By using a sample of Chinese A-share manufacturing listed companies and considering firm heterogeneity, this paper applies generalized propensity score methods to explore whether government subsidy can stimulate firm R&D investment and quantifies the dynamic effect range. We find that the subsidy effect presents interval features when subsidy intensity is beyond a certain range, the substitution effect will beyond the incentive effect. At present, the direct subsidies have stimulated effect on manufacturing listed companies in China and the average subsidy intensity is far below the threshold. We also find that the subsidy effect and the optimal subsidy range present heterogeneity of ownership, firm size and industries.