The nudging effect of technology reward on R&D input: empirical evidence from machine learning

Purpose Traditional policy interventions, typically involving economic incentives or mandatory regulations, are often constrained by fiscal limitations and resource misallocation. This study utilises a technology award system, namely the China Patent Award (CPA), to assess whether such awards are associated with low-cost nudge effects on corporate innovation. Design/methodology/approach Empirically, we manually match award-winning patent information with publicly listed company data. Based on counterfactual frameworks, double machine learning and causal forest model are employed for evaluation. Findings The results suggest a positive effect of CPA on firms' research and development (R&D) input, exhibiting a nudge-like pattern. This is evident in two aspects: first, even when considering income from intellectual property utilisation, the resource effect is much smaller than the certification effect, satisfying the feature of shaping behaviour without markedly altering economic incentives. Second, the estimated effect operates partly through alleviating financing constraints and increasing revenue, without raising government subsidies. Unlike financial instruments that may create dependency, the CPA appears to support a more self-reliant pattern of corporate innovation activities, aligning with the principles of libertarian paternalism. Additionally, we provide the distribution of individual treatment effects (ITEs), revealing significant associations between ITEs and firm characteristics. Notably, firms with lighter asset bases but larger workforces, and those in non-high-tech or polluting industries, tend to experience stronger policy effects. Originality/value This study examines the effect of CPA on firms' R&D behaviour through a nudging framework.