Do Chinese Industry Spur Innovation to Embrace the Challenges of the Current Technological Revolution: Data Envelopment Approach

doi:10.21203/rs.3.rs-1965062/v1

Download PDF

Research Article

Do Chinese Industry Spur Innovation to Embrace the Challenges of the Current Technological Revolution: Data Envelopment Approach

https://doi.org/10.21203/rs.3.rs-1965062/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The capacity base of the manufacturing sector plays a pivotal role in the process of wealth creation and economic growth of any country. The industrial innovative performance of an economy depends on the rigorous application of new ideas, methods of production, and modern techniques. This study uses the Malmquist productivity index of Data Envelopment Analysis (DEA), and the generalized ordinal model to investigate the actual innovation growth of China’s industrial enterprises. It also analyses the possibilities of the low innovative industries under study transforming into high innovative industries. The findings highlight that, although China’s industrial enterprises above high designated are on average experiencing a negative growth rate in terms of technical progress, yet they have spurred innovation with an average growth of 7% in total factor productivity. It concludes that there is a general need for the authority concern to lay more emphasis on reducing technical inefficiency, thereby improving technological innovation level and subsequent increase in overall efficiency scale.

Microeconomics

Innovation

High designated Industries

High Innovation industries

Faced with the opportunities and challenges brought about by the world's new scientific and technological revolution, innovation has become a nuclear weapon central to promoting economic and social development. Presently, countries around the globe are leveraging a number of innovation activities including capital, human resources, and technology to improve the performance of their operational industries (Lu and Huang 2010). Recently, China unveiled its ambitious “Made in China 2025” (MIC₂₅) industrial strategy with the aim of shifting its manufacturing industrial enterprises from being the world’s largest manufacturer to the world’s best. This means innovation activities are needed to accomplish the desired industrial upgrade and restructuring, thereby improving the added value generated by a different unit. Although relevant industrial enterprises have started changing the landscape of China manufacturing at a remarkable level of automation across catalyzing mass production industries in both automotive and non-automotive industries, yet the way to high-level innovation leadership is still very long (McKinsey report, 2017; Teece, 2019).

Technically, production is possible with variations in factor intensities and production technologies. However, there are substantial inter-industrial differences in the application of science and technology among China’s industrial enterprises that are above high designated size (Teece, 2019). Overall, to some degree, such disparities have impeded the harmonious growth of the Chinese economy (Guan & Liu, 2005). One of the key issues facing policymakers in China is how to build and strengthen the innovation structures of the lagged industries, for entire industries to develop their innovation capacity at relatively the same pace, and thus, enable them to achieve the MIC₂₅ agenda. Consequently, it is of paramount importance to empirically examine China's industrial enterprise's innovation performance at the firm level and explore the following questions. Do China's industrial enterprises spur innovation to embrace the challenges of the new scientific and technological revolution in the world? To what extent do Chinese industrial enterprises convert innovation inputs into outputs with varying degrees of efficiency? Therefore, this study considers China’s industrial enterprises that are above designated size by the industrial sector as the study case.

One of the most common measures of innovation performance of an industry is its efficiency in resource utilization (Abel et al., 2017). This is because innovation is not a linear operation that simply converts input into output; rather, it relies on the industrial capacity to convert innovation input into innovation output. Thus, analysis of industrial performance is necessary to distinguish between efficiency in innovation in order to obtain the desired information for necessary action. However, this study considers empirical study at the firm level favorable to investigate Chinese national innovation systems, thereby investigating and analyzing industrial enterprise innovation transition processes and innovation performance variations among industries. This is due to the fact that the measure of innovation performance entails comparing peer performance in highly environmental settings and assessing trends over time.

However, a number of pieces of literature exist that highlighted the importance of innovation efficiency in the Chinese context and other countries of the world, at different levels, using national and regional level datasets. For instance, (Yam et al., 2010; Mazur 2017), study of innovation measurement finds the effect of technology and non-technological innovation on a firm’s performance. (Evangelista and Vezanni, 2010; Yang et al., 2013; Wei, et al., 2017), a cross-country analysis of national innovation efficiency (Athawale and Chakraborty, 2011), and regional studies (Zhao and Cheng, 2013; Hailing et al., 2013). Nonetheless, the existing literature has clear limitations as many of the analysis focuses on what innovation measurement should be, how innovation should be measured, and on comparative innovation performance studies based on cross-sectional studies at regional and country level. Thus, less attention is paid to measuring the innovation output of a given economy and to finding the appropriate innovation inputs that match the industry at its specific level of success in innovation, which is the focus of this study. The efficiency of innovation performance of an industry can be estimated by comparing the composite scores of the inputs-outputs. The underlying argument regarding the theoretical models of the innovation process makes the formulation of clear inputs-outputs functional relationships quite difficult. Thus, this study employed the use of the non-parametric approach of Malmquist Productivity Index techniques of data enveloping analysis (DEA) inter-alia, at the onset of the analysis. The use of DEA is based on its number of applications in empirical research in various studies of performance evaluation (Chen and Guan, 2012; Hollanders, 2019). Therefore, the findings of this study are set to guide policymakers in the management of the industries examined and the government of China on the best practices to achieve the government's ambitious innovation agenda. It will also proffer some suggestions on optimizing the current innovation challenges raised by the world’s new science and technology revolution. In addition, the conceptual and analytical contexts could be of application to other fields of innovation performance evaluation.

This study begins with the measurement of the technological progress (TFP), which reflects the degree of innovation spurs, or industrial progress at the frontier of technological innovation, and the variability in industrial innovation efficiency. Subsequently, the study uses generalized ordinal logistic regression to further examine the impact of measurement variables on the industries at their different level of innovation performance. This will help in understanding the industries' innovation dynamics on the basis of three categories: low, moderate, and high innovative industries. The second section reviews related literature and theoretical concepts, the third section presents the methodology, followed by empirical analysis in the fourth section and the final section offers conclusion, policy, and managerial implication.

Concept of Innovation and Prior Research on Innovation Performance Measurement

Innovation means value creation through the transformation of knowledge and ideas (Dvir and Pasher, 2004). Innovation is possible through the combination of inputs to produce outputs. It is a process of transformation in which the innovation production unit uses science and technology resources from a variety of channels to generate new knowledge and increase economic wealth (Chen and Guan, 2012). Once there is innovation, there is a need to know how much a firm grows in that process which brought about innovation efficiency. At the industry level, continuous application of innovation in the production process is expected to reflect on overall productivity and efficiency performance over time. Importantly, the industry should measure the extent of growth in innovation at its own level which is ultimately driven by innovation efficiency. Innovation efficiency implied the firm’s capacity to translate innovation inputs into innovation outputs (Hollanders, 2019).

Classically, the concept of innovation measurement was first forward in the literature of management and economics. However, there is a stark difference in their perspective on innovation measurement. The economic perspective of innovation is to explore the rationale and driving forces behind industrial innovation, along with the macro-economic impact of innovation and barriers to innovation on economic agents. In contrast, the management perspective is based on how innovation can alter an industry's market position and how innovative ideas can be generated. Some of the famous available literature on innovation are; the idea of "creative destruction" coined by Schumpeter (1934), which explained innovation as the disruption of existing economic activity by inventing new ways of producing goods and services. The diffusion theory (Rogers, 1962) explores the mechanisms through which innovations are transmitted and adopted by the participants in a social system over time. The evolutionary theories (Nelson and Winter, 1982) found innovation as a path-dependent mechanism through which inventions are created by interactions between different actors and then evaluated on the market.

Table 1 presents a summary of some of the studies that addressed ‘innovation efficiency’ and their method of analysis, the sample size used, and the main idea. The main thrust of these studies (see Table 1), is on the analysis of what should be the innovation measurement, how to measure innovation, and comparative innovation performance studies based on cross-section studies at regional and country levels. Thus, little evidence is found on measuring the innovation output of a given economy on one hand, and on finding the appropriate innovation inputs that match the industry at its specific level of success in innovation. This is a black box that this study is intended to open.

Table 1

Conceptual and Empirical findings of Innovation Efficiency Literatures
Authors	Technique	Sample	Main idea
Athawale and Chakraborty, (2011)	Comparative studies	10 MCDM	Focuses on the comparison of the relative efficiency of the ten most famous multi-criteria decision-making MCDM methods for the rankings of alternative robots observed for a given place-pick operation.
Evangelista and Vezzani, (2010)	Empirical studies	A Stratified sample data representing all Italian service and manufacturing firms with more than 9 employees.	Investigates the impact of technological and non-technological innovations on firms’ performances, and found these types of technological innovation to be present and vital in all industries but have different influences on a firm’s efficiency.
Hailing et al., (2013)	Comparative studies	29 provincial data for eight years.	Comparison of technology innovation efficiency based on time and region of Chinese large and medium-sized industrial enterprises. It found eastern china’s region's technological efficiency to be high compared to the north and central region.
Mazur, (2017)	Conceptual framing	125 research papers	Classifies innovation performance measures implemented in the IT research sector. It found innovation performance to be considered in two ways; inventive performance (referring to workforce creativity and knowledge sharing) and Technology performance (referring to R&D outputs and inputs like patents and new products.
Wei, et al., (2017)	Action research	China Firm-level data 1995–2014	Evaluates China’s innovation policy, and conclude that, to improve the efficiency of resource allocation, policy changes should perhaps lead to a balancing of the playing fields for businesses across all levels of ownership, with corresponding cuts indirect subsidies and taxes.
Yam et al. (2010)	Modeling	200 manufacturing firms	Develops a framework that enables the link between innovation capabilities and innovation performance. They argued that any innovation efficiency measurement shall cover the size of the innovation potential of three actions: financial, strategic, and achievement.
Yang et al. (2013)	Empirical studies		Investigates innovation status of two sub-stages during innovation process of high tech industries in China.
Zhao and Cheng (2013)	Empirical studies		Built evaluation index of innovation capability of high tech industries, and further evaluates china’s high tech innovation efficiency at the regional level. It concludes that there is a series innovation efficiency gap at the regional level and the gap is widening.

This study employed an output-oriented DEA model to examine the nature of returns to scale in the industries and variation in their levels at turning innovation inputs into innovation outputs with varying efficiency. The output-oriented DEA model is deemed apt in this analysis because of its advantages in the analysis of industry-level data. Understanding the actual innovation of a firm lies in its technical progress. Thus, the non-parametric Malmquist Productivity Index (MPI) is often applied in the literature for evaluating technical progress and technological change or productivity growth over time. This technique has the advantage of judging the stability of each assessment unit in terms of efficiency and also playing a critical role in the assessment of trends in units’ efficiency value. MPI can be used to measure changes in technological innovation growth of China industrial enterprises above designated from period (t) to (t&txhinsp;+ 1). The MPI based on output orientation is given by (Fare et al 1992) and is written in form of algebraic equation as;

\({{{M}_{o} ({y}^{t+1},x}^{t+1}, {y}^{t},{x}^{t})=\left[\frac{ {d}_{o}^{t}\left({x}^{t+1},{y}^{t+1}\right)}{{d}_{o}^{t}\left({x}^{t},{y}^{t}\right)} \times \frac{ {d}_{o}^{t+1}\left({x}^{t+1},{y}^{t+1}\right)}{{d}_{o}^{t+1}\left({x}^{t},{y}^{t}\right)}\right]}^{1/2}\) …………………. (i)

This study assumes that \(x\) stands for inputs, and \(y\) stands for outputs. \({d}_{o}\) is the output distance function, \(o\) subscript stands for output orientation, \({d}_{o}^{t}\left({x}^{t},{y}^{t}\right)\) indicates output distance function assessing period t data of \(\left({x}^{t},{y}^{t}\right)\) concerning technology in the t + 1 period. \(M\) Stand for the recent productivity \(\left({x}^{t+1},{y}^{t+1}\right)\) using period (t+1) technology with respect to the earlier point of production \(\left({x}^{t},{y}^{t}\right)\) using period (t) technology.

\({{if M}_{o} ({y}^{t+1},x}^{t+1}, {y}^{t},{x}^{t})>1\) Means there is positive innovative growth between the periods

\({{if M}_{o} ({y}^{t+1},x}^{t+1}, {y}^{t},{x}^{t}) ˂ 1\) Signifies decline/negative innovation growth is occurring

\({{M}_{o} ({y}^{t+1},x}^{t+1}, {y}^{t},{x}^{t})=1\) Means stagnation/No technological innovation growth

MPI is the product of technical efficiency change (TEC) referred to as managerial efficiency and technical progress (TC). Form equation (ii) the first expression in the equation represents technical efficiency, and the second expression means technical progress. (Fare et al 1992). The Malmquist productivity index can be decomposed into two terms:

\({{{M}_{o} ({y}^{t+1},x}^{t+1}, {y}^{t},{x}^{t})= \frac{ {d}_{o}^{t+1}\left({X}^{t+1}, { Y}^{t+1}\right)}{{d}_{o}^{t}\left({X}^{t}, {Y}^{t}\right)}\left[\frac{ {d}_{o}^{t}\left({x}^{t+1},{y}^{t+1}\right)}{{d}_{o}^{t+1}\left({x}^{t+1},{y}^{t+1}\right)} \times \frac{ {d}_{o}^{t}\left({x}^{t},{y}^{t}\right)}{{d}_{o}^{t+1}\left({x}^{t},{y}^{t}\right)}\right]}^{1/2}\) …. (ii)

\(Technical Efficiency change = \frac{ {d}_{o}^{t+1}\left({X}^{t+1},{Y}^{t+1}\right)}{{d}_{o}^{t}\left({X}^{t},{Y}^{t}\right)}\) ………………………. (iii)

\(Technical Progress= {\left[\frac{ {d}_{o}^{t}\left({x}^{t+1},{y}^{t+1}\right)}{{d}_{o}^{t+1}\left({x}^{t+1},{y}^{t+1}\right)} \times \frac{ {d}_{o}^{t}\left({x}^{t},{y}^{t}\right)}{{d}_{o}^{t+1}\left({x}^{t},{y}^{t}\right)}\right]}^{1/2}\) …………………. (iv)

The technological changes occur when there is an entire shift of the production frontier between the two periods (t and t + 1). The efficiency change in the study is examined as the degree to which an individual enterprise improves or worsens its efficiency. In other words, whether there is an innovation growth, stagnation, or decline. Therefore;

If (i) TEC \(>\) 1 means improvement in managerial efficiency

(ii) TE \(<\) 1 means decline in managerial efficiency

However, as per Technical Change, TC \(>\) 1 means Technological progress, and TC \(<\) 1 indicates Technological regresses. And TC \(=\) 1 means No change in technical innovation

Data Collection Techniques with Input and Output dimensions

This study data were mainly obtained from the China statistical Book of Science and Technology from 2016 to 2019, published by the National Bureau of Statistics and Ministry of Science and Technology. A central database is commonly considered to be valid and publically available for statistical records. These industries are chosen based on their significant role in the advancement of science and technology in China. The size of the industries was decided on following (Cooper et al, 2001), that they must be three times the number of inputs and outputs. The study period 2016–2018 was chosen based on the MIC₂₅ policy enacted period and data availability.

The industrial levels analysis requires input and output variables for computing the change in a firm’s technology. From the available data, this study chooses five variables as supported by the literature. The innovation performance measurement indicators of this study include both measure indicators for research and development (RD) activities and commercial activities indicators. The output indicator comprises, number of patent applications (PAT) (Grillices, 1990; Kou et al., 2016), and Sales amount through innovation (SAI) (Guan and Chen 2010; Mazzola et al. 2012). The study employs both sales and patents as its output innovation measure indicators because patents or sales alone cannot sufficiently explain the nature of China's industrial enterprise innovation efficiency. Input indicators include the number of R&D dedicated personnel (RDP) (Nasierowski and Arcelus 2003; Lee, Kim, and Choi, 2019), expenditure on research and development R&D (EXPRD) (Park, Kim, and Sung 2017), the expenditure on new project development (EXPNPD), and a number of R&D projects (NRDP) (Parida et al. 2012; Goedhuys and Veugelers 2012).

Table 2 presents descriptive statistics of the input and output variables data for China’s high designated industries. (A) Under the output variables; patents (PAT), sales accrued through innovation (SAL), and a number of successful research projects (NPRD) from 2016 to 2018 have an average value of 9.0140, 16.5386, and 8.5190 respectively. The maximum output variables are 15.74, 10.87, and 19.87 respectively, and the lowest output variables are 5.31 for patents, 5.46 for successful research projects, and 10.05 for sale amount accrued through innovation. (B) The average value of input variables is 10.48, 14.03, and 14.02 for full-time R&D dedicated personnel, expenditure on R&D, and expenditure on new projects respectively for the year 2016–2018.

Table 2

Descriptive statistics
	Inputs
	FTRDP	EXPRD	EXPNRDP	PAT	NRDP	SAL
Max	19.06	16.94	17.25	15.74	10.87	19.87
Min	7.71	9.35	10.59	5.31	5.46	10.05
Mean	10.4827	14.033	14.0279	9.0140	8.5190	16.5386
Std. Div.	1.6646	1.5325	1.62761	1.68343	1.40304	1.85034
Valid Number	114	114	114	114	114	114

Table 3

The correlation coefficient among Inputs and Outputs
	Inputs
Outputs		FTRDP	EXPRD	EXPNRDP
Outputs	PAT	0.790**	0.689**	0.752**
	SAL	0.720**	0.800**	0.845**
	NRDP	0.860**	0.802**	0.850**
**Correlation coefficient sig. (2 tailed)

Furthermore, it is vital to consider the effect of time lag when evaluating the transformation of innovation inputs into innovation output. This is because the innovation process takes time to reflect on overall productivity. Many studies have measured time lag using a different methodology. For example input and output variable correlation tests and regression (Kaihua and Mingting 2014). However, lack of theoretical development and quality of innovation data have ruled out the accuracy in measuring time lag (Nasierowski and Arcelus 2003), thus there is no generally accepted approach for the measurement of time lag of innovation (Grillices, 1990). Hollandes and Esser (2007) have empirically found that the effect of time lag on estimating innovation performance was not substantial. In line with the extant literature, correlation analysis was conducted in this study using input and output variables at all periods. The result presented in Table 3 shows that input and output indicators hold an isotonic relationship which provides justification for including the variables in the model.

Statistical Regression Model

The technological change index derived from the Malmquist DEA analysis is the dependent variable for the estimation of the parametric model. The independent variables comprise of EXPRDsal, EXPRDpat, FTRPaaemp, PATemp, and SALexpnpd. The definitions of the model’s indicators chosen for the analysis are presented in Table 4. The target here is to explore the measure that best suits the low, moderate, and high innovative industries by further investigating the impact of the study measurement indicator on the different levels of innovation performance of the study industries.

Table 4

OLM Model Estimation Variables
Variables names	Meaning
EXPRDsal	Expenditure on R&D related to sales amount accrued through innovation, signifying continuity in technology innovation
EXPRDpat	Expenditure on R&D related to patent application signifying continuity in technology innovation
FTRPaaemp	R&D personnel related to the average annual number of employment, signifying investment in human capital
PATten	Patent application divide ten thousand number of employees, signifying independence research in technological progress
SALexpnpd	The ratio of sales output related to expenditure on new project development, signifying fertility in continuous technological innovation developments

The dependent variables of the study are categorized into high, Moderate, and low Innovative industries. Since the dependent variable of the study has ordinal values, the ordinal logistic model (OLM) is thus used in the analysis. The OLM model is written as:

\({\delta }_{j }= \frac{{x}^{\propto j}+ {\beta }_{{j}^{2}}}{\sum _{n}{x}^{\propto h}+{\beta }_{{j}^{x}}}\) ……………………….. (v) j = 1, 2, … (v)

In this study, we denoted the probability of high innovative industries as a baseline category by \({\delta }_{o }\)and the estimate by\({\delta }_{oo }\).

The moderate innovative industry by \({\delta }_{1}\) and the estimates\({\delta }_{11 }\). The low innovative is denoted by \({\delta }_{3 }\) and the estimate \({\delta }_{33. }\)by the response probability satisfying with our base category (high innovative), we can calculate these probabilities as follows.

\(Log \left(\frac{{\delta }_{11 }}{{\delta }_{oo }}\right)\) , \(\left(\frac{{\delta }_{22 }}{{\delta }_{oo }}\right)\), \(Log \left(\frac{{\delta }_{33 }}{{\delta }_{oo }}\right)\) ………………. (vi)

Since we have three categories, (j = 3) thus we have the following equations:

Lets \({y}_{1}= Log \left(\frac{{\delta }_{11 }}{{\delta }_{oo }}\right)\), \({y}_{2}= Log \left(\frac{{\delta }_{11 }}{{\delta }_{oo }}\right)\), \({y}_{3}= Log \left(\frac{{\delta }_{11 }}{{\delta }_{oo }}\right)\) …………… (vii)

To calculate the base of the system of the actual logarithm we have;

\({\delta }_{1}= \frac{exp \left({y}_{1}\right)}{1+ exp \left({y}_{1}\right)+ exp \left({y}_{2}\right)+ exp \left({y}_{3}\right)}\) …………. (viii)

\({\delta }_{2}= \frac{exp \left({y}_{1}\right)}{1+ exp \left({y}_{1}\right)+ exp \left({y}_{2}\right)+ exp \left({y}_{3}\right)}\) ………… (ix)

\({\delta }_{3}= \frac{exp \left({y}_{1}\right)}{1+ exp \left({y}_{1}\right)+ exp \left({y}_{2}\right)+ exp \left({y}_{3}\right)}\) …………. (x)

Where the first term in each denominator and numerator reflects changes in \(\text{exp}({\alpha }_{0}+ {\beta }_{0})\) for\({\alpha }_{0}= {\beta }_{0}=0\)

Innovation Performance across China Industrial Enterprises above High Designated Size by Industrial Sector

This section starts with the analysis of the output-oriented DEA model. Table 5 and Table 6 presents the optimal efficiency solution and technological change of the 38 selected industrial enterprises above high designated for the year 2016–2018. Table 5 shows the High Designated Industries (HDI) nature of scale economies: Constant Return to Scale (CRS), Variable Return to Scale (VRS), Scale (R-Scale), and Decision. The R-scale value was obtained by dividing the CRS by the VRS. The letters (icr, cns and dcr) in the decision column represent a return to scale translated as innovation efficiency of the industries as increasing, constant and decreasing based on R-scale. The Malmquist output presented in Table 6, comprises changes in technical efficiency (ECCH), changes in technical progress (TCCH), changes in pure technical efficiency (PECH), changes in scale technical efficiency (SECH), and technological innovation change or total factor productivity change (TFPCH). MPI tends to calculate productivity changes over time that may result from changes in technical efficiency (EC) and technical progress (TC). The evaluation of MPI is based on the value of one (fare et al 1992). Any value greater than one means that industry for a given (t + 1) period has incurred positive growth compared to time (t). Value less than one means a decline in productivity. Value equals one means productivity is stagnant.

Table 5

Malmquist Index of firm means
Industries	CRS	VRS	SCALE	R-Scale
HDI1	0.976	0.982	0.995	drs
HDI2	0.986	0.987	1.000	cns
HDI3	0.983	0.984	1.000	cns
HDI4	0.999	0.999	1.000	cns
HDI5	0.995	0.996	0.999	drs
HDI6	0.995	0.995	1.000	cns
HDI7	0.965	0.993	0.972	dcr
HDI8	0.963	0.999	0.964	dcr
HDI9	0.981	1.000	0.981	icr
HDI10	0.974	0.981	0.993	dcr
HDI11	0.975	0.985	0.990	dcr
HDI12	0.983	0.992	0.991	dcr
HDI13	0.993	0.993	1.000	cns
HDI14	0.967	0.983	0.983	dcr
HDI15	0.975	0.995	0.980	dcr
HDI16	1.000	1.000	1.000	cns
HDI17	0.996	0.999	0.997	dcr
HDI18	0.994	0.999	0.995	dcr
HDI19	0.988	0.993	0.995	dcr
HDI20	0.989	1.000	0.989	icr
HDI21	0.996	0.997	0.999	dcr
HDI22	0.985	0.988	0.997	dcr
HDI23	0.986	0.989	0.997	dcr
HDI24	0.984	0.986	0.998	dcr
HDI25	0.991	0.994	0.996	dcr
HDI26	0.986	0.987	0.999	dcr
HDI27	0.997	0.998	0.999	dcr
HDI28	0.992	0.994	0.998	dcr
HDI29	0.994	0.995	0.999	dcr
HDI30	0.989	0.995	0.993	dcr
HDI31	0.975	0.984	0.991	dcr
HDI32	0.979	0.987	0.992	dcr
HDI33	0.982	0.991	0.990	dcr
HDI34	0.968	0.984	0.984	dcr
HDI35	0.974	0.979	0.994	dcr
HDI36	0.976	0.987	0.989	dcr
HDI37	0.957	0.980	0.977	dcr
HDI38	0.986	0.987	0.998	dcr
Mean	0.984	0.991	0.992

Table 6

Malmquist Index of firm means
Industries	TECH	TCCH	PECH	SECH	TFPCH
HDI1	1.029	0.991	1.029	1.000	1.019
HDI2	1.018	1.012	1.018	0.999	1.030
HDI3	1.025	1.006	1.024	1.000	1.031
HDI4	1.000	1.020	1.000	1.000	1.020
HDI5	1.000	0.997	1.000	1.000	0.997
HDI6	0.996	0.998	0.996	1.000	0.993
HDI7	1.051	1.071	1.008	1.042	1.125
HDI8	1.052	0.983	0.998	1.055	1.034
HDI9	1.029	0.977	1.000	1.029	1.005
HDI10	1.039	0.981	1.028	1.010	1.019
HDI11	1.000	0.997	1.017	1.011	1.025
HDI12	1.028	1.030	1.012	1.015	1.057
HDI13	0.999	0.965	0.990	1.000	0.955
HDI14	1.032	0.967	1.008	1.024	0.988
HDI15	1.009	0.984	0.992	1.016	0.992
HDI16	1.000	0.972	1.000	1.000	0.972
HDI17	1.002	1.019	1.002	1.000	1.020
HDI18	0.999	1.014	0.999	0.993	1.006
HDI19	1.004	1.015	1.011	0.993	1.018
HDI20	0.992	1.021	1.000	0.992	1.013
HDI21	0.994	1.014	0.996	0.998	1.008
HDI22	1.012	1.008	1.011	1.001	1.020
HDI23	1.006	0.985	1.006	1.000	0.991
HDI24	1.020	1.020	1.019	1.001	1.040
HDI25	0.991	0.955	0.994	0.977	0.946
HDI26	0.983	0.970	0.985	0.998	0.953
HDI27	0.995	0.984	0.996	0.999	0.980
HDI28	0.997	1.003	0.994	1.003	1.004
HDI29	1.001	1.003	1.001	1.000	1.004
HDI30	0.986	1.035	0.996	0.990	1.020
HDI31	0.992	1.026	0.998	0.994	1.017
HDI32	1.016	1.010	1.014	1.003	1.027
HDI33	1.013	0.983	1.011	1.001	0.996
HDI34	1.009	0.977	1.013	0.996	0.985
HDI35	1.011	0.981	1.013	0.998	0.992
HDI36	1.007	0.981	1.019	0.988	0.988
HDI37	1.000	0.985	1.031	0.970	0.985
HDI38	1.020	0.996	1.017	1.003	1.015
Mean	1.009	0.998	1.006	1.003	1.007

Clearly, from the nonparametric DEA analyses presented in Table 5, among the entire thirty eights industries examined, only two industries have exhibited increasing returns, hence, have their innovation efficiency improved. Evidently, a constant return to scale was evident in six industries. This means that industries with constant returns to scale have exhibited stagnation in mean efficiency. In contrast, the remaining industries have exhibited a decrease in their scale efficiency. Therefore, based on the DEA result, only eight industries have better outcomes on investment on innovation levels (R&D personnel, expenditure on R&D, and expenditure on new project development) under the period of study. This indicates a wide gap among the industries under study, which is due to technical differences.

From Table 6, the average growth rate of technological change (TFPCH) is 0.7%, with an average growth rate of technical efficiency (TECH) of 0.9%, and an average negative growth rate in technical progress (TCCH) of -2%. The result indicates that innovation is essential to total factor productivity (TFP) growth. However, the positive average TFP growth rate of 0.7% was not brought by technical progress but rather a technical efficiency. This demonstrates that management efficiency exceeds technological efficiency during the study period. And that China industrial enterprises above high designated are on average experiencing a negative growth rate in terms of technical progress for the period under study.

Similarly, the contribution of technical progress was lower than anecdotal, based on the overall industrial efficiency. Overall, the industries have exhibited technical inefficiency within the range of 1.9–18.8% for the technical progress during the study period. Based on the current world’s science and technology revolution, it means that Chinese industries have a low level of technological innovation, but working for the improvement of technology is quite promising. However, despite the decline in TCCH, there is a positive increase in TECH and TFPCH of 0.9 and 0.7 percent respectively, which signifies that China’s industrial enterprises above high designated are spurring innovation during the study time frame. However, more initiatives are needed to sustain the pace of innovation for the industries to be better equipped to withstand the innovation challenges of the world of science and technology revolution.

By decomposing the mean of technical progress inefficiency (0.998) into pure technical efficiency mean and scale of technical efficiency, it can be observed that the overall technical inefficiency of Chinese industrial enterprises is primarily due to pure technical efficiency. It suggests that management of China’s industrial enterprises should concentrate on reducing technical inefficiency thereby improving technological innovation in their industries first, and then move on to increase their efficiencies on the scale. Regarding pure technical inefficiency, it is found that on average industries can improve their technical innovation by the same level of measured inputs with 0.7% less output keeping the current ratio of input constant. Meaning, the inefficient industries have the potentials to improve their level of innovation output for them to perform similarly to the most efficient industries.

The average innovation performance trend of the study industry’s efficiency scores is shown in Fig. 1. It comprises TECH, TCCH, and TFP. It has been seen that there is a high change in technical efficiency in low innovative performance industries like HDI13, HDI15, HDI33, HDI35, and HDI37. And invariably a high changes in technical progress among high innovative output industries such as HDI17, HDI11, and HDI25. These disparities indicate a low technical innovation for the low innovative industries. Similarly, the trend of TECH was shown to be more close to TFPCH, indicating an excellent level of managerial skills in the resource allocation and utilization of each factor of production. Overall, the China industrial enterprise should focus on improving their technical innovation in order to meet the current challenges posed by the world's new science and technological revolution.

Insert Fig. 1

Discussion of Empirical Result

In this section, the present study introduced ordinal regression analysis to further explain the variation of innovation performance of the industries under study. Reference from the output-oriented DEA result of the technological change index obtained from Table 6, the study ranks the innovation performance of the study industries into three categories. High, moderate, and low innovative industries, based on the average efficiency score of total factor productivity change of industry. A score level less than 1 means low innovative, a scores level falling between (1-1.20) represents moderate innovative, and a score level falling between (1.21–1.60) means high innovative. In essence, these categorizations formed the dependent variable of the study estimation of the parametric model. The independent variables cover EXPRDsal, EXPRDpat, FTRPaaemp, PATemp, and SALexpnpd (see Table 4).

Table 7 shows the case processing summary of the ordinal regression result, the proportion of cases falling at each level of independent variables are 26, 12, and 38 for the Low, Moderate, and High Innovative industry respectively. The result of the likelihood Chi-square ratio reveals that there is a significant improvement in the fit of the final model over the null model [\({x}^{2}\left(3\right)=21.628, p<0.001\)].

Table 7

Case process summary and Model fitting
Category	Level	Falling No	Marginal percentage
Low Innovative	˂ 1.00	26	34.2
Moderately innovative	1-1.20	12	15.8
High Innovative	1.21–1.60	38	50.0
Valid		76	100%

The estimation technique of ordinal logistic regression (OLR) assumes that the relationship between the independent variables is the same across all possible comparisons (Osborne, 2017, p. 147) involved in the dependent variable - an assumption referred to as proportional Odds. The study result of the test of parallel lines reveals that the assumption is satisfied (p = 0.743) and the Pseudo R-square value is 0.28.6. The Pearson Chi-square test stood at [\({x}^{2}\left(3\right)=156.90,p=0.236, p>0.001\)] and deviance test [\({x}^{2}\left(3\right)=131.13, p=0.789, p>0.001\)] were both non-significant that the study model is of good fit. The non-significant text result of the goodness of fit model indicates the model fits the data well (Field, 2018), thus, the variables used in this study serve as good measurement indicators for innovation performance of the industries under study.

Table 8 depicts regression coefficients and significance tests for each independent variable in the model. The Exp(B) column contains odds ratios signifying the multiplicative change in the odds of being in a highly innovative category, for every unit increase on the innovative indicator used in this study holding the remaining indicators constant. The odds ratio greater than 1 means a high probability of being at a high level, less than 1 means less probability of being at a high level. The odds ratio equals unity means no predicted change. As observed, for the most part, the p-value from both tables is very consistent.

Table 8

Model Estimates
Parameters		β	Std. Error	Exp(β)	Sig.	95% Confidence Interval
Parameters		β	Std. Error	Exp(β)	Sig.	Lower Bound	Upper Bound
Threshold	[Low Innovative = 1]	23.411	7.467	1.470E + 10	.002	8.777	38.045
Threshold	[Moderate Innovative = 2]	24.247	7.505	1.390E + 10	.001	9.538	38.955
Location	EXPRDsal	6.182	6.163	484.096	.316	-5.897	18.262
	EXPRDpat	.083	.042	255228.62	.047	.001	.166
	FTRPaaep	− .019	.010	0.982	.050	− .037	1.455E-5
	PATten	12.450	4.126	4.367	.003	4.363	20.537
	SALexpnpd	1.474	.471	1.087	.002	.550	2.398
Link function: Logit.

From the model estimate in Table 8, the result shows that for every single unit increase of (EXPRDsal and EXPRDpat) there was a projected increase of 6.182 and 0.083 in the odds that the industry will be in a highly innovative category of innovation. This shows that any industrial enterprises falling into the category of high-innovative are likely to resolve the ongoing revolution in new science and technology in the world within the stipulated period of the government MIC₂₅ plan. The (EXPRDsal and EXPRDpat) odds ratio reveals that, the odds of being in a higher category of innovation increases by a factor of 484.10 and 255228.62 for every one-unit increase on R&D expenditure per sale and R&D expenditure per patent respectively. The coefficient of research and development expenditure per sale is not statistically significant indicating the existing gap in technical progress across the study sample. However, expenditure per sale (EXPRDsal) and expenditure per patent (EXPRDpat) are referred to the continuous development of technology innovation in this research (Shi and Zhang, 2010).

R&D personnel (FTRDPemp) is a non-significant predictor in the model. This indicates the need for less use of human capital to overcome the current technological innovation. The result shows that every one unit increase in (FTRDemp) would lead to a projected decrease of (-019) in the chances of a moderate and low innovative category falling into the category of high-innovative industry and achieving the desired MIC25 plan. The odds ratio of (FTRDemp) shows that the chances of a moderate and low innovative category falling into the high innovative category increases by a factor (0.982) for each unit increase in R&D workers per average annual number of employees. This result is in line with the Chinese government MIC₂₅ plan, and the current world’s new science and technology revolution. That is production of goods and services to be fully automotive with less or without any human effort. Consequently, this shows the need for the change in innovation strategic formula for the industries under study, from their old strategic formula of winning through cheap labor to winning through technological innovation.

The patent per thousand employees (PATten) represents the role of independent research in technological progress for the industries under study. It was found to have a positive significant indicator of innovation in this model. It shows that a one-unit increase in patents would lead to a predicted increase of 12.45 in the odds of being in the category of the high-innovative industry by the moderate and low-innovative category. PATten odd ratio was found to be 4.367. This reveals high odds of falling into a highly innovative industry with a one-off rise in independent research in technological progress by the moderate and low innovation industries. However, this result reveals no evidence of the importance of research collaboration in technological progress, which is contrary to the existing nature of the world’s new science and technology revolution. The nature of the current science and technology revolution advocates the need for industries to design and develop an innovative experience environment based on collaborative ICT-infrastructures that can enable network organization and customer communities to interact (Tzu-Ying, 2004).

(SALexpnpd) represents the ratio of sales accrued from innovation related to the expenditure of new project research development. It means fertility in continuous technological innovation developments. It was also found to be a significant positive predictor of innovation. The result shows that for every one-unit increase on SALexpnpd, there was a predicted increase of 1.474 in the odds of an industry being in a higher (as opposed to lower) category innovation industry. Meaning, an industry with high performance in innovation is more likely to quickly overcome the new challenges brought by the world’s new science and technology revolution, and thus achieve China’s government MIC₂₅ plan within the stipulated period. The ratio of sales accrued through innovation related to the expenditure of new project research development helps the industry to foster innovation and innovate more. Therefore, the result shows the advantage taken by high innovative industries to wider their innovation gap with moderate and low innovative industries. The SALexpnpd odds ratio reveals that the odds for low and moderate innovative industries falling into the high innovative category increases by a factor of 1.087 for every one unit increase in sales accrue from innovation.

Findings and Conclusion

This study uses Malmquist productivity index DEA and generalized ordinal regression to evaluate whether Chinese industrial enterprises above high designated spur innovation to embrace the challenges of the world’s new scientific and technological revolution, and the degree of variation in innovation efficiency across the industries. Since the study is at the firm level thus, a three years panel data for the industries understudy was formed (2016–2017). Interestingly, the non-parametric DEA model results presented in Table 8 show that the level of innovation growth differs substantially among the industries under investigation. Perhaps, this may be due to China's unique market size and the nature of its development, which is not the subject of this study.

However, the study also examines the level of variation in innovation efficiency and the impact of innovation measurement indicators on the different industries at their varying degrees of innovation. The total number of the industries under investigation were categorized into high, moderate, and low innovative industries based on the average efficiency score of total productivity change, obtained in the DEA model. Subsequently, generalized ordinal regression based on maximum likelihood was applied to obtain the parameter estimates of the model. The empirical results obtained have been found (EXPRDpat, PATemp, and SALexpndp) to be significant positive predictors of innovation. (FTRDPemp) is found negative but significant predictor of innovation. Although EXPRDsal was found positive but is a non-significant predictor of innovation. The study odds ratio derived from the model estimates reveals high probability that, non-high innovative class industries will overwhelm the new world science and technology revolution and transform into high-innovative industries.

Policy Implication

This study fills the existing literature gap by measuring the innovation output of a given economy, which represents the first attempt in the literature. It was found that there were appropriate levels of innovation inputs that match each cluster of industries at their respective levels of innovation prospects. Therefore, the findings provide policymakers in major emerging economies with some new insights. First, policymakers need to recognize the positive effect of science and technology activities on the performance of innovation in the industry. New industrial policies should be conceptualized aiming at boosting the prospects of firms' indigenous technical know-how such as through intramural spending on science and technology activities, and encouraging firms to be more responsive and collaborative for innovation. This would enable firms to integrate expertise from diverse sources with regional technologies into commercial goods.

Secondly, the results of the DEA study show that there was significant variation in the level of innovation efficiency across the industries examined. Therefore, the study suggests the need for customized R&D planning in line with the technological characteristics of the organization and innovation performance of the industries. As a result, more efficient progress in innovation can be realized through this process. Also, the existing gap in innovation efficiency between the industrial enterprises in China can be minimized if not eliminated, over a reasonable time. This study provides sufficient evidence on the crucial role of technical efficiency in the innovation level of the industries.

Managerial Implication

The new science and technology of the world today is an innovation-driven development strategy through the use of technology. It requires the achievement of industrial output growth through technical progress as opposed to the obtained DEA result of this study. The China industrial enterprises above high designated have on average failed in the use of technological innovation to meet or shift their production frontier to achieve technological change. Therefore, the management of China’s industrial enterprises above high designated size should stay focus on improving their technical innovation, for improving their innovation output to meet the contemporary challenges prompted by the new science and technology revolution of today’s world.

However, the study found that on average, the industry can improve its technical innovation using the same level of measured inputs with 6% less output keeping the current input ratio constant. Therefore, this result would help the industries in acting fast by applying the right mix of strategies to squarely face the challenges of the world’s new science and technology revolution, and in achieving the desired target of the Chinese government MIC₂₅ innovation agenda. However, the management concern of China industrial enterprises above high designated should concentrate on improving their productivity level through an increase in technical innovation not an increase in technical efficiency.

Competing interests: The authors declare no competing interests.

Abel S., Alex B., and Alex B., 2017. “Decomposition of the Technical Efficiency : Pure Technical and Scale Efficiency of the Financial” Economic Research South Africa. June, p. 1–27.

Athawale, V.M., and Chakraborty S. 2011. “International Journal of Industrial Engineering Computations,” International Journal of Industrial Engineering Computation vol. 2, p. 831–50.

Cooper WW, S Li, Seiford LM, Tone K, Thrall RM and Zhu J., 2001. Sensitivity and stability analysis in DEA: Some recent developments. Journal of Productivity Analysis, 15(3), 217–246.

Coelli, T. J., Rao, D. S. P., O'Donnell, C. J., and Battese, G. E. 2005. An introduction to efficiency and productivity analysis. USA: Springer.

Chen, K., and Guan J. 2012. “Measuring the Efficiency of China’s Regional Innovation Systems: Application of Network Data Envelopment Analysis (DEA).” Regional Studies, vol. 46, no 3, p. 355–77.

Dvir R. and Pasher E., 2004. Innovation engines for knowledge cities: an innovation ecology perspective, Journal of Knowledge Management 8, 16–27.

Evangelista, R., and Antonio V., 2010. “The Economic Impact of Technological and Organizational Innovations. A Firm-Level Analysis.” Research Policy, vol. 39, no 10, Elsevier B.V., p. 1253–63.

Färe R., Grosskopf S., Lindgren B., Roos P. (1992) Productivity changes in Swedish pharamacies 1980–1989, Journal of Productivity Analysis 3 (1): 85-101.

Guan, J. and Chen K. 2010, “Measuring the Innovation Production Process: A Cross-region Empirical Study of China’s High-tech Innovations,” Tech-novation 30, pp.348-358

Guan J.C and Liu S.Z., 2005. Comparing regional innovative capacities of PR China-based on data analysis of the national patents, International Journal of Technology Management 32, 225–245.

Goedhuys M, and Veugeelers R., 2012. Innovation strategies, process and product innovation, and growth: firm level evidence from Brazil. Struct. Change Econ. Dyn. 23(4), 516-529.

Griliches Z., 1990. Patent statistics as economic indicators: A survey Journal of Economic Literature, doi: 10.1016/S01697218 (10)020095.

Hailing H., Shi T., and Wu H. 2013. “Measurement On Technological Innovation Efficiency And Comparison Based On Time And Area In The Region Of China’s Large And Medium- Sized Industrial Enterprises,” Springer Science and Business Media Deutschland GmbH 545–49.

Hollanders H. and Esser C.F., 2007. Measuring innovation eﬃciency. INNO-Metrics Thematic Paper.

Hollanders, Hugo. 2019. “Disclaimer :” no. January 2007.

Kaihua C. and Minting K., 2014. Staged eﬃciency and its determinants of regional innovation systems: A two-step analytical procedure. The annals of Regional Science, 52(2), 627-657. Doi: 10.1007/s00168-014-0604-6.

Kou M., Chen K., Wang S., and Shao Y., 2016. Measuring efficiency of multi-period and multi-division systems associated with DEA: An application to OECD countries’ national innovation systems. Expert system with applications 46, 494-510. doi:10.1016/j.eswa.2015.10.032.

Lee, J., Kim, C., and Choi, G., 2016. The effect of external R&D on the innovation efficiency: An empirical study of manufacturing industries in Korea. Journal of Society of Korea Industrial and Systems Engineering, 39(4), 125–136

Lu W, Pan T. and Hung S., 2010. DEA Performance Measurement of the National Innovation System in Asia and Europe. Asia-Paciﬁc Journal of Operational Research, 27(3), 369–392.

Mazzola E., Bruccoleri M. and Perrone, G., 2012. The effect of inbound, outbound and coupled innovation on performance. International Journal of Innovation Management, 16(6), 1.

Mazur K., 2017. “Methodological Aspects of Innovation Performance Measurement in the IT Sector,” Management vol. 21, no 2, p. 14–27.

McKinsey Global Institute. 2017. “A Future That Works: Automation, Employment, and Productivity.” McKinsey Global Institute. http://www.mckinsey.com/global-themes/digital-disruption/ harnessing-automation-for-a-future-that-works.

Nasierowski, W, and F J Arcelus. 2003. “On the Efficiency of National Innovation Systems,” Socio-Economic Planning Sciences 37 (3), 215–234

Nelson, R. and Winter S., 1982. An Evolutionary Theory of Economic Change, Harvard University Press, Cambridge, MA.

Parida, V., Westerberg, M., and Frishammar, J., 2012. Inbound open innovation activities in high-tech SMEs: the impact on innovation performance. Journal of Small Business Management, 50(2), 283–309.

Park J., Kim J., and Sung S., 2017. “Performance evaluation of research and business development: A case study of Korean public organizations,” Sustainability, vol. 9, 2017.

Rogers E., 1962. Diffusion of Innovations, Free Press, New York.

Schwab K., 2017. “The fourth industrial revolution”, textbook published pp. 42-51

Shi, X. and S. Zhang (2010), “Trade Credit, Credit Constraints and Impacts on Eﬃciency: Empirical Evidences from China,” Economic Research Journal, pp.102-114

Teece D.J 2019. “Dialogue, Debate, and Discussion China and the Reshaping of the Auto Industry : A Dynamic Capabilities Perspective,” Management and Organization Review March 2019, 177–199.

Tzu-Ying, C. and Jen-Fan, L. (2004). “A Comparative Study of Online User Communities Involvement in Product Innovation and Development”, 13th International Conference on Management of Technology

Wei S.J, Xie Z., and Zhang X. 2017. “From ‘Made in China’ to ‘Innovated in China’: Necessity, Prospect, and Challenges.” Journal of Economic Perspectives, vol. 31, no 1, p. 49–70.

Yam R.C.M., Lo W., Tang E.P.Y. and Lau A.K.W., 2010. Technological innovation capabilities and firm performance. World Academy of Science, Engineering and Technology, 66(10), 1009-1017.

Yang, G.L., Shen, W.F., Zhang, D.Q., Liu, W.B., 2013. Extended utility and DEA models without explicit input. Journal of the Operational Research Society, in press. DOI:10.1057/jors.2013.68.

Zhao Y. and Cheng P., 2013. An empirical analysis of technological Innovation capacity of Hi-tech industry on provincial scale in china. Journal of Business Economics

Download PDF

Version 1

posted

You are reading this latest preprint version

Do Chinese Industry Spur Innovation to Embrace the Challenges of the Current Technological Revolution: Data Envelopment Approach

Status:

Version 1

Abstract

Figures

Introduction

Literature Review And Theoretical Concept

Concept of Innovation and Prior Research on Innovation Performance Measurement

Methodology

Data Collection Techniques with Input and Output dimensions

Statistical Regression Model

Result Analysis

Innovation Performance across China Industrial Enterprises above High Designated Size by Industrial Sector

Insert Fig. 1

Discussion of Empirical Result

Findings and Conclusion

Policy Implication

Managerial Implication

Declarations

References

Status:

Version 1