Abstract:  Firms′ vertical integration strategies to deal with the challenge of external environments may profoundly influence their innovation activities. Theoretically, vertical integration can stimulate firm innovation and the effect may be reversed if the extent of vertical integration goes too high. Empirically, existing studies on vertical integration and firm innovation fail to reach a consensus, and three different patterns have been found: positive relationship, negative relationship, and inverted-U-shaped relationship. One limitation of these studies is that they focus on a specific industry, thus weakening the generality of their findings. Additionally, short examination period, small sample size, and unidentified causality are some other common problems in these studies either abroad or at home.
Our paper overcomes the above problems by investigating all manufacturing listed firms on the A-share main board in China for a long period from 2000 to 2012, and we use generalized method of moments (GMM) to solve the potential endogeneity problem. In addition, due to the availability of patent citation data, domestic studies on firm innovation usually use the number of patents to measure the innovation output, which is subject to measurement errors as patent quality is highly heterogeneous. Thanks to a new patent data set, this paper is able to use patent citation counts to measure firms′ innovation output. Following the literature, we use the value-added method (VA/S) to measure the degree of vertical integration. Theoretically, vertical integration can promote firm innovation by eliminating opportunistic behaviors in knowledge transactions, coordinating R&D investment plans, and ensuring R&D exclusivity. However, with the increase in the extent of vertical integration, the coordination cost within the firm is rising and the strategic flexibility is declining. Accordingly, the effect of vertical integration on firm innovation can be reversed.
This paper thus focuses on such a nonlinear relationship between vertical integration and firm innovation. For high-tech firms, on the one hand, the synergy of knowledge associated with vertical integration makes them more adaptive to the industrial environment characterized by rapid technological innovation. On the other hand, the innovation disadvantages brought in by a high degree of vertical integration tend to slow them down and lose the capability of prompt reactions to technological changes. Therefore, we postulate that compared to non-high-tech firms, the strength and weakness of vertical integration on innovation are both more pronounced in high-tech firms.
State-owned enterprises (SOEs) are difficult to benefit from the innovation advantages of vertical integration as SOE administrators cannot fully benefit from but have to suffer from the cost of innovation. Moreover, due to the superiority of access to external resources, SOEs are immune to both advantages and disadvantages of vertical integration. We thus postulate that the relationship between vertical integration and firm innovation is more pronounced among non-SOEs than that among SOEs. With control for various firm characteristics, year fixed effects, industry fixed effects, and city fixed effects, our OLS estimation results show that the relationship between vertical integration and firm innovation is significantly inverted-U-shaped; that is, firm innovation is first increasing and then decreasing as the extent of vertical integration increases.
Our calculation reveals that on average the turning point of the inverted-U-shaped relationship appears when the VA/S value is equal to 0.21. This means that when the degree of vertical integration exceeds 0.21, its marginal impact on firm innovation turns from positive to negative. After solving the endogeneity problem by using the GMM method, this relationship persists. Additionally, consistent with our predictions, this inverted-U-shaped relationship is more pronounced among high-tech firms and among non-SOEs. Last, this inverted-U-shaped relationship only exists in the eastern and central regions but not in the western region.
In practice, our study is valuable both for firm administrators to choose the appropriate organization boundary and for government agents to make related policies. First, as shown, in the early stage of vertical integration, it benefits firm innovation, indicating some extent of vertical integration should be encouraged. However, given the existence of a turning point of this strategy, firms must carefully choose the degree of vertical integration to better cater to their innovation activities. For firms whose degree of vertical integration has exceeded the turning point, a more reasonable strategy is to shrink the boundary appropriately to better serve its long-term growth. Nevertheless, it should be noted that the turning point varies over industries and over firm ownership.
Second, admittedly, vertical integration can expand a firm′s boundary and enhance its market power. However, as we have found, an excessively high degree of vertical integration impedes firm innovation. It results in a group of large firms with inefficient innovation capabilities, thereby retarding the improvement of economic efficiency in the long-term. To this end, the government can implement non-discriminated R&D subsidies and encourage banks to provide low-interest loans to stimulate firm innovation and facilitate the transformation of firms′ strategy from size-driven to innovation-driven. It will not only help firms better balance between being bigger and being stronger, but also contribute to the long-term development of the whole economy.

Key words: vertical integration, firm innovation, patent output