The Influence of Alliance Innovation Network Structure upon Enterprise Innovation : A Case Study of China ’ s Energy-Saving and Environment-protection Industry

The Energy-saving and environment-protection industry, an important strategic and emerging industry in China, will develop into a pillar industry. In view of global climate change, environmental pollution, resource depletion and the defects and deficiencies in traditional technology, technology and product innovation constitute the lifeline of energy-saving and environment-protection industry. The alliance network of enterprises will influence, stimulate, and regulate enterprise innovation greatly. A comprehensive analysis of alliance data of China's energy-saving and environment-protection industry from 2000 to 2013 by using Ucinet software can reveal the network structure parameters such as degree, clique number, average path length, clustering coefficient, and betweenness centrality, which reflects different types of enterprise networks and different positions of enterprises in different types of networks. A negative regression analysis of enterprise patent data and network structure parameters by using Stata software can make some conclusions that the influences of network characteristics on enterprise innovation reach the maximum in the second year of the window period end, that innovation accumulation, clustering coefficient, betweenness centrality are related to the enterprise innovation, that clique number, network density are negatively related to the enterprise innovation, and that there is an inverted U relationship between average path length and enterprise innovation. It is suggested to increase the accumulation level of innovation, appropriately control the network density, reduce the average path length, improve the betweenness centrality and clustering coefficient of enterprises, so as to improve the overall innovation level.


Research Background
Innovation is the main way for enterprises and countries to gain competitiveness, and the main means for enterprises to obtain excess profits.Many enterprises lack innovation and core technology, and are under the control of other countries in these respects.Many Chinese enterprises with foreign businesses are not in the key link of the global industry chain, and their product pricing ability is weak and the added value of production is low.As a result, they are subject to other enterprises.In recent years, scholars have begun to explore the influencing factors of innovation in various aspects, among which the enterprise alliance cooperation mechanism has also attracted attention.

Innovation Costs and Risks Require Enterprise Alliance Cooperation
With the continuous progress of technology, the difficulty and risk of innovation further increase.For enterprises, innovation based on their internal resources has been unable to meet the needs of market competition.Enterprises gain competitive advantage by their technology and innovation.It requires a lot of money, manpower and time.At the same time, enterprises bear huge market risks.Innovation costs and risks cause enterprises to cooperate in alliances.With the deepening of postwar economic globalization, more and more enterprises begin to form strategic alliances to share resources and increase mutual innovation advantages.Enterprise strategic alliances help to enhance the core competitiveness of enterprises, economies of scale and scope, reduce operational risks and prevent excessive competition (Zhang, 2001).Strategic alliances can improve the competitiveness of enterprises in the areas of institutional and organizational (Zhou, 2000).The knowledge alliance helps enterprises not only acquire explicit knowledge, but also learn tacit knowledge and create new capabilities (Shen Zuzhi, 2003).It can update or create its core competence through strategic management .

The Innovation Behavior of Enterprises Is Stimulated, Influenced and Restricted by the Alliance Innovation Network
In the process of innovation, various kinds of alliances and cooperation, formed by enterprises, is called alliance innovation network.From a global perspective, a certain amount of enterprise innovation alliance forms a sparse alliance innovation network, which stimulates, influences and restricts the innovative behavior of enterprises.More and more scholars have proved that network organization is beneficial to enterprise technological innovation.Network practice, management orientation, external knowledge and network embeddedness have influence on innovation performance (Chen, 2016(Chen, , 2017)).There is a main effect between the local and super local double embeddedness of cluster enterprises and the promotion of innovation capability, network strength, persistence and network diversity have a significant impact on the innovation capability of cluster enterprises (Wei, 2014(Wei, , 2016)).The strength of alliance relations has an inverted U effect on corporate innovation performance, and the quality of alliance relationship has a positive impact on enterprise innovation performance (Xie, 2016(Xie, , 2017)).In the process of technological innovation in the industry alliance, relational embeddedness brings benefits, in which weak relationship has a positive impact on technological innovation, while strong ties have a U effect on technological innovation (Wang, 2017).The alliance network characteristics of small and micro technology enterprises play a significant role in promoting innovation performance (Zhang, 2016).Researching and developing cooperation can significantly promote the enterprise innovation, and the close relationship of alliance network is an important means to promote radical innovation (Gao, 2016).

Research Methods: Social Network Analysis
From the earliest interpersonal network to the later social network research, multidisciplinary integration formed a complete social network theory, methods and techniques.Granovetter (1984) points out that the theoretical hypothesis of traditional economics, pure economic relations, does not exist in real life, and that economic activities can not bypass social relations.He opened up a new field of economic sociology and established network analysis methods.In recent years, with the development of computer technology and of network analysis software, the analysis of the social network with highly complex structure has gradually become an important research object and methods.At the same time, social network and whole network analysis are gradually permeating into the field of economic management.This trend has been around the world for more than 10 years.In recent years, scholars have begun to try to build an alliance network, and they extends the interrelated analysis layers from isolated individual to an interrelated network.Zhao Yan et al. think that the network centrality of enterprises has an lagged positive effect on enterprise innovation (Zhao, 2017); the small world of strategic alliance network positively influences innovation performance (Zhao, 2013); the non redundant links and aggregation of enterprise alliance network nodes have a potential impact on enterprise innovation (Zhao, 2013); the structure holes of the alliance network can significantly promote the innovation performance of enterprises (Zhao, 2012); faction and knowledge flow have positive influence on alliance network innovation performance (Zhao, 2016).

Research Purpose and Significance
Energy-saving and environment-protection industry has become one of the strategic emerging industries to accelerate the cultivation and development of China.A Development Plan of the National Strategic Emerging Industry From 2016 to 2020 proposed that we should focus on the construction of ecological civilization and climate change, comprehensively promote energy-efficient and advanced environment-protection and resource-recycling industry system construction, and promote energy-saving environmental protection industry to become a pillar industry.Energy-saving and environment-protection enterprise's research and developing high-tech equipment should be in accordance with the specific conditions of environmental pollution and our national requirements for environmental protection, energy consumption, and resource consumption (Note 1), which requires a large number of innovative behavior.At the same time, there are many strategic alliances between enterprises in order to promote technological cooperation and accelerate innovation.At present, there is no literature published about the relationship between alliance structure and enterprise innovation behavior of China's energy-saving and environment-protection industry.It is necessary for us to do pioneering research on it.Through this study, we should determine whether there is a connection between these two and their specific forms (mathematical expressions).On this basis, we can draw some useful inferences and suggestions for the government, industry and enterprises.

Enterprise Innovation Function
Enterprise's innovation ability can be restricted by many factors.We simply divide these factors into two aspects: network factors (Song, 2014) and other factors.Enterprise innovation function can be put forward as follows: "Y" represents the total amount of innovation, and "i" represents the ith company.The network factors include the following indexes: degree, betweenness centrality, local efficiency, network density, clustering coefficient, average path length, cliques numbers, core values, etc.Other factors include individual factors such as enterprise scale, enterprise nature, industry characteristics, enterprise culture, enterprise history, Researching and developing investment intensity, management characteristics, enterprise strategy choice and macro factors, such as international politics, economy, national law, social development, government policy and so on.
This paper focuses on the influence of alliance network on enterprise innovation, and we use the "accumulation of innovation" variables to measure all other factors as a whole.In later calculation, we can find that "accumulation of innovation" is an important explanation for enterprise innovation.At the same time, a number of network factors also make partial explanations for enterprise innovation.
When individual factors of enterprise and network factors affect the innovation linearly, the function can be written in the following form: Considering the possible cross effects between factors:

The Nature of the Single Factor Innovation Function
This paper next analyzes the nature of f j (X ji ) (the influence function of network factors on enterprise innovation) when j ≠ 0.
Network factors affect enterprise innovation mainly through two aspects which are the availability and convenience of innovation resource acquisition (AC), and the management cost of alliance relationship (Cost) (Zhong Shuhua, 1998).
Next, we analyze the property of this function by using the simplest network structure parameter (degree).
The first part of Equation ( 4) plays a positive role in promoting enterprise innovation.With the increase of alliance relationship, the status, importance and centrality of enterprises in the network are increasing, and the availability and convenience of innovation resources are also increasing.It is obvious that the AC function goes through the origin and rises monotonically.When the enterprise has only a few alliances, the enterprise is on the edge of the alliance network, and the information is limited by the core enterprises, and lacks the right to speak.The benefits of the alliance relationship are not easy to show.With the increase of alliances, enterprises are gradually approaching the central position of the network, and they can get more information and speak more right.According to economics theory, when the resource inputs began to gradually increase from zero, it initially manifests as economies of scale (increasing marginal revenue).After reaching a certain critical value, it manifests as economies of scale (decreasing marginal revenue).It is assumed that the slope of the AC function increases first and then decreases, and eventually approaches zero.That is to say, with the increase of the number of alliances, the unit income first rises gradually, then decreases gradually until it disappears.The second part of equation ( 4) has negative influence on enterprise innovation, and its absolute value increases with the increase of alliance relationship.It is obvious that the Cost function is also monotonically increasing through the origin, and its slope has no obvious tendency to change.We assume that the slope is fixed (Jin, 2005;Fan, 2003;Gu, 2001).
Obviously, when degree reaches the maximum theoretical value of Dmax (enterprise sets up alliances with all other enterprises), the Cost function value should be greater than that of AC function, and obviously no enterprise the maxim alliances F

Note. The
The abov agree wit be cross e

The Pr
We choos terms.

Data
The   er than the e negative ND was highly correlated with CN.CV, D, and E were highly correlated with each other.In the latter model estimation process, variables E, ND, and CV are excluded.

Build Models and Estimate
After the establishment of enterprise alliance, it takes a certain period of time to produce obvious effect on innovation.In order to investigate the length of the lag period, we establish the negative binomial models with different lag stages, and use Stata to estimate them.The results are shown in table 2.

Model 6 Is Relatively Optimal
On the basis of model 6, this paper tries to cross terms, but no significant items appear.We can see that although the R-squared is small, the F statistic of the equation and the T statistic of each coefficient pass the significance test.We can think that the following relationship exists: ijbm.ccsen The uppe

Analys
Observe t

Accum
In all the innovatio relationsh

The I Alliance
In the ful N p2 >N p1 > sorted lik growing reaches it Next, we

From
The clust small gro network d The close each ente elicitation while sin closed th innovativ Figure 2. Note.The d using the N enterprises h Figure 3. D lliances in the cha loop, the informa uming the innova oliferation of inno =49.5(CC=1); 1+ Figure 5 taking into account other network structure parameters, enterprise characteristics variables and macro factors.

Table 3 .
Stata negative binomial regression models with cross item (obs=706) .3.4We Introduce Square Terms One by One to Model 6, and Find That Only the Square Term Of APL Is Significant * p < 0.1, ** p < 0.05, *** p < 0.01.If the cross terms Cc*cn and a*cn are added to the model, the result cannot be calculated.3