تصمیم نوآوری: تجزیه و تحلیل اقتصادی

Studies of the determinants and effects of innovation commonly make an assumption about the way in which firms make the decision to innovate, but rarely test this assumption. Using a panel of Irish manufacturing firms we test the performance of two alternative models of the innovation decision, and find that a two-stage model (the firm decides whether to innovate, then whether to perform product only, process only or both) outperforms a one-stage, simultaneous model. We also find that external knowledge sourcing affects the innovation decision and the type of innovation undertaken in a way not previously recognised in the literature.

How do firms make the decision to innovate and how do they choose between product and process innovation, or doing both? Are these decisions sequential or simultaneous? Despite the substantial body of research on the determinants and effects of innovation, surprisingly little is known about the decision-making process of the innovation decision. Since the early observations by Abernathy and Utterback (1982) that the distribution of firms’ resources between product and process innovation depends on the market phase of the relevant technology, there has been an advance in the theoretical literature which examines the product/process mix (Klepper, 1996; Yin and Zuscovitch, 1998; Rosencranz, 2003). But this has not been reflected in the empirical work on the microeconomics of innovation, where it is still common to proceed as if firms concentrate wholly on product innovation. This is most evident in the recent empirical literature examining the ‘knowledge production function’, which invariably either ignores process innovation or treats it as a subsidiary issue with no explicit consideration of how firms make the relevant decisions (e.g. Crépon et al, 1998; Lööf and Heshmati, 2001 and Lööf and Heshmati, 2002; Klomp and Van Leeuwen, 2001). Virtually the only empirical study that explicitly deals with the choice between product, process or both is Cabagnols and Le Bas (2002), which examines the determinants of 12,779 French innovating firms’ product innovation and process innovation during 1987–1992. Cabagnols and Le Bas (2002) analyse how French innovating firms’ choices among product innovation, process innovation and both are determined. However, they investigate innovating firms only, which excludes a crucial part of the innovation decision: whether the firm decides to innovate at all. This begs the question of whether the innovation decision is a one-off or a two-stage process. The answer to this question is important both conceptually and econometrically: if it is a two-stage decision we need a better understanding of the process in order to distinguish and differentiate the determinants of firms’ innovation strategies in each stage. On the other hand, if it is a one-off decision, excluding non-innovating as one potential choice unquestionably violates the exhaustiveness assumption of multiple-choice model, such as the multinomial logit model used by Cabagnols and Le Bas (2002), suggesting that more caution should be exercised in future research. This paper directly addresses the issue of the nature of the innovation decision from an economic perspective. More specifically, we consider two competing models of firms’ innovation decision process involving the choice of undertaking: (1) no innovation; (2) product innovation; (3) process innovation; and (4) both product and process innovation, and test which is statistically more reliable. The purpose of the analysis is therefore to examine the economics behind the innovation decisions actually made by firms, rather than to offer advice to management on how the innovation process should be structured. Throughout the analysis we use a panel dataset of Irish manufacturing firms spanning the period 1994–2002, which allows us to capture the information on firms’ business environment, knowledge sourcing activities, absorptive capacities and other firm characteristics.

Studies of the determinants and effects of innovation commonly make an assumption about the way in which firms make the decision to innovate, but rarely test this assumption. Using a panel of Irish manufacturing firms we test the performance of two alternative models of the innovation decision, and find that a two-stage model (firm decides whether to innovate, then whether to perform product-only, process-only or both) outperforms a one-stage, simultaneous model. This has both conceptual and econometric dimensions: as we point out in the introduction, omitting the non-innovate choice in a one-stage model (e.g. Cabagnols and Le Bas, 2002) violates the exhaustiveness assumption of multiple-choice models. We also find that external knowledge sourcing affects the innovation decision and the type of innovation undertaken, but in subtle ways not previously recognised in the literature, and that absorptive capacity and resource considerations have a systematic impact. The finding that the two-stage model is statistically more reliable is clearly specific to the sample studied. However, since other empirical evidence on the subject is lacking, this has implications for the way in which studies of innovation are set up, and suggests that a two-stage model should be considered in economic research on the determinants and effects of innovation. Consider, for example, the increasingly large body of research relating internal and external knowledge sourcing to the innovation performance of firms, and relating this in turn to the productivity and growth of enterprises (e.g. Crépon et al., 1998; Lööf and Heshmati, 2001 and Lööf and Heshmati, 2002; Klomp and Van Leeuwen, 2001). Although such research is increasingly sophisticated in terms of econometrics, there is rarely any explicit consideration given to the nature of the innovation decision itself. The results discussed above suggest that a more detailed consideration of this part of the ‘knowledge production function’ would be worthwhile. For example, a clearer understanding of how the innovation decision evolves might lead to further insights into issues such as nature of the interaction between internal and external sources of knowledge: does the impact of different knowledge sources occurs at the same stage of the innovation decision process, as is often assumed, or might some knowledge sources affect principally the innovate/not innovate decision while others influence the type of innovation undertaken? While this paper examines the innovation decision from an economic point of view, further research of this type would not only further understanding about the early stages of the knowledge production function, but also might help provide policymakers managerial implications on which to base policy interventions designed to enhance innovation and competitiveness.