Abstract:  Traditional foreign direct investment theory holds that only enterprises with high productivity and competitiveness will invest overseas. However, in recent years, the outward foreign direct investment of emerging economies has increased substantially, and the proportion of investment flows in global flows has increased year by year. From 2003 to 2015, China′s outward FDI reached an average annual growth rate of 38.8%, far exceeding the global average. At the same time, China′s innovation capability have also been rapidly improved. The National Innovation Index Report showed that China jumped from 38th place in 2000 to 18th place in 2015. The Global Innovation Index′s evaluation of China′s innovation capability was 43rd in the world in 2010, and entered the top 25 for the first time in 2016. Whether China′s outward FDI has technology-seeking motivation and whether China′s technological innovation capability have benefited from cross-border investment has aroused widespread concern in academia.Although many scholars have carried out a lot of tests on the existence of reverse technology spillover effect of outward FDI, no unified conclusion has been formed so far. Two notions of "significant technology spillover" and "no spillover effect" coexist. For the differences in conclusions, scholars have analyzed from the perspective of sample selection, technology measurement methods, regional characteristics, industry categories and so on. Surprisingly, from the literature we hold, there is currently no scholars to explain it from the perspective of dual margin of investment. In fact, the divergent conclusions may also be due to the neglect of the structural characteristics of outward FDI. Thus, we attempt to decompose the new outward FDI into two investment margins, defining the investment of newly increased overseas enterprises as the extensive margin and the investment of existing ones as the intensive margin, and then explore and compare the reverse technology spillover effects of two different investment margins.Based on the international technology spillover theory framework of Bodman and Le(2013), we extend the R&D spillover calculation model of LP(2001), and decompose the total technology spillover effect obtained by outward FDI(ODI) into previous ODI spillover effect and new ODI spillover effect. The latter is further decomposed into the intensive margin′s and extensive margin′s. Additionally, we identify and elaborate the impact mechanisms of outward FDI with different stages and different ways on domestic technology, and use data from 30 provinces in China from 2007 to 2014 for empirical tests.The results show that, firstly, China′s outward FDI has a positive reverse technology spillover effect on the whole. Secondly, the foreign R&D capital obtained by the previous ODI and the new ODI can both enhance technological innovation ability of our country, but the former is stronger. The possible reasons are as follows: on the one hand, it takes time for overseas subsidiaries or branches to develop new technology independently or cooperatively with the host countries; on the other hand, parent companies will also need time to absorb and digest the newly acquired technology that overseas subsidiaries or branches return back. Thirdly, only the extensive margin from the new ODI can significantly help promote China′s technology innovation while the intensive margin cannot. The possible reasons are as follows: the intensive margin makes excess profits by expanding production scale and invest some in the domestic R&D, but it also crowds out domestic investment. These two impacts are opposite on domestic technology innovation. In addition, although the extensive margin will also squeeze domestic investment, it can quickly obtain the advanced technology by cross-border mergers and acquisitions from the acquired enterprises. And the strengthening awareness of competition will promote multinational enterprises to increase R&D efforts as well.The above conclusions have important policy implications. First of all, our government should encourage competent enterprises to "go global". It is beneficial for technology-seeking multinational enterprises to establish R&D institutions in developed countries with better R&D environment. They can make full use of the human capital and innovation environment there, and then return as many R&D outcomes as possible from overseas subsidiaries or branches to the parent companies, which will finally enhance the technological innovation capability of our country. For multinational enterprises, they should optimize investment methods, allocate funds rationally, and establish more overseas subsidiaries or branches. That allows them to make links with more host countries and thus have more opportunities to acquire advanced technology from different countries which is significant to technological progress promoting. At the same time, decentralized investment can also reduce the investment risk.There are three main contributions in this paper. (1) Theoretical mechanism. Previous studies on the relationship between outward FDI and domestic technology are mainly based on the aggregate analysis, that is, focusing only on the overall reverse technology spillover of outward FDI and ignoring the impact of its structural characteristics. This paper makes up for this disadvantage by identifying and systematically elaborating the impact mechanisms of previous ODI and new ODI including the intensive and extensive margins on domestic technology, which will help deepen our understanding of the inherent mechanism of reverse technology spillover of outward FDI. (2) Extension of LP model. We insist that the technology protection of host countries will hinder the technology acquisition of investor countries, so that we introduce the factor of technology protection into the traditional LP model. Besides, considering the structural characteristics of outward FDI, the LP model is further extended to measure the reverse technology spillover effects of ODI at different stages and in different ways. Last but not least, we also provide the selection criteria for the destination countries of technology-acquired ODI in LP model. (3) Empirical findings. We build an econometric model to test the impact of outward FDI on domestic technology, and provide empirical evidence from China. We use generalized method of moments (GMM) to correct the endogenous problem in the model. It is found that the extensive investment margin has a significant promotion effect on domestic technology, but the intensive margin has not. The introduction of the dual margin perspective allows us to explain the coexistence of the two views above under a unified analytical framework, which is a new discovery.

Key words: outward FDI, intensive margin, extensive margin, reverse technology spillover