Evaluating the indirect effects of cluster-based innovation policies: the case of the Technological Research Institutes in France

When it comes to evaluating the causal effect of public policies on corporate performance, most studies tend to focus exclusively on targeted firms, as if these firms have no relationship to the rest of the economy. Yet, public policies are highly likely to influence non-targeted firms indirectly due to the relationships they have with targeted firms. This paper aims to fill this gap by evaluating the indirect causal effect of a new French cluster-innovation policy on the financial and employment outcomes of non-targeted companies. To do so, it focuses on French Technological Research Institutes, which are science-industry collaborations based on technology platforms that bring together SMEs, large companies, universities, and public research bodies with the goal of accelerating the transfer of knowledge towards firms and generating spillovers (indirect effects) inside and outside the scheme. Based on the literature on spillover effects and agglomeration economies, it can be assumed that industry-specific spillovers tend to be spatially concentrated. By comparing a non-targeted firm located in the NUTS-3 regions within which the policy was implemented (referred to as "treated regions"), to a non-targeted firm outside of these "treated regions", using a difference-in-differences method with fixed effects applied to panel data (2008–2016) combined with a double matching at the NUTS-3 region and firm level, we find that non-targeted firms located in the "treated regions" significantly improve their financial performance (turnover, financial autonomy) compared to control firms located in the NUTS-3 control regions. The dynamics of employment outcomes are ambiguous. A negative significant effect is observed on the proportion of managers at the beginning of the policy and a positive significant effect is noted later, at the end of the period of observation. An analysis of the dynamics of the effects indicates that performance does not improve immediately after the policy, but later in time.