Purpose of the paper
This paper offers an analysis of the adoption of three new process technologies (numerically controlled machines, robotics and computer aided design) at the firm level (inter-firm diffusion). Our intention is to contribute to the literature in three ways. First, we expect to narrow the gap detected between theory developments and empirical research on the literature on diffusion (Karshenas and Stoneman, 1993). Second, although we control for the importance of market characteristics, such as concentration, we pay especial attention on analysing the impact of firm features at determining the decision to adopt. Finally, we also assess the interactions that may arise between different technologies and that may contribute to the explanation of the adoption decision. More precisely, our analysis considers the idea that the decision to invest in a new technology is affected by the simultaneous use of complementary or substitute technologies by the firm (Stoneman and Kwon, 1994).
Theoretical framework
The inter-firm diffusion of new technologies proceeds as each firm takes the decision to invest into a new technology. The literature has contended that this investment is motivated by four main types of factors: rank, stock, order and epidemic effects (Karshenas and Stoneman, 1993). Recent research on the diffusion of innovations has added three additional elements to the analysis of the decision to adopt a new technology. On the one hand, alternative explanations for the effect of firm size Astebro (2002) and (Fuentelsaz, Gómez and Polo, 2003) have been proposed. On the other hand, with the advent of the resource based view of strategy some papers have refined the idea of a pure epidemic effect by relating the amount of information available on a new technology to a firm’s capability to interpret and respond to it (Cohen and Levinthal, 1990; Srinivasan, Lilien and Rangaswamy, 2002 and Zahra and George, 2002). Additionally, the fact that the implementation of new technologies may be influenced by managers and end-users (Leonard-Barton and Deschamps, 1988) widens the array of firm specific variables explaining adoption. A third feature is the consideration of interdependencies among technologies when explaining the decision to adopt (Stoneman and Kwon 1994).
In this paper we take into account these considerations at the time of designing and measuring our hypothesis and when developing our estimations. Although we control for market characteristics we make special emphasis on the importance of firm specific characteristics influencing the decision to use a new technology. We attempt to distinguish between size and financial structure effects and consider the absorptive capacity concept at determining the adoption behaviour of firms. Importantly, the availability of longitudinal data allows us to consider additional firm specific and non-observed variables involved in the decision through the estimation of a fixed effects model.
Method
The data set used for this study is the Survey of Industrial Strategic Behaviour carried out by the SEPI Foundation. Although the survey is administered to firms annually from 1990, questions related to the adoption behaviour of firms are only included in the questionnaire every four years. Concretely, the design of the questionnaire leaves us with information that refers to years 1994, 1998 and 2002. This selection of the sample and the rejection of cases with missing data in the basic variables results in 4594 observations that will be used in the analysis. To contrast our hypothesis we will use a logit model.
Results
The estimations confirm the role of firm size, absorptive capacity and market concentration at explaining the adoption behaviour of manufacturing firms in Spain. As it is also evident in previous research, our results confirm that larger firms are more likely adopters of new process technologies. Although the availability of data did not allow us to search for alternative explanations that could explain this association (Astebro, 2002), we were able to independently assess the role of financial structure. This variable only contributed to the explanation of diffusion in the case of robotics, not being relevant for the other two technologies. The capacity of a firm to absorb new technology did play a significant role at explaining adoption patterns. Those firms investing a relatively larger amount of money in R&D activities were the ones also showing a larger likelihood of having adopted the new process technologies. Finally, market concentration did have a negative impact on the introduction of numerically controlled machines, robotics and computer aided design
Interestingly, the characteristics of our sample have allowed us to test for the existence of interdependencies and non-observable fixed effects at explaining diffusion. In the first case, our results provide clear evidence underpinning the hypothesis that the adoption of one of the technologies analysed here is positively related to the introduction of the other two. This result is in line with the one previously obtained by Stoneman and Kwon (1994). In the second, our results only show evidence of misspecification in the estimation that refers to the use of CAD in manufacturing. In other words, firm specific effects are not at operation in the case of the adoption of CNC and robotics.