Terminological distinctions of ‘control’: a review of the implications for management control research in the context of innovation

The articles analyzed in this review were acquired by systematically searching for literature within the discipline of business and management in August 2016. By using the search mode Boolean/Phrase the following string was applied for searching within publication abstracts: AB (AB “management control” AND AB innovation) OR AB (AB “management control” AND AB “research and development OR r&d”) OR AB (AB “management control” AND AB “new product development”). The criteria for the inclusion and exclusion of search terms in the field of management control and innovation were defined as follows. While focusing on management control, the search terms management accounting, performance measurement and budgeting as potential synonyms for MCS were not included. As indicated in the introduction, the article follows an Anglo-American understanding of MCS where not only the design of an appropriate accounting system is central (Guenther 2013). Instead, it takes a wider view and includes more indirect control mechanism as well (Haustein et al. 2014). For this reason, articles which refer to management accounting and management accounting systems solely were excluded, for instance those by Ferreira et al. (2010) or Ittner and Larcker (1998). Moreover, since innovation is defined in this article as an outcome of processes of invention, studies setting NPD or research and development instead of innovation central are included in the literature review such as Bonner (2005). Articles emphasizing a processual perspective on innovation such as learning (e.g. Widener 2007) were not taken into account. There are a number of empirical studies pinpointing the consequences of management control namely formal, personnel, behavioral control for individual creativity and intrinsic motivation in uncertain, innovative environments (for instance Adler and Chen 2011; Amabile 1998; Breunig et al. 2014; Collins and Amabile 1999; Randle and Rainnie 1997). These studies deliver reasonable insights into the discussion of management control and innovation both by emphasizing the autonomy from managerial control as critical to intrinsic motivation and by realizing that some forms of control like personnel control are most effective in environments of high task uncertainty and complexity. However, the appearance and change of managerial control and related systems in the different contexts of innovation as well as in which ways managerial control intersects with innovating practices, decision making and power structures can hardly be explored by studies at the level of individual creativity and motivation only (see for a similar discussion cf. Hopwood 1983 and Hopwood 1978). For this reason, these studies were not included.

The database UniSearch was approached as the main search entry. UniSearch is a library all-in-one service for searching articles, books and more through a single access point. By this means, the search string does not need to be adapted to suit the specific requirements of the different databases. The material types searched were academic journals and books. The search returned 104 results in the three databases Business Source Premier, EconLit and ScienceDirect. “Appendix 1” provides an overview of the journals that revealed results according to the search string. In a next step UniSearch allows to limit the results to the publication language English. Furthermore, it permits to specify further ‘subjects’ which are keywords defined by authors or are controlled vocabularies defined by publishers and librarians. By choosing the subject ‘empirical studies’ it was possible to detect the studies of interest for the article. At this stage, the sixty-six remaining publications from the systematic database search were manually assessed. They were finally limited to nineteen by excluding publications for which the topic product innovation and management control was not the central research question (e.g. innovation strategy, innovation culture, innovation values, managerial innovation, service innovation or control solely understood as hierarchy) or if the publication was still of pure conceptual nature.

In addition to the systematic database search several journals were manually approached since it was known that these sources had published papers on the topic or were recommended by experts in the field: Industrial Marketing Management, Organization Science, Scandinavian Journal of Management, International Journal of Networking and Virtual Organisations, Review of Managerial Science and Omega. This search revealed another seven articles. In total, the following literature review is based on twenty-five publications (study sample).

Observing the empirical field of management control in contexts of innovation without distinctions is impossible. Once a distinction is set it delivers on the one hand a decision which information shall be gained and transformed and which information not (the alternative would be chaos) (Luhmann 1988). An observation is a differentiation in which some things are omitted (unmarked state, ‘blind spot’) and others are highlighted (marked state) (Spencer Brown 1972). The delimitation of a marked state as well as the insinuation of an inner order is always dependent on the observer (here: the researcher in management control) and on his or her motives and objectives (Spencer Brown 1972). Benefits and pitfalls of “draw a distinction” (Spencer Brown 1972, 3) for the acquisition and transformation of empirical data are not easily visible. The marked state produces asymmetries which are widely taken for granted. These asymmetries are only reversible by replacing the underlying distinctions with novel distinctions which produce, of course, new asymmetries themselves. Hence, research in management control in the context of innovation needs to take distance, for instance, in literature reviews to challenge established distinctions.

As it counts for this article, distinctions are often presented as categories. By defining clear criteria for membership categories consider objects as being equivalent and tend to be viewed as separated from each other (Rosch 1998). They provide information so “that the perceived world comes as structured information rather than as arbitrary or unpredictable attributes” (Rosch 1998, 252). Authors usually introduce categories by prototypical members (Rosch 1998). Categories have a crucial function. They predetermine results and recommendations of empirical and conceptual studies whilst an unmarked state is inevitably inscribed. For exploring the marked and unmarked state it is necessary to have a closer look at the categorizations being used in empirical studies of management control and innovation. A categorization is thereby an entirety of categories, for instance, market, accounting and non-financial measures or for-profit and not-for-profit organizations.

Three major categorizations with different explanatory power were found in empirical studies of management control and innovation producing different kinds of marked states. Categorization 1 (Cat1) highlights studies which investigate the fit of different types of managerial control for the context of innovation. Categorization 2 (Cat2) represents studies which distinguish between the design mode and the use mode of managerial control instruments in the context of innovation. Categorization 3 (Cat3) summarizes studies that are referring to the enabling and constraining character of MCS in the context of innovation. Except for one article, these empirical studies either base their findings on Cat1 or on Cat2. An extended overview with sector and country investigated, empirical methods used in detail, their thematic focus and important findings besides their categorization can be found in “Appendix 2”.

Following, prototypical members representing the respective categorization are presented to define its marked and unmarked state.

As depicted in Table 1, for categorization 1 (Cat1), there were twelve articles found which base their empirical investigations on the distinction of different types of control.

In general, Cat1 studies assume that the implementation of a specific control element or control package causes a specific innovation performance. They trust in the fit of different types of managerial control elements or packages for the context of innovation. Cat1 studies refer to contingency in a way that associates the mission of management control with the identification and matching of specific MCS elements or packages to certain defined circumstances.

To exemplify how the ‘ideal fit logic’ of the Cat1 studies works some of the studies are presented in more detail in the following. Bonner (2005), for instance, investigated the degree of customer interaction when output control, process control and team rewards control are used in NPD projects. Bonner (2005) follows the idea that each type of control directs project teams in a specific way: either to organizational processes or to customers. According to his findings, customer interactivity is increased by output control and control via team rewards. By contrast, the interaction with customers is decreased when organizations decide for a strong focus on process control in NPD projects without combining it with output control. In an earlier study, Bonner et al. (2002) investigated the relation between different control mechanism (formal control, output control, interactive control, process control) and NPD project performance expressed by the degree of the NPD project’s innovativeness and the degree how the project contributes to the company’s product program. The findings of the study show inter alia that there is a negative relation between process control and NPD project performance, no relation between output control and project performance but a positive influence on NPD project performance when interactive control was installed at the beginning of the project.

Cardinal (2001) distinguishes between input, behavior and output types of control. The main finding is that input, behavior, and output control enhance radical innovation whereas input and output controls also enhance incremental innovation. Consequently, she states that incremental and radical innovation should be managed in the same way.

In a collaborative situation of open innovation, Kuschel et al. (2011) recommend to release direct forms of control like an intellectual property strategy in favor of more indirect ways of control, for instance by an active facilitation of spin-offs that might be re-acquired when innovations are mature enough, by open standards and generative technological IT-platforms or by mobilizing a critical mass of active and participating users to create value of innovative products.

Four propositions about the impact of informal control exercised by a company’s top management were formulated by Richtnér and Åhlström (2010). Informal control can positively affect knowledge creation in the later phases of a NPD project and the creation of explicit knowledge. It can negatively affect knowledge creation in the early phases of a NPD project as well as the creation of tacit knowledge.

Rijsdijk and van den Ende (2011) point out that the combination of output and clan control is the most effective in positively influencing the performance of NPD processes. Informal and formal controls are explicitly not seen as potential substitutes. They sound a cautionary note about combinations of different kinds of control which may result in synergistic or conflicting effects in view of NPD outcomes.

In summary, supporting Moll (2015) and her concerns about the low theoretical explication of management control and innovation, there are only three of the Cat1 studies which mention explicitly their grounding in an organization theory. Cardinal (2001), Richtnér and Åhlström (2010), Rijsdijk and van den Ende (2011) take up a contingency approach (originally Burns and Stalker 1961; Lawrence and Lorsch 1967; Woodward 1965). Even if not explicitly stated, the remaining studies take on the same perspective.

Cat1 studies refer to different types of control as a starting point for empirical investigations, taking up an argumentation of the following kind: “too much of the wrong type of formal control may constrain the team’s creativity, impede their progress, and injure their ultimate performance” (Bonner et al. 2002, 234).

As Fig. 1 generalizes, empirical investigations of Cat1 studies assume that an ideal fit between situational requirements and MCS is necessary for the success of innovation. The contingency paradigm emphasizes that there are specific situational factors that can affect the direct relationships between independent variables (MCS) and dependent variables (innovative performance) in studying organizational behavior. In general, there are two types of fit – horizontal and vertical fit (Wright and Snell 1998). The horizontal fit describes the fit among the management control elements within a MCS (MCS as a package; Otley 1980). Cat1 studies focus on the vertical contingency fit that covers the alignment of MCS with the strategic objectives of the firm (e.g. product differentiation). Contingency fit means that there is for each contingency variable a MCS package or element that constitutes fit and delivers the highest possible innovation performance for this contingency variable (Van de Ven and Drazin 1985).

The contingency perspective as interpreted by Cat1 studies in the field of management control and accounting has already been discussed by Otley (2016). By stating that contingency studies leave “no obvious starting point for an explanation of an increasing body of often contradictory results” (Chapman 1997, 189), reviewers started to discover the unmarked state of this theoretical perspective in management control (Argyris 1972; Child 1972; Hendry 1979; Smith and Nichol 1981). In particular, Bisbe and Otley state that contingency research provides “inconsistent findings regarding the relationship between formal MCS and product innovation” (Bisbe and Otley 2004, 710). Management accounting and control systems are “often operationalized as a purely formal and routine technology, with expectations of it often couched in terms of its presence in stable settings, or absence in unstable one” (Chapman 1997, 190). Bisbe and Otley even argue that “some authors have pointed to the different styles of use of formal MCS... or the different roles of MCS... as explanations for these apparently inconsistent studies” (Bisbe and Otley 2004, 710). As far as studies of management control in the context of innovation are concerned, the observation that cumulative findings and replications studies are rare cannot be justified simply by an imbalanced use of qualitative and quantitative studies (as Otley (2016) states for management control in general) as “Appendix 2” proves.

In my reading, Cat1 studies have a limited contribution to the theoretical corpus in management control in the context of innovation for other reasons. On the one hand, contingency is epistemologically seen the knowledge about all knowledge being relative—‘it can always be completely different’. This is an important lesson for the field of management control and shows that closed and simultaneously universal theories are not possible in themselves (Luhmann 1984). Hence, Cat1 studies remind the community of management control researchers that the design of MCS as such matters and influences the behavior in and of organizations.

On the other hand, however, doubts remain. First, within the ‘ideal fit’ interpretation of the contingency perspective in management control it is left open how innovating organizations can be described by an inevitably limited number of variables (contingency factors) considering (a) the bounded rationality of human cognition (Simon 1991) and given that organizations are (b) in essence open systems which consist by definition of more variables than one can assume (Berry et al. 2009; Thompson 2003). Innovation and the research on contingency factors for an ideal fit seem to be mutually exclusive. Contingency approaches obviously work for closed systems where a limited number of controllable variables and a predictable global and task environment enable a determined system (Berry et al. 2009; Thompson 2003). For the context of innovation described as a complex and uncertain environment the ‘ideal fit’ of contingency seems to be an inappropriate description of organizations as open systems.

Second, the question appears to which extent a specific organizational behavior is in fact inscribed in the design of MCS (for instance, result control enhances innovation). Designers of MCS suppose a certain impact of the MCS on behavior in its context of use (Fried 2010; Fried et al. 2013). However, Cat1 studies hardly take into consideration that the design and the actual use of MCS in the context of innovation differ and assume that a MCS constantly produces the same output. This way, the focus of Cat1 studies constitutes a deterministic view on managerial control and innovation. MCS are seen as an independent variable which are not subject of enactment or can ‘drift’ as Quattrone and Hopper (2001) describe. The following categorization 2 takes up however the analysis of the use mode and frames research in the field of management control and innovation from a voluntarist point of view.

In categorization 2 (Cat2), there were twelve empirical studies found focusing either on the use or explicitly on the distinction between the design and the use of managerial control instruments in the context of innovation (see Table 2).

Eight of the Cat2 studies (Artto et al. 2011; Bedford 2015; Bisbe and Malagueño 2009; Bisbe and Otley 2004; Chiesa et al. 2009; Davila et al. 2009a; Lopez-Valeiras et al. 2015; Pfister 2014) base their empirical efforts explicitly on the Levers of Control framework developed by Simons (1995). The Levers of Control framework distinguishes two different modes of MCS’ usage: diagnostic use and interactive use. It emphasizes that interactive control systems are measurement systems that are used to direct attention to changing information and their strategic assessment by senior managers. Interactive control systems shall support the formation of emergent strategies by intensifying opportunity-seeking and learning (Bisbe and Otley 2004). In contrast, diagnostic control systems are formal feedback systems and “the backbone of traditional management control” (Simons 1995, 59) to guarantee the implementation of the intended strategy. Diagnostic control systems are used to monitor organizational outcomes and to ensure the correction of deviations (Simons 1995). Thus, interactive MCS are considered to be perfect for the context of innovation.

Besides the eight Cat2 studies which put Simons (1995) central, Janssen et al. (2011), Davila (2000), Davila et al. (2009a), Kivisaari (1991) and Westling (2002) also pay attention to the use mode in the context of innovation and the “actual utilization of information obtained” (Janssen et al. 2011, 107) from MCS. In summary, Cat2 studies seem to either concentrate on the question which kind of innovation management mode, intensity or project characteristics suits the interactive use. Or, they investigate different elements of MCS and their interactive usage.

Bedford (2015), for instance, examines Simon’s framework across different modes of innovation known as exploration and exploitation. His main findings are that diagnostic and interactive control approaches independently enhance performance of organizations specialized in one of the innovation modes whereas they have complementary effects in ambidextrous organizations.

Bisbe and Malagueño (2009) find out that the choice of an interactive MCS depends on the company’s innovation management mode. The product innovation output as such is influenced by the innovation management mode and by interactive management accounting and control systems (MACS) featuring similar cognitive models and information priorities. Eventually, similarity in patterns between the innovation management mode and MACS tend to not apply for the innovation output.

The specific relationship between the interactive use of MCS and successful innovations was identified by Bisbe and Otley (2004) as an important research question. They found that an interactive use of MCS most likely favors innovation only in low-innovating firms, while the effect is the opposite in high-innovating firms. Moreover, Bisbe and Otley describe two ways how the interactive use of MCS can be realized: in a moderating or mediating style of use. A moderating style of use means that MCS serve as a forum to involve organizational members in an ongoing face-to-face dialogue to cope with non-routine, multi-disciplinary tasks in processes of NPD. When MCS are used in a mediating way they shall avoid dysfunctional extremes of innovative activities (too much innovation or too low innovation). According to their study, a positive impact on product innovation was found only for the moderating style of use.

Davila (2000) could show a strong diversity in the usage of MCS depending on the characteristics of product development projects in the medical devices industry. Characteristics of product development projects depend on their technology focus, market and project scope. According to Davila’s case studies, prototyping replaces MCS when uncertainty evolves from the technology focus. When uncertainty elicits from the project scope or the marked scope MCS are a suitable tool to lower uncertainty by proving information instead of measuring goal divergence. Beyond this, Davila (2000) points out in his findings that project managers rely mainly on non-financial measures and less on traditional accounting measures to reduce uncertainty.

Janssen et al. (2011) focus on the design and use of innovation metrics in product innovation processes and state that the coherence and adaptation of innovation metrics have no significant impact on their conceptual (thus, interactive) use. Top performers measure immaterial inputs, quality aspects, project progress, output and outcome more often than low performers. Kivisaari (1991) concentrates on dilemmas in control of divisional R&D activities in large companies. Reinforcing central control in R&D prerequisites high quality interaction between corporate and business management. In traditional practices and structures not supporting change, strict central control over R&D is difficult to realize.

Lopez-Valeiras et al. (2015) run a study similar to Bedford (2015) about different types of innovation with the result that interactive MCS are able to synchronize and adapt process innovations with a positive effect on financial performance. Interactive MCS have a positive influence on organizational innovations but these innovations do not have a provable effect on financial performance.

Beside the empirical objects of innovation metrics, budget and decentralization in the context of innovation Westling (2002) asks about the balance of control and innovation, in particular the importance of informal and formal face-to-face meetings in complex product developments. Informal meetings facilitate the perception of complex and ambiguous issues. Formal meetings elevate and enter the defined issues into a formal existence. In line with the arguments scrutinizing the contingency perspective of Cat1 studies, the readers’ attention should be on Westling (2002). Westling refers to the social-technical discourse on non-routine work. In doing so, he challenges the dichotomous distinction between formal and informal types of control by illustrating how meetings can be identified along a continuum that varies with the amount of discipline that is expected from a participating individual. Based on Goffman’s work on social order (Goffman 1963), Westling describes how some meetings “are elevated into systems of substantive rules, and become law. Others, equally binding when it comes to face-to-face interaction, govern the ceremonial behavior towards others that is encoded etiquette. The ultimate penalty for breaking the rules is harsh” (Westling 2002, 22). By using a different ontological orientation, this study is an important example for how the contest of existing categorizations can lead into a multifaceted research.

Chiesa et al. (2009) investigate the front-end of radical innovation projects. They found flexible and social control management systems as most suitable in the early phases of the innovation process whereas at later stages of the innovation process diagnostic control activities were preferably in use. Remarkably, Chiesa et al. (2009) discuss the impact of contingencies as the Cat1 studies do but understand contingency more in its initial interpretation. Instead of asking for the ideal fit of contingency factors and MCS for the best innovation performance possible they investigate four innovation projects and turn the logic of contingency by identifying those variables which distinguish the cases from each other. Four contingency variables (size, origin of innovation, organizational structure and governance approach) are specified and limit the comparability of the case. The interpretation of contingency takes here into account that there is no universal solution for management control in the context of innovation.

Like the other Cat2 studies, the study by Chiesa et al. (2009) can serve as an example for a voluntarist point of view in the discourse of management control and innovation. In contrast to the determinist perspective which believes that MCS directly shape behavior, Cat2 studies seem to agree with the voluntarist viewpoint that humans can also intend as well as influence the social effects of a MCS. MCS are seen as a dependent variable. Thought ahead, the consequence is that a focus on MCS as such is as important as the enactment of them. Consequently, as for instance Bisbe and Malagueño (2009) and Bisbe and Otley (2004) started to investigate, more empirical studies should pay attention to the ways how organizations make MCS suitable for the context of innovation.

In a nutshell, one can conclude that Cat2 studies fill the unmarked state of Cat1 studies and vice versa. There is, interestingly, only one Cat2 study that points out in regards to innovation metrics that the “findings not only show the importance of the relationship between the design of innovation metrics and their conceptual use, but also underline the value of empirically validating the design factors of formal performance management systems within product innovation” (Janssen et al. 2011, 109). This is an important point to consider that the design mode as well as the use mode of MCS matter in the context of innovation. The challenge of research in the field of management control and innovation becomes now obvious. It lies in the analysis of both structure (determinism) and action (voluntarism) as well as micro- (individual) and macro-focused (organizational) analysis without giving primacy to either.

As shown in Table 3, two Cat1 studies (Cardinal 2001; Richtnér and Åhlström 2010) and five Cat2 studies (Bisbe and Malagueño 2009; Bisbe and Otley 2004; Davila 2000; Pfister 2014; Westling 2002) can additionally be classified as being part of categorization 3 (Cat3). The article by Jørgensen and Messner (2009) is the only empirical study listed in Table 3 which was neither part of the Cat1 nor Cat2 studies.

Cat3 studies encompass the distinction between the enabling and constraining characteristics of MCS. On the one hand, Cat3 studies sort the existing literature according to the contributing as well as counterproductive role of MCS in the context of innovation. Usually, a traditional view is thereby introduced which “mainly characterize[s] MCS as hindering or, at most, being irrelevant in R&D settings” (Davila 2000, 384; cf. Bisbe and Otley 2004). In contrast to the traditional view, it is stated that a “growing number of studies have concluded that formal MACS [management accounting and control systems] may effectively contribute to the innovation effort” (Bisbe and Malagueño 2009, 371f.; cf. Bisbe and Otley 2004; Pfister 2014). Consequently, controlling of innovation in organizations is not seen any more as a paradox but as a necessity to balance the tension between efficiency and effectiveness. Likewise, managerial control can thereby enable or constrain.

Those Cat3 studies which are Cat1 studies likewise explain the enabling or constraining character of MCS based on the belief that MCS directly shape behavior in a certain way (Cardinal 2001; Richtnér and Åhlström 2010). Thus, the enabling or constraining character of a MCS mainly depends on its ideal fit. There are MCS and related elements which seem to be appropriate or inappropriate for the context of innovation. Conversely, those Cat3 which are Cat2 studies likewise may instead argue that “under some styles of use, formal MACS may be dynamic, flexible and adaptive to changing environments, whilst at the same time being stable enough to frame cognitive models and communication patterns” (Bisbe and Malagueño 2009, 372). Cat2 studies accept from a voluntarist perspective that humans influence as well as change the organizational effects of a MCS. Thus, human beings with different interests, knowledge and opportunities are responsible for the enabling or constraining character of a certain MCS in the context of innovation.

In Cat3 studies, so far the constraining or enabling effects of managerial control are mainly investigated in settings which are seen as innovative per se. Mainly, product development professionals, R&D directors, R&D project leaders in high-tech companies and innovation managers are asked to respond to surveys or are interviewed and observed in their daily routine. Yet, there is little research on how the same MCS is enacted in innovative contexts compared to non-innovative contexts. Research designs of this kind can open potential to understand the impact of the design of MCS in different contexts.

Furthermore, it is not entirely clear yet what empirical studies in management control and innovation try to achieve regarding the complexity and uncertainty of innovation processes. Complexity and uncertainty are by definition constitutional for innovation processes. However, it seems complexity and uncertainty shall be either reduced by identifying the ideal contingency fit or, on the contrary, MCS shall harness uncertainty and complexity of the innovation process as such.

Given the bounded rationality of human cognition, the handling of uncertainty and complexity becomes an essential question and needs to result in novel research questions for management control in the context of innovation. It begins by taking serious that no universal solutions to problems of control in the context of innovation are possible and may end by understanding the contingency perspective of organizations as a search for ‘quasi fits’ (Donaldson 2001, Volberda et al. 2012) of MCS in the context of innovation. The ‘quasi fit’ understanding of MCS takes into account that a control system in the context of innovation can never be a perfect solution in all situations at all times. They achieve their functionality not before practice, they are enacted and adapted there and undergo at best continuous modification to suit innovating activities—taking into account that a once assumed ideal fit of MCS for a certain task environment at a certain point in time can be prejudicial in the future. Jørgensen and Messner (2009) deliver an example and report from repair efforts (Adler and Borys 1996) of engineers in a company for automated analytical solutions to adjust management control systems to their needs, to understand the origin of production costs, to satisfy top managers’ expectations and finally to make a good case for their products. The example illustrates an implication of a ‘quasi fit’ understanding of MCS: it could fix MCS failures or to intervene if misuse of MCS becomes apparent.

The following section grasps at the implications for future research based on the marked and unmarked states of the three categorizations and based on the insights from the suggested ‘quasi-fit’ interpretation of contingency presented above.