86

This paper deals with the persistence of innovation in low-tech industries. Since innovation persistence is mainly explained in the literature by the sunk cost hypothesis, it is important to investigate the existence of such a phenomenon in this kind of industry. Using an innovation panel data set on French dairy industry for the period 2010–2017, we modeled two processes related to the persistence phenomenon: the innovation decision and its intensity. A sample selection model for count data accounting for individual effects and handling the initial conditions problem is estimated. The econometric results show that past innovation experience is an important factor for future innovation and thus confirm the hypothesis of true innovation persistence. In addition, the results highlight the importance of internal funds and knowledge accumulation (internal and external) in the ability of French dairy firms to innovate continuously over time.JEL Codes: O31, L20, C33, C39.