The Sustainable Efficiency Improvement of Internet Companies under the Background of Digital Transformation

Abstract

The digital economy is a driving force for economic growth in various countries. When the digital economy is deeply integrated with Internet companies, it also brings about great challenges to corporate innovation. This paper used the DEA-Malmquist model to measure the efficiency of the science and technology investment of 30 Internet-listed companies in China from 2011 to 2019, constructing a long-distance function and displaying the dynamic changes in the comparative efficiency of time and space. This research evaluated the digital maturity and digital experience of Internet companies under the continuous investment of science and technology, concluding that the investment in science and technology under the digital economy can improve the innovation efficiency of China’s Internet companies. Research shows that technology investment has a significant positive impact on the digitalization of Internet companies, but there is heterogeneity among different companies. Based on the theoretical research in this article and the empirical experience of in-depth digitalization, it also provides path suggestions for the in-depth development of digitalization of Internet companies. This research has important theoretical significance and practical value for further promoting the development of Internet companies and promoting the application of the digital economy.

1. Introduction

In the past 20 years, the digital economy has become increasingly important worldwide. With the emergence of digital technologies, including the Internet of Things (IoT), big data, artificial intelligence, cloud computing, and blockchain [1,2,3,4], the digital economy empowering enterprise transformation has become an inevitable competitive strategy for most enterprises, and it is also a key research issue that scholars are paying attention to. According to the China Internet Development Report released by the China Academy of Information and Communications Technology (CAICT), the scale of the Chinese digital economy accounted for 29% of its GDP in 2019. Within this, the service industry, industry, and agricultural digital economy penetration rates were 37.8%, 19.5%, and 8.2%, respectively. The digital economy has become a new engine for domestic economic growth [5]. The digital economy uses technological inputs and technological changes to provide innovative services, as well as to promote the innovation and transformation of organizational systems through productivity improvements and technological tool changes [6]. Digital transformation is defined as “the use of new digital technologies (such as social media, mobile analytics, embedded devices, etc.) to help main business operations improve efficiency (such as improving customer service operations, process-oriented operation management, or creating new business models)” [7,8]. At the same time, digital technology has contributed to companies’ innovation and management capabilities, while also promoting corporate performance [9]. Digital transformation means combining digital technology and business processes in the digital economy to achieve organizational transformation [10]. This process also includes key business operations, products, and the process of reshaping business models [11].

Some foreign scholars have analyzed from a related perspective that in addition to e-commerce, the digital economy also includes information technology [12], corresponding information and communication technology (information communications technology, referred to as ICT) infrastructure, the IT (information technology) industry itself and its commodities, digital transmission of services, and retail sales of tangible goods supported by IT [13]. Rose [14] and Schwab [15,16] believe that the fourth industrial revolution should create a new digital economy by improving the super-physical space system of intelligent manufacturing , transportation services, and even biological systems. Some domestic scholars believe that the essence of the digital economy is informatization [17], which includes information technology industrialization, traditional industry, infrastructure, and lifestyle [18]. Some scholars have proposed that ICT is the foundation of the digital economy, which is the digitization of transactions, exchanges, and cooperation through the Internet, mobile communication networks, and the Internet of Things [19]. E-commerce and the information technology industry that directly supports its development are both within the scope of the digital economy [20,21].

The Chinese Internet industry has become the backbone of China’s economic development [22]. In 2019, the business revenue of the top 100 Internet companies increased significantly in comparison with the year 2018, reaching CNY 2.75 trillion, which is 8.8% of the total digital economy in China. It further drove the growth of the digital economy by about 2% and became a solid foundation for the rapid development of the Internet industry. The Internet industry has penetrated every corner of people’s lives, and all aspects of the economy, culture, sports, entertainment, and lifestyle are enjoying the convenience brought about by Internet innovation [23]. However, although Internet companies provide convenience through innovation, they also involve huge risks and challenges. Repeated investment, waste of resources, information security, organizational changes, and personnel management risks, among others, are forcing Internet companies to transform and pay attention to and reshape the efficiency of input and output in the digital economy [24,25,26]. In view of this, this paper used the DEA model to measure the total factor productivity of 30 listed Internet companies in China from 2011 to 2019, constructing a total factor productivity analysis model. Research shows that, on the one hand, this improvement promotes the improvement of technical efficiency and scale efficiency through technological breakthrough innovation and market breakthrough innovation by Internet companies. Prior research has found that different Internet companies have different characteristics in the context of the digital economy [27,28,29]. This paper conducted research on the efficiency of science and technology investment in the context of the digital economy. Internet companies that have contributed major strengths were used as research objects to measure and evaluate how companies that already have Internet plus measure and evaluate their operational efficiency in the process of Smart plus, and at the same time explore what factors restrict their further development.

The contribution of this article is mainly reflected in the following aspects: First, there are only a few documents analyzing the impact of the digital economy on the efficiency of Chinese Internet companies’ investment in science and technology from a theoretical level. This paper starts with innovative theory, reveals the mechanism of the digital economy in improving the efficiency of Internet companies’ technology investment, explores the impact of Internet companies’ heterogeneity on the acquisition of digital economic dividends, and enriches the theoretical research on the relationship between the digital economy and Internet companies. Second, in view of the availability of data, most of the existing literature focusing on the relationship between the digital economy and China’s Internet companies are case studies [30] or have used provincial or city-level data to conduct empirical studies [31,32]. Our empirical and quantitative investigations at a different level enrich the empirical evidence on the digital economy and Internet companies. Third, this article identifies the relationship between the heterogeneity of Internet companies affecting the efficiency of technology investment in the context of digitalization, which helps to clarify the heterogeneous effects of the digital economy in different types of Internet companies and inspires companies to adjust their business models and innovation paths according to their own characteristics, and effectively seek competitive advantages to obtain core competitiveness, which has important theoretical significance and practical value for further deepening the innovation of Internet companies.

The structure of this paper is as follows. We present the theoretical background, describe the DEA-Malmquist methodology used, and develop our research model in Section 2. We present the data and variable selection in Section 3. We show our main results and discuss further clustering classification, primarily aimed at addressing the heterogeneity of different kinds of Internet companies in Section 4. Finally, we discuss our findings and conclude in Section 5.

2. Theoretical Analysis and Models

2.1. Quantitative Model: Concept and Analysis Model

Schumpeter first proposed an innovative theory in his Theory of Economic Development in 1912. In his subsequent works, he expanded the theory of innovation and the content of the theory of innovation into the concepts of innovation, economic growth, and economic cycles, emphasized the relevance of technology and the economy, and regarded innovation as the core of economic growth, including product innovation, technological innovation, market innovation, resource allocation innovation, and organizational innovation [33]. Leifer [34] divides technological innovation into breakthrough and incremental, according to different degrees of innovation. Christensen [35] first proposed disruptive technologies in his book The Innovator’s Dilemma: When New Technology Bankrupts Large Companies. Simplicity, convenience, and cheapness are regarded as the stage characteristics of the initial formation of destructive technologies. Zhou [36] divides breakthrough innovation into technology-based breakthroughs and market-based breakthrough innovations. These concepts are hereinafter referred to as technical breakthrough innovation and market breakthrough innovation.

Christensen also analyzed the characteristics of disruptive innovations in new markets and innovations from low-end markets that disrupt the current business model [37], namely: (1) Whether the target customers targeted by the innovations were unable to complete the corresponding work by themselves due to lack of money and technology in the past. Many of the most successful destructive growth businesses are to provide people with direct products and services that are mainly occupied by complex functions and expensive prices, products, and services. (2) Whether the customers targeted by the innovation are those who like simple products. Disruptive products must be technically simple and easy to understand and must target customers who are willing to use simple products. The resource allocation procedures of incumbent companies often require quantification of the size and possibility of innovation opportunities so that potentially disruptive innovations are forcibly incorporated into obvious, measurable, and existing market applications. This puts disruptive innovations on the existing market to compete with sustaining innovations. For example, the innovation and entrepreneurship team formed in a large company not only costs a lot but also has a very high probability of failure. (3) If innovation can help customers more simply and effectively accomplish a task. The digital economy has an impact on the innovation of Internet companies and their total factor productivity through breakthrough innovation. Breakthrough innovation is a fundamental change to existing technology, which usually involves exploring new knowledge, spending a lot of time and money, and taking huge technological risks [38]. The digital economy relies on its advantages in cross-temporal information dissemination, data processing, and near-zero cost of information acquisition, and the utilization of information technology such as big data, cloud computing, and artificial intelligence can alleviate the contradiction between supply and demand of R&D elements and product supply and demand in enterprise innovation. It also reduces the difficulty of research and development and increases the tendency for breakthrough innovation [39].

Based on Zuo [40] we draw Figure 1, which shows how the investment of Internet companies in the digital economy can improve the industry’s total factor productivity through two channels from the perspective of technology research, development processes, and market products. The first channel is technological breakthrough innovation. The enterprise creates discontinuities in technological development through creative destruction caused by technological mutations, replacing existing mainstream technologies [41] so that product architecture is innovated and has an impact on the existing technology market. The second channel is market breakthrough innovation, which includes market characteristics, breakthrough marketing, and business models. It focuses on customers in the high-end market and gradually penetrates the low-end market. The breakthrough product value is accepted by customers and gains rapid growth. To a large extent, it will be affected by the market size, structure, customer demand preferences, industry, and geographic distribution of the market [42]. Breakthrough marketing can find leading users, help companies develop breakthrough technologies, improve product concepts and designs, listen to the opinions of leading users in the trial production stage to improve products, and provide customers with more support to help them accept breakthroughs in the large-scale sales stage. Breakthrough marketing participates in the process of value discovery, creation, delivery, and recovery of breakthrough technology, creatively develops breakthrough product markets, influences business models, and promotes breakthrough technology innovation performance improvement [43]. The business model is a system about value realization that embodies the business logic implicit in the enterprise. It is the logical carrier of breakthrough technology, breakthrough marketing, and venture capital business ideas and commercial value creation in breakthrough technological innovation. The increase in total factor productivity will promote technical efficiency and the scale efficiency of Internet companies. The result is that the increase in total factor productivity stimulates Internet companies to increase investment in science and technology, causing a positive feedback mechanism.

2.2. Quantitative Model: Data Envelopment Analysis (DEA)

In 1978, American scholars Charnes and Cooper [44] first proposed data envelopment analysis (DEA). It is based on the concept of relative efficiency evaluation, a new method for efficiency analysis that evolved after continuous revision and innovation. It mainly uses valid sample data to measure production effectiveness according to different classifications (called decision-making units, or DMUs). First, this method needs to establish the frontier of production efficiency, place all relevant input and output indicators into a unified computing system, measure the maximum or minimum value in the decision-making unit, and find out the output and input based on the calculation results. The optimal ratio is ranked in order of magnitude after comparing the efficiency values of other decision-making units. The value of the decision-making unit is positioned on the production frontier, indicating that the efficiency of the decision-making unit is the most effective, represented by an efficiency value of 1. Correspondingly, the decision-making unit that cannot be positioned on the production frontier indicates that the efficiency value is invalid. The efficiency of its decision-making unit will be between 0 and 1. Fare and Grosskopf [45] applied DEA to estimate the factor productivity index, technical efficiency, and scale efficiency. Based on the concept of long-distance functions, the Malmquist index model can measure total factor productivity (TFP) [46]. Following John’s [47] analysis, the subject of super efficiency, slacks-based measure, network DEA, dynamic DEA, as well as banking and environmental studies are discussed in DEA application, research fronts are all focused on methodologies and techniques, including bootstrapping and two-stage analysis, undesirable factors, cross-efficiency and ranking, and network DEA, dynamic DEA, and SBM [48]. Banking performance has been examined with DEA, especially by combining DEA analysis with the Malmquist index [49,50,51]. Recently, scholars have made important breakthroughs in efficiency measurement. Based on the partial normal panel data model, Ye [52] uses SBM with unexpected output to calculate the green economic efficiency of 29 provinces in China. He and Zhao [53] calculated and evaluated the sustainable development efficiency of the Beijing–Tianjin–Hebei Region by using the super efficiency of the CCR-DEA model and the Malmquist index. The boot-strap DEA model was used to measure 31 Chinese Universities’ technology research efficiency and discuss their heterogeneity in the same area [54], and also compared with the traditional DEA model [55], which the result shows that the Bootstrap-DEA model is more effective. Gong [56] analyzed the logistics efficiency of the six Chinese provinces from the dynamic and static perspective by the DEA-Malmquist model. In summary, there is less research using the DEA-Malmquist index on Internet enterprises' efficiency; therefore, relevant research needs to be further discussed. Referencing the banking application of DEA, we use the DEA-Malmquist index model to examine the science and technology investment of Internet enterprises in the digital economy to measure their efficiency and find their heterogeneity among different industries.

The DEA-Malmquist index model is a non-parametric estimation method for measuring total factor productivity, that is, the Malmquist productivity index. When Malmquist productivity is greater than or less than 1, total factor productivity has a positive or negative growth rate from t to t + 1.

In the context of the digital economy of Internet companies, the purpose of this article is to evaluate the impact of technological investment on the efficiency of Internet companies in the context of the digital economy, as well as to estimate changes in the efficiency of Internet companies through the DEA-Malmquist index model. The model considers the changes in efficiency at different stages. The index model’s return to scale variable, input–output direction, and time t and t + 1 are all expressed by the following formula:

$$M\mathit{\text{(}}{y}_{t+1},x_{t+1},y_1,x_1\mathit{\text{)}}\mathit{=}\sqrt{\frac{D^t\left(x_{t+1,}{y}_{t+1}\right)\mathit{*}{D}^{t+1}\left(x_{t+1,}{y}_{t+1}\right)}{D^t\left(x_{t,}{y}_t\right)\mathit{*}{D}^{t+1}\left(x_{t,}{y}_t\right)}}$$

(1)

where x represents the input variable, y represents the output variable, and D (x, y) represents the independent evaluation unit. M ($y_{t+1},x_{t+1},y_1,x_1$) is the Malmquist index, which represents the change in total factor productivity (TFP) from t to t + 1. If M is larger (or smaller) than 1, it means that the unit production efficiency increases (or decreases) from t to t + 1. $D^t\left(x_{t+1,}{y}_{t+1}\right)$ represents the t + 1 stage efficiency level under technical factors at time t, and $D^t\left(x_{t,}{y}_t\right)$ represents the current level of efficiency at t technical factors. Using the DEA model, we calculated four indices, ($D^{t+1}\left(x_{t+1,}{y}_{t+1}\right)$, $D^t\left(x_{t,}{y}_t\right)$, $D^t\left(x_{t+1,}{y}_{t+1}\right)$, and $D^{t+1}\left(x_{t,}{y}_t\right)$, in model (1). Applying linear programming, the special model is as follows:

$${\left. [D^t\left(x_{t+1,}{y}_{t+1}\right)\right]}^{-1} =ma{x}_{\varphi \lambda }\varphi , \mathit{s.}\mathit{t.}-\mathit{\varphi }{y}_{i,t+1}+y_{0,t}\lambda \mathit{\ge }\mathit{0}, x_{i,t+1}-x_{0,t}\lambda \mathit{\ge }\mathit{0}$$

(2)

$${\left. [D^{t+1}\left(x_{t,}{y}_t\right)\right]}^{-1}=ma{x}_{\varphi \lambda }\text{ϕ}, \mathit{s.}\mathit{t.}-\mathit{\varphi }{y}_{i,t}+y_{0,t+1}\lambda \mathit{\ge }\mathit{0}, x_{i,t}-x_{0,t+1}\lambda \mathit{\ge }\mathit{0}$$

(3)

In the meantime, production efficiency can be divided into scale efficiency and technical efficiency, which are measured separately.

$$M_0=effch\times techch=pech\times sech\times techch$$

(4)

where effch represents the increase in productivity and measures the relative efficiency from t to t + 1. Techch involves the upgrading of technology, pech is the efficiency of pure technical factors, and sech is the change of scale efficiency.

3. Data Source and Variable Selection

3.1. Research Samples and Data Sources

This article is mainly based on the list of China’s top 100 Internet companies as published by the Internet Society in 2020, and at the same time, consults the CSMAR database and the annual reports of each company. We obtained 30 Internet companies as the research sample. The sample interval was 2011–2019 and included 14 Internet information service companies, 8 game media companies, 4 financial technology companies, and 4 manufacturing companies. The selected companies are shown in

Table 1. Sample enterprises.

Specific Field	Company Name	Amount (No.)	Proportion (%)
Information service	Lianluo Information, Century Huatong, Sanqi Interactive Entertainment, Ganglian E-commerce Holdings, Furun Digital Technology, Bestton Holding, Marine Technology, Zhewen Interactive Group, Tianjin Marine Shipping, Sinnet Technology, Focus Media Information, Er San Si Wu Network Holding Group, Tianrong Technology, Focus Technology	14	47
Game media	Youzu Interactive, Kingnet Network, Wangsu Science and Technology, Giant Network Group, Yaoji Technology, Perfect World, ZheShu Culture, Rastar Group	8	27
Financial technology	Spic Dongfang New Energy, Rendong Holdings, East Money Information, RoyalFlush Information Network	4	13
Manufacturing	Leo Group, Great Chinasoft Technology, Boton Technology, Oceanwide Holdings	4	13
Total	-	30	100

.

3.2. Index Selection

As shown in

Table 2. The input–output index system of Internet enterprise operation efficiency.

Indicator Type	Serial Number	Specific Indicators
Investment index	X1	Number of R&D employees (a)
	X2	Net fixed assets (CNY 100 million)
	X3	Net intangible assets (CNY 100 million)
Output indicators	Y1	Total revenue (CNY 100 million)
	Y2	Profit (CNY 100 million)

, the selected companies cover 30 companies in information services, game media, financial technology, and industrial manufacturing, which help to fully reflect the overall operating efficiency of the industry.

The basic idea of DEA analysis is to analyze the input and output of the production unit and then obtain the corresponding efficiency value. Therefore, this article refers to the research of Qiu [57] and Chan [58], among others, comprehensively considering the availability of Internet industry data, and the selected input and output indicators are shown in Table 2. The investment indicators of Internet companies were selected as the number of R&D employees (X₁), the net fixed assets (X₂), and the net intangible assets (X₃). All three indicators can represent the investment of Internet companies in the technology industry. The output variables selected total income (Y₁) and profit (Y₂). These two indicators represent the technological output factors of Internet companies.

4. Empirical Analysis Results

Based on Internet enterprise data, the Malmquist index represents the total factor productivity from t to t + 1, and dynamically takes a measurement of the continuous evolution of technology input efficiency in the digital economy. In the above empirical analysis, we found that the difference in the efficiency of Internet companies’ investment in science and technology influences their overall efficiency and income. The remainder of this section discusses specific analyses.

4.1. Analysis of Changes in Overall Dynamic Efficiency

Table 3. The 2011–2019 total factor productivity of Chinese enterprises′ digital economy and its decomposition.

Time	Technical Efficiency Change Index	Technology Progress Change Index	Change Index of Pure Technical Efficiency	Scale Efficiency Change Index	Total Factor Productivity Index
	effch	techch	pech	sech	tfpch
2011–2012	2.464	0.445	1.340	1.839	1.095
2012–2013	0.403	2.891	0.430	0.936	1.164
2013–2014	1.596	0.651	1.431	1.115	1.040
2014–2015	0.617	4.308	0.829	0.744	2.658
2015–2016	1.529	0.881	1.438	1.063	1.347
2016–2017	0.572	2.457	0.510	1.123	1.405
2017–2018	1.469	0.672	1.399	1.050	0.987
2018–2019	0.554	1.562	0.736	0.753	0.866
Mean	0.956	1.301	0.921	1.038	1.243

represents the empirical analysis of the total factor productivity of Chinese enterprises in the digital economy from 2011 to 2019 with the help of DEAP 2.1, and the reasons for the changes were analyzed based on the empirical results.

It can be seen from Table 3 that in the nine years from 2011 to 2019, the average growth rate of total factor productivity of Chinese enterprises was 24.3%. From the changes in the growth rate during this period, the efficiency of Chinese enterprises′ digital economy investment has increased rapidly. The technological progress change index and scale efficiency change index showed an upward trend, while the technical efficiency change index and the pure technical efficiency change index have declined. From 2011 to 2015, the overall trend of China’s corporate digital economy investment efficiency increased. From 2014 to 2015, China’s corporate digital economy investment efficiency increased by 165.8%, mainly due to the first year of the mobile Internet era in 2013, wherein the basic pattern of China’s digital economy took shape and entered a mature period. Traditional industries have begun to become Internet-based, and Internet model innovations have continued to emerge, injecting new vitality into China’s digital economy [59]. The possible reason is that the Guiding Opinions of the State Council on Actively Promoting the Internet + Action was issued in July 2015. Xi Jinping, General Secretary of the Central Committee of the Communist Party of China (CPC), delivered a series of important speeches on digital economy-related issues, and at the same time, various ministries and commissions intensively introduced the encouragement of the digital economy-related policies and guidance for development (accessed on 1 April 2020). From 2015 to 2016, total factor productivity dropped by 131.1% compared with the previous year, showing a downward trend from the previous year with technological progress changes. The main reason may be that after a short love period, a group of Internet-based industries represented by Internet medical services entered a disillusionment period [60]. From 2016 to 2017, total factor productivity showed a certain rebound compared with the previous year. The technology change index rose by 145.7%, and the scale efficiency change index rose by 12.3%. The reason for this may be that Premier Li Keqiang mentioned the digital economy in the government work report in March 2017 (accessed on 19 March 2020). It further reflects China’s high attention to the digital economy at the national level, and at the same time shows that the development of the digital economy has risen to the height of the national strategy [61]. From 2017 to 2018, the total factor productivity of Chinese enterprises’ digital economy input dropped by 0.13%, and the technological progress change index dropped by 32.8%. From 2018 to 2019, total factor productivity continued to drop by 13.4%, of which the technical efficiency change index, pure technical efficiency change index, and scale efficiency change index fell by 44.6%, 26.4%, and 24.7%, respectively. The cause may have been that demand growth slowed sharply, and the era of incremental competition entered the era of stock competition, causing some companies to reduce technological investment and technological innovation [62].

In May 2018, the National Internet Information Office of China released the Digital China Construction and Development Report (2017). The report data show that in 2017, the Chinese digital economy reached CNY 27.2 trillion, with a year-on-year increase of 20.3%, accounting for 32.9% of the GDP, becoming an important driving force for economic transformation and upgrading (accessed on 10 May 2018). Figure 2 shows the change curve of the technology investment efficiency of 30 Internet companies from 2011 to 2019.

4.2. Analysis of Dynamic Efficiency of the Internet Enterprises

This paper presents the DEA-Malmquist index and the average results of each decomposition index of 30 Internet companies in China from 2011 to 2019. Through the analysis of each decomposition index, we can further understand the dynamic change trend of technology investment in the background of the corporate digital economy.

From

Table 4. Malmquist index of each company.

Serial Number	Enterprise	Technical Efficiency Change Index	Technology Progress Change Index	Change Index of Pure Technical Efficiency	Scale Efficiency Change Index	Total Factor Productivity Index
		(effch)	(techch)	(pech)	(sech)	(tfpch)
1	Oceanwide Holdings	1.000	1.175	1.000	1.000	1.175
2	Oriental Energy	0.947	1.173	1.089	0.870	1.111
3	Focus Media Information	1.427	1.341	1.067	1.338	1.914
4	Leo Group	1.080	1.236	0.963	1.122	1.336
5	Youzu Interactive	0.776	1.211	0.912	0.851	0.940
6	Er San Si Wu Network Holding Group	1.268	1.194	1.011	1.254	1.513
7	Tianrong Technology	0.835	1.466	0.818	1.021	1.224
8	Lianluo Information	0.728	1.196	0.730	0.996	0.870
9	Focus Technology	0.921	1.360	0.743	1.239	1.253
10	Great Chinasoft Technology	0.724	1.424	0.675	1.073	1.031
11	Kingnet Network	0.624	1.259	0.619	1.008	0.786
12	Sanqi Interactive Entertainment	1.411	1.147	1.234	1.144	1.618
13	Giant Network Group	1.117	1.218	1.095	1.020	1.361
14	Century Huatong	1.000	1.294	1.000	1.000	1.294
15	Yaoji Technology	0.858	1.295	0.916	0.936	1.110
16	Perfect World	1.130	0.977	1.035	1.091	1.104
17	Rendong Holdings	0.554	1.068	0.609	0.909	0.592
18	Wangsu Science and Technology	0.750	1.193	0.733	1.024	0.895
19	Boton Technology	0.972	1.485	0.937	1.037	1.444
20	RoyalFlush Information Network	1.059	1.392	1.054	1.005	1.474
21	Rastar Group	0.694	1.258	0.745	0.932	0.874
22	East Money Information	1.000	1.288	1.000	1.000	1.288
23	Ganglian E-commerce Holdings	1.764	1.516	1.266	1.394	2.676
24	Furun Digital Technology	1.095	1.474	1.020	1.073	1.614
25	Zheshu Culture	0.699	1.324	0.822	0.851	0.925
26	Bestton Holding	1.097	1.520	1.065	1.030	1.667
27	Marine Technology	0.776	1.371	0.723	1.073	1.063
28	Zhewen Interactive Group	1.040	1.336	1.150	0.904	1.389
29	Tianjin Marine Shipping	0.990	1.715	0.943	1.049	1.697
30	Sinnet Technology	1.312	1.368	1.174	1.118	1.795
	Mean	0.956	1.301	0.921	1.038	1.243

, the average value of the total factor productivity index of digital economy input of 30 listed companies was 1.243. 14 companies, including Oceanwide Holdings, Rastar Group, Youzu Interactive, Kingnet Networks, Wangsu Science and Technology, and ZheShu Culture, were below the average level, indicating that the efficiency of game media Internet companies’ digital economy investment and utilization of funds is not high. During this nine-year period, the average Malmquist index of 23 companies, such as Focus Media Information, Er San Si Wu Network Holding Group, and Ganglian E-commerce Holdings, was greater than 1. Among them, 21 companies, such as Spic Dongfang New Energy and Leo Group, had a growth rate of more than 10%. In general, the financial technology investment of companies in technology is rising overall, and the digital economy of those Internet companies is developing well. Among them, technological progress and technical efficiency have improved, and they have played a role in promoting the overall efficiency of the companies’ digital economy, including 14 information service and financial technology companies such as Sanqi Interactive Entertainment and Zhewen Interactive Group, indicating the investment in digital economy technology in these industries. The funds are reasonably allocated and are fully utilized, promoting the transformation of scientific and technological achievements. A total of 15 companies, including Oriental Energy and Marine Technology, exist due to technological advances in the digital economy of Internet companies. Because these companies are equipped with relatively complete ICT infrastructure and relatively developed Internet information technology, they can effectively use the digital economy policy support, which is conducive to the improvement of the efficiency of digital economy investment by Internet companies. The digital economy efficiency of some information technology companies such as Focus Media Information and Great Chinasoft Technology even exceeds that of online game companies such as Kingnet Network and Wangsu Science and Technology. This shows that the construction of infrastructure, network integration, and the support of digital economic policies have produced positive results. The impact of the digital economy has promoted the flow of more digital economy resources to high-tech enterprises. These funds are effectively used by technology manufacturing enterprises to promote technological innovation, making the effect of digital economy development gradually prominent. Among the seven companies with a total factor productivity of less than 1 in the digital economy, this is mainly caused by the decline in the efficiency of technological changes. The management efficiency of technology-based businesses is low, and the development potential of other companies is insufficient. The development of the digital economy is not optimistic, and the efficiency of technological input and innovation is low. These companies should increase investment in the corporate digital economy, expand digital infrastructure construction, broaden digital integration channels, transform traditional management and operation methods, adjust business structure layout, promote the rational allocation of digital resources, and increase production capacity.

According to partitioning, density and model clustering algorithms can be divided into multiple types [63].

Table 5. Basic information table of clustering categories.

Clustering Category	Frequency	Percentage (%)
Cluster 1: Leading Group	2	6.67%
Cluster 2: Following Group	8	26.67%
Cluster 3: Ordinary Group	14	46.67%
Cluster 4: Backward Group	6	20.00%
Total	30	100%

uses cluster analysis to classify the samples. Using the K-means cluster analysis method, we divided the samples into four groups. The leading group comes from the information service industry, which had the highest efficiency in its science and technology input. The following group includes eight enterprises with relatively high efficiency in financial technology and information service industries. The ordinary group possesses average efficiency in game media and manufacturing industries, while several enterprises are in information service industries. The backward group has sux enterprises with low overall efficiency in game media and manufacturing industries. As can be seen from Figure 3, each group represents the number of companies ranked in the quartile of high technology input total factor productivity among the 30 companies in the digital economy background, each accounting for ratios of 6.67%, 26.67%, 46.67%, and 20%. Figure 4 further supports the conclusion that TFP significantly differs between cluster groups. Specifically, for the highest item, the effch has the highest value, the Internet enterprises possess continuous investment in science and technology, they improve technical efficiency through innovation, which can be seen through their innovative service and products. While techch, the backward item, contains largely game media and manufacturing enterprises that likely cannot afford expensive investment or do not prepare well for digitalization transformation, technology progress could not make a greater contribution.

After obtaining the cluster analysis shown in

Table 6. Analysis and comparison results of cluster classification variables.

Malmquist Index	Comparison Results of Variance Analysis of Cluster Categories (Mean ± Standard Deviation)				F	p
	Cluster_1 (N = 2)	Cluster_2 (N = 8)	Cluster_3 (N = 14)	Cluster_4 (N = 6)
Technical efficiency change index	1.60 ± 0.24	0.71 ± 0.09	1.11 ± 0.13	0.87 ± 0.11	35.39	0.000 **
Technology progress change index	1.43 ± 0.12	1.23 ± 0.08	1.27 ± 0.14	1.47 ± 0.13	5.56	0.004 **
Change index of pure technical efficiency	1.17 ± 0.14	0.76 ± 0.12	1.06 ± 0.08	0.81 ± 0.11	22.864	0.000 **
Scale efficiency change index	1.37 ± 0.04	0.94 ± 0.07	1.05 ± 0.10	1.08 ± 0.08	13.81	0.000 **
Total factor productivity index	2.29 ± 0.54	0.87 ± 0.15	1.41 ± 0.21	1.29 ± 0.25	23.246	0.000 **

** p < 0.05.

, the system derived a strip of data to represent each cluster group. The K-means algorithm has advantages in simple mathematical ideas, fast convergence, and easy implementation in the partition-based clustering algorithm than others [64]. To explain the characteristics of each classification, we used variable analysis to study the differences of each classification group, and finally, in combination with each group’s own characteristics, each group was named, and variable analysis was used to reveal the different characteristics of each group [65]. It can be seen from Table 6 that each type of grouping is very important for the research item (p < 0.05), which means that each item in the four groupings has very different characteristics in terms of the research item.

The cluster center is the mathematical theory or intermediate process index of the clustering algorithm, which is of little practical significance for analysis. First, the initial clustering refers to the first cluster center value obtained by the algorithm clustering; second, the final cluster center refers to the final cluster center determined after multiple iterations of the algorithm.

Figure 5 presents an efficiency analysis of pure technical efficiency and scale efficiency of 30 Internet companies, and it can be seen that the trend of scale efficiency is more stable among the 30 enterprises, and nearly half of the enterprises have a scale efficiency of more than 1. The pure technical efficiency has achieved a very high level among a few enterprises, and the rest of them are very low.

Table 7. Cluster centers.

Term	Initial Cluster Center				Final Cluster Center
	Cluster_1	Cluster_2	Cluster_3	Cluster_4	Cluster_1	Cluster_2	Cluster_3	Cluster_4
Technical efficiency change index	1.597	−1.559	0.395	0.703	2.281	−1.044	0.462	−0.445
Technology progress change index	−0.568	−1.461	−1.444	1.204	0.787	−0.552	−0.252	1.062
Change index of pure technical efficiency	0.354	1.231	0.444	−1.601	1.269	−0.987	0.697	−0.733
Scale efficiency change index	0.346	0.458	0.133	2.554	2.449	−0.818	−0.003	0.28
Total factor productivity index	3.064	0.14	1.099	0.123	2.407	−1.034	0.264	−0.038

Sum of squares of error SSE: 50.400.

illustrates the great differences in the efficiency of Internet companies’ technological investment. Information service companies such as Sanqi Mutual Entertainment and Shanghai Steel Union have reached the forefront of productivity, and Zhewen Internet and Haike B have achieved higher pure technical efficiency, 10% higher than other companies. Zhejiang Furun’s scale efficiency was found to be 10% higher than other companies. The game media and financial technology companies, including Oceanwide, Perfect World, Flush, Oriental Fortune, Baotong Technology, and Yaoji Technology, have great potential to improve their pure technical efficiency and scale efficiency. The efficiency index of Kaiying Network, Chinasoft Technology, and Rendong Holdings is at the lowest level. The low investment in digital economy technology of these companies means that these companies should increase investment in scale and operational efficiency. Internet companies increase their investment in digital economy infrastructure, ICT information technology, and 5G networks, among others, which can improve their operational, production, and management efficiency, and therefore can also promote the efficiency of products and services.

These results show the difference between the level of investment in technology and the overall efficiency of Internet companies in the context of the digital economy, which will ultimately affect their economic performance and productivity. The increasingly fierce competition among enterprises has also greatly promoted enterprises to pay attention to business innovation and improve operation, management, and production efficiency through technological investment.

We chose pure technical efficiency and scale efficiency as the focus of our comprehensive analysis. Through these analyses, we were able to ascertain that technical efficiency has an impact on the efficiency of Internet companies. As shown in Figure 5, we were able to intuitively see the difference between technical efficiency and scale efficiency of different types of Internet companies.

5. Conclusions and Suggestions

As a continuation of the information technology revolution, the digital economy has brought about huge impacts on business models and production efficiency, which provide Internet companies with opportunities for in-depth development. Integrating and developing new digital technologies is obviously one of the common challenges faced by many organizations at this stage. Based on the theory of innovation, this article systematically established the impact of technological investment on enterprise innovation efficiency in the context of the digital economy, obtaining a sample of 30 Internet companies from 2011 to 2019 based on the list of the top 100 Internet companies published by the Internet Association (2020) and the annual reports of each company.

After finding the heterogeneity of different types of Internet companies, we made the following conclusions: (1) Under the digital economy, investment in science and technology has promoted the increase in the total factor productivity of Chinese Internet companies. (2) Different types of Internet companies have differences in terms of the efficiency of digital economic investment: technology-based information service companies have the highest efficiency in terms of digital economy technology investment, followed by manufacturing companies and financial technology companies, and game media companies are relatively weak. However, even in the same industry, some enterprises possess high efficiency of science and technology in the digital economy while others do not. That all depends on their ICT infrastructure construction, technology application, and effective use of the digital economy policy support to improve technological progress and technical efficiency. Based on these conclusions, we put forward the following policy recommendations:

First, Internet companies should persist in breakthrough innovation and gradually deepen their digital transformation and upgrade. In recent years, Internet companies have significantly reduced transaction costs and information asymmetry with their unique Internet plus advantages, creating huge social benefits. However, with the gradual rise of the Internet in all walks of life, the Internet bubble has become increasingly more obvious. The value created is also gradually decreasing. Therefore, the digital economy promotes the industrial Internet, and while the Internet industry is deepening innovation, it should continue to strengthen the awareness of breakthrough technological and market innovation, establish effective innovation mechanisms, deepen corporate organizational and management changes, and achieve in-depth transformation of the digital economy as well as contribute Smart plus to corporate innovation and development.

Second, Internet companies should eliminate the original curse of focusing on business model innovation instead of investing in technological innovation. Through the innovation of business models, Internet companies have defeated the industry monopoly in the past, injected new vitality into the development of the Internet, and at the same time planted a long-term development restriction trap. In the era of the digital economy, Internet companies cannot rest on their laurels. They should actively seek change; endogenously transform the challenges of the digital economy as the driving force for their own reforms; comprehensively promote technological innovation, product innovation, and service innovation, and use breakthrough innovative thinking to adapt to intelligence trends of the era. At the same time, Internet companies also objectively recognize their own advantages in internalization and informatization, learn from each other’s strengths, build diversified platform service models, quickly respond to market demand, and continuously improve their overall strength.

Third, the government should vigorously provide announcement products that support the development of the digital economy and continue to promote China’s digital economy transformation process in addition to policies. On the one hand, it is necessary to start with industry norms and to promptly introduce relevant laws and regulations and management regulations to promote the sustainable and healthy development of the digital economy; on the other hand, it is necessary to focus on infrastructure construction and increase the focus on 5G communication technology, Beidou positioning, the Internet, and Blockchain. The basic research and development of other technologies ensures the application of landing scenarios and guarantees that the development of the digital economy is not restrained by the lagging development of basic information technology.