The Choice of Interactive Control Systems under Different Innovation Management Modes

Abstract

This paper contributes to the recent levers of control (LOC) literature on the relationships between innovation and management accounting and control systems (MACS) by emphasising the importance of the choice by which individual MACS are selected for interactive use. Using survey data collected from 57 medium-sized Spanish firms, we find evidence supporting (1) the choice of individual MACS selected for interactive use is associated with a firm's innovation management mode (IMM), and (2) the level of product innovation output is influenced by whether or not IMM and interactive MACS feature similar cognitive models and whether the sophistication of the information contents provided by the interactive MACS responds to the priority needs perceived in the IMM. Our findings further indicate that similarity in patterns between IMM and MACS does not lead to a beneficial impact on the level of innovation outputs, suggesting instead that it may induce the replication of existing dysfunctional trends caused by innovation momentum.

Acknowledgements

The authors would like to thank the guest editors Antonio Davila and Daniel Oyon as well as the two extremely thorough anonymous referees for their extensive comments and suggestions. The second author is grateful to the Comissionat per a Universitats i Recerca del Departament d'Innovació, Universitats i Empresa de la Generalitat de Catalunya for its financial support.

Notes

Management Accounting and Control Systems (MACS) refer to the set of procedures and processes that managers use in order to provide valuable information in decision-making, planning, monitoring and evaluation and, ultimately, to ensure the achievement of their goals and the goals of their organisations. MACS comprise multiple formal and informal individual control systems that operate collectively and interdependently to constitute control packages (Otley, 1980, 1999; Chiapello, 1996; Merchant and Otley, 2007; Malmi and Brown, 2008).

Ambiguous findings may potentially be explained by differences in the conceptualisation of what constitutes an ICS. If the definitions of what constitutes an ICS include only a narrow subset of its theoretical properties, subsets may vary across studies, and ICS may be mistaken for other constructs such as mere intensive use or mere participative use (Bisbe et al., 2007).

According to the LOC framework, while any individual MACS can potentially be used diagnostically as well as interactively, individual MACS present in the control package of a given firm are used, except in rather exceptional circumstances, either diagnostically or interactively (Simons, 1995, pp. 103, 120; 2000, pp. 124, 208, italics in the original). Nevertheless, some authors (i.e. Tuomela, 2005; Widener, 2007) have pointed out that individual MACS can simultaneously be used both in an interactive and in a diagnostic manner. Based on LOC theory, potential explanations for this discrepancy include the following: (1) the object of analysis covers only one individual MACS, which makes it virtually impossible to comparatively detect diagnostic uses in some individual MACS and interactive uses in other individual MACS; (2) the nature of the MACS under analysis (e.g. performance measurement systems) is too broad (with many subsystems within, some used interactively, some diagnostically); and (3) the conceptualisation of the constitutive properties of ICS is not stringent enough (see Note 2).

The scope of a particular MACS refers to focus and quantification (we have ignored the time horizon sub-dimension in this study). Narrow (broad) scope MACS provide information that is internally focused and financial (related to both the internal and the external environment and including both financial and non-financial measurements). Aggregation refers to the degree to which data are processed and summarised to provide summated information. Hence, low (high) aggregation refers to systems that only provide basic raw, unprocessed data at lower-level units of analysis (systems that provide processed data that are aggregated in higher-level units of analysis). Finally, integration refers to the provision of information as to how the decisions made in one department or area may influence the performance of other departments, areas or activities (Chenhall and Morris, 1986; Bouwens and Abernethy, 2000).

Hereafter, we use ‘balanced scorecard’ (BSC) to refer to multi-perspective performance measurement systems in generic terms. Therefore, BSCs as defined here do not need to follow the exact procedure as suggested by Kaplan and Norton (1996, 2000). For the purposes of this study, summarized, multi-perspective sets of both financial and non-financial indicators that aim to capture the extent to which strategic objectives are being achieved, are labelled as BSC.

Organisational and managerial processes refer to the routines, operating principles and patterns of practice within a firm. Rather than modular or loosely-coupled entities that operate in isolation, organisational and managerial processes are better understood as operating in organisational configurations, each configuration representing a coherent multidimensional constellation of conceptually distinct routines, operating principles and patterns of practice (Meyer et al., 1993; Fiss, 2007).

While we extend R&D and R&D departments in Roussel et al.'s framework to innovation and innovative units, we acknowledge that innovation is not necessarily originated or developed within an R&D department or from R&D activities (Von Hippel, 1988; Escorsa and Valls, 2003). In fact, studies on technological innovation are experiencing a paradigm shift from R&D management to knowledge management. Even so, frameworks that link knowledge management, R&D management and innovation management (e.g. Nieto, 2002; Park and Kim, 2005) have often drawn upon the Roussel et al. (1991) typology. Some authors have proposed refinements or extensions of Roussel et al.'s typology (Rogers, 1996; Liyanage et al., 1999; Miller and Morris, 1999; Park and Kim, 2005), but consensus regarding these adaptations, and consequently their influence, is still limited.

Even though typologies of R&D modes and IMM can be interpreted as chronologies of generations associated to specific periods of time in an evolutionary process, they can be alternatively interpreted as maps of configurations that can coexist at a given moment in time across firms that follow different organisational patterns. For example, consistent with this latter approach, Roussel et al. (1991, p. 25) point out that ‘as we look on today's industrial scene, we see [the] three generations of R&D management in practice’. In this study, we adopt this latter approach, and therefore expect the different IMM to coexist contemporarily in the industrial setting.

In order to control for undesired effects related to relationships with headquarters, subsidiaries of multinational companies (MNC) with headquarters outside Spain were excluded, since most often these companies do not locate research centres and innovation activities in Spain. Even though there are some significant exceptions, most Spanish subsidiaries of MNC headquartered outside Spain engage in advanced manufacturing or commercial activities related to innovative advanced products developed abroad rather than developing their own innovations (Buesa and Molero, 1998; Hermosilla, 2001).

The firms in the resulting useable sample represent a variety of industries, including chemical and pharmaceutical (11 firms), textile (seven firms), food and beverages (six firms), manufacturing of mechanical equipment (six firms), metal manufacturing (six firms), manufacturing of electrical equipment (five firms), automobile supplies and parts (four firms) and miscellaneous (12 firms). Average sales are €62.8 million (minimum €18.63 million, maximum €165.28 million) and the average number of employees is 374 (minimum 204, maximum 800).

Hierarchical agglomerative techniques using Ward's method indicated similar percentage changes in the agglomeration coefficient across the relevant range of number of clusters. Visual inspection of the dendrograms did not provide either a clear-cut basis for selecting a number of clusters to be formed. Therefore, cluster centroids from the hierarchical results for 2, 3, 4 and 5 clusters were respectively used as initial seed points for the respective non-hierarchical clustering procedures. The four-cluster solution was selected since it was consistent with the results of the analysis and was theoretically interpretable. In order to test robustness, we engaged in a two-step sensitivity analysis. First, we ran six alternative hierarchical clustering procedures, excluding one of the six variables used to form the clusters, at a time. Based on the agglomeration coefficients and the dendrograms, in five out of six computations the four-cluster solution surfaced as appropriate and was theoretically interpretable. Second, we replicated the non-hierarchical procedures (pre-specifying four clusters) with subsamples resulting from a random split. We evaluated the congruence between the assignment of observations to clusters using the randomly split subsamples and the assignment to clusters in the full-sample solution, obtaining 74% coincidence in the assignment of observations to clusters (Z = 3.56; p < 0.001).

IMMs and ICS have different natures. While IMM represent a limited number of equilibrium states that are largely path-dependant (Roussel et al., 1991; Park and Kim, 2005), the interactive use of a particular MACS can be adjusted or fine-tuned incrementally in a continuous progression (Simons, 1995). Consequently, we ruled out a configuration approach to fit and the predicted association between IMM and ICS is specified as a Cartesian fit (Gerdin and Greve, 2004, 2008).