How control systems influence product innovation processes: examining the role of entrepreneurial orientation

Abstract

This paper yields insights into the channels through which management accounting and control systems (MACS) exert an influence on product innovation by examining the extent to which different forms of control (i.e. value systems (VS), diagnostic control systems and interactive control systems (ICS)) are directly associated with the distinct phases of innovation processes. Using survey data collected from 118 medium and large Spanish companies, we find that (1) VS and ICS have significant main effects on the creativity, co-ordination and knowledge integration, and filtering (sub-)phases of innovation processes and (2) the significance and direction of these influences vary depending on the entrepreneurial orientation (EO) of firms. By highlighting the relevance of EO in shaping the influence of MACS on product innovation processes, this study calls for caution in generalising the expected effects of MACS on innovation.

Keywords:

• management accounting
• control systems
• innovation
• creativity
• conversion ability

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The relationships between (sub-)phases of innovation processes and innovation outputs have been extensively examined and tested in previous literature (e.g. Parthasarthy and Hammond 2002, Gomes et al. 2003, Hirunyawipada et al. 2010, Baron and Tang 2011). These relationships, which are not covered in the current paper, highlight the relevance of creativity and the sub-phases of conversion ability under consideration in order for innovation efforts to be successful.

2. ‘Phases’ and ‘sub-phases’ of innovation processes tend to be conceived and deployed sequentially (Parthasarthy and Hammond 2002, Tidd and Bessant 2009). However, given the often interwoven and iterative nature of innovation processes (Amabile 1996, Cooper 2008, Jorgensen and Messner 2009, Revellino and Mouritsen 2009), the (sub-)phases can also be construed as distinct modes or activities in which firms engage. Still, creativity and conversion ability conceptually represent two distinct sets of mental, motivational and social mechanisms – and are consequently considered to be meaningful categories for research (Trott 2008, Davila et al. 2009, Hennessey and Amabile 2010, Adler and Chen 2011).

3. Even if idea generation is central to the creativity phase, conversion ability may involve some generation of new ideas that help formulate alternatives in turning previously generated concepts into marketable products (Revellino and Mouritsen 2009). Similarly, the creativity phase may involve structures and processes for co-ordinating and integrating ideas and knowledge in collective action as well as screening devices for helping organisational members assess the value of possible solutions before concepts are generated – and so trigger self-restraint (Amabile 1996, Cooper 2008). As defined in this paper, the generation of ideas after concept generation and during project implementation is not included in the scope of the creativity phase. Analogously, the CKI as well as the filtering sub-phases here relates only to conversion ability, which extends from selection and prioritisation by senior managers of previously generated concepts to launch and commercialisation (Sheremata 2000, Chandy et al. 2006).

4. Typologies of controls are rarely exhaustive. In our case, for example, the proposed typology does not encompass administrative controls or ways of effecting personnel controls such as selection, placement, training or job design.

5. Ongoing debates around the conceptualisation of EO refer to whether EO should be conceptualised from a dispositional or behavioural point of view (Voss et al. 2005, Covin and Lumpkin 2011), what are the characteristics to include under the EO construct (Lumpkin and Dess 1996, Hult et al. 2007) and whether EO should be conceptualised as a latent construct or as an index (George 2011, George and Marino 2011).

6. The argumentation that follows and that leads to our hypotheses assumes unidirectional causality from forms of control to innovation phases. We build on the well-established stream of prior theoretical developments and field evidence that have examined the influence of forms of control on innovation-related variables using unidirectional models (e.g. Henri 2006, Chiesa et al. 2009, Adler and Chen 2011). The implicit assumption that we borrow from these studies is that the causal interval of a potential relation (innovation-related variables → forms of control) is longer than the causal interval of the relation (forms of control → innovation-related variables) and that forms of control precede in time the effects on innovation phases (Luft and Shields 2003, pp. 193–5). While one cannot rule out that innovation-related variables affect forms of control, this potential causal direction has been less examined in prior literature and it is not contemplated here.

7. We used SABI 2008. Surveys were administered between March and June 2010. All firms that were object of the survey with one single exception were unlisted in 2008 and remained so at least until end of 2010.

8. We excluded eight firms for at least one of the following reasons: the firm had not been involved in new product development in the last three years, the respondent was not a senior manager, or the respondent had been in the company for less than one year. Fourteen of the returned questionnaires had missing items. After Little's Missing Completely At Random test, imputation of missing data was computed through Maximum Likelihood Estimates using the expectation maximisation algorithm.

9. The original instrument for creativity developed by Lee and Choi (2003) with five questions was reduced to four questions as the wording of two of them overlapped in the Spanish version of the questionnaire. The original CKI instrument contained eight items. However, one of the eight items was dropped from the final construct because of a high volume of missing data (only item with >10% of missing data), which was considered an indication of ambiguity or lack of clarity in its wording.

10. For models reported in Tables 7–9, a two-stage approach was used to estimate the parameters of second-order formative constructs in our hierarchical latent variable models. We first applied the repeated indicators approach to obtain the latent variable scores for the lower order components. These scores were saved and subsequently used in a second stage in which they served as manifest variables in the measurement model of the higher order component (Henseler and Chin 2010).

11. In order to test the significance of difference in corresponding path coefficients across sub-samples, we use the following t-statistic: where m is the sample size of the first sub-sample, n the sample size of the second sub-sample, with (m+n−2) degrees of freedom.

12. Through this second angle, our research design provides insights on the interrelationship between MACS across phases, that is, how the main effects in one phase are integrated with the effects of other MACS in other phases. Nevertheless, the cross-sectional nature of our research does not allow us to take an in-depth longitudinal view of the interplay between MACS across phases.

13. Further tensions are likely to be introduced through interactions between control systems within each (sub-)phase (Henri 2006, Mundy 2010). These sources of interplay have not been investigated in this study.