Using firm-level panel data on fiscal subsidies and R&D investment, this study employs a propensity score matching method to obtain a matched dataset. After a parametric estimation of the impact of different categories of subsidies (purchase subsides or R&D subsides) on innovation, the author finds that purchase subsidies for EVs have a significantly larger positive effect on firms’ R&D investment than R&D subsidies for non-EVs. The mechanism of the impact of different categories of subsidies on innovation has also been investigated. Generally, the impact of EV subsidies on firms’ intangibles is found to be significantly weaker than R&D subsidies for non-EVs whereas their effect on labor cost is significantly more pronounced. Results indicate that, purchase subsidies for EVs stimulate more R&D investment; however, the incentive of firms’ seeking demand-side subsidies affects their allocation of private R&D expenditure (research activities or development activities), which results in lower level of basic research activities in terms of intangible assets investment.