Based on a spatial extension of an R&D investment model, this paper measures the macroeconomic impact of the French R&D policy mix on business R&D using regional data. Our measure takes into account not only the direct effect of policies but also indirect effects generated by the existence of spatial interaction between regions. Using a unique database containing information on the levels of various R&D policy instruments received by firms in French NUTS3 regions over the period 2001–2011, our estimates of a spatial Durbin model with structural breaks and fixed effects reveal the existence of a negative spatial dependence among R&D investments in regions. In this context, while a-spatial estimates would conclude that all instruments have a crowding-in effect, we show that national subsidies are the only instrument that is able to generate significant crowding-in effects. On the contrary, it seems that the design, size and spatial allocation of funds from the other instruments (tax credits, local subsidies, European subsidies) lead them to act (in the French context) as beggar-thy-neighbor policies.