Diversified Agglomeration, Specialized Agglomeration and Innovation Efficiency of Pharmaceutical Manufacturing

Based on the panel data of 31 provinces in China from 2009 to 2016, this paper uses DEA model and panel Tobit regression method to estimate and test the effect of diversified agglomeration and specialized agglomeration on the innovation efficiency of pharmaceutical manufacturing. The results show that diversified agglomeration significantly improves the innovation efficiency of pharmaceutical manufacturing industry. Specialized agglomeration is obviously not conducive to the improvement of innovation efficiency of pharmaceutical manufacturing industry; management capacity, innovation institution penetration rate and government innovation subsidies all significantly promote the innovation efficiency of pharmaceutical manufacturing industry. The effect of cooperative innovation intensity and industrial structure on the innovation efficiency of pharmaceutical manufacturing industry is not obvious.

community generally believes that the two agglomeration models have great differences in the path and impact of industrial innovation. Yang and Li believe that diversified agglomeration can promote innovation by optimizing industrial structure and rationally arranging industries [3]. Liu and Wang believe that diversified agglomeration provides cross-industry technical support, strengthens the integration of knowledge and technology in different fields, promotes innovative thinking, and is more conducive to innovation than specialized agglomeration [4]. At the same time, Lu and others believe that specialized agglomeration strengthens the exchange of similar innovation subjects, generates technological spillovers, and promotes innovation [5]. Zhu believes that professional agglomeration attracts high-level innovative talents, strengthens technology and knowledge spillovers, and is significantly conducive to innovation [6]. Moreover, the role of specialized agglomeration for enterprises to save the cost of innovation factors and strengthen the dissemination of silent information has been widely recognized. Cheng and Liu believe that enterprises in specialized clusters can feel the fierce competition in the market, which will promote enterprises to reduce production costs and increase labor productivity, and promote enterprises to upgrade and innovate technology [7]. From the research conclusions, Chai and Xiang [8], Lu and Shang [9] all believe that specialized agglomeration is more conducive to innovation and efficiency improvement than diversified agglomeration; Zhang and Zhao [10], Peng and Jiang [11], Hong [12] both believe that specialized agglomeration is not conducive to innovation, while diversified agglomeration promotes innovation [13]. From the above literature analysis, it is known that the effects of diversified agglomeration and specialized agglomeration on industrial innovation have not formed a consistent conclusion. This may be due to differences in research objects and agglomeration patterns that favor industrial innovation. Then, in the pharmaceutical manufacturing industry, which is related to the national health but is rarely studied, what kind of agglomeration model is more suitable for industrial innovation? This is the question that this article tries to answer, and it is also the research topic of this article.
Another striking innovation in this paper is that in the literature on the effects of diversified agglomeration and specialized agglomeration innovations, there are more research on innovative output, such as the value of new product output and the number of patent applications. But there are less research focusing on the study of innovation efficiency. Innovation efficiency is the ratio of innovation output to corresponding input of innovative production factors in a given time. Under the national accelerated development goal, attaching importance to the issue of innovation efficiency and improving the utilization rate of innovative resources is an objective need to accelerate the advancement of technology.
In addition, the key reason why this paper chooses pharmaceutical manufacturing industry rather than other industries as research objects is that the progress versified agglomeration and specialized agglomeration on the innovation efficiency of pharmaceutical manufacturing. The research in this paper will help pharmaceutical manufacturing enterprises to gather in the best mode, achieve optimal innovation resource allocation, and achieve maximum innovation output under the established innovation investment. These will optimize the innovation efficiency of pharmaceutical manufacturing industry. Innovative medicines which are produced at the fastest speed will further protect Chinese residents from overcoming the threat of illness, living and working healthily, thus promoting the accelerated development of the Chinese economy.

1) Diversified agglomeration and innovation efficiency of pharmaceutical manufacturing
First of all, in terms of the innovation efficiency of pharmaceutical manufacturing industry measured by the commonly used input and output, the factors affecting the innovation efficiency of pharmaceutical manufacturing industry fall into two categories, one is the saving of innovation investment in pharmaceutical manufacturing industry, and the other is the growth of innovative output of pharmaceutical manufacturing industry [14]. Then, we will mainly analyze the process and mechanism of diversified agglomeration affecting the innovation efficiency of pharmaceutical manufacturing industry from these two aspects.
First, about the savings in innovation investment in the pharmaceutical manufacturing industry. The savings in innovation investment in the pharmaceutical manufacturing industry in the diversified agglomeration areas are mainly reflected in the efficiency of the use of innovative resources. The innovative resource allocation of pharmaceutical manufacturing enterprises in diversified agglomeration areas is more autonomous, flexible and efficient [15]. What's more, using different but related knowledge improves production technology and management level, make processes upgrade, achieve the efficiency of resource allocation, reduce the investment in innovation required for a given innovation. At last, it will achieve savings in innovative investment in pharmaceutical manufacturing. Second, about the growth of innovation output in the pharmaceutical manufacturing industry. Compared with the specialized agglomeration area, the growth of the innovation output of the pharmaceutical manufacturing industry in the diversified agglomeration area is mainly reflected in the success probability of product innovation. In the diversified agglomeration area, pharmaceutical manufacturing enterprises benefit from the technological environ-ment with greater knowledge width and technical potential [3], which is less affected by the lock-up of specialized agglomeration [16], and enhances the innovation of pharmaceutical manufacturing products as a whole. The probability of success is to achieve the growth of innovative output in the pharmaceutical manufacturing industry. However, the development of things has two sides. The higher the degree of diversification, the higher the investment in innovation in the pharmaceutical industry may be. In the diversified agglomeration area, the technical thresholds of non-similar enterprises are large, the cost of cooperative innovation is high, and the cost of talent and capital flows is large. They will all contribute to increasing the investment in innovation in pharmaceutical manufacturing.
2) Specialized agglomeration and innovation efficiency of pharmaceutical manufacturing Specialized agglomeration also affects the innovation efficiency of pharmaceutical manufacturing through the two channels of savings in innovation in pharmaceutical manufacturing and growth in innovation in pharmaceutical manufacturing.
First, about the savings in innovation in pharmaceutical manufacturing. Compared with diversified agglomeration areas, specialized agglomeration affects the savings of innovation in pharmaceutical manufacturing from two aspects, one is capital, and the other is labor. The savings of specialized agglomeration on capital are mainly reflected in the cost of capital use. In the case of designated innovations, pharmaceutical manufacturing companies located in specialized clusters require less capital. This is because, compared to non-aggregated areas, pharmaceutical manufacturers in the agglomeration area have lower prices for collecting and acquiring innovation-related information, lower prices for public infrastructure [17], and lower prices for purchasing innovative elements. The price of obtaining external capital is also lower [18]. The concentration of specialized agglomeration on labor is mainly reflected in the improvement of human capital. Labor has investment value, and high levels of human capital have higher labor productivity than low-level human capital. Specialized agglomeration promotes the quality of labor. Technology and management experience through knowledge spillover. They raise the level of human capital, reduce the low level of human capital and the number of machines required for established innovation. In particular, high levels of human capital can accomplish innovative work that takes less time and can't be completed with low levels of human capital. Thus, when the degree of specialized concentration is strengthened, pharmaceutical manufacturers usually only need less capital and labor input to obtain the established innovation. The savings on the cost of innovation resources translates into savings in innovation inputs, and reductions in innovation inputs under established innovation outputs, thereby enabling improvements in the innovation efficiency of pharmaceutical manufacturing. terprises [19], and promote the participation of pharmaceutical manufacturing enterprises in the innovation production process, so as to achieve the growth of innovation output. On the other hand, the level of patent protection in the agglomeration area [20] and the development of the technology market [21] are significantly better and faster than the non-aggregation areas. They play an important role in the transformation and industrialization of innovation achievements in pharmaceutical manufacturing enterprises. They are conducive to the increase in innovation output. However, there are also views that the degree of specialization is too high, and its negative crowding effect will reduce the innovation efficiency of the pharmaceutical manufacturing industry [22], that is, under the premise of innovation factors and land air supply, excessive professional concentration is likely to cause input costs. The increase and deterioration of the ecological environment [23] have led to the escape of high-level innovation elements, which has made the innovation and efficiency of the pharmaceutical manufacturing industry less efficient.

Model Setting and Variable Selection
In the empirical part, we will first use the DEA model method to Finally, the value of the innovation efficiency of the pharmaceutical manufacturing industry will be in the 0 -1 range. The larger the value, the higher the efficiency. The innovation efficiency of the pharmaceutical manufacturing industry will also be divided into scale efficiency and technical efficiency to explore whether the improvement of efficiency comes from the expansion of scale or the advancement of technology.

2) Panel Tobit regression
The general model of the panel Tobit is as follows:

3) Indicator selection
The explanatory variable is the innovation efficiency of the pharmaceutical  Table 1 reports the basic statistical characteristics of the variables.

1) General estimation results
Firstly, for the panel model of the merged data, the mixed Tobit regression is performed using the clustering robust standard error. Secondly, for the fixed-effect Tobit model, the conditional maximum likelihood estimation cannot be performed because the sufficient statistics of individual heterogeneity cannot be found. If the dummy variables of the panel unit are added directly to the mixed Tobit regression, the resulting fixed effect estimates are also inconsistent.
Therefore, in general, only the Tobit model of random effects is considered. As for whether you should choose mixed regression or random effects. Table 2 shows that the LR test rejects the null hypothesis of mixed regression at the significance level of 0.05, so it is considered that there is an individual effect, and the Tobit regression model of random effects should be used.  The panel Tobit regression model's estimations of random effects are listed in Table 2. It shows that diversified agglomeration significantly promotes the in- From the perspective of control variables, the pharmaceutical manufacturing industry management ability is promoting the innovation efficiency of pharmaceutical manufacturing. If the industrial management ability is stronger, the choice of innovative projects will be more sensible. The supervision of innovative projects will be more comprehensive. The efficiency of innovation resource utilization will be higher, and they will contribute to the improvement of the

2) Robustness test
In this paper, the sample segmentation and substitution variables are used to test the robustness in Table 3: First, sample segmentation. The data of 31 provinces and cities in 2009 were removed, and the Tobit regression method was used again to estimate the effect of diversified agglomeration and specialized agglomeration on the innovation efficiency of pharmaceutical manufacturing.
Second, replace the variable. Use production value instead of employment to measure diversified agglomeration and specialized agglomeration. The test results are shown in Table 3. Although the sizes of the diversified agglomeration and the specialized agglomeration coefficients have changed, the sign of the coefficients has not changed, and its significance remains unchanged or even more significant. This shows that the conclusion of this paper is that diversified agglomeration significantly promotes the innovation efficiency of pharmaceutical

Conflicts of Interest
The author declares no conflicts of interest regarding the publication of this paper.