By using the stochastic frontier analysis model, based on the manufacturing panel data of provincial level after the first round of “shortage of migrant workers”, and by considering the time lag of innovation input and output, this paper calculates the technical efficiency of Chinese 30 provinces after the first round of “shortage of migrant workers”, and analyzes the influence of labor cost on technology innovation efficiency. The results show that: firstly, after 2004, the first round of "shortage of migrant workers", the labor cost in the provinces of China has been increased rapidly, which constitutes a positive effect on manufacturing innovation efficiency objectively, but it has obvious time lag, time lag is about three years. Secondly, on the whole, the average innovation efficiency of China's manufacturing industry is only about 0.4987, it means that the efficiency loss is about 50%, and there are significant differences in regional innovation efficiency, there exists the imbalanced contradiction of regional development. Thirdly, the interaction term of human capital, ownership and labor costs also have obvious indirect impact on enterprise innovation efficiency, there are significant differences between different regions while the influence of labor cost and firm scale on innovation efficiency is not significant. In addition, the scale of foreign investment, the market competition degree also have very different influence on innovation efficiency between different regions.