Introduction

Employment quality stands as the cornerstone of people’s livelihood and is intrinsically linked to a nation’s development. Advancing full employment is vital for enhancing public welfare, impacting family happiness and the long-term economic growth and social stability of a country. The current downward trend in the global economy makes it more challenging to achieve full employment1, and improving employment quality has become a critical issue that needs to be solved urgently2. However, the employment environment has been partially improved with the rapid development of the digital economy, and new employment spaces continue to emerge under the regional digital transformation3. Specifically, digitalization has given rise to new and more creative social forms, inspired workers to upgrade their skills to meet the development requirements continuously, improved the employment situation4, and become one of the critical elements of employment quality improvement. It can be said that regional digital transformation optimizes employment quality. Therefore, in the background of the rapid development of the digital economy, optimizing the labor force structure, as well as using regional digital transformation to improve employment quality, has important theoretical significance and practical value.

The research on the coordinated development of regional digital transformation and employment quality explores the impact of digital transformation on employment quality based on the theoretical level3and researches the mechanism analysis and development path and other aspects of the effects from the empirical level. A combination of the literature shows that the “digital divide” and other unbalanced development phenomena caused by regional digital transformation will have a certain degree of impact on employment5. Based on existing research, this study explores regional digital transformation’s role in influencing employment quality and related mechanisms. It analyzes the realization paths to promote the development of the level of employment quality. The marginal contribution may lie in the following. First, the index system of employment quality is constructed, and the employment quality is scientifically measured, further improving the comprehensiveness of the evaluation dimension of employment quality. Second, it examines the impact of regional digital transformation on employment quality, finds the “U-shaped” trend of digital transformation on employment quality, and verifies its robustness with various methods. Third, from the perspective of labor force structure, we consider employment quality and labor force structure as two completely independent but related variables. The mechanism of regional digital transformation on employment quality is explored from the industry level, sector level, and skill level, respectively, and the non-linear mechanism is measured by using the instantaneous mediation effect. It provides strategic recommendations for advancing regional digital transformation and is instrumental in enhancing employment rates. Furthermore, it executes and refines the implementation of regional employment policies to augment their efficacy.

Theoretical basis and research assumptions

Direct impact of regional digital transformation on employment quality

The theory of skill-biased technological progress proposed by Acemoglu states that technological progress does not have the same impact on all types of workers and that it is more inclined to enhance the productivity of highly skilled workers6. This theory is further validated in the context of regional digital transformation. Digitalization increases the demand for high-skilled labor while low-skilled labor is even replaced. This skill-biased technological advancement leads to a non-linear change in employment quality. The employment quality of high-skilled labor is enhanced while low-skills may decline. In addition, according to Schumpeter’s theory of “creative destruction”, digital transformation is an innovative process usually accompanied by the elimination of old technologies and the rise of new industries. The impact mechanism is distorted when the original labor force structure equilibrium is disrupted. On the one hand, the innovation and application of digital technology promote the development of new industries and create a large number of employment opportunities; on the other hand, the decline of traditional industries and the substitution of technology also lead to unemployment. Therefore, the impact of regional digital transformation on employment quality presents a “U”-shaped characteristic: at the initial stage, the employment quality of part of the workforce may decline due to the impact of new technologies and substitution effects; however, the development of emerging industries and the popularization of technologies will gradually improve the overall employment quality as time goes by.

The existing literature needs to be more consistent in exploring the role of regional digital transformation on employment quality. In terms of positive effects, digital transformation is having a facilitating impact on employment quality. First, the infrastructure development driven by digital transformation improves the regional employment situation and increases the corresponding income7. Second, developing the digital economy increases self-employment opportunities, provides more private-sector jobs, and improves the employment environment8. Once again, digital transformation leads to the digitalization of industries, which directly expands the demand for labor and can continuously attract the inflow of workers9. Finally, integrating digital transformation with essential public services can improve the job market10. Overall, digital transformation optimizes the efficiency of labor allocation, shortens the gap between time, income, and other employment benefits11,12, and although there is a particular impact and substitution on the labor force, it promotes the upgrading of workers’ skills13, and plays a role in human capital14. In terms of the reverse effect, digital transformation may also have a dampening effect on employment quality. Issues such as the “digital divide” caused by digitization among employment groups have resulted in significant disparities in employment quality across regions and sectors5. For example, high-skilled groups have a high degree of adaptation to the changes brought about by digitization. In contrast, low-skilled groups have difficulty adapting to the complex requirements of digitized employment in the short term. It is necessary to explore how to avoid the expansion of the employment gap15,16, and to thoroughly study the substitution and compensation effects of regional digital transformation on employment17.

In summary, in the early stage of regional digital transformation, Some workers are unfamiliar with digital operations and technologies and cannot adapt to the needs of digital transformation, which will have an impact on the development of the employment market and the level of employment quality may show a short-term downward trend, expressed as the left side of a “U-shaped” curve; with the deepening of the regional digital transformation, through the strengthening of skills training, optimizing the matching of demand on the supply side of the labor force. After a short-lived employment shock, new market demand for labor emerges, and sustainable employment development has been achieved. As a result, employment quality shows an upward trend, expressed as the right side of a “U-shaped” curve. Based on the above theoretical analysis, this study has the following hypotheses:

H1: Regional digital transformation negatively impacts employment quality in the early stage, after which employment quality will gradually increase, showing a “U-shaped” curve.

 Indirect effects of regional digital transformation on employment quality

Technological innovations and applications driven by digital transformation will inevitably trigger significant changes in employment demand across sectors and industries. These changes affect employment quality by adjusting the labor force structure, including industry, sector, and skill level reconfigurations. Thus, labor force structure plays a crucial mediating role in this chain of influence, allowing us to understand more fully the complex mechanisms of digital transformation on employment quality10. Some traditional sectors may reduce employment opportunities, while emerging sectors that utilize digital technologies generate new employment opportunities. Such changes affect the labor force structure at the industry level, sector level, and skill level18, which in turn has a direct impact on employment quality. For example, some emerging fields that urgently need high-skill requirements will provide high-paying jobs to attract laborers and improve their employment quality. The decline of some traditional industries has shrunk the demand for low-skilled workers and reduced their employment quality.

This study analyzes the indirect impact of digital transformation on employment quality from the perspective of labor force structure. First, from the industry level, the substitution effect of digital technology on the labor force will be accompanied by the complementary effect; the labor-intensive secondary industry dominates the substitution effect, and the high-end service and technology-intensive industry dominate the complementary effect3, advanced digital technology will gradually replace the labor force, and the production and sales of factories will move towards automation, which inevitably eliminates the low and medium-skill jobs and forces the labor demand to decline20. Second, the analysis of the labor force structure at the sector level is reflected in the fact that the regional digital transformation has increased the employment quality and size of public organizations. It significantly impacts the labor force structure at all levels, tilting the overall labor force structure toward manufacturing, high-tech, and high-skill, showing a “U-shaped” adjustment process. Finally, from the skill level labor force structure analysis, the development of digital technology has increased the number of high-skill jobs, promoting the transformation of the labor force structure from low-skill to high-skill employment21. The development will accelerate the change of skilled labor force structure22, which can contribute to productivity development by optimizing the corresponding structure23. Regional digital transformation affects employment trends and changes the direction of labor force structure. To test the possible nonlinear mediation effect of labor force structure, this study proposes the following hypothesis:

H2: Regional digital transformation indirectly affects employment quality through labor force structure. There is a nonlinear mediating effect of labor force structure between regional digital transformation and employment quality.

From the perspective of labor force structural perspective18, this study divides labor force structure into three dimensions, including industry level, sector level, and skill level, according to which the following three hypotheses are proposed:

H2a: Regional digital transformation indirectly impacts employment quality via nonlinear mediation of industry-level labor force structure.

H2b: Regional digital transformation indirectly affects employment quality through the nonlinear mediation of sector-level labor force structure.

H2c: Regional digital transformation indirectly affects employment quality through the nonlinear mediation of skill-level labor force structure.

The overall logical structural content of the study is shown in Fig. 1.

Fig. 1
figure 1

The overall structure of the thesis.

Full size image

Construction and measurement of regional digital transformation and employment quality indicator system

Data source and preprocessing

Restricted to the scientificity and comprehensiveness of each data in the construction of the regional digital transformation indicator system, this study selected the data of 31 provincial-level provinces in mainland China for the period of 2013–2022 from the website of the National Bureau of Statistics of China, China Statistical Yearbook, China Labor Statistical Yearbook, and China Population and Employment Statistical Yearbook. Some data needed to be filled in using linear interpolation and regression methods. All variables were normalized by polar deviation to reduce the volatility of data and the influence of extreme values and non-normal distribution on the model.

Construction of the indicator system

Construction of regional digital transformation index system

Existing studies often use the level of digital facilities, digital application level, and other related indicators to measure the degree of regional digital transformation16, taking into account the level of platform construction as an essential carrier of regional digital transformation so that it will be included as a complementary indicator. The digital penetration level indicator is added to the regional digital transformation penetration rate measurement. In summary, this study constructs a regional digital transformation level indicator system from the four dimensions: facility level, application level, platform level, and penetration level. Accordingly, the development level of regional digital transformation is comprehensively evaluated, and the weights of the indicators at various levels are solved, as shown in Table 1. To eliminate the influence of subjective factors, this study uses the entropy weight TOPSIS method to measure the level of regional digital transformation and employment quality.

Table 1 Indicator construction for the level of Regional Digital Transformation.
Full size table

Construction of employment quality indicator system

Employment quality refers to the comprehensive status of working conditions, remuneration level, social security, and labor rights protection that workers receive during employment. It reflects the level of economic and social development and policy orientation. This study establishes a system of indicators of employment quality from three dimensions13, divided explicitly into the employment environment, employment remuneration, and employment protection. Accordingly, the employment quality is comprehensively evaluated, and the weights of each level are calculated, as shown in Table 2.

Table 2 Indicator construction of employment quality system.
Full size table

The direct impact of regional digital transformation on employment quality

Model construction

In order to test hypothesis H1 and explore the mechanism between regional digital transformation and employment quality. This study uses employment quality lagged 1 period as an instrumental variable22, and constructs a dynamic panel regression model. Where i is the province, t is the year, the dependent variable Qualyit represents the employment quality, Qualyit−1 represents the first-order lag term of employment quality, the independent variable Dig is the regional digital transformation, Dig2 represents the square of the regional digital transformation, Control is a series of control variables, vi represents the province fixed effect that does not vary with t, and εit is the random disturbance term.

$$Qual{y_{it}}={\alpha _0}+\varphi Qual{y_{it - 1}}+{\alpha _1}Di{g_{it}}+{\alpha _2}Di{g_{it}}^{2}+{\alpha _3}Contro{l_{it}}+{\upsilon _i}+{\varepsilon _{it}}$$
(1)

In order to test hypothesis H2, examine whether changes in labor force structure play a mediation effect in the impact of regional digital transformation on employment quality, and explore the detailed mechanism, this study draws on the “U-shaped” relationship test method of Haans et al27., calculates the confidence interval of the instantaneous mediation effect by bootstrap test method, and uses the nonlinear model to test the mediation effect, LaborStrucit represents the labor force structure, the model is set as follows:

$$LaborStru{c_{i,t}}=\theta +{\theta _1}Di{g_{i,t - 1}}+{\theta _{\text{2}}}Dig_{{i,t - {\text{1}}}}^{{\text{2}}}{\text{+}}\sum {{\theta _{\text{j}}}CV{\text{s}}+\sum {Year+\sum {Ind} } } +\tau$$
(2)
$$Qual{y_{i,t+1}}={\varphi _{\text{0}}}{\text{+}}{\varphi _{\text{1}}}Di{g_{i,t - 1}}{\text{+}}{\varphi _{\text{2}}}Dig_{{i,t - {\text{1}}}}^{{\text{2}}}{\text{+}}{\varphi _{\text{3}}}LaborStru{c_{i,t}}{\kern 1pt} {\kern 1pt} {\kern 1pt} {\text{+}}\sum {{\varphi _{\text{j}}}CV{\text{s}}+\sum {Year+\sum {Ind} } } +\varepsilon$$
(3)

In addition to being affected by regional digital transformation, changes in employment quality are affected by many other factors. Therefore, control variables are introduced, including: (1) Capital intensity (Intensity), measured using the level of fixed asset investment per capita. (2) Human capital stock (Stock), measured using China’s human capital index based on the J-F method at each provincial level. (3) Research and Development Intensity (Rd), measured using the share of research and experimental development expenditures of industrial enterprises above designated size in the technology market turnover of each region. (4) Government Expenditure Intensity (Govern), using the share of the sum of local fiscal science and technology expenditures and education expenditures in the general budget expenditures of local finance to measure the digitization expenditures of the government in order to control for the differences in government impacts across regions. (5) Income differences (Income), using the deflated calculated per capita disposable income of residents (with 1978 as the base period for calculating the actual per capita disposable income) to measure the income differences of residents in each region. (6) Regional Economic Development Level (Economy), using GDP per capita to measure each region’s economic size and development level.

The direct impact of regional digital transformation on employment quality

This study uses the panel data model based on the following reasons: first, due to the unbalanced development between regions, some indicators are missing, the use of panel data can increase the number of samples, avoiding the degree of freedom is too low and affecting the validity of the coefficient estimation; second, the panel data has the advantage of cross-sectional data and time series data, which can better reflect the dynamics of the model. In addition, considering that employment quality may be potentially affected by market trends in the preceding period, we incorporate Qualyit−1, which lagged for a certain period, into the model to enhance the dynamic explanatory power of the model and reduce endogenous disturbances in the model estimation. Since the data are short panel types, the systematic GMM method is more efficient and can remove the endogenous effect. Finally, the models all control for region-fixed and year-time effects; the results are shown in Table 3.

Table 3 Impact of Regional Digital Transformation on Employment Quality.
Full size table

The test results support the hypothesis that there is no second-order serial correlation in the regression equation. Meanwhile, the validity of instrumental variables is analyzed using the Hansen over-identification constraint test, and the p-values of the Hansen test are all greater than 0.1, proving that the instrumental variables are valid. The Wald test for joint significance of coefficients rejects the original hypothesis that the coefficients of the explanatory variables are zero at 1% significance level, which proves the reasonableness of the whole model setup.

As shown in Table 3, M(1) represents a linear model of regional digital transformation and employment quality, with a regression coefficient of 0.153 for Dig, which raises output and expands the job market. And as regional digital transformation deepens, the trend of employment quality may change non-linearly. Therefore, a quadratic term for Dig is introduced in the model. According to the regression results of M(2), the coefficient of Dig2 is 2.539, which makes it reasonable to measure the relationship between regional digital transformation and employment quality with a nonlinear model. The results of the M(3) model with control variables show that the coefficient of Dig2 is 1.644, and the coefficient of Dig is − 0.806. This infers a U-shaped trend whereby employment quality decreases and then increases as regional digital transformation improves. In order to deeply analyze the data characteristics, the M(4) and M(5) models in Table 3 show two different simple OLS estimation results: M(4) does not incorporate region and time as dummy variables to control, while M(5) includes explicitly these variables as control terms, to comparatively analyze the specific effects of the region and time factors on the model estimation results, and the model results similarly reveal the non-linearities of the variables.

Lind and Mehlum24 argued that judging a non-linear relationship based only on the significance of the secondary term may lead to a misjudgment of a “U-shaped” relationship. It can be considered more accurately by the Utest, first, that the coefficients of the primary term be significantly negative and the coefficients of the secondary term be significantly positive; second, that the slopes of the two endpoints of the curve be significantly negative and significantly positive, respectively; third, the value of the turning point must be located in the range of values of the variable. The results show that the coefficients of the main terms of regional digital transformation are all significantly negative, and the coefficients of the squared terms are all significantly positive. After substituting the β1β2 coefficients for the slopes of the left and right endpoints of Dig, the slopes are significantly non-zero, the slopes of the left endpoints are all negative, and the slopes of the right endpoints are all positive, which is in line with the “U-shaped” curve. In addition, the p-value is 0.018, which rejects the original hypothesis that the model is not “U-shaped” at the 5% significance level. The value of the turning point of the “U-shaped” curve of the impact of regional digital transformation on employment quality is 0.245, which is located in the range of regional digital transformation (0.037–0.659), confirming the non-linear “U-shaped” relationship.

Overall, the impact of regional digital transformation on employment quality shows a “U-shaped” trend. There is a slight decrease in the first period, but after regional digital transformation develops to a certain extent, employment quality shows significant growth. In the early stage, although workers have more access to employment information, which is beneficial to solving frictional unemployment, some of them are less likely to have the employment skills required by digitization, which will easily lead to an adaptation period and a “U-shaped” decline. In the later stages of the digital transition, as digital technology improves productivity and spawns a series of new industries, the need for labor is expanded, and low-skilled groups can also learn digital skills to enhance employment competitiveness, narrowing the digital divide among workers and showing a “U-shaped” upward trend.

Endogeneity issues

To ensure the accuracy of the findings, the following robustness tests were conducted separately in this study. The results of the reanalysis are shown in M(1)- M(5) in Table 4.

  1. 1.

    Replacing the measures of explanatory variables. Employment efficiency is calculated using the DEA-Malmquist index. Human capital stock, per capita disposable income of residents, per capita GDP, capital intensity, government expenditure intensity, and R&D intensity are used as input variables and urban registered unemployment rate, the average wage of urban employed persons, employment speed, and vocational training attendance are used as output variables.

  2. 2.

    Replacing the measures of core explanatory variables. In the Critic evaluation method, the degree of conflict between indicators with strong positive correlations is lower, which helps to evaluate the variables comprehensively. The Critic evaluation method was used to recalculate the level of regional digital transformation in each region.

  3. 3.

    Although several variables have been controlled, factors that are difficult to measure may interfere with the interpretation of inter-variable relationships. Therefore, this study incorporates the lags of explanatory variables into the model and re-runs the regression analysis to minimize problems such as endogeneity due to reverse causality.

  4. 4.

    During the period of the new crown epidemic within mainland China, the input of regional digital transformation may be volatile; considering that the relevant fluctuations may have an impact on the research results, the data for 2020–2022 are excluded and re-estimated.

  5. 5.

    Replace the panel Tobit model. Since the explanatory variables are all greater than zero, considering the emergence of extreme values, data set limits, truncated tail censoring, and other problems, using the panel data Tobit model can alleviate these problems.

The coefficient of the squared term of regional digital transformation is positive and significant, and the coefficient of regional digital transformation is negative and significant and passes the Utest. Most of the variables were able to be significant at the 10% level of significance. Although the model coefficients differ from the benchmark regression, the “U-shaped” impact trend remains unchanged.

Table 4 Robustness test of the impact of Regional Digital Transformation on Employment Quality.
Full size table

The study found that there is a “U-shaped” relationship between regional digital transformation and employment quality. In the early stage of regional digital transformation, employment quality tends to decline in the short term, and after a certain threshold, regional digital transformation can significantly improve employment quality. The above relationship is valid after replacing the core variables, changing the model estimation method, and other robustness tests.

Analysis of the mechanism of the impact of regional digital transformation on employment quality

Nonlinear mediation model of labor force structure

Although it has been confirmed that there is a “U-shaped” relationship between regional digital transformation and employment quality, the specific mechanism between the two has not yet been identified and tested. Therefore, this study introduces the labor force structure as the mediating variable and divides it into industry level, sector level, and skill level18, respectively, to test whether there is a mediation effect of different levels of labor force structure between regional digital transformation and employment quality. Among them, the ratio of employment in the tertiary industry to that in the secondary industry is used to evaluate the industry-level labor force structure, sector-level labor force structure is assessed through the share of employment in high-end manufacturing and high-end services in overall employment, and the ratio of employment with tertiary education and above to that with tertiary education and below is used to evaluate the skill level labor force structure.

Due to the possible time lag in the transmission of the mediating variables, the impact of regional digital transformation on employment quality needs to be reflected over a period of time, such as the absorption of technology, the cultivation of talents, and the adjustment of the market, so it is treated as a two-period interval. The Qualy is front-loaded by one period, while the Dig is lagged by one period, and the rest of the variables are kept in the current period to overcome the possible reverse causality between variables to a certain extent. Equation (4)(5) are constructed to express the relationship of the variables separately. The existence of the indirect effect can be confirmed by testing the new parameter IND in Eq. (6), which is generated by the product of the coefficient of the squared term of regional digital transformation in Eq. (4) and the coefficient of the labor force structure in Eq. (5), and represents the strength of the curvilinear indirect effect of regional digital transformation on employment quality through the labor force structure. Therefore, the significance of IND can be an essential basis for testing the mediation effect. The nonlinear mediation model is set up as follows:

$$LaborStru{c_{i,t}}=\theta +{\theta _1}Di{g_{i,t - 1}}+{\theta _{\text{2}}}Dig_{{i,t - {\text{1}}}}^{{\text{2}}}{\text{+}}\sum {{\theta _{\text{j}}}CV{\text{s}}+\sum {Year+\sum {Ind} } } +\tau$$
(4)
$$Qual{y_{i,t+1}}={\varphi _{\text{0}}}{\text{+}}{\varphi _{\text{1}}}Di{g_{i,t - 1}}{\text{+}}{\varphi _{\text{2}}}Dig_{{i,t - {\text{1}}}}^{{\text{2}}}{\text{+}}{\varphi _{\text{3}}}LaborStru{c_{i,t}}{\kern 1pt} {\kern 1pt} {\kern 1pt} {\text{+}}\sum {{\varphi _{\text{j}}}CV{\text{s}}+\sum {Year+\sum {Ind} } } +\varepsilon$$
(5)
$$IND={\theta _2} \cdot {\varphi _3}$$
(6)

The reasons for choosing labor force structure as a mediating variable are: first, the labor force structure at the industry level reflects the distribution of the number of laborers in the three major industries, and with the upgrading of the industrial structure, there is a clear trend of labor migration, which is especially obvious in the flow of laborers from the secondary industry to the tertiary industry. Regional digital transformation can accelerate market mobility and optimize the industry-level labor force structure, thus indirectly affecting employment quality; secondly, it reflects the trend of labor migration from low-skill to high-skill sectors, which is particularly obvious in labor-intensive sectors, and digital transformation creates more employment opportunities and improves the efficiency of factor matching, which has a significant impact on the quality of employment; thirdly, The skill level labor force structure is reflected in a substantial increase in the share of employment of highly skilled people. With the continuous promotion of technological innovation, the demand for high-skilled labor is also rising, expanding the employment gap, which in turn triggers changes in the labor force structure. The mediation path test for different labor force structure levels is shown in Table 5.

From the industry-level analysis, the upgrading and optimization of labor force structure can significantly alleviate the problem of employment changes caused by industrial migration, as shown M(1) M(2) in Table 5, the regression coefficient of IndStrucof regional digital transformation on the labor force structure at the industry level is − 0.438, and the regression coefficient of the quadratic term of the regional digital transformation is significant as a positive value of 0.358. It passes the Utest test, which indicates that the impact of regional digital transformation on the labor force structure at the industry level presents a “U-shaped” trend. In the process of the gradual replacement of the old sector structure by the new one, the relevant employees will experience a short period of pain before adapting to the changes in the new sector structure. The penetration of the early stages of the development of regional digital transformation contributed to the creation of jobs. It relatively reduced the size of employment in the secondary sector, with a slight reduction in the labor force structure at the sector level, which is on a downward trend. However, as the degree of regional digital transformation accelerates, the match between labor supply and demand improves, and employment quality is enhanced by improving the labor force structure at the industry level, showing an upward trend. In the Bootstrap test of industry-level labor force structure, the 95% confidence intervals of the direct and indirect effects do not contain 0, indicating a significant mediation effect. The mediation effect at the industry level is calculated according to the regression results using the “U-shaped” relationship test method of Haans et al27.

Table 5 Identification of the “U-shaped” mediating variable mechanism of Regional Digital Transformation affecting employment quality.
Full size table
$$\frac{{\partial IndStruc}}{{\partial L.Dig}} \times \frac{{\partial F.Qualy}}{{\partial IndStruc}}= - 0.{\text{438}}\left( {{\text{1}}{\text{.289}} - {\text{5}}{\text{.534}}IndStruc} \right)$$
(7)

Analyzing from the sector level, the innovation of production organization and the increase of social human capital at the sector level will enhance the matching level of labor supply and demand, and the regional digital transformation can realize the transfer of labor from low-skilled sectors to high-skilled sectors. From the results of M(3) and M(4) in Table 5, the regression coefficient of regional digital transformation on the labor force structure at the sector level is − 0.352, and the regression coefficient of the quadratic term of regional digital transformation is significantly positive at 0.187 and passes the Utest test. Therefore, the digital transformation shows a “U-shaped” trend in the labor force structure at the sector level. In the early stage of digitization, they are mainly affected by the pattern of “machines replacing people”. With the deepening of digitalization and creative destruction, a large number of new high-end manufacturing and high-end services employment demand is re-created. Regional digital transformation is usually accompanied by the application of automation and robotics, resulting in the automation of some repetitive tasks. This may lead to a reduction in work, but with the development of regional digital transformation, a part of the automated system to create jobs, manufacturing, and service industry employment ratio gradually increased, and the development of high-end sectors can promote the improvement of employment quality. In the indirect effect test of sector-level labor force structure, the bootstrap test 95% confidence interval does not contain 0, indicating that the mediation effect is significant. The value of the mediation effect at the sector level is calculated:

$$\frac{{\partial \operatorname{Sec} Struc}}{{\partial L.Dig}} \times \frac{{\partial F.Qualy}}{{\partial \operatorname{Sec} Struc}}= - 0.35{\text{2}}\left( {1.{\text{117}} - {\text{4}}{\text{0.094}}\operatorname{Sec} Struc} \right)$$
(8)

Analyzing the skill level, skill inequality increases in the pre-digital transformation period, with higher-skilled workers better adapting to the work environment under new technologies and lower-skilled jobs increasing the risk of unemployment. As a result, employment quality declines in the short term, forming the left side of a “U-shaped” curve. Digital development first replaces groups of people in repetitive jobs. After a period of sustained regional digital transformation, the digital skills of the low-skilled workforce can be upgraded, improving the mismatch between skills and employment. With the wide application of digital technology, the demand for high-skill jobs continues to rise, accelerating the prosperity of the labor market and thus improving the level of employment quality. From the results of M(5) and M(6) in Table 5, the regression coefficient of regional digital transformation on skill level labor force structure is − 0.728. The regression coefficient of the quadratic term of regional digital transformation is significantly positive at 0.021. It passes the Utest test, so the regional digital transformation on skill level labor force structure shows a “U-shape”. In the test of the mediation effect of skill level labor force structure, the bootstrap test 95% confidence interval does not contain 0, indicating that the mediation effect is significant. The mediation effect of skill level is calculated:

$$\frac{{\partial SkillStruc}}{{\partial L.Dig}} \times \frac{{\partial F.Qualy}}{{\partial SkillStruc}}= - 0.{\text{833}}\left( {0.{\text{021}} - {\text{1}}{\text{0.456}}SkillStruc} \right)$$
(9)

Instantaneous indirect effects of labor force structure

In order to further quantify the level of effect of labor force structure as a nonlinear mediator, to measure the value of the instantaneous indirect effect in the above three model formulas, which is a function of the three dimensions of the labor force structure, the nonlinear mediating effect is subject to two conditions: first, there is at least one set of nonlinear relationships between the explanatory variables and the mediating variables, and between the mediating variables and the explanatory variables; second, the mediating effect value is non-zero, the product of the rate of change of the explanatory variables and the mediating variables is zero. Stolzenberg and Land28 pointed out that if the independent variable has a nonlinear effect on the dependent variable through the mediating variable, the change in the dependent variable caused by the independent variable’s change in the mediator variable and thus the indirect rate of change in the dependent variable is calculated as shown in Eq. (10). Combined with the actual situation, the regional digital transformation has a “U-shaped” impact on employment quality, and the indirect rate of change in employment quality caused by changes in labor force structure due to the regional digital transformation is deduced from Eq. (11) and Eq. (12), as shown in Eq. (13).

$$\theta {\text{=}}\frac{{\partial LaborStru{c_{i,t}}}}{{\partial Di{g_{i,t - 1}}}} \times \frac{{\partial Qual{y_{i,t+1}}}}{{\partial LaborStru{c_{i,t}}}}$$
(10)
$$LaborStru{c_{i,t}}={i_{\text{1}}}{\text{+}}aDi{g_{i,t - 1}}$$
(11)
$$Qual{y_{i,t+1}}={i_2}+{c_1}^{\prime }Di{g_{i,t - 1}}+{c_2}^{\prime }Di{g_{i,t - 1}}^{2}+{b_1}LaborStru{c_{i,t}}+{b_{\text{2}}}LaborStru{c_{i,t}}^{2}{c_1}^{\prime }$$
(12)
$$\theta =\frac{{\partial LaborStru{c_{i,t}}}}{{\partial Di{g_{i,t - 1}}}} \times \frac{{\partial Qual{y_{i,t+1}}}}{{\partial LaborStru{c_{i,t}}}}=a\left( {{b_1}+2{b_2}LaborStru{c_{i,t}}} \right)$$
(13)

Next, this study uses the method proposed by Preacher29 by calculating the instantaneous indirect effect, assigning a specific value x to the independent variable, calculating the corresponding indirect rate of change, and employing the Bootstrap method to test the significance of the instantaneous mediation effect corresponding to different xvalues. By using the MEDCURVE program30, the instantaneous mediation effect of labor force structure located at the industry, sector, and skill level between regional digital transformation and employment quality was tested separately. If zero is not included in the confidence interval, then it means that the instantaneous indirect effect of labor force structure is significant, and there is a mediation effect.

According to Table 6, it can be seen that the instantaneous mediation effect of industry-level labor force structure is − 0.4038, − 0.3166, and − 0.2294; all confidence intervals do not contain 0, and the indirect effect is negative and gradually weakening. With the deepening of regional digital transformation, the adjustment of the labor force structure at the industry level is stabilizing, and the labor force gradually adapts to this change. The labor force structure at the industry level has a significant nonlinear mediation effect between regional digital transformation and employment quality, and hypothesis H2a holds.

The instantaneous mediation effect of labor force structure at the sector level is − 0.4066, − 0.3585, and − 0.3104, and all confidence intervals do not contain 0. The regional digital transformation pushes the adjustment of labor force structure in the sector, and the negative impact is weakening, indicating that the labor force is gradually adapting to this change, and the labor force structure at the sector level has a significant nonlinear mediation effect between regional digital transformation and employment quality. Hypothesis H2b holds.

While the skill level shows a different trend, the instantaneous indirect effect is 0.5405, 0.3872, and 0.3568; all confidence intervals do not contain 0. The indirect effect is positive and gradually weakening, indicating that the increase in demand for high-skilled labor has a positive impact on the enhancement of employment quality. Still, this marginal effect might gradually decrease, and the labor force structure at the skill level has a significant nonlinear mediating effect between regional digital transformation and employment, as hypothesis H2c holds. In summary, labor force structure under different micro-levels plays an important mediating role in the impact of regional digital transformation on employment quality.

Table 6 Test results for Instantaneous Indirect effects of Labor Force structure.
Full size table

Conclusions and discussion

Research conclusion

In essence, one of the significance of regional digital transformation and development is the ability to closely integrate social and employment development trends, formulate appropriate policies and plans, and play a key role in driving the growth of digitization and employment quality, and thus social optimization. This study uses a nonlinear model and an instantaneous mediation effect model to quantitatively assess the path between regional digital transformation and employment quality. A comprehensive study that considers industry level, sector level, skill level, and other dimensions of refinement demonstrates the role of labor force structure as a bridge between the two. This multidimensional approach deepens the understanding of the impact of the double-edged sword of digitization and provides valuable insights into the perspective of improving employment quality.

From the perspective of labor force structure, this study analyzes the nonlinear influence mechanism of regional digital transformation and employment quality, scientifically understands the trends and causes, stabilizes employment quality in the early stages of regional digital transformation, raising employment levels in the middle and late stages, and constructing a labor resource system with high quality, stable allocation, and even distribution. Firstly, the direct path of the impact of regional digital transformation on employment quality and the indirect path of the effect through labor force structure is proposed, and the relevant hypotheses are put forward on this basis. Secondly, this study constructs an evaluation index system of regional digital transformation and employment quality and comprehensively evaluates the level of regional digital transformation and employment quality of 31 provinces in mainland China from 2013 to 2022. Finally, this study establishes a relevant empirical model and analyzes the impact mechanism of regional digital transformation on employment quality and so on. The conclusion shows that:

Regional digital transformation has a direct impact on employment quality and with the gradual development of regional digital transformation, employment quality roughly shows a “U” curve, with a short-term decline in employment quality at the initial stage due to the difficulty of some workers in adapting quickly, and an upward trend in the later stage as they adjust to regional digital transformation. At the same time, a nonlinear mediation effect model is established to verify the indirect path, and regional digital transformation affects the change of employment quality through the labor force structure at the industry, sector, and skill level, respectively, and shows a “U”-shaped change trend.

Discussion

This research has implications for the social field of regional digital transformation and labor resources. Trends in the impact of regional digital transformation and employment quality are provided, and the indirect effect of different levels of labor force structure is analyzed. Rationalized recommendations for further optimization of employment quality are provided. However, there are some differences and improvements in the findings of this study compared with previous studies. While there have been many research topics related to the digital economy and employment quality, for example, For example, Si and Li (2022)30and Qi et al. (2020)31discuss the role of digitization in enhancing employment quality, exploring the positive impact of both. Balsmeier and Woerter (2019)14examined the creative and destructive impacts of digitization on employment, Frey and Osborne (2017)32explored future trends in computerized employment, and Borland and Coelli (2017)33analyzed the impacts of the impact of digital robots on employment. However, these studies also do not explore the mechanism of action between the two in more detail, and this study gives factual evidence of the important non-linear mediating influence of labor force structure in the process of marginal contribution and empirical argumentation. To better explain the mechanisms involved, this study constructs mediating variables of labor force structure at sector, sector, and skill levels.

This study empirically analyzes the trend of employment quality from the perspective of regional digital transformation, which is helpful for both practice and theory construction in related fields. However, the limitations of this study should also be noted. In the future, it is necessary to refine different scenarios, analyze the U-shaped trend in various regions and at different stages, and put forward targeted recommendations for different sectors and industries to promote the regional digital transformation and, at the same time, protect the industries and people damaged by the “creative destruction”. Expand research directions to more comprehensively explore the impact of regional digital transformation on employment quality. Promote the deep integration of digital technology and the labor market.

This study aims to provide a new direction of employment quality for the development of countries, emphasizing the strategy of changing trends in employment quality-oriented to regional digital transformation. Good employment quality is closely related to people’s well-being. Therefore, in the process of promoting digital upgrading, we need to give full consideration to how to satisfy people’s employment quality, improve their quality of life, and promote the overall progress of society. We hope that these studies will provide valuable references and lessons for countries to integrate digitization and employment more scientifically and rationally.

Policy Recommendations.

1. Optimize and upgrade the regional digital transformation infrastructure.

The regional digital transformation of various industries should be sustained so that it becomes a driving factor for economic growth, and policies on digital infrastructure construction should be improved to ensure a stable external construction environment. Strengthen direct input, rationalize planning, and steadily expand investment in regional digital transformation to promote digital development.

2. Rationally control structural unemployment and optimize labor force structure.

In the initial period, the use of digital technology to replace some of the middle and low-skilled workers resulted in some groups being unable to adapt to the work under regional digital transformation in the short term, triggering the phenomenon of structural unemployment. It is necessary to improve the skill level of the labor force to make it more adaptable to market demand. Provide flexible work opportunities through three levels of labor force structure.

3. Deepen the penetration of digital technology to promote employment quality.

To promote regional digital transformation and upgrading, firstly, it is necessary to build a compatible and integrated model with digital technology; secondly, we should play the full role of network platforms. Through long-term and sustained efforts to increase regional digital transformation, we will continue to enhance employment quality at the right end of the “U-shaped” curve. It promotes higher labor productivity, reduces the need for repetitive labor, and improves workers’ skills in adapting to the digital work environment.