Efficient industrial policy for innovation: standing on the shoulders of hidden giants

Research and development is underprovided whenever it creates knowledge spillovers that drive a wedge between its total and private economic returns. Heterogeneity in the intensity of this market failure across technological areas provides an argument to vertically target public support for R&D. This paper examines potential welfare gains of such vertical industrial policy for innovation. It develops measures of private and spillover value of patented innovations using global data on patents and their citations. Our new method identifies a large number 'Hidden Giants' - i.e. innovations scoring higher on our new spillover measure than on the traditional forward citation count measure - which are shown to be particularly prevalent among patents applied for by universities. The estimated distributions of private values by technology area are then used to parameterize a structural model of innovation. The model permits estimation of the marginal returns to technology-area-specific subsidies that reduce innovators' R&D costs. Marginal returns are high when knowledge spillovers in the technology area are valuable, when private innovation costs are low, and when private values in a technology sector are densely distributed around the private cost. The results show large variation in the marginal returns to subsidy and suggest that targeted industrial policy would have helped mitigate underprovision of R&D over the time period studied. Variation in the extent to which knowledge spillovers are internalized within countries also makes a compelling case for supranational policy coordination, especially among smaller countries.