The effectiveness of R&D and external interaction for innovation: Insights from quantile regression

This paper utilises censored quantile regression techniques to analyse the impact of various forms of innovation inputs on the innovation output of a sample of Irish firms, using data from the Irish Community Innovation Survey 2008-2010. While there is a substantial literature on the drivers of innovation, there is a new and growing research interest in the application of quantile regression in the context of innovation. The advantage of quantile regression is that it moves beyond the typical assumption of variation around a mean, and allows for insights into the changing effectiveness of innovation inputs across the full innovation distribution. However, most papers treat innovation output as a continuous variable, when in fact it is more accurate to treat this variable as censored. Therefore, this paper applies a censored quantile regression estimator to evaluate the impact of innovation inputs on innovation output and to assess whether the effectiveness of these inputs varies, depending on how innovative a firm is. The key results of the paper are that both intramural and extramural R&D decline in effectiveness as firms become more innovative. We also find evidence that external networking is more important for less innovative firms.