innovation efficiency industries technological achievements transformation high

Efficiency of Two-Stage Technological Innovation in High Patent-IntensiveIndustries That Considers Time Lag: Research Based on the SBM-NDEA ModelConsidering time lag and accumulation of inputs and outputs, this paper adoptsthe superefficiency data envelopment analysis (DEA) model to study thetechnological innovation efficiency of high patent-intensive industries usingpanel data from 2007 to 2017. Given the characteristics and the actualcircumstances of the industries, the innovation process is divided into twostages, and an input-output indicator system is established. The results showthat the overall innovation efficiency level of high patent-intensiveindustries in China is increasing. However, the R&D achievements in technologyare not quickly applied or sufficiently transformed.#### 1. IntroductionWith the advance of the economy and technology, knowledge and technologicalinnovation have become major driving forces of economic growth in the 21stcentury. The United States has proposed the “National Strategic Plan forAdvanced Manufacturing,” Germany has proposed “German Industry 4.0,” and Chinahas proposed the “Innovation-Driven Development Strategy” and the “Made inChina 2025 Strategy.” The “National Intellectual Property Strategy”promulgated in China in 2008 has promoted the development of the nation’spatent-intensive industries. In October 2016, the State Intellectual PropertyOffice released the “Statistical Report on the Patent-Intensive Industries inChina” and the “Catalogue of Patent-Intensive Industries (for trialimplementation).” These publications indicate the national and social focusthat has been placed on the development of patent-intensive industries. Thetwo documents also reveal that patent-intensive industries play anincreasingly significant role in strengthening China’s economiccompetitiveness and boosting technological and economic development. “Patent-intensive industry” [1] refers to industries in which the intensity and scaleof invention patents reach a certain standard that rely on intellectualproperty rights to participate in market competition and that are in line withinnovation-oriented development. Eight industries that exceed the averagelevel of such industries are identified as high patent-intensive industries.These eight industries are pharmaceutical manufacturing, electrical equipmentmanufacturing, computers and communications equipment manufacturing, and fiveother industries. According to “China Statistics Yearbook on High TechnologyIndustry” [2], the internal R&D expenditure of China’s patent-intensiveindustries in 2016 was 291575 million yuan. The number of patent applicationswas 185913, and the sales of new products totaled 4792424 million yuan. Theadded value of patent-intensive industries was 15379.6 billion-yuan,accounting for 20.8% of GDP. Each year, on average, 13420 thousand employmentopportunities are created. In China, high patent-intensive industries havebecome the leading industries of technological advancement and pillarindustries of economic development.From the perspective of previous literatures, much attention has been paid onthe study of technological innovation, but there are heterogeneous betweendifferent studies not only in the results of studies but also in the settingof the statistical models. Hence, a nonparametric model may be a better choicefor the study of technological innovation and its impact factors.Against this backdrop, there are two main purposes and contributions in thisresearch. First, it is imperative to investigate and evaluate the innovationefficiency of the high patent-intensive industries, determine the problemsconfronted in technological innovation, analyze the factors that affectinnovation efficiency in these industries, and establish effective means toimprove innovation efficiency in such industries. These measures are ofsubstantial significance to improve the efficiency of technological innovationin high patent-intensive industries. Second, the time lag factor should beconsidered so that the relationship between impact factors and technologicalinnovation could be more clear, and new policy suggestion could be putforward.#### 2. Literature ReviewSeveral foreign scholars have defined patent intensity and conducted studieson patent-intensive industries. Siwek [3], an American economist, provided thefirst definition of the “intellectual property industry.”Examining the American pharmaceutical manufacturing industry, Mazzucatoet [4]conducted research on high patent-intensive companies. Among experts, there isno consensus on the concept of technological innovation. Afriat [5] providedthe first conceptual description of the efficiency of technologicalinnovation. As research has increased, more scholars have begun to focus onthe technological innovation efficiency of high patent-intensive industries.Research topics include the evaluation of technological innovation efficiency,the comparison of technological innovation efficiencies, and the analysis ofthe factors that affect technological innovation efficiency [6–8].BangRae Lee, EunSoo Sohn, DongKyu Won, and WoonDong Yeo [9] analyzed the R&Dinvestment efficiency of 23 industries in the field of precision medicine inSouth Korea. They found that investment in these technologies can produce goodbenefits.Kwangsoo Shin, Eungdo Kim, and EuiSeob Jeong [10] analyzed the relationshipsamong knowledge, innovation ability, technological innovation, and financialperformance. They concluded that transformation ability, the connectionability of technological innovation, innovation ability, and absorptiveability have direct and indirect effects on the financial performance oftechnological innovation.Eetal [11] applied the DEA model to analyze panel data from 185 regions in 23European countries and to investigate innovation efficiency in these regions.Domestic scholars have conducted a series of studies on high patent-intensiveindustries. Xu and Jiang [12] measured patent-intensive industries based on acombination of research results from home and abroad and studied the R&Dperformance of patent-intensive industries in China.Scholars have adopted various research methods to study innovation efficiency.Pan and Yang [13] used the TOPSIS method and the entropy method to assesspatent-intensive industries. Han [14] used stochastic frontier analysis (SFA)to study the technological innovation efficiency of China’s high-techindustries. These studies found that the overall innovation efficiency ofChina’s high-tech industries is improving. A number of scholars have employeddata envelopment analysis (DEA) to study technological innovation efficiency,successfully analyzing the technological innovation efficiency of high-techindustries using a variety of models. Chen et al. [15] adopted the DEA-Malmquist model, whereas Feng [16] employed the DEA-SBM (data envelopmentanalysis-slack-based measure) model, which considers undesirable outputs.To study innovation efficiency, scholars divide the industrial innovationprocess into different stages. Among such researchers, Jiang [17], Hu, andZhou [18] regarded the industrial innovation process as one stage. Wen-jing etal. [19] and Bao et al. [20] divided the industrial innovation process intotwo stages. Liu et al. [7] and Kang et al. [21] divided the industrialinnovation process into three stages to measure high patent-intensiveindustries.In summary, a considerable number of scholars have studied the technologicalinnovation efficiency of high patent-intensive industries from variousaspects. The research on the technological innovation efficiency of suchindustries has primarily focused on the evaluation of influencing factors andthe overall efficiency of technological innovation. Efficiency evaluation ofsubindustries is rarely conducted, and there is little analysis of theinnovation process. In studying innovation efficiency, scholars have adoptedvarious methods, whereby increasingly more researchers have adopted DEAmodels. Drawing on previous research, this article starts from the innovationprocess and uses the superslack-based measure network data envelopmentanalysis model (SBM-NDEA), dividing the technological innovation process intotwo stages: technological R&D and achievement transformation. With thisapproach, we can understand the deficiencies in the innovation process ofChina’s patent-intensive industries and establish the breakthrough points forefficient development. Additionally, the SBM-NDEA model with relaxationvariables introduced in our approach can better restore the two-stagecharacteristics of the innovation process.#### 3. Research MethodData envelopment analysis (DEA) is a new, widely applied and effectivenonparametric method to measure technological innovation efficiency. There aretwo types of DEA. One is the traditional DEA model. The other is the networkDEA model. The traditional DEA model does not consider the relationshipbetween decision-making unit (DMU) and input resources and views DMU as a“black box.” In contrast, the network DEA model divides the process oftechnological innovation into several subprocesses, thus transforming the“black box” into a “gray box.” Guan and Zuo [22] analyzed innovationefficiency in 35 countries using a two-stage network DEA model. Yu Wenjing, Maet al. [19] adopted a two-stage serial DEA model to analyze the efficiency ofChinese high-tech enterprises at the provincial level. Liu [23] used a three-stage DEA model. Tone and Sahoo et al. [24] introduced variable relaxation asobjective function and constructed a superslack-based measure (SBM) model. Liuet al. [25] adopted the superefficiency DEA model to perform a comparativeevaluation of the efficiency of S&T innovation in several Chinese provinces.Chen et al. [26] used a superslack-based measure network data envelopmentanalysis (SBM-NDEA) model to study the technological innovation efficiency ofhigh patent-intensive industries. This model not only considers the problem ofvariable relaxation but also distinguishes the size of the effective DMU.In Equation (1), Xi is the input of subprocess 1, the number of which is m;i.e., Xi=(xi1,xi2,…,xim); fi is the output of subprocess 1, the number ofwhich is p; i.e., fi=(fi1,fi2,…,fip); zi is the input of subprocess 2, whichincludes the outputs of subprocess 1 and additional inputs, the number ofwhich is q; i.e., zi=(zi1,zi2,…,ziq); y is the output of subprocess 2, thenumber of which is h; i.e., yi=(yi1,yi2,…,yih); and λi is the weight of theith DMU. Then, the set of production possibilities based on variable returnsto scale is as follows:The restriction of intermediate variable f is a free link, which can bechanged freely.The restrictive conditions are as follows:Based on the preceding assumption, the linear representation of a super-SBMmodel with variable return is as follows:The constraints are as follows:The efficiency value of each DMU can be greater than 1, and the effective DMUscan be sorted accordingly. Considering the relaxation of input and outputvariables, Grosskopf and Fare [25, 27] regarded the process of technologicalinnovation as an interrelated network, which also includes intermediatevariables. The complex and intermediate process from input to output and itsmodel is shown by equation (5). Here, k represents the relative weight of thekth stage of the DMU, and (i = 1,2,3, …, I), (p = 1,2, …, P), and (h = 1,2, …,H) represent the I-th input, the P-th intermediate variable, and the H-thoutput, respectively. α represents the efficiency value of the technology R&Dstage, β represents the efficiency value of the achievements transformationstage, and represents the best intermediate variable. However, according toChiu et al. [26, 28], this assumption assumes that all the inputs in theachievements transformation stage come from the outputs in the technology R&Dstage without considering the subsequent inputs. Therefore, the model must bemodified. The modified model is as follows:where represent the inputs in thetechnology R&D stage and the inputs in the achievements transformation stage,respectively. This paper uses the NDEA model to study the technologicalinnovation efficiency of high patent-intensive industries.#### 4. Empirical Research##### 4.1. Indicator SelectionBased on previous research [29, 30] and the purpose of this paper, theinnovation process is divided into two subprocesses: the technology R&D stageand the achievements transformation stage. An input-output indicator system isalso constructed (Table 1).|* * * — Technology R&D stage| Input indicators| Full-time equivalent (FTE) of R&Dpersonnel R&D capital stock Output indicators| Number of patent applications Number of authorized invention patents * * * Achievements transformation stage| Input indicators| New product developmentcosts Number of patent applications Number of authorized invention patents Output indicators| New product sales revenue Number of newly developed projects * * * The innovation of labor input is typically based on the number of employees orwage costs. However, it is difficult and inaccurate to obtain the exact levelof wage costs in different regions and periods. Regarding the availability andaccuracy of the acquired data, this paper adopts the full-time equivalent(FTE) of R&D personnel for the innovation of labor input.Therefore, the full-time equivalent (FTE) of R&D personnel and R&D capitalstock are selected as input variables.It is widely recognized that the output indicators and patents of the R&Dstage are important indications of technological innovation. Patents includeinvention patents, utility model patents, and design patents. The number ofannual patent applications refers to the number of patent applicationssubmitted to and accepted by the State Intellectual Property Office in a year,which is the basis of R&D innovation. “Authorized patent” refers to a patentauthorized by the State Intellectual Property Office and represents animportant achievement of technological innovation.Therefore, in this paper, the number of patent applications and the number ofauthorized invention patents are chosen as the output indicators of innovationin the technology R&D stage.The input indicators in the achievements transformation stage include thenumber of patent applications that include output indicators and the number ofauthorized invention patents in the technology R&D stage. However, consideringother additional inputs, such as the cost of technology import in the stage ofachievements transformation, it is more reasonable to count the additionalcosts as the inputs of new product development costs.Therefore, in this paper, the input indicators of the achievementstransformation stage include new product development costs, number of patentapplications, and the number of authorized invention patents.The output indicators in the achievements transformation stage, the salesrevenue of new products, and the number of newly developed projects are thekey indicators reflecting the economic benefits of innovative activities ofhigh patent-intensive industries.On the one hand, the output indicators are used to measure the innovativeproduction capacity. On the other hand, they can reflect the economic value ofinnovation achievements affected by the market environment and other factors.Therefore, it is reasonable to adopt the new product sales revenue and thenumber of newly developed projects as output indicators in the achievementstransformation stage.According to recent research [7, 31–33], the input indicators of most studieson the technology R&D stage primarily include two aspects: capital inputinnovations and labor input innovations. Capital input innovation has animpact on not only the current period but also the later period due to capitalprecipitation, which also refers to time lag and accumulation. Thus, if onlyR&D expenditure is used as the capital input indicator, innovation efficiencycannot be accurately reflected. Therefore, this paper uses capital stock as aninput indicator and adopts the method of perpetual inventory to calculate theR&D capital stock. Considering the patent application cycle and output cycle,it is assumed that the lag period is one year [34]. The calculation formula isas follows:where represents the t-th R&D capital stock of the i-th DMU andUi(t-1) represents the internal expenditure on regional R&D in the year t − 1.δ represents the depreciation rate. Based on recent research results, δ is setat 15% in this paper.##### 4.2. Data SourceFormula (7) is obtained by combining recent research results, thecharacteristics noted in national industrial classification statistics, thestatus of Chinese high patent-intensive industries, and the relevant data ofthe report “Main Statistical Data Report of China’s Patent IntensiveIndustries” issued by the State Intellectual Property Office.The intensity of industrial invention patents equals the quotient of the totalnumber of authorized invention patents in the industry within 5 years and theaverage number of employees in the industry within 5 years.Formula (7) is as follows:According to the concept of the intensity of industrial invention patents, theaverage value of patent intensity and the respective patent intensities of allChinese industries are obtained by calculating the relevant data from 2007 to2017, whereby the average values of patent intensity in 9 industries exceedthose of all other national industries. The nine industries are pharmaceuticalmanufacturing; arts and crafts manufacturing; oil and gas exploration;electrical machinery manufacturing; chemical manufacturing; computers,communications, and other electronic equipment manufacturing; specialequipment manufacturing; instrument manufacturing; and tobacco processing.According to the research content, the data related to patents from 2007 to2017 are selected. Considering the cycle of patent outputs and achievementstransformation, this paper sets the time lag as one year. Therefore, theoutput indicator data lag one year behind the input indicator data. Based onthe accumulation of input, the input indicator is readjusted. The relevantdata are obtained by referring to “China Statistical Yearbook (2007–2017)” and“China High-Tech Statistical Yearbook (2007–2017).”##### 4.3. Empirical AnalysisIn this paper, the input and output indicator data are imported into Max DEAPro 6.9 A software. The network relationship and unexpected output indicatorsare set, and the evaluation results are calculated and obtained. Then, thecurrent status of the technical efficiency evaluation of high patent-intensiveindustries is further analyzed based on the evaluation results.###### 4.3.1. Technical Efficiency Analysis of High Patent-IntensiveIndustriesFrom the operation results for the innovation efficiency data of the eightindustries in 2016 (Table 2), it can be observed that the average values ofinnovation efficiency in the stage of technology R&D and the stage ofachievements transformation are 0.896 and 0.850, respectively. These outcomesindicate that more attention is paid to resource inputs in the stage oftechnology R&D than in the stage of achievements transformation. In terms oftotal efficiency value, the average efficiency value of the eight high patent-intensive industries is only 0.873, indicating a large margin for additionalinnovation efficiency improvement. However, the efficiency values among theindustries are not balanced, and the differences are obvious. The efficiencyvalues of certain industries are higher than those of others.|* * * — Industry| Tech R&D| Achi. trans| Efficiency * * * Pharmaceutical manufacturing| 0.832| 0.827| 0.830 Crafts manufacturing| 0.954| 0.897| 0.926 Oil and gas exploration| 0.781| 1.000| 0.891 Electrical machinery manufacturing| 1.000| 0.726| 0.863 Chemical manufacturing| 0.704| 0.802| 0.753 Computer, communications, and other electronic equipment manufacturing| 1.000|1.000| 1.000 Special equipment manufacturing| 0.911| 0.764| 0.838 Instrument manufacturing| 0.985| 0.787| 0.886 Mean value| 0.896| 0.850| 0.873 Standard deviation| 0.104| 0.098| 0.068 * * * By observing the innovation efficiency values of each stage, we find that theaverage efficiency value of the technology R&D process is 0.896 and that thestandard deviation is 0.104. These outcomes indicate that although theoriginal resource input is reduced by 10%, the expected output level can stillbe achieved. The average efficiency value of high patent-intensive industriesin the stage of achievements transformation is 0.850, and the standarddeviation is 0.098. This value is lower than the average value of the totalefficiency and the average efficiency value in the technology R&D stage. Thisresult indicates that the low innovation efficiency of the achievementstransformation stage affects the innovation efficiency of the entire industry.As shown in the table, China’s technology R&D is more stable than thetransformation of achievements, and the national policy support and R&Dprocedures are more mature. However, the achievements transformation ofChinese high patent-intensive industries is far from sufficient. If the numberof patent applications is valued and the transformation of achievements isignored, the role of technology in economic development cannot be fullyexploited, and the value of technological development is thus lost. Inaddition, the transformation of achievements is substantially affected bymarket factors. Good market and economic and financial environments are ofsubstantial help to improve the transformation of achievements (Figure 1).###### 4.3.2. Analysis by Industry of Technical Efficiency of High Patent-Intensive IndustriesBased on the data in Table 2, this paper analyzes the characteristics of thetechnology R&D stage and the achievements transformation stage as well as theoverall efficiency of 8 types of high patent-intensive industry in China in2016. To further analyze these characteristics and determine the evolutionpath, this paper calculates the industry data of the 8 industry types from2007 to 2018 and obtains the overall efficiency and the efficiency of 2substages of technological innovation of these industries for the past 10years (Table 3).|* * * — Industries| 2007| 2008| 2009| 2010| 2011| 2012| 2013| 2014| 2015| 2016 e1| e2| E0| e1| e2| E0| e1| e2| E0| e1| e2| E0| e1| e2| E0| e1| e2| E0| e1|e2| E0| e1| e2| E0| e1| e2| E0| e1| e2| E0 * * * Pharmaceutical manufacturing| 0.631| 0.881| 0.756| 0.697| 0.897| 0.797| 0.742|0.912| 0.827| 0.856| 0.865| 0.861| 0.768| 0.876| 0.822| 0.732| 0.761| 0.747|0.681| 1.000| 0.841| 0.821| 1.000| 0.911| 0.835| 1.000| 0.918| 0.832| 1.000|0.916 Crafts manufacturing| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 0.876| 1.000|0.938| 1.000| 0.866| 0.933| 1.000| 1.000| 1.000| 0.935| 1.000| 0.968| 0.769|1.000| 0.885| 0.981| 0.814| 0.898| 0.867| 0.962| 0.915| 0.954| 0.897| 0.926 Oil and gas exploration| 0.316| 0.788| 0.552| 0.346| 0.865| 0.606| 0.256|0.972| 0.614| 0.396| 1.000| 0.698| 0.384| 1.000| 0.692| 0.368| 1.000| 0.684|0.396| 1.000| 0.698| 0.632| 1.000| 0.816| 0.763| 1.000| 0.882| 0.781| 1.000|0.891 Electrical machinery manufacturing| 0.884| 0.814| 0.849| 1.000| 0.824| 0.912|1.000| 0.744| 0.872| 1.000| 0.793| 0.897| 1.000| 0.781| 0.891| 1.000| 0.702|0.851| 1.000| 0.816| 0.908| 1.000| 0.842| 0.921| 1.000| 0.871| 0.936| 1.000|0.886| 0.943 Chemical manufacturing| 0.412| 0.724| 0.568| 0.441| 0.786| 0.614| 0.368|0.756| 0.562| 0.421| 0.684| 0.553| 0.396| 0.762| 0.579| 0.536| 0.857| 0.697|0.545| 0.863| 0.704| 0.713| 0.842| 0.778| 0.721| 0.806| 0.764| 0.704| 0.851|0.778 Computer, communication, and other electronic equipment manufacturing| 1.000|1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000|1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000|1.000| 1.000| 1.000| 1.000| 1.000| 1.000| 1.000 Special equipment manufacturing| 0.545| 0.846| 0.696| 0.616| 0.715| 0.666|0.632| 0.768| 0.700| 0.684| 0.700| 0.692| 0.701| 0.768| 0.735| 0.810| 0.786|0.798| 0.841| 0.778| 0.810| 0.891| 0.868| 0.880| 0.902| 0.823| 0.863| 0.911|0.814| 0.863 Instrument manufacturing industry| 0.415| 0.831| 0.623| 0.620| 0.703| 0.662|0.684| 0.698| 0.691| 0.702| 0.732| 0.717| 0.641| 0.706| 0.674| 0.734| 1.000|0.867| 0.786| 0.922| 0.854| 0.896| 0.846| 0.871| 0.967| 0.852| 0.910| 0.985|0.877| 0.931 * * * The analysis by different industries reveals that the efficiency value of eachindustrial technology R&D stage and achievements transformation stagefluctuates in accordance with the total efficiency fluctuation rule. Thefluctuation of the efficiency value of the achievements transformation stageis much more obvious. The efficiency of the technology R&D stage of mostindustries is higher than that of the achievements transformation stage. Itcan be observed that achievements transformation remains the focus of highpatent-intensive industries. Three industries display high efficiency in thetechnology R&D stage: handicraft manufacturing, computer and electronicsmanufacturing, and electrical equipment manufacturing. The efficiency of thetechnology R&D stage in these industries is significantly higher than that ofthe achievements transformation stage, which indicates that the R&D of thesethree industries has reached a higher level but that achievementstransformation remains insufficient. In the process of technologicalinnovation, particularly in the electrical equipment manufacturing industry, adisconnection remains between R&D and achievements transformation. The averageachievements transformation efficiency value in the past ten years is only0.807, an outcome that requires further study and support. In pharmaceuticalmanufacturing and oil and gas exploration, the efficiency of achievementstransformation is higher than that of technology R&D. The respective averageefficiency values of these two industries in the stage of achievementstransformation are 0.919 and 0.963, respectively. These outcomes indicate thatthe achievements transformation levels of these two industries are alreadyhigh, reflecting the urgent social demand for the products of these twoindustries. The reason for the low overall innovation efficiency can be foundin the low efficiency of the technology R&D stage. Therefore, it is imperativeto strengthen technology R&D in these two industries. Low innovationefficiency occurs in two substages in chemical manufacturing, specialequipment manufacturing, and instrument manufacturing. This outcome indicatesthat both technology R&D and achievements transformation are insufficient inthese three industries. Steps should be taken to increase inputs, addresstechnical problems, and improve the level of achievements transformation.According to the efficiency values in Table 2, the 8 industries can beclassified as follows (Figure 2):Category I: industries with dual low innovation efficiency, i.e., lowinnovation efficiency both in technology R&D and in achievementstransformation. These industries are chemical manufacturing, special equipmentmanufacturing, and instrument manufacturing.Category II: industries with low-to-high level innovation efficiency, i.e.,low innovation efficiency in technology R&D and high innovation efficiency inachievements transformation. These industries are pharmaceutical manufacturingand oil and gas exploration.Category III: industries with high-to-low level innovation efficiency, i.e.,high efficiency in technology R&D and low efficiency in achievementstransformation. These industries are arts and crafts manufacturing, computerand electronics manufacturing, and electrical equipment manufacturing.##### 4.4. Research Deficiencies and ProspectsFirst, this paper is mainly based on the relevant theory of technologicalinnovation efficiency while studying the innovation efficiency of high patent-intensive industries in China. In spite of the fact that complex relationshipamong different stages, time lag, resource accumulation, and other aspects areall considered, the disturbing factors that affect efficiency still need to becomprehensively dived in for further study. Secondly, the selected statisticaldata are based on “China Statistical Yearbook” and “China Statistics Yearbookon High Technology Industry,” so the data are lagging. Thirdly, the mechanismand path of influencing factors on technological innovation need to beclarified. Next, scholars can carry out further research on technologicalinnovation mechanisms, conduct in-depth enterprise researches, and conductmore targeted researches.#### 5. Conclusions and LimitationsBased on the above research, this paper starts from the innovation process andadopts SBM-NDEA to divide the technological innovation process into twostages: technology R&D stage and achievement transformation stage, andconducts researches on both stages. According to the panel data of China hightechnology industry from 2007 to 2016, the technological innovation efficiencyof high-patent intensive industries and subindustries was evaluated. The studybelieves that the overall technological innovation efficiency of China’s high-patent intensive industries is gradually increasing. The technological R&Defficiency has improved, but the rapid absorption and transformation of R&Dachievements are insufficient; the development of technological innovationefficiency in each subindustry is uneven, and the gap is relatively obvious.It is suggested that the government departments should further promotecollaborative innovation and coordinated development of industry, academia,and research; improve the market structure and give play to the market’sguiding role in resource allocation and transformation of results; andincrease intellectual property protection, increase government support for theindustry, and promote the sustainable development of high patent-intensiveindustries.#### Data AvailabilityThe data used to support the findings of this study are available from thecorresponding author upon request.#### Conflicts of InterestThe authors declare that they have no conflicts of interest.#### AcknowledgmentsThe research presented in this paper is part of a research project sponsoredby the Youth Fund Project of the National Natural Science Foundation(71603048) and the National Social Science Fund (14BGL007).