then conduct an empirical study to establish which arguments have greater validity in our setting.

The typology suggests two dimensions that address different facets of ‘difficulty’ for those seeking to introduce a management innovation to an organization. In other words, for the internal change agents in question, radical (rather than incremental) innovations require a greater departure from the current way of working, and systemic (rather than single function) innovations require more parts of the organization to buy into the new way of working. However, it is not entirely obvious whether these more difficult innovations are helped or hindered by external involvement. Two contrasting arguments can be put forward.

On the one hand, more difficult and complex innovations benefit from external involvement, because they require a greater breadth of knowledge inputs to be conceived and implemented (Nelson & Winter, 1982). There are many empirical studies suggesting that external knowledge sources provide such breadth (e.g. Laursen & Salter, 2006; Leiponen & Helfat, 2009), as well as providing access to multiple organizations (Garud & Kumaraswamy, 1995). While management innovation differs from large-scale technological innovation in that the innovations concerned are not products to be sold to customers, we argue that the same principle holds true: when firms pursue more radical and/or more systemic innovations, they benefit from having access to a diversity of external knowledge sources.

On the other hand, there are circumstances in which the relationship between external input and innovation is negative. For example, the literature on organizational complexity suggests that there are often so many interdependencies within a system that additional inputs from the outside can compromise or slow down the innovation process. According to this worldview, new insights from the outside are not a bad thing per se, but the costs of integrating them with the existing structures and systems of the organization often outweigh the benefits (Ethiraj & Levinthal, 2004). A separate point that applies particularly to the role of external change agents is that such individuals have their own motivations, such as reducing the risk of failure, and applying their insights about what works from other organizations they have worked with (Kaplan, 1998). Both these arguments suggest that external involvement may sometimes dampen the extent to which a management innovation is radical or systemic in its scope.

Rather than choose between these two lines of argument, we prefer to set up the following research question that we then address in our empirical analysis:

Our second research question is concerned with the interaction between different forms of external involvement. On an intuitive basis, one might expect that ‘more is better’ and therefore that organizations using multiple forms of external involvement would benefit more than those using one form. However, empirical evidence suggests there is not a single factor that drives the adoption of management innovations (Damanpour & Evan, 1984), and in more recent work (Mol & Birkinshaw, 2009) it has been shown that external knowledge sources and internal knowledge-producing characteristics, such as having a highly educated workforce, can be traded off against each other, i.e. they act as substitutes for each other. Using the behavioral theory of the firm, Mol and Birkinshaw (2009, p. 1273) argue that there is a cost involved in any of the mechanisms that help generate management innovations, especially more active mechanisms such as the search for external knowledge sources and the use of external change agents, and internal change agents will only bear that cost if it is required. This suggests that the three different modes of external involvement (external experience, external knowledge sources and external change agents) may not actually have a cumulative effect in the production of radical and systemic innovations. Hence our second research question:

Methods

When looking for a sample of management innovation processes to analyse, we needed to study ‘completed’ innovations where processes and outcomes were relatively clear, because of our interest in categorizing management innovations as radical/incremental and systemic/single function. This meant using archival sources of data, which has significant advantages and disadvantages (Ketchen, Ireland & Baker, 2013; Myers, 2013; Yin, 2003). In terms of advantages, it allowed us to compare and contrast important historical management innovations (e.g. brand management, the M-Form) that have had a significant impact on the advancement of the field of management. Moreover, the literature sources we used had been through a quality assurance process (i.e. peer review and subsequent citation of work), allowing us to be more confident about the quality of descriptions of innovation processes. Furthermore, we were able to identify detailed quotes that closely align with our theoretical constructs, overcoming the problem of external validity that plagues many archival studies in strategic management (Ketchen et al., 2013).

In terms of disadvantages, this historical record clearly has a pro-innovation bias and is far from comprehensive in terms of the innovations it documents and the aspects of those innovations it acknowledges. By restricting ourselves to English language archives, we potentially missed out on some important cases. With all such biases, the key concern is whether they are likely to systematically influence our findings, and our view is that they will not (i.e. our interest here is not in innovation success but rather in the type of innovation produced). We believe that the sample presented here is representative of the major management innovations recognized in scholarly investigation. We focused on the last 150 years, although of course there were innovations even before that (Greif, 1996), but the record on such innovations is far less detailed.

To the best of our knowledge nobody has attempted to create a comprehensive list of key historical management innovations. Hence, we created our own list for this study. This list had to comply with several objectives. First, the innovations had to be completed, so that their outcomes were known. Second, for practical purposes, we restricted ourselves to innovations that were well-known in the English language literature, even though they might have originated from outside the English-speaking world. Third, a specific instance of implementation had to be described in the literature. Fourth, it had to be possible to find descriptions with some level of detail about the process through which the innovation was implemented.

We first constructed a long list of possible management innovations by consulting a variety of mostly academic sources. Some of these sources contained lists of management innovations in particular subfields (e.g. Kossek, 1987), others merely a few examples (e.g. Abrahamson, 1996). We used various sources to triangulate information (including Abrahamson, 1991, 1996; Barley & Kunda, 1992; Carson, Lanier, Carson & Guidry, 2000; Clark, 2003; Davenport & Prusak, 2003; Ettlie, 1988; Georgantzas & Shapiro, 1993; Kimberly & Evanisko, 1981; Kleiner, 1996; Kossek, 1987; Knights & McCabe, 1998; Lillrank, 1995; Nickell, Nicolitsas & Patterson, 2001; Pascale, 1990; Rigby, 2001; Staw & Epstein, 2000; Voss, 1995; Wolfe, 1995). Next we interviewed four functional experts at a leading global business school, who were asked to comment and to add further innovations. Through this process we generated a list of 181 potential innovations. ¹

The second major stage of defining the sample involved narrowing that list using various filters. First, we tested whether a case met our definition of a management innovation presented above. In the very few cases where we did not know the term at all, we either asked a colleague with substantial knowledge of the specific area or ran a preliminary search to identify its meaning. For example, at this stage we dropped ‘globalization’ because it is a broad term, not a specific practice. For the remaining cases we searched for a formal definition as well as an early example or a first implementation, which led us to drop further cases. ² During this stage, 90 cases were dropped, leaving us with 91 potential innovations.

The next step was to create a survey featuring the remaining 91, their formal definitions, and an instance of their first implementation. This survey was distributed among a panel of eight academic innovation experts who were not part of the research project. This panel was asked to comment on whether they thought these were management innovations, bearing in mind our definition, and to provide insights into the examples. Based on the assessments of these experts another 17 terms were dropped. Terms dropped for instance included ‘empowerment’, as it was seen as representing a movement rather than a specific practice. We then investigated the remaining 75 cases. For 36 out of 75 it proved difficult to find detailed information on the first implementation. For example, Celanese is mentioned as a firm where the practice of outplacement might have originated, but we could not find sufficient detail on what actually happened. Thus we collected data on the remaining 39 innovations from a variety of sources as described below, but in many of these cases we could not obtain sufficient data on variables of interest in this paper. For example, we were not able to trace the external experience of the creators of revenue (or yield) management at American Airlines. We ended up with a sample of 23 innovations and view these cases as representative of the population of major management innovations covered in the English language literature over the last 150 years, rather than comprehensive.

Results

We start our analysis by describing, in Table 2, the names of all 23 cases of management innovation, the organizations they took place in, when they first emerged, which of the three kinds of external involvement they featured, whether they were radical and/or systemic and during what specific stages external change agents were involved. One case had no external involvement. We retain it because the absence of external involvement could well have shaped the innovation. A total of 14 innovations were coded as systemic, while 17 innovations were coded as radical. From this table it can be observed that for only one of these 23 cases (activity-based costing) we could find no evidence of external involvement (Kaplan and Cooper were strongly involved, but only in stage 4 by theorizing around activity-based costing and promoting it to a wider audience). Note that we do not code for the presence of external change agents if there is ‘stage 4’ (Birkinshaw et al., 2008) involvement only, as it occurs largely or entirely outside the focal organization and is present for all of the major innovations sampled here. This is the phase where management innovations are prepared for diffusion and possibly become management fashions (Abrahamson, 991).

OPEN IN VIEWER

Table 2.

Innovations in Sample and Coding.

Name	Year started	Company	Sources	Experience	Agents	Radical	Systemic	Stages
1. Activity-based costing	1985	John Deere	0	0	0	1	1
2. Market segmentation	1921	GM	0	1	0	1	0
3. Return on investment	1912	DuPont	0	1	0	1	0
4. M-Form	1921	GM	0	1	0	1	1
5. Scientific management	1883	Midvale, esp. Taylor	0	1	0	1	1
6. Corporate welfarism	1836	Krupp	1	0	0	1	0
7. Business process reengineering	1983	Ford	1	0	0	1	1
8. Mass customization	1968	Lutron	1	0	0	1	1
9. Matrix organization	1959	McDonnell	1	0	1	1	1	1
10. Scenario planning	1971	Shell	1	1	0	0	0
11. Benchmarking	1979	Xerox	1	1	0	0	1
12. Industrial research	1890	Bayer	1	1	0	1	0
13. Lean manufacturing	1948	Toyota	1	1	0	1	0
14. Moving assembly line	1913	Ford	1	1	0	1	0
15. Strategic planning	1946	Ford	1	1	0	1	0
16. Balanced scorecard	1987	Analog Devices	1	1	0	1	1
17. Brand management	1931	P&G	1	1	0	1	1
18. Professional managers	1846	LNW Railway	1	1	0	1	1
19. T-Groups	1946	NTL and Bayway	1	1	1	0	0	1, 2, 3
20. Management by objectives	1952	GE	1	1	1	0	1	1, 2, 3
21. Six Sigma	1986	Motorola	1	1	1	0	1	1
22. Supply chain management	1982	Booz Allen with Philips	1	1	1	0	1	3
23. Total Quality Management	1950	Toyota & Matsushita	1	1	1	1	1	1, 2

0 = absent, 1 = present, stages are numbered as indicated in text

For six innovations we found evidence of external change agents. One such case is MBO: when GE’s Harold Smiddy implemented MBO in the late 1940s, management thinker Peter Drucker was very heavily involved in these efforts, perhaps to the extent that it was really Drucker who invented MBO and Smiddy who assisted him, and in the process used GE as an experimentation ground (Greenwood, 1981). As Greenwood (1981, p. 226) puts it:

When Harold Smiddy first came to the General Electric Company in 1948, he introduced the monthly letter concept to two operating divisions; but it was not until 1952, when he first began to have the corporate philosophy of management written, that Drucker was able to convince him of the ease with which the monthly letter concept could act as a foundation for an MBO philosophy of management.

And while it was Smiddy who used these monthly letters, it was Drucker who eventually coined the term MBO. Drucker was equally involved in some of the raw ideas that led to MBO in the earlier stages of the innovation process, for instance by introducing the notion that objectives are neither given nor obvious or ubiquitous (Greenwood, 1981).

For 18 cases there was clear evidence of use of external knowledge sources that helped to inspire the innovating firm. In the case of brand management, it was Neil McElroy at P&G who, while in England working on the introduction of the Oxydol brand,

… observed that the diverse operations of European soap and margarine giant Unilever competed directly with one another, but in an inefficient fashion … McElroy believed that each P&G brand should have its own brand assistant and managers dedicated to the advertising and other marketing activities for the brand. (Low & Fullerton, 1994, p. 180)

So this knowledge about outside examples provided McElroy with both an idea about what might be possible, to let different brands compete head-on, as well as a warning sign about the ineffectiveness of the method used elsewhere, meaning that if P&G were to successfully introduce brand management it needed to do so differently from Unilever. Yet sometimes external knowledge sources were absent: Alfred Sloan, for example, explicitly said he did not draw on any significant external sources when inventing the M-Form at GM, and DuPont’s simultaneous invention of a divisional structure was a separate development (Chandler, 1962).

Finally, there were also 18 cases of internal change agents with prior experience. In the cases of the M-Form and brand management, for example, this experience came primarily from the training backgrounds of the internal change agents. Neil McElroy, prior to joining P&G as an advertising department mail clerk in 1925, had graduated from Harvard Business School and used this training in his work on brand management (Schisgall, 1981). Alfred Sloan at GM had trained as an engineer at MIT, using this focus to design the structure he proposed in his famous ‘Organizational Study’ (Chandler, 1962). He was also previously the president of Hyatt Roller Bearing and United Motors (Chandler, 1962).

When it comes to MBO, the external experience came from the work experience of Smiddy. As Greenwood (1981, p. 226) argues:

Harold Smiddy, before coming to GE, had been a partner at Booz, Allen, and Hamilton. There he had learned the concept of the ‘manager’s letter’ from one of his associates. Simply put, the ‘manager’s letter’ required a job holder (manager or individual contributor) to write a letter to his or her superior indicating what the goals for the next period of time were, how the goals would be met, and what standards were to be expected. When the superior accepted this letter — usually after editing and discussion — it became the work ‘contract.’

In Table 3 we present the correlations. With the number of observations available here it is very difficult to obtain significant correlations, and any tests are thus conservative. The first point to note is that the correlation between the radical and systemic variables is close to 0, suggesting the dimensions of the framework in Figure 1 appear to be orthogonal. In terms of addressing research question 1, we find that the correlation with radical innovation is negative for external change agents, knowledge sources and experience. On the other hand, the correlation with systemic innovation is strongly positive for external change agents, marginally positive for external knowledge sources, and negative for external experience. Finally, the correlation with innovations that were radical and systemic (referred to here as radical*systemic) is somewhat negative for external change agents and negative for external knowledge sources and external experience. In other words, and without drawing strong conclusions due to the small sample size, the overall pattern suggests that external involvement is negatively associated with radical innovation, especially when external change agents are involved, but more positively associated with systemic innovation. This is a somewhat surprising finding, as it suggests that the role of external involvement varies depending on which of the two dimensions of the typology we are concerned with.

OPEN IN VIEWER

Table 3.

Polychoric Correlations Between Variables (N = 23).

	Radical	Systemic	Radical*systemic	Change agents	Knowledge sources	Experience
Radical	1
Systemic	−.0.12	1
Radical*systemic	0.67*	0.77*	1
Change agents	−0.77*	0.49	−0.21	1
Knowledge sources	−0.70	0.02	−0.29	0.98	1
Experience	−0.73	−0.37	−0.62	0.15	−0.04	1

*

= significant at 5% level

It is worth considering a couple of the cases to exemplify what is going on here. For example, we characterized the invention of industrial research by Bayer in 1890 as a radical break with the past but focused on a single function rather than the enterprise as a whole, yet it was put in place entirely through internal change agents (Meyer-Thurow, 1982, p. 369):

By about 1890 Bayer had recognized the distinctive role of the research chemist and made research a salaried, lifelong, specialized occupation. Yet many difficulties remained. The laboratories were scattered all over the plant, located where there happened to be a free room. (Beer, 1958, p. 129)

Shortly after, the company decided to further formalize industrial research and committed to the construction of a purpose-built laboratory:

In 1890 the directors of the plant voted to house the new research division in a laboratory costing 1½ million marks. This modern building incorporated some original features of interior arrangement, conceived by Duisberg, that have been widely copied by industrial laboratories ever since. (Beer, 1958, p. 129)

In contrast, we characterized the development of T-Groups in NTL in 1946 as incremental in nature, yet there was a very high level of external influence from external agents, in this case practically minded academics, who brought with them ideas about what had worked in their prior engagements:

Although the history of the T-group is usually traced to Kurt Lewin and the serendipitous events in New Britain, a number of experiments with group processes preceded the Connecticut workshop. In fact, Lippitt and Bradford had earlier experimented with confrontational meetings among nurses and doctors at the Freedmen’s Hospital in Washington while they were both at the Federal Security Agency. Much earlier, the German psychologist Jacob Moreno had developed the concept of encounter, which emphasized breaking social constraints and dealing honestly with others … More immediate influences on the ideas of the T-group developers were the projects conducted during World War II by the Tavistock Institute in England and the Office of Strategic Services (OSS) in the United States. (Highhouse, 2002, p. 280)

Similarly, in the case of MBO, as Drucker argued in an interview, although the concept had not been proposed in the manner he put forward,

A good many people in earlier times managed by objectives. Sloan was the first one I met, for the simple reason that he was the first significant figure in management I met personally. But I would imagine that Pierre Du Pont, before him, started out with objectives — perhaps even more clearly than Sloan did … In other words, to manage by objectives is nothing new. (Greenwood, 1981, pp. 229–230)

These examples reinforce that very few novel practices are truly new to the world: some innovations are ‘new to the state of the art’ (Birkinshaw et al., 2008), and the vast majority are best designated as being only ‘new to the organization’. Internal and external change agents, in this case Smiddy and Drucker, learnt from others before and during the creation of management innovations, and therefore, even in our sample of the most well-known management innovations, a good number of innovations were not radically new.

Moving now to research question 2, we correlate the different types of innovation with the number of forms of involvement (anywhere from 0 to 3), using standard Pearson correlations. This produced some clearer findings. In particular, the number of forms of external involvement is negatively correlated with the production of radical innovations (significant at 5%) but not correlated significantly with systemic innovation or radical*systemic innovations. In line with the notion of substitution between forms describe above, we calculated a correlation between the types of innovation and the curvilinear (squared) term for the number of forms of involvement. This produced an even stronger negative correlation (significant at 1%) with radical innovation. This tentative finding is interesting, as it suggests there may be some ‘optimal level’ of external involvement, beyond which an innovation is less likely to be radically new.

Moving on to the QCA analysis, we begin by looking for sufficient conditions for radical, systemic and radical*systemic innovations. Sufficiency can be observed directly from truth tables. Although there are eight possible configurations, our dataset only includes cases for six of these. The cut-off point for assessing whether a condition is sufficient for the dependent phenomenon to occur is a consistency of 0.80 (see Ragin, 2008). In Table 4 we summarize the results, using recently developed conventions (Fiss, 2011). We report which sets of conditions, whether present or absent, are associated with the outcome variables, distinguishing between core conditions, which are part of the parsimonious and intermediate solutions, and peripheral conditions, which are eliminated in the parsimonious solution.

OPEN IN VIEWER

Table 4.

Configurations Associated with Radical, Systemic and Radical*Systemic Innovations in the QCA Analysis.

	RADICAL		SYSTEMIC		RADICAL and SYSTEMIC
Condition	S1	S2	S1	S2	S1	S2
Change agents	Ө		•	Ө	Ө	•
Knowledge sources	Ө	•	•	Ө	Ө	•
Experience		Ө		Ө	Ө	Ө
Observed cases	5	4	6	1	1	1
Consistency	1.00	1.00	0.83	1.00	1.00	1.00
Raw coverage	0.29	0.24	0.36	0.07	0.10	0.10
Unique coverage	0.29	0.24	0.36	0.07	0.10	0.10
Solution consistency	1.00		0.86		1.00
Solution coverage	0.53		0.43		0.20

• = core condition when present; • = peripheral condition when present; Ө = core condition when absent; Ө = peripheral condition when absent

The truth table suggested that in this dataset radical innovations come about when one of four configurations occurs: (1) no external involvement at all; (2) only external knowledge sources; (3) only external experience; (4) external change agents and external knowledge sources. Our intermediate solution reflects this, as it produces two combinations of sufficient conditions, namely ~sources*~agents (this means not sources and not agents), with a raw coverage of 0.29, and ~experience*sources (not experience and sources), with a raw coverage of 0.24. ⁴ Coverage tells us to what degree the solutions that are produced explain the outcomes. The fact that a condition is necessary, however, does not provide information on whether or not that condition is sufficient and therefore further tests are warranted (Schneider & Wagemann, 2012). There is no single condition that occurs in both solutions and we run a separate necessity test, which reveals that none of the conditions has a consistency score above 0.90. Thus our conditions are sufficient but not necessary.

One example of the ~sources*~agents configuration is activity-based costing. Its motivation and development was entirely internal, as noted by Kaplan and March (1987, p. 7):

Keith Williams had been aware that the existing cost system, although satisfactory at an aggregate level, was ineffective for costing and bidding individual parts. He was experimenting with other ways to apply overhead to products. When Maxwell called him in November 1984, Williams realized that the situation at Gear and Special Products provided an opportunity to demonstrate the weaknesses of the current system and to develop a new approach that would be more useful for decision making.

From this internal development, a radical innovation emerged, that according to Anderson and Young (2001, p. 20) ‘revolutionized the field of management accounting, elevating cost analysis from a tactical exercise in management evaluation and control to a key component of planning’.

In the truth tables systemic innovations came about when one of three configurations is present: (1) no external involvement; (2) change agents and knowledge sources; (3) all three forms of involvement. The intermediate solution again produces two necessary conditions, which are ~experience*~sources (not experience and not sources), with a raw coverage of 0.07, and sources*agents (sources and agents), with a raw coverage of 0.36. Similar to the above, there is no single condition that occurs in both solutions and the necessity test did not throw up conditions with a score above 0.90 and these conditions are sufficient but not necessary.

A well-known example of the sources*agents configuration is the development of total quality management, which is systemic because (Ebrahimpour, 1985, p. 421):

Implementing quality control effectively necessitates the cooperation of all people in the company, involving top management, managers, supervisors, and workers in all areas of corporate activities such as market research, research and development, product planning, design, preparations for production, purchasing, vendor management, manufacturing, inspection, sales and after-service, as well as financial control, personnel administration, and training and education.

The innovation as developed relied heavily upon the involvement of external changes agents, especially Joseph M. Juran, Armand V. Feigenbaum and Philip Crosby (Ross, 1999, p. 4), and external knowledge sources were used too, as TQM as developed in Japan drew explicitly upon earlier developments in the United States (Juran, 1995, p. 556):

The middle 1920s witnessed the first significant wave of so-called ‘Statistical Quality Control’ (SQC). It had its origin in the Bell System. It was initiated in 1926 when a team from the Bell Telephone Laboratories proposed that the Hawthorne Works of Western Electric Company (the manufacturing arm of the Bell System) apply certain tools of statistical methodology to the control of quality of manufactured telephone products.

Finally, radical*systemic innovations occur when one of two configurations is present: (1) no external involvement; (2) change agents and knowledge sources. Once more, the analysis produces two solutions, ~experience*~sources (not experiences and not sources), with a raw coverage of 0.10, and ~experience*agents (not experience and agents), also with a raw coverage of 0.10, which is a substantially lower coverage than we obtained for the two underlying dimensions and therefore provides less of an explanation for the outcome variable. Since ~experience appears in both solutions it could be a necessary condition, but its consistency score in a test of necessity is only 0.40, indicating this is not the case. The results for the QCA analysis are broadly consistent with the correlations presented earlier, but provide additional insights, and the quotes elucidate why we observe these patterns, meaning our methods are complementary.

Discussion and Limitations

We can summarize the findings from this multi-method study as follows. Research question 1 asked how external involvement varied depending on the extent to which the innovation was radical (rather than incremental) and systemic (rather than single-function). Our evidence suggests the presence of external change agents is associated with systemic (not single-function) and incremental (not radical) innovations; that the absence of external experience is associated with systemic and radical innovations; and that the presence or absence of external sources of knowledge have no clear effect one way or the other.

The key insight, then, is that external involvement appears to have a dampening effect on the radicalness of the management innovation. Or to say it slightly differently, when we see truly radical management innovations, they transpire primarily through internal change agents (typically the top executives) acting on their own initiative, and without much outside help. Of course, these individuals may well have been inspired indirectly by things they observed beyond their firm’s boundaries, but in cases where they directly received external input, their management innovations were more likely to be incremental.

In contrast, the effect of external involvement on the single function/systemic dimension appears more mixed. External change agents are seen more in systemic innovations, we believe because a significant part of their role is to act as ‘process consultants’ who take responsibility for implementing changes across multiple parts of the organization. External experience, on the other hand, is associated more with single-function innovations, perhaps again because such experience pushes internal change agents towards less ambitious change programmes.

Research question 2 asked how the three types of external involvement interact in the management innovation process, and the answer was that they act to a large degree as substitutes rather than complements in the creation of more complex forms of management innovation. This suggests that the optimal number of forms of involvement, to the extent that more radical and systemic innovations are sought, may be an intermediate number.

Conclusions

In this paper we theorized about three forms of external involvement that affect the management innovation process, developed a typology of management innovation, and investigated how the forms of involvement relate to the type of innovation by embarking on an exploratory historical study. Our discussion and empirical analysis focused on the use by internal change agents of external experience and external knowledge and the direct involvement of external change agents. The typology of management innovation applies the dimensions of radical versus incremental and systemic versus single-function innovations. The results suggest that external involvement is frequently associated with less radical but more systemic innovations, suggesting that external change agents, in particular, play a more nuanced role in the management innovation process than was previously recognized. We also showed that the three forms of external involvement are for the most part substitutes not complements. We conclude by expressing the hope that the management innovation research agenda will continue to be explored further in future work.