Introduction

At present, mankind is experiencing a new round of technological revolution with DGPTs as the core. The progress of digital technology and its series of innovative applications have not only become an important driving force for economic development and industrial upgrading, but also driven the global economy into the era of digital economy. As the digital economy exhibits a unique strategic position in economic development and participation in global competition, more and more countries foresee that the development of the digital economy will drastically change the original pattern of distribution of world interests and reshape the main forces in international competition. Against this backdrop, more and more countries are seizing the opportunities for the development of the digital economy, strengthening the construction of digital infrastructure, promoting domestic digital transformation, and attaching importance to inter-country competition in the digital industry1. The digital economy has gradually become the core area and strategic high ground of the new round of international competition. Specifically, From the “Made in China 2025” strategy to Germany’s “Industry 4.0 Plan”, the United States’ “Industrial Internet”, and the United Kingdom’s “UK Industry 2050 Strategy”, DGPTs is developing dramatically and widely used around the world. From the perspective of technology application, DGPTs, as a class of digital technologies with broad coverage, wide application scenarios, strong complementarity and complex evolutionary characteristics2, are the key technological basis for the current digital economy. Among them, mobile Internet, as a DGPT with high penetration rate and extremely wide application scenarios, has gradually replaced broadband Internet and become an indispensable tool for economic society and people’s life. According to The State of Mobile Internet Connectivity Report 2024 published by the Groupe Speciale Mobile Association (For more details visit: https://www.gsma.com/r/somic/), by the end of 2023, the number of people using mobile Internet increased to 4.6 billion people (57% of the global population). Although not all of them are active users, it still shows the high penetration of mobile Internet applications in the world. As for the national level, China, for example, which ranks among the world’s top countries in terms of the size of mobile Internet users, mobile Internet has achieved leapfrog development in the past decade: as shown in Fig. 1, from 2014 to 2021, the number of 4G base stations in China jumped from 0.85 million to 5.9 million; the scale of mobile Internet data traffic usage grew from 210 million GB to 22.16 billion GB; the average mobile data usage per household per month increased from 0.2 GB/household·mouth to 13.4 GB/household·month. The widespread availability of mobile Internet has given rise to many innovative mobile applications and services, and the scale of information consumption has surged, significantly boosting the development of China’s digital economy. Consequently, as a crucial new-generation DGPT to promote the digital upgrading of the real economy, the impact of the mobile Internet on the economic society has also been the focus of attention from all walks of life in recent years. The leapfrog development of China’s mobile Internet in the last decade has also provided a natural testing ground for scholars to explore the economic effects of the mobile Internet.

Fig. 1
Fig. 1
Full size image

4G base station construction and mobile Internet data traffic usage in China during 2014–2021.

The global economy is currently struggling to recover from multiple negative factors such as COVID-19 and the Ukraine crisis and is showing resilience, but growth is still slow and still faces considerable downward pressure, which has led to a continued rise in employment concerns. Not only that, the current progress of DGPTs and the development of the digital economy are also driving the economic society to undergo a systemic change, which will inevitably have significant shocks in many areas such as investment, production, distribution and consumption. In the process of this transformation, enterprises in all sectors will have to respond and adjust their business strategies, either proactively or reactively, which will undoubtedly have remarkable impacts on their labour demand. Enterprises, as microcosmic bodies that carry the employment of the working population, are crucial to creating employment and stabilizing people’s livelihood. In the current wave of the new digital revolution with digitalization, networking and intelligence as the main features, how digital technology applications affect firm labor demands has not only caused extensive discussions in academia, but also become the focus of government authorities. As a key DGPT supporting the digital economy, mobile Internet not only provides an important basic digital environment for firms’ digital transformation, but also the new business model and new economy generated by its series of innovative applications have profound impacts on employment environment and employment form, which has become a crucial factor affecting the current employment situation. Consequently, in the above development context, it is of great theoretical and practical reference value to analyze the effect of mobile Internet on firm labor demands, and to empirically test their correlation paths between them.

There are two distinct views on the impact of digital technologies on employment that have long existed in academia. One is that digital technologies will reduce employment by saving labor and displacing some jobs (e.g., Refs.3,4),the other is that digital technology applications, while displacing some jobs, simultaneously create more new jobs and thus increase the amount of employment in the economy (e.g., Refs.5,6). Digital technologies, as the widely used skill-biased technological change (SBTC) at this stage, scholars also realize that in addition to affecting the aggregate employment level, digital technologies also provide new opportunities for restructuring and changing the employment structure. Some scholars, based on the SBTC hypothesis, pointed out that digital technology adoption increases the demand for high-skilled labor but reduces the size of low-skilled labor employment in firms7,8. Some scholars have also argued from the routine-biased technological change (RBTC) hypothesis that RBTC represented by digital technologies will replace jobs in routine tasks and increase labor demand in non-routine tasks, thus causing structural changes in employment9,10. However, current research findings are not consistent as to whether this process will promote or reduce employment. Some scholars argue that while digital technology applications somewhat replace labor in routine tasks, they also contribute to net employment growth by creating more new jobs11,12. Some scholars also found that the employment destruction effect of digital technologies on the employment restructuring process is greater than its creation effect, thus reducing the employment level of the economy13,14.

As a new-generation DGPT, the economic effects of mobile Internet development have also attracted the interest of scholars. Currently, the positive impacts of mobile Internet on economic growth have been confirmed by scholarly studies15,16. In addition, scholars have explored the impact of mobile Internet penetration and related innovative applications on labor productivity17,18, interpersonal digital divide19, amelioration of poverty20, and farmers’ income21. Extant studies provide a valuable reference for understanding the economic effects of the mobile Internet. However, it is worth pointing out that studies on the economic effects of mobile Internet are still lacking and scattered. Although extant studies have continuously focused on the relationship between digital technologies and employment, no research has been conducted in the literature on the employment effects of the mobile Internet. Therefore, there is a lack of corresponding empirical analysis of the actual impact of the current rapid diffusion of mobile Internet on the employment volume and employment structure. Lastly, in terms of the practical significance of the research, the conclusions of extant studies on the employment effects of various digital technologies are inconsistent, so it is hard to provide a theoretical reference for answering the employment effect of the mobile Internet and the guiding significance for the real situation is weak. Consequently, how does the mobile Internet affect employment? What are its main influencing paths? What are the patterns and heterogeneity of its impact on employment structure? These questions have not been answered by existing studies and need to be further analyzed. The above limitations of existing studies provide an extension of this article.

Compared to existing studies, the marginal contributions of this study are in the following three aspects: First, in terms of research perspective, there is still a gap in the study of the impact and path of mobile Internet on employment. This paper attempts to conduct research from theoretical mechanisms and empirical analysis, and to make a useful supplementary to the research in the field of employment effects of digital technology applications from the micro level. Second, in terms of research methodology, taking China’s A-share listed firms as the research object of employment results, we apply the city statistics to construct a mobile Internet development index for each city based on the characteristics of mobile Internet development in China, and identify and quantify the causal relationship between mobile Internet development and firms labor demand by selecting relevant historical data as instrumental variables (IV). Third, in terms of research content, after evaluating the average effect of mobile Internet on firm labor demands, the paper further explores the actual impact path of mobile Internet on firm labor demands, analyzes the impact of mobile Internet on firms’ labor demand structure, so as to provide more comprehensive empirical evidence.

The remainder of this studies is structured as follows: The next section is the theoretical mechanism analysis. The third section provides details on the methodology and data used. The results of the main regression analysis are reported and discussed in "Empirical results and discussions". The fifth section presents and discusses the results of a series of impact path analyses, employment structure analyses. The final section concludes and presents the implications of economic policies.

Mechanism analysis and research hypothesis

Analysis of the employment effects of mobile internet development

Mobile Internet is not only the key DGPT to promote the current digital upgrading of the economic society, but the mobile Internet economy spawned by its deep integration with the real economy is also an essential part of the digital economy. The development of the mobile Internet has brought about an intelligent and convenient digital life, and significantly reshaped business models and market patterns, which directly or indirectly leads to significant changes in the employment environment and employment structure, thus profoundly affecting the level of employment in the economy. Firstly, based on the SBTC hypothesis, as the widely applied SBTC currently, mobile internet development may directly change the ratio of inputs to capital to labour in the economy, thus changing the level of labour demand22. Moreover, mobile Internet applications may be skill-biased, which can have different degrees of creation and destruction effects on labour demand for different skill types7, thus modifying the structure of firms’ labour demand at the micro level. Secondly, according to the theory of industrial structure evolution, mobile Internet, as a new generation of DGPT widely used currently, can effectively promote the digital upgrading of industries and the efficiency of industrial organization operation, thus promoting the upgrading of industrial structure1, which significantly changes the labor demand of different industries. Thirdly, according to transaction cost theory, mobile Internet applications have contributed to making market information more open and transparent, significantly reducing the cost of information search and processing for both transaction parties and improving the information asymmetry of transactions20, which has helped to reduce employment market friction and improve employment efficiency. As a result, the current rapid development of the mobile Internet will significantly affect the level of employment in the economy. Specifically, the development of the mobile Internet will have employment creation and employment destruction effects.

As the DGPT is currently widely applied, the employment creation effects of mobile Internet development are mainly reflected in the following aspects: First, the current prosperous development of the mobile Internet and the rapid growth in the scale of mobile users have greatly facilitated the development and expansion of new industries and businesses with the mobile Internet at their core, which has not only generated many new labour demands but also created a series of new jobs23. Secondly, the integration of the mobile Internet with the real economy further promotes the process of industrial digitization, which not only leads to an increasing labour demand for digital industries related to the mobile Internet, but also gives rise to a large number of related jobs in the emerging industries derived from the integration of the mobile Internet with traditional industries, thus raising the economy’s labour demand for digital skills. Thirdly, the popularity of mobile Internet technology allows people to access the latest information on the job market more conveniently and quickly on the mobile, reducing the cost of obtaining employment information and the cost of searching for jobs, improving the employment opportunities of job seekers, and thus increasing the total employment level of the economy.

Although the development of the mobile Internet contributes to the digital upgrading of industries, it has a certain negative impact on some employment positions in traditional industries and potentially changes the skill structure of labour demand in firms, thus inevitably generating employment destruction effects. It is specifically reflected in the following three aspects: Firstly, the development of mobile Internet has promoted the digital upgrading of traditional industries, leading to an increase in the labor demand for digital skills in these industries, but it will also have a substitution effect on non-digital skill positions in some traditional industries, thereby reducing the level of labor demand in these industries. Secondly, due to the biased characteristics of mobile Internet technology3, with the penetration of mobile Internet technology, routine tasks in many industries can be transferred to mobile terminals and even some tasks can be completed entirely on mobile terminals, which accelerates the replacement of traditional routine work tasks and jobs, and reduced the scale of people engaged in routine tasks in the economy. Thirdly, mobile Internet applications have accelerated the digital upgrading of enterprise management modes, improving the management efficiency of enterprises by promoting the transfer of daily management and work handover to mobile terminals, thus facilitating enterprises to save labour costs. It led enterprises to optimize their employment structure by cutting redundant staff, which reduced their labour demands4.

Based on the above analyses, the main features of the employment effects of mobile Internet development can be summarized by Fig. 2. Further, we propose the following hypothesis:

Fig. 2
Fig. 2
Full size image

Main impact mechanisms of mobile internet development on employment.

H1: Mobile Internet development will promote employment growth when its employment creation effect is larger than its employment destruction effect.

H2: Mobile Internet development will reduce employment when its employment creation effect is smaller than its employment destruction effect.

Analysis of the impact path of mobile internet development on employment

Mobile internet development is driving local digital upgrades and providing a fundamental digital environment for firms’ operations, which has a profound impact on business decisions. Broadly speaking, mobile internet development may significantly change firm labor demands through the following channels:

  1. (1)

    Facilitating the digital transformation of firms.

Mobile Internet development effectively enhances the accessibility of mobile application services, reduces application costs, and improves the supply quality of mobile service. Consequently, the greater the investment related to mobile Internet technology in the process of digital upgrading of enterprises, the greater the impact of regional mobile Internet development will be. In addition, the rapid upgrade of mobile smart terminals in recent years has made mobile Internet functions and applications more mature and diversified, promoting the upgrade and increasing demand of mobile Internet applications in multiple fields. With the enrichment and expansion of mobile Internet application scenarios, especially the emergence of the mobile Internet of Things and its innovative applications, the network externality of the mobile Internet has increased rapidly, which has significantly increased the potential benefits of enterprise management models to implement digital transformation based on the mobile Internet, thus driving the digital transformation of enterprises24.

The relationship between the digital transformation of enterprises and their labor demands has been studied and discussed in the literature. The implementation of digital transformation by firms will have two different effects on their labor demands. On the one hand, the digital transformation of enterprises will reduce their labor demand. According to the SBTC hypothesis, the application of digital technology can generally significantly improve the marginal productivity of highly skilled workers who can use digital technology, so that enterprises will reduce the labor demand for low-skilled workers8. According to the RBTC hypothesis, enterprise digital technology access will replace the routine task works with repetitive and proceduralization, and will complement non-repetitive and non-programmatic work9. Therefore, digital transformation will reduce the employment volume of routine tasks of enterprises and correspondingly increase the labor demand for non-routine tasks, so that it may generally reduce the employment volume of enterprises.

On the other hand, in the process of integrating mobile Internet and business operation, although some jobs will be replaced, digital technology will also improve the productivity of low-skilled labor and improve the output efficiency of enterprises, so that enterprises will still employ a certain number of low-skilled laborers to engage in non-digital skills-related jobs. In addition, the digital transformation of enterprises will increase the demand for high-skilled labor, so the scale of employment in enterprises may be comparable to the original or even increase. For example, Pantea et al.25 found that ICT investments by manufacturing and service firms in seven EU countries did not reduce the amount of labor they employed. Balsmeier and Woerter11 found that firms’ access to digital technology generally increases their demand for high-skilled labor and reduces the demand for low-skilled labor, but the net effect is slightly positive. Besides, as the digital transformation deepens, the gradual shift of enterprises’ management models and processes to mobile will replace some jobs while also creating new tasks and job content that match mobile applications, which in turn will generate new jobs and increase firm labor demands26. Based on the above analysis, we hypothesize that:

H3: Mobile Internet development can affect the labor demand of enterprises by facilitating their digital transformation.

  1. (2)

    Improving the productivity of firms.

First, mobile Internet makes information receiving and sending free from time and space constraints, which greatly enhances people’s information accessibility, improves the real-time and interactive information exchange, reduces information asymmetry and information exchange costs27, thus helping enterprises achieve cost reduction and efficiency improvement. Second, by transferring daily management and business processes to mobile, enterprises can improve the efficiency of corporate communication and inter-departmental business cooperation, and improve the efficiency of business management practices and organizational coordination, making business operations more rational and efficient28.

Regarding the impact of the productivity improvement of enterprises on their labor demands, there are mainly the following two directions: The first is the negative employment substitution effect. An increase in firm productivity, which means a decrease in the unit time or unit cost required to produce a given quantity of goods or services, will reduce its labor demands to further increase economic profits without a significant expansion in the size of output3. The second is the positive employment creation effect. Because of the complementary effects of mobile Internet applications and high-skilled labor, to further strengthen the productivity effects of digital transformation, firms tend to increase their demands for high-skilled labor. Especially when the cost of DGPTs access is low, the increase in firms’ productivity will cause the increase in labor demand to be significantly larger than its substitution effect on labor, thus contributing to firms’ employment growth29. Based on the above analysis, we put forward the following hypothesis:

H4: The development of mobile Internet can change firms’ labor demands by promoting the improvement of their productivity.

  1. (3)

    Promote the expansion of firms’ production scale.

First, in the early stage of access to mobile Internet, firms need to invest in the hardware and software of mobile Internet applications and allocate corresponding human capital, resulting in a surge in the operating costs, which is not conducive to the improvement of enterprises’ performance and output scale. Consequently, firms may reduce the total operating costs by reducing the size of employees, which will have a destructive effect on different types of jobs. However, with the completion of the digital upgrading of enterprises, mobile Internet applications will reduce operating costs by eliminating some old technologies and make full use of existing resources to expand production scale28. The expansion of the production scale of firms will lead to an increase in labor demand. Furthermore, firms will reduce costs and increase efficiency by accessing mobile Internet, which makes the products have a more competitive advantage in the market, thus expanding their market share30. Therefore, firm’s production scale will be further expanded, and the corresponding labors demand of firms will further increase. Through the above analysis, the following hypothesis is proposed in the paper:

H5: Mobile Internet development can contribute to the growth of firm labor demand by expanding its production scale.

  1. (4)

    Promote firms to expand the scope of business.

Frist, mobile Internet applications can reduce transaction costs by saving people’s information processing time and improving the timeliness of various trading activities. The reduction of transaction costs can promote enterprises to break through the original boundaries of production possibilities and start new businesses, thus expanding the business scope of enterprises24. Second, the popularization of mobile Internet improves the transparency of market information and facilitates consumers’ feedback on product experience, which can not only provide more timely information for firms to improve their original products, but also help firms better understand the specific needs and characteristics of customers and realize the division of consumer groups, thus helping firms to develop differentiated and personalized products. Third, mobile Internet access greatly improves firms’ information collection abilities, enables firms to understand the dynamic changes in the market in a more timely and accurate manner, and further establishes the direction of product design and R&D according to the needs of users, which can reduce the trial-and-error cost of innovation and encourage firms to develop new products, thus improving the willingness of firms to expand new businesses.

The business scope of the firm directly affects its labor demand. On the one hand, with the expansion of the business scope of the firm, it is difficult for the original number of employees to fully cope with the rapid increase in workload after this expansion. To ensure the efficient operation of the firm, the firm needs to recruit new labor force to expand the workforce, thus promoting the growth of the employment scale of the firm. On the other hand, in order to improve internal resource integration ability and operating efficiency, the enterprises will also set up new jobs according to the newly launched business type after expanding their business scope, so as to create more employment opportunities for the economy. For example, Harrison et al.31 found that the development of new products by enterprises in France, Germany, Spain and the United Kingdom between 1998 and 2000 had a significant positive impact on their labor demands, that is, the employment of enterprises increased by an average of 4.8–8.0%. From the above analysis, we propose the following hypothesis:

H6: The development of mobile Internet will promote the expansion of enterprises’ business scope, thus increasing the scale of their employment.

On the basis of the above analyses, the main paths of mobile Internet development influencing firm labor demands can be summarized in Fig. 3.

Fig. 3
Fig. 3
Full size image

The main paths of mobile Internet development on firm labor demands.

Methodology and data

Baseline econometric model specification

To quantitatively evaluate the impact of mobile Internet development on firm labor demands, based on econometric theory, a fixed effects panel data model we establish is as follow:

$${Labor}_{ijt}={\beta }_{0}+{\beta }_{1}{Mobile}_{it}+{{{\varvec{X}}}_{{\varvec{i}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{1}+{{{\varvec{Z}}}_{{\varvec{i}}{\varvec{j}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{2}+{\alpha }_{i}+{\tau }_{t}+{\varepsilon }_{ijt}$$
(1)

in model (1), Labor represents firm labor demands; Mobile stands for regional mobile development; The vectors X and Z denote a set of city-level control variables and firm-level control variables, respectively. The subscripts i, j and t denote the city, the firm and the year, respectively. \({\beta }_{k}\) and \({{\varvec{\Gamma}}}_{{\varvec{k}}}(k=\text{0,1},2,\dots )\) refer to the coefficients of independent variables. \({\mu }_{i}\) indicates city fixed effects, while \({\tau }_{t}\) denotes time fixed effects. \({\varepsilon }_{it}\) indexes the residual stochastic disturbance term.

Mechanism verification method

Based on the previous mechanism analysis, we intend to perform empirical tests to explore the channels through which the mobile Internet affects firm labor demands. Drawing on the research ideas of existing studies32,33, this paper uses the mediating effect model to carry out the empirical analysis, and the model constructed is shown below:

$${Path}_{ijt}={\beta }_{10}+{\beta }_{11}{Mobile}_{it}+{{{\varvec{X}}}_{{\varvec{i}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{11}+{{{\varvec{Z}}}_{{\varvec{i}}{\varvec{j}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{12}+{\alpha }_{i}+{\tau }_{t}+{\varepsilon }_{ijt}$$
(2)
$${Labor}_{ijt}={\beta }_{20}+{\beta }_{21}{Mobile}_{it}+{{{\varvec{X}}}_{{\varvec{i}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{21}+{{{\varvec{Z}}}_{{\varvec{i}}{\varvec{j}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{22}+{\alpha }_{i}+{\tau }_{t}+{\varepsilon }_{ijt}$$
(3)
$${Labor}_{ijt}={\beta }_{30}+{\beta }_{31}{Mobile}_{it}+\gamma {Path}_{ijt}+{{{\varvec{X}}}_{{\varvec{i}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{31}+{{{\varvec{Z}}}_{{\varvec{i}}{\varvec{j}}{\varvec{t}}}}{\prime}{{\varvec{\Gamma}}}_{32}+{\alpha }_{i}+{\tau }_{t}+{\varepsilon }_{ijt}$$
(4)

where Path indicates the path variables, which is then explicitly defined according to the content of the subsequent empirical tests; \(\gamma\) denotes the coefficient to be estimated for Path; the remaining variables are defined in line with the previous section. It should be noted that since the path variable Path in model (4) may be endogenous, we draw on Nunn and Wantchekon’s32 research idea to focus only on the correlation between Path and Labor in model (4) after identifying the causal relationship between Mobile and Path through model (2), and does not overemphasize the validity of the empirical test, but treats it as some kind of tentative evidence or circumstantial evidence. Consequently, this study will combine the results of this test with the previous mechanism analysis to analyze the channels through which mobile Internet development indirectly affects firm labor demands in an integrated manner.

Data sources

The data set used for the empirical research in the paper is composed of city-level statistics in China and the data released by China’s A-share listed companies. The statistical data of the cities come from China City Statistical Yearbook and the statistical yearbooks of China’s provinces. The data of China’s A-share listed companies were taken from the annual reports of each listed company and the CSMAR and Wind databases.

To ensure the accuracy and reliability of the empirical results, referring to existing studies12,27, initial data were pre-processed as follows: (1) we exclude “ST” (special treatment) and “*ST” company samples, (2) considering the special characteristics of the financial industry and its differences in accounting standards, we omit financial firms; (3) the samples that have been listed for less than a year are omitted; (4) we removed samples of data missing for many years; (5) exclude companies with financial abnormalities such as total assets less than 0, net assets less than 0, operating income less than or equal to 0, and gearing ratio greater than 100%; (6) delete companies with less than 1 employee. Through the preprocessing of the data, our research sample includes 285 prefecture-level cities and 3,840 China’s A-share listed enterprises in 2015–2019.

Variable measurement

  1. 1.

    Scale of Firm labor demand (Labor). This study applies the natural logarithm of the total number of employees + 1 for each listed firm as the measure of Labor.

  2. 2.

    The level of mobile Internet development (Mobile). Regarding the measurement of mobile Internet development, Bertschek and Niebel17 used the percentage of employees whose companies were equipped with mobile terminals (e.g., smartphones, tablets, and laptops) to access the Internet via cellular networks as a measure of mobile Internet access. Del Gaudio et al.34 used the number of mobile cellular network subscriptions per 100 people to measure the level of mobile Internet access in EU countries. Khan et al.21 applied the natural logarithm of the number of mobile cellular subscribers as a measure of mobile phone usage.

The existing literature provides a reference basis for this study to scientifically measure the level of regional mobile Internet development, but they still have limitations. This study considers that the development of mobile Internet in a region is a systematic project, covering numerous dimensions such as infrastructure supply, application scale, and popularization rate. Consequently, it is not allowed to accurately and comprehensively measure the level of regional mobile Internet development with a single indicator. Thus, this study measures the mobile Internet development level of each prefecture-level city by constructing a more systematic and operable regional mobile Internet development comprehensive evaluation index system, as shown in Table 1.

Table 1 Comprehensive evaluation system of regional mobile Internet development.
Full size table

To avoid the subjectivity of assigning weights to the sub-indicators in the above evaluation system and to ensure the rationality of assigning weights which assign higher weights to indicators with a larger amount of information, following the practice of extant studies33, we apply the entropy method to assign objective weights to the above sub-indicators in Table 1.

  1. 3.

    Control variables (X, Z). To ensure the accuracy and robustness of empirical results, this paper draws on the existing research (e.g., Refs.5,12,13,14,27) to control the factors that may affect firm labor demands from the urban and enterprise levels respectively:

    1. (1)

      Broadband Internet penetration (Internet). Considering that extant studies have confirmed that Internet penetration is an important factor affecting employment5,12, this factor is controlled in the empirical analysis, and the natural logarithm of the number of Internet broadband access users is applied as a measure.

    2. (2)

      Industrial structure (Industry). The ratio of added value of tertiary industry to added value of the secondary industry in each city is used as a proxy for industrial structure.

    3. (3)

      Foreign investment openness (FDI). In this paper, the share of FDI in GDP is used as a proxy.

    4. (4)

      Investment intensity (Invest). We adopt the ratio of fixed asset investment to GDP as a measure.

    5. (5)

      Level of economic development (lnY). The natural logarithm of real GDP per capita is used to control the economic development of each city.

    6. (6)

      Human capital supply (HR). Under the constraint of data availability, the paper uses the natural logarithm of the number of college students in each region to measure its human capital supply.

    7. (7)

      Government behavior (Govern). This study adopts the existing literature’s usual practice of using the ratio of local general budget expenditures to GDP as a measure.

    8. (8)

      Firm age (Age). We take the logarithm of the current year of operation of the enterprise minus the year of its establishment + 1 as a measure.

    9. (9)

      Production scale (Scale). Due to the lack of output data of various firms, this paper takes the natural logarithm of total operating income as a measure.

    10. (10)

      Firm productivity (TFP). We apply LP method, a semi-parametric method proposed by Levinsohn and Petrin35, to measure TFP of each firm.

    11. (11)

      Firm’s ownership (Ownership). Measured by: if an enterprise is a state-controlled enterprise, the variable is assigned to 1, otherwise 0 is taken.

    12. (12)

      Number of shareholders (Shareholder); we use the natural logarithm of the number of shareholders of each listed firm as a measure.

    13. (13)

      Board membership (Board). This study applies the number of board membership of each listed firm as a measure.

Empirical results and discussions

Baseline regression analysis

Based on model (1), Table 2 reports the estimated results of model coefficients. Column (1)–(3) contain only the fixed effects, the fixed effects effect and the basic characteristics of the city, and the fixed effect and the basic characteristics of the city and the firm, respectively. It can be seen that the coefficients of Mobile are all positive and significant at 1% level, revealing that mobile Internet development contributed to promote the growth of firm’s labor demands. We can also see that the coefficient of Mobile in column (3) is relatively small, which reflects if the economic factors of cities and the heterogeneity of firms are not fully considered in the regression analysis, the positive impact of mobile Internet on firm labor demands will be overestimated. Specifically, after controlling for the series of control variables at the city and firm levels, other things being equal, every 0.1 increase in the development index of mobile Internet, the number of firm labor demands increases by approximately 3.93%. Consequently, based on the above empirical analysis at the firm level, it reveals that the employment creation effect of the current mobile Internet development is greater than its employment destruction effect, thus contributing to the growth of firm’s labour demand. Namely, the above analyses support H1 proposed in this paper while rejecting H2.

Table 2 Regression results of baseline model.
Full size table

Robustness test

To ensure the robustness of the regressions of the baseline model, a series of robustness tests need to be carried out. First, we test whether the outlier value in the sample will influence the robustness of the estimated results. Specifically, to avoid the interference of outlier values on the estimated results, we conduct empirical analysis by winsorizing the maximum and minimum 3% of Labor in the data set respectively. The corresponding estimations are reported in columns (1) and (2) of Table 3. It can be seen that the coefficients of Mobile are all positive at 5% significance level, which is consistent with the estimations in Table 2. Therefore, the empirical results remain robust when the outliers of the samples are removed.

Table 3 Robustness test results.
Full size table

Second, we explore whether there is a lag between the effects of the variables. This paper lags the city-level explanatory variables, including Mobile, by one period to conduct empirical analysis to examine whether there is a lag in the impact of mobile Internet on firm labor demands and to alleviate to some extent the endogeneity problem. The corresponding regressions are reported in columns (3) and (4) of Table 3. It can be seen that the coefficients of Mobile are positive at 1% significance level, supporting the estimations of the baseline model. In addition, it can be found the coefficient of Mobile in columns (4), indicating there is a certain lag effect of the marginal impact of mobile Internet on firm labor demands.

Third, we perform robustness tests by adjusting the measurement of main explanatory variable. On the one hand, to avoid interference with the basic findings due to the different ways of assigning objective weights, we use the coefficient of variation method to determine the objective weights of each sub-indicator in Table 1 and use the index obtained from it as a measure of Mobile. The corresponding results are reported in column (5) of Table 3. On the other hand, considering that the indicator of the number of cell phone subscribers can more intuitively reflect the scale and popularity of mobile Internet applications in cities, we apply the natural logarithm of the number of cell phone subscribers as a measure of Mobile to conduct a regression analysis to quantify the marginal impact of mobile Internet penetration on firm labor demands. Column (6) of Table 3 reports the regression results. We can see from column (5) that the coefficient of Mobile is positive at 1% significance level, indicating that changing the method of assigning objective weights to the sub-indicators in Table1 does not affect the basic findings of this paper. Furthermore, as shown in column (6), the coefficient of Mobile is positive at 1% significance level, confirming mobile Internet penetration contributes to firm employment level. Specifically, all else being equal, a 1% increase in the number of mobile Internet users in a city is associated with a 0.21% increase in the employment size of local listed firms.

The discussion of endogeneity

In this study, the dependent variables are micro-level firm data, while the main independent variables are city-level statistics, thus there is no reverse causality problem in the regression in terms of the logic of the empirical analysis. However, there may still be endogenous due to omitted variables. In view of this, this section applies the IV approaches to alleviate potential endogeneity issues to improve the credibility of the empirical results.

Regarding the selection of IV, given that the development and application of regional ICT is usually influenced by the path dependence of the development and applications of its related technology, and thus the relevant ICT applications in the history may exert effects on the current development of new ICTs. In line with this idea, we adopt the number of post and telecommunications offices per million people in 1984 for each prefecture-level city as IVs of Mobile. Besides, the IVs satisfies exogeneity. Since the use of fixed-line telephones and the supply of postal and telecommunication infrastructure mainly serve the daily communication needs of the people, the historical penetration of them in each city will not directly affect employment in each city. In summary, the IV of Mobile selected in this study is reasonable and feasible. However, since the above historical data are cross-sectional which cannot be directly applied to regression analysis of the panel data. In this regard, follow the practice of Nunn and Qian36, we construct an IV for the panel data by introducing a time-varying variable based on the historical data described above. Specifically, we use the above cross-sectional IVs multiplied by the natural logarithm of the amount of national IT service revenue in year t-1 as the IVs of Mobileit. Based on the method described above, IV regressions are carried out and the corresponding results are showed in Table 4.

Table 4 IV estimation results.
Full size table

We can see that Anderson canon. corr. LM statistics both passed the significance test at 1% level, rejecting the original hypothesis of insufficient identification of IV; Cragg-Donald Wald F statistics both rejected the original hypothesis of weak IV at 10% significance level. Therefore, the IV of Mobile constructed in this study is reasonable and effective. Furthermore, the coefficients of Mobile are all positive at 1% significance level, indicating that mobile Internet development significantly promotes firm’s employment size, which is consistent with the previous estimations. In short, through the above analyses, it can be concluded the basic findings of this study remain robust and reliable. That is, in general the employment creation effect of mobile Internet development is greater than its employment destruction effect, thus promoting employment growth.

Further analysis and discussion

Mechanism analysis

To better understand the impact paths of mobile Internet development on firm labor demands and the specific effects on different paths, this section will carry out an empirical analysis of the series of impact paths based on the model of testing mechanism constructed in "Methodology and data".

The path of facilitating the firm’s digital transformation

First, this section examines whether mobile Internet development can affect the labor demands of firms by effectively promoting their digital transformation. Regarding the measure of digital transformation of firms (DT), following prior literature28, we apply textual analysis to reflect the progress of digital transformation by counting the frequency of keywords related to digital transformation mentioned in the annual reports of each listed company in the sample period. The specific process is as follows: First, the paper uses the data crawler function of Python software to crawl the text information of annual reports and announcements released by Chinese A-share listed companies. Secondly, based on extant studies28 and taking into full consideration the characteristics of digital technology development and application at this stage, the paper constructs the information base for firm digital transformation progress whose keywords directly related to the digital transformation and digital technology application of firms (The key words are: artificial intelligence, cloud computing, big data, Internet+, information construction, digital technology, information technology, informatization, digitalization, intelligence, and smart. In addition, it should be noted that the annual reports of China’s A-share listed companies collected in this study and the constructed information base are all in Chinese, thus the above keywords may not be fully consistent in the English context). Finally, we apply Python software to match the crawled text information with the constructed information database and count the frequency of keywords related to the digital transformation of firms, and use it as the measure of DT. Table 5 presents the corresponding estimation results.

Table 5 Mechanism analysis 1 (Firm digital transformation as an intermediary variable).
Full size table

According to the results in column (1), the coefficient of Mobile is positive at 1% significance level, indicating that mobile Internet significantly contributes to the digital transformation of firms. This result highlights that the booming mobile Internet in China has become an important engine driving the digital upgrade of firms in recent years. The results in columns (2)–(3) show that the coefficients of Mobile are all positive at 1% significance level, but the coefficient value of Mobile in column (3) is about 9.4% smaller than that in column (2). In addition, the coefficient of DT is positive at 1% significance level, indicating a significant positive correlation between the digital transformation of firms and their employment size. Therefore, from the above results, it is clear that the employment promotion effect of mobile Internet on firms is partly absorbed by the digital transformation effect, so mobile Internet will propel the growth of firm labor demands by promoting their digital transformation, which suggests that H3 proposed in this paper is valid. In summary, it is found through the test that the employment creation effect generated in the process of digital transformation of firms promoted by mobile Internet development is significantly larger than its employment destruction effect, which ultimately promotes firm labor demands.

The path of improving the productivity of firms

Second, we test whether mobile Internet affects firm labor demands by influencing firm productivity. Based on the constructed mediating effects model, we conduct an empirical analysis using TFP as the path variable, and the corresponding estimations are reported in Table 6.

Table 6 Mechanism analysis 2 (Firm productivity as an intermediary variable).
Full size table

Firstly, we can see from the column (1) that the coefficient of Mobile is positive at 1% significance level, indicating mobile Internet contributes to firm productivity improvement significantly. Second, the results in columns (2)–(3) show that the coefficients of Mobile are positive at 5% and 1% significance levels, respectively, consistent with previous benchmark regressions. However, a comparison reveals that after the model controls the path variable TFP, the coefficient of Mobile increases by 12.2%, indicating if the differences in productivity between firms are ignored, the marginal impact of Mobile on Labor will be underestimated. Finally, from column (3), it can be seen the coefficient of TFP is negative at 1% significance level, reflecting the firm productivity is negatively related to its employment scale. Since after the model controls TFP, the coefficient of Mobile rises obviously and the coefficient of TFP is significantly negative, it can be concluded: mobile Internet development will reduce firm labor demands by improving firms’ productivity, that is, firm’s productivity is an important path for the mobile Internet to affect firm labor demands. Although in theory, the productivity promotion effect of mobile Internet may either promote the employment expansion of firms by improving their performance, or reduce their employment scale in order to optimize labor productivity, through the above empirical analysis, it can be seen the current development of mobile Internet mainly reduces firm labor demands through the path of improving firms’ productivity, which indicates that H4 proposed in this paper is valid.

The path of promoting expansion of production scale

In this section, we continue to test whether mobile Internet affects firm labor demands by affecting their production scale. We perform the empirical analysis using Scale as the path variable, and the corresponding regressions are reported in Table 7.

Table 7 Mechanism analysis 3 (Firm production scale as an intermediary variable).
Full size table

As shown in column (1) of Table7, the coefficient of Mobile is significantly positive at 1% level, meaning mobile Internet development has a significant positive impact on the expansion of firm production scale. The results of columns (2)–(3) show the coefficients of Mobile are positive at 5% and 1% significance levels respectively, but the coefficient of Mobile in column (3) decreases by 24.7% compared to that in column (2). Combined with the result that the coefficient of Scale is positive at 1% significance level, it can be concluded the promotion effect of mobile Internet on firm labor demands is somewhat absorbed by its growth effect of production scale, that is, promoting the expansion of production scale is an important path for the mobile Internet to change firm labor demands. In summary, through the above tests, the paper verifies that the mobile Internet development can influence firm labor demands through the channel of facilitating the expansion of their production scale. Thus, H5 proposed in this paper is confirmed.

The path of facilitating the expansion of business scope

This section will examine whether mobile Internet development can indirectly affect the size of a firm’s employment by promoting the expansion of business scope. Regarding the measurement of a firm’s business scope (Scope), due to data availability, we adopt the number of main product businesses presented by each listed company in its annual report as its measure. Based on this, Table 8 shows the corresponding regressions.

Table 8 Mechanism analysis 4 (firm business scope as an intermediary variable).
Full size table

The results in column (1) show the coefficient of Mobile is positive at 5% significance level, implying mobile Internet has a significant positive impact on expansion of firms’ business scope. Combining the results in columns (2)–(3) shows the coefficients of Mobile are both positive at 1% significance level, but the value for the latter is 13.1% smaller than that for the former. In addition, the coefficient of Scope is positive at 1% significance level, which shows firms with a wider business scope may have stronger employment absorption capacity. Combining the above results, it is clear the promotion effect of mobile Internet development on firms’ labor demands is partially absorbed by the expansion of their business scope, which means facilitating business scope expansion is an indirect path for mobile Internet to influence firms’ labor demands. Therefore, H6 proposed in this paper is held.

Analysis of the effect on the labor demand structure

In this section, we will further explore the impact of mobile Internet development on the firm’s labor demand structure. Specifically, we intend to explore the impact of mobile Internet on the employment of firms in different industries, the labor skill preferences of firms, and the labor demand for different positions in firms, so as to clarify the effect of mobile Internet on the labor demand structure of firms.

Analysis of the labor demand structure of the industry

Based on the industry classification results of listed companies of the China Securities Regulatory Commission (CSRC) in the first quarter of 2020 (For more information, see: http://www.csrc.gov.cn/csrc/c100103/c1451995/content.shtml), we divide the firms in the sample into three groups: agriculture, manufacturing and service industries, and then carries out a regression analysis to investigate the impact of the mobile Internet on the employment of firms in different industries. The corresponding estimations are presented in Table 9.

Table 9 Analysis results of the employment structure of the industry.
Full size table

As can be seen from Table 9, in the group of agriculture, the coefficient of Mobile is not significantly negative and its absolute value is relatively small, reflecting the employment creation effect of mobile Internet on agricultural enterprises is generally comparable to that of its employment substitution effect, so the aggregate employment effect is not obvious; In the group of manufacturing, the coefficient of Mobile is significantly positive at 1% level, showing mobile Internet applications have significantly increased the labor demands of manufacturing firms; In the group of service, the coefficient of Mobile is positive at 10% significant level and its absolute value is significantly greater than that of the other two groups, indicating that the mobile Internet has the greatest positive impact on the growth of firm labor demands in the service industry. In general, mobile Internet development is conducive to the expansion of employment in service and manufacturing firms, but to a certain extent reduces the employment scale in agricultural firms, thus contributing to the movement of China’s employed population toward an advanced industrial structure.

Analysis of the skill structure of firm labor demands

To examine the impact of the mobile Internet on the skill structure of firm labor demands, drawing on existing research37, we take the education level of firm employees as the basis for division, select the logarithm of the number of employees with a bachelor’s degree or above to measure the employment of high-skilled workers in firms, the logarithm of the number of employees with high school education and below to measure the employment of low-skilled workers in firms, so as to examine the changes in the demand for different skilled labor demands of firms with mobile Internet development. The corresponding subsample regressions are reported in Table 10.

Table 10 Analysis results of labor demand skill structure.
Full size table

It can be seen that the coefficients of Mobile are significantly positive in subgroups, indicating that mobile Internet significantly contributes to the demands of firms for both high-skilled and low-skilled labor. However, the value of the coefficient shows that mobile Internet contributes more to the growth of demand for low-skilled labor in firms. The reasons may be: due to the bias of high-skilled labor, mobile Internet applications are promoting the demands for high-skilled labor in firms, but since high-skilled labor and low-skilled labor have some complementarity in the production process7, the expansion of employment scale of high-skilled labor in firms will also drive the increase of labor demands for low-skilled labor to some extent.

Analysis of the positions structure in firms

We further analyze the impact of mobile Internet on the employment of different position types in firms. According to the position types and characteristics of the listed firms in the sample, the paper divides the positions of firm employees into the following 8 employment types: production, sales, management, customer service, technology, finance, staff administration and others. Table 11 reports the corresponding results.

Table 11 Analysis results of position structure of firms.
Full size table

As can be seen from Table 11, the coefficient of Mobile is negative in the groups of production, customer service, and finance, but only Mobile in the group of customer service is statistically significant, indicating that the mobile Internet has a negative impact on the labor demands for the positions of production, customer service, and finance in firms, but this impact is greatest in the customer service positions. In the groups of sales, management, technology, staff administration and other, Mobile has positive coefficients, while it is statistically significant in the groups of management and technology. It means mobile Internet promoted the labor demands for sales, management, technology, staff administration and other types of positions in listed firms, with the employment creation effect of the mobile Internet on positions in the technology and management being more evident.

Conclusions and policy implications

Conclusions

As a widely applied DGPT at present, the effects of mobile Internet development on firm labor demands deserves attention. Based on the microdata from China’s A-share listed firms and the macrodata for prefecture-level cities, we empirically analyze the impact of mobile Internet on labor demands of local listed firms. This article first constructs comprehensive evaluation system of regional mobile Internet development from four dimensions, and then calculates the level of mobile Internet development in each city in China by using the entropy method. Second, the OLS regressions and IV estimations are adopted to quantify the effects of mobile Internet development on firm labor demands. Third, on the basis of mechanism analysis, we examine the main channels through which the current mobile Internet development changes firm labor demands by apply the mediating effect model. Lastly, this study further examines the changes of mobile Internet development on the structure of firm labor demands, that is, we quantified the impact of mobile Internet on firm labor demands in different industries, changes in firm labor demands for different skills, and labor demands for different positions, respectively. The main research findings are as follows.

First, mobile Internet development can significantly increase the labor demand of firms, which is not only statistically significant, but has a considerable actual effect. And the effects remain valid after a series of robustness tests and the alleviation of potential endogeneity issues. This finding reveals at the micro level that the positive effect of mobile Internet technology, one of the key DGPTs underpinning the current digital economy, on expanding employment.

Second, the empirical tests of the impact path found mobile Internet development can reduce the labor demands of firms by contributing to their productivity improvement; In addition, mobile Internet development can increase the labor demands of firms by facilitating the expansion of production scale, digital transformation process and the extension of business scope. The above findings demonstrate that although the development of mobile Internet will reduce the labor demand of firms by promoting the growth of productivity, it contributes to expand the labor demand of firms in other ways. Consequently, the above empirical analyses not only reveal the paths of the development of mobile Internet affecting the labor demand of firms, but further clarify that the impact of the development of mobile Internet on the employment level of firms is generally that the employment creation effect is greater than the employment destruction effect.

Third, the impact of mobile Internet on the employment of firms in different industries is heterogeneous: it has the greatest positive effect on the employment of firms in the service industry, followed by those in the manufacturing industry, and has a certain negative effect on those in agriculture, which indicates that mobile Internet development has to some extent promoted the flow of the working population toward the advanced industrial structure. This result is also more in line with the reality of China’s economic development at this stage: Manufacturing, as an important part of the real economy, is also one of the most important industries absorbing employment in the process of China’s industrialization. However, as industrialization enters the late stage, the service sector is also gradually becoming the main industry to accommodate labour transfer, coupled with the fact that most of the mobile Internet applications are in the service sector, and thus the employment creation effect generated by the mobile Internet development is stronger in the service sector.

Fourth, the mobile Internet has effectively boosted the demands for both high-skilled and low-skilled labor, with a greater marginal impact on the former than on the latter. Thus, this positive effect on the demands for labor of different skills avoids to some extent the “technological polarization” of employment. Although it reflects to a certain extent that due to a certain degree of complementarity between high-skilled labour and low-skilled labour, mobile Internet development will to a certain extent lead to an increase in the demand for low-skilled labour while promoting the expansion of the scale of high-skilled labour employment, it is worth pointing out that the marginal impact of the current mobile Internet development on the employment of high-skilled labour in firms is smaller than that of low-skilled labour employment, possibly due to the fact that: The current mode of training high-skilled labour force in China does not fully match digital economy development, resulting in the complementarity between mobile Internet technology application and high-skilled labour force not being given full play to, and thus the promotion effect of mobile Internet development on the employment of high-skilled labour force in listed firms is lower than that of the employment of low-skilled labour force.

Fifth, in general, mobile Internet has a significant creation effect on positions with strong relevance to digital applications, such as those in technology and management, but shows a certain destruction effect on positions with the characteristics of routine tasks, such as those in customer service, production and finance, which reveals the biased nature of the employment effect of mobile Internet development. The above findings further reflect that the mobile Internet will change the original position structure of firms, which means that as mobile Internet applications continue to integrate with the real economy, firms will increase the number of digitally related positions and reduce the number of positions for routine-type tasks.

Policy implications

On the basis of the research findings in this article, mobile Internet technology is not only an important DGPT that propels the current digital economy, but helps to improve firm labor demands, which reveals the possible ways to promote employment. The following policy recommendations are put forward.

First, according to the empirical results, the development of the mobile Internet promotes labour demands at the micro level, which is an important digital factor in expanding economy’s employment level currently. However, as the current level of mobile Internet development in various regions in China still requires further improvement, thus to give full play to the employment-promoting effect of the mobile Internet, it is necessary to make efforts from the supply side of mobile Internet development: The government authorities can, on the one hand, strengthen the construction of mobile Internet infrastructure to narrow the inter-regional digital divide and further improve the level of access and coverage of mobile networks; on the other hand, promote the upgrading of mobile Internet technology to provide solid basis for the current digital economy, so as to fully exert the employment effect of the integration of mobile Internet and the real economy and improve the employment level of the economy and society.

Second, this study finds that mobile internet development can change the labour demands by facilitating firms’ digital transformation, improving the productivity of firms, propelling the expansion of business scope and promoting the production scale of firms, and that all three paths except the first one are contributing to promoting firms’ labour demands. The above findings reveal that the development of mobile internet in not only contributes to the business benefits and digital upgrading of firms, but is also an important driver of the expansion of their labour demands. Hence, there are the following policy implications: The government authorities can introduce corresponding policies to encourage and promote the innovation of mobile Internet application scenarios, support various enterprises to develop new mobile Internet applications and new businesses, and enrich mobile Internet application services in various fields of the economy, thus expanding new space for employment while enhancing social productivity.

Finally, as a biased technological progress, the mobile Internet not only directly changes labor demands of firms, but also has a significant impact on the labor demand structure, which has been confirmed by this study. Thus, the government authorities also need to pay attention to the fact that the series of innovative applications of mobile Internet not only promote the digital upgrading of the economic society, but are changing the structure of labor demands of firms. In the face of the structural changes in employment and labor demands in the digital economy era, the government should actively integrate and mobilize resources to strengthen the training of new digital talents, such as supporting firms to strengthen on-the-job training of digital skills, especially to increase the popularization of skills training among middle and low-skilled labor groups, and guide them to upgrade and transform into high-skilled labor, so as to stimulate the potential of domestic digital economic development and promote the optimization and upgrading of the employment structure.

Further discussion

With the current rapid development of the digital economy, the existing literature has continued to address the relationship between various types of DGPTs and employment, but the conclusions of the studies are still not consistent. Moreover, as a new generation of DGPT that is currently widely applied globally, there are still few and scattered studies on the economic effects of mobile Internet development, in particular, there has been no literature on the relationship between mobile Internet and employment, and thus there is a lack of corresponding empirical evidence on its actual impact on the current level of employment and the structure of employment. On the basis of theoretical analysis, this paper analyses the impact of mobile Internet development on employment from the firm level in the context of the current rapid development of mobile Internet in China, which is a meaningful supplement to the existing research. Admittedly, there are still some limitations to this study that need to be improved. First, due to the multiplicity of impact paths of mobile internet development on employment, we failed to construct a theoretical model to clearly clarify the impact path of mobile internet on employment. Second, given that mobile Internet development is a systematic project, we tried to evaluate the level of mobile Internet development in each region by constructing a comprehensive evaluation system, but under the constraint of data availability, we fail to get the mobile Internet data traffic in each region, which makes the comprehensive evaluation system we constructed still need to be improved. Third, we identify the impact paths of mobile Internet development on employment by constructing a mediated effects model, and although the identification strategy is relatively simple and efficient, the quantitative accurateness is not sufficient. In the view of this, further research may address and supplement the limitations of this study38.