Does servitization of industrial structure adversely affect the quality of economic development? Evidence from China

Abstract

China is shifting from focusing on speed of economic development to quality development. As the proportion of service industry increases, the impact of servitization of industrial structure on high-quality development urgently needs to be clarified. This paper for the first time uses the panel data of 280 prefecture-level cities ranging from 2003 to 2019 in China, applies the dynamic panel Generalized Method of Moment (GMM) estimation method, employs labor productivity, average wage, and environmental pollution as intermediary variables, and analyzes how servitization of industrial structure affects the quality of economic development. The results show that there is a U-shaped effect for servitization of industrial structure on quality of economic development. The higher the initial level of servitization of industrial structure, the weaker the negative effect for servitization of industrial structure on the quality of economic development. When the initial level of the former is high enough, it can promote the latter. Moreover, the effects differ by region. Furthermore, negative effect of the service-oriented trend of industrial structure on economic development quality is weakened by improving the overall labor productivity and reducing environmental pollution, while it is strengthened by reducing the average wage level. As such, we propose that it is necessary to further promote the rationalization of the industrial structure, promote the optimization of the development of the industrial structure, and to promote the effective allocation of resources.

1. Introduction

Since the reform and opening up, China's economy has maintained rapid growth for more than 40 years, and its per capita GDP entered the ranks of upper middle-income countries in 2009. The rapid economic development is due to the continuous progress of reform and opening up, as well as the continuous increase in the proportion of non-agricultural industries, namely secondary and tertiary industries. By 2035, China will be in a new stage of development to build a modern country in an all-round way. At this stage, China faces two major tasks: first, to accelerate the progress of upper-middle-income countries towards high-income countries; Second, we will vigorously promote high-quality economic development. Now the question is, the increase of the proportion of secondary and tertiary industries, especially the proportion of manufacturing, is conducive to the rapid development of the economy. How does the change of the internal structure of secondary and tertiary industries, especially the increase of the proportion of service, affect the economic development? How does industry servitization affect the quality of economic development?

In 2012, China's proportion of the tertiary industry in GDP exceeded that of the secondary industry, and in 2015, its proportion of the tertiary industry in GDP exceeded 50% for the first time. However, the GDP growth rate has been on a sustained and significant decline since 2010, from 10.6% in 2010 to 6.1% in 2019.1 The prevailing explanation for the current economic downturn is a “structural deceleration ". It is argued that the increase in the proportion of the service industry and the concentration of employment in the tertiary industry leads to further expansion of the tertiary industry with a low labor productivity growth rate. This leads to a reduction in the overall labor productivity of the whole society. Therefore, the reason for China's slowing economic growth rate is attributed to its service-oriented industrial structure. However, some scholars still disagree with this view [1–4]. The development experience of several other countries also shows that service-oriented industrial structure will lead to more distinct differentiation and efficiency differences in the international economy. When the service industry becomes the leading industry in the national economy, developed countries in Europe and North America pay attention to efficiency improvement, both capital deepening and consumption structure optimization, recognize the advanced internal structure of the service industry, and finally bring about the sustained efficiency of economic development. Latin American countries, on the other hand, blindly pursue the expansion of the service industry in terms of scale, while ignoring the transformation and upgrading of the service industry. As a result, their service industry has been dominated by the traditional low-end consumer service industry for a long time. With the increase in the proportion of the tertiary industry, the economy has been stagnant for a long time, accompanied by Baumol's cost disease [5].

This study will verify the impact of the industrial structure service-oriented trend on the quality of economic development. Our study will clarify the following questions: How does the service-oriented trend of industrial structure affect the quality of economic development? Is the effect nonlinear? What are the channels of influence? Does the initial level of service-oriented industrial structure have a moderating effect on this impact? The explanation of these problems and relevant policy suggestions will provide theoretical support and policy suggestions for China to achieve high-quality development in the process of industrial servitization.

This study finds that there is a nonlinear effect of servitization of industrial structure on the quality of economic development: The higher the initial level of servitization of industrial structure, the weaker the negative effect of servitization of industrial structure on the quality of economic development; when the initial level of the former is high enough, it can promote the latter; the effects differ by region. Furthermore, we verify that the negative effect of the service-oriented trend of industrial structure on economic development quality is weakened by improving the overall labor productivity and reducing environmental pollution, while it is strengthened by reducing the average wage level.

The remainder of this study is organized as follows. Section 2 reviews the existing literature. Further, Section 3 presents the research hypotheses. Section 4 describes the data and the study's methodology. Moreover, Section 5 presents the results and discusses the results. Finally, the last section concludes the study and provides policy advice.

2. Literature review

There are two opposing views about how will servitization of industrial structure affect economic development in academia: “promoting theory” and “suppression theory”. In the history of economic thought, scholars such as Petty, Clark, Kuznets, Channary and Hoffman put forward a series of theories on the evolution of industrial structure successively. Afterwards [6,7], introduce industrial structure factors into the growth theoretical model, and prove that industrial structure is an important variable of economic growth using empirical evidence. In the 1940s, economists had proposed that with the continuous development of the national economy, the industrial structure would shift from being dominated by the primary sector to being dominated by the secondary sector, followed by a shift from the secondary sector to the tertiary sector [8,9]. [10,11] have shown that the industrialization mode of labor and capital shifting from low-productivity industries to high-productivity industries significantly improves the efficiency of resource allocation, thus bringing about rapid economic growth. However, with the continuous advancement of the industrial structure, the trend of service-oriented development is constantly strengthened, and the technical progress of the service industry is wholly insufficient compared with that of the manufacturing industry. Therefore, when the industry gradually gives way to the service industry, it will restrict the improvement of the overall labor productivity of society at large, thus inhibiting and hindering its economic growth [12].

The “promoting theory” is mainly based on the structural dividend in the industrialization stage. This theory suggests that the flow of input factors from sectors with low productivity and low productivity growth to sectors with high productivity and high productivity growth lead to the increase in the productivity level for society at large, resulting in a “structural dividend” [13, 14, 15, 16, 17, 18]. A “structural dividend” sustains economic growth and is the core reason why the industrial structural change promotes economic growth [14,15]. [16] also confirm that the structural dividend of China's transformed industrial structure is the key factor and driving force for its rapid economic growth. [17] also points out that in the process of industrial structure adjustment in China, the inter-regional and inter-departmental flow of labor force is an important factor in the improvement of efficiency. More and more domestic scholars have explored the impact of industrial structure on economic growth from the perspective of structural dividend, but it does not reach a complete consensus. [18] believes that China's economy is undergoing the industrial and urban transformation experienced by developed countries several decades ago, and that the “structural acceleration” caused by China's industrialization led to the rapid growth in the past 30 years. [13] show that the rationalization and upgrading of industrial structure will have a positive impact on economic growth. Some scholars also believe that although industrial structure has a positive impact on economic growth, such a structural dividend is gradually weakening with the advance of reform. The research of [19] show that the labor factor has a more obvious structural dividend than the capital factor, but that the industrial structure effect of labor tends to weaken during the process of reform. [20] also reach a similar conclusion. By decomposing the structural and technological progress effects from factor productivity growth rates, they find that the structural effect has made a significant contribution to the economic growth in China over the past three decades. But with the increasing of marketization, the structural effect of industrial structural change has gradually given way to the technological progress effect and its contribution to economic growth has been weakening.

“Inhibition theory” is based on Baumol's cost in the service-oriented stage [12]. It argues that the labor productivity of the service industry is generally lower than that of the manufacturing industry, and that the increase of the proportion in the service industry will bring down the labor productivity [18, 21, 22, 23]. [21] finds that after World War II, the labor force in developed countries shift from the industrial sector with a higher labor productivity growth rate to the service sector with a lower labor productivity growth rate. This leads to a decline in the social labor productivity growth rate, providing evidence for Baumol's disease. Based on the Maddison statistics database [18], studies factors of economic growth in developed countries, and finds the labor productivity changes can explain about 85% of the per capita GDP growth. As the employment rate gradually shifts to the tertiary industry, high proportion of employment and the expansion of the low labor productivity growth in services hinder the progress of society wide labor productivity growth, resulting in “structural deceleration”. [22] find that the structural expansion of the tertiary industry would restrain the positive effect of the first and second industries on the economic scale, eventually leading to economic deceleration. [23] finds that the economic growth in eastern China has entered a structural deceleration stage, while the central and western regions have experienced structural acceleration due to the continuous development of industrialization. “Inhibition theory” scholars insist on structural deceleration as a primary explanation for the economic downturn. They argue that the economic deceleration of developed countries after the 1970s is closely related to the deceleration of productivity growth and that the deceleration of productivity is caused by the structural negative interest of service-oriented industrial structure. They argue that another reason for this slowdown of China's economic growth is due to the service-oriented economy. However, scholars still have different views on the reasons for the economic slowdown. [2] believe that the simultaneous decline in the contribution rate of total factor productivity of the three industries towards economic growth is the real reason for the slowdown of economic growth. They thus believe that there is insufficient evidence for the “structural deceleration” of China's economy. [3], based on the proportion of the output value of the three industries measured by constant price, believe that China is not in the condition of “structural deceleration " of economic service and point out that the technical attribute of the service industry is not immutable, which will become an important strategic variable for China's economy to maintain a medium-high growth rate in the future. [4] also find out that the development of the service industry is not in conflict with economic growth, that the industry in China's “structural deceleration” phase is not caused by the service industry, and that the decline of China's economic momentum is rooted in its major structural imbalance and its difficulty in sustaining Chinese-style industrialization. [1] believes that the total factor productivity of service industry is underestimated due to the neglect of labor force heterogeneity, which provides unreliable evidence for “structural deceleration ".

The rapid economic growth provides the material foundation for the improvement of people's living standard; however, the economic speed growth is not equal to the economic quality growth. The rapid economic growth may be accompanied by many problems that affect the economic quality, especially in developing countries, such as: lager economic fluctuations, serious inflation problems, asset prices fluctuating, higher housing prices, serious resource and environmental damage, and so on. Therefore, we should not only pay attention to economic growth, but also pay attention to the quality of economic development. More importantly, we should pay more and more attention to the quality of economic development, especially in developing countries.

There are two opposing views in academia on the impact of industrial structure changes on economic growth. Will the impact of service-oriented industrial structure on the quality of economic development be similar to the impact on economic growth? There are few quantitative studies on how the industrial structure affects the quality of economic development in China. [24] uses the panel data of 30 provinces between the years 1995 and 2018 in China which analyzes the total production factor of tertiary industry. They find that the adjustment cost of the industrial service structure and the ratio of production service to the tertiary industry will minimize the improvement of the total production factor. [25] uses the panel data of 30 provinces between the years 2000 and 2017 in China which empirically investigates the influence of industrial structure change on China's high-quality development. Their results show that the change of industrial structure is conducive to promoting high-quality development in China. [26] show that China's science and technology innovation not only promotes the improvement of economic quality in the region, but also has positive spatial spillover, leading to the improvement of economic quality in neighboring regions. They find that industrial structure upgrading is an important transmission path for science and technology innovation to promote sustainable and high-quality economic development. [27] find that the upgrading of industrial structure itself is not conducive to high-quality development, but the upgrading of industrial structure promoted by the development of digital economy is conducive to high-quality development, and digital economy can effectively overcome the low efficiency of service industry in the process of structural transformation.

This study for the first time uses the panel data of 280 prefecture-level cities between the years 2003 and 2019 in China, applies the dynamic panel GMM estimation method, and analyzes how servitization of industrial structure affects the quality of economic development. The results of these studies are very interesting and many of them are first found. We find that there is a U-shaped effect for servitization of industrial structure on the quality of economic development. When servitization of the industrial structure is low, it has a negative effect on the quality of economic development. However, when the former exceeds a certain threshold, it may have a positive effect on the latter. The higher the initial level of servitization on the industrial structure, the weaker the negative effects of servitization on the industrial structure relative to the quality of economic development. When the initial level of the former is high enough, it can promote the latter. Moreover, the effects differ by region. Furthermore, we use labor productivity, average wage, and environmental pollution as intermediary variables to analyze the influence mechanism, which servitization of industrial structure affects in relation to the quality of economic development. We find that negative effect of the service-oriented trend of industrial structure on economic development quality is weakened by improving the overall labor productivity and reducing environmental pollution, while it is strengthened by reducing the average wage level.

3. Research hypotheses

[15] proposes that the main channels through which industrial structure changes and affects economic development include: (1) Structural dividend. Throughout economic development, because more productive sectors pay higher wages, higher wages attract highly skilled workers to move between sectors. The flow of labor from sectors with low productivity or low productivity growth to sectors with high productivity or high productivity growth increases overall labor productivity and productivity growth. (2) Structural negative interest. Baumol's unbalanced growth theory states that under the premise of constant demand, each industry has a different ability to improve its own labor productivity, and an increasing share of labor force shifts from progressive industries with higher productivity growth to stagnant industries with lower productivity growth, thus reducing the overall productivity level. (3) Externality. The externality effects of different industrial sectors are different. Economies with a higher proportion of spillovers of producers and consumers are more likely to achieve faster growth. Producer spillover refers to the easier diffusion of productive knowledge under the spatial proximity and the same institutional basis, while consumer spillover refers to the positive externalities generated by the use of certain products or services.

In the process of economic development from industrialization to service, some scholars believe that the service of industrial structure brings about “structural deceleration”. They use “structural deceleration” to explain economic downturn. However, some scholars believe that the service-oriented degree of China's economy is still low, and that there is insufficient evidence of “structural deceleration”. They believe that the development of the service sector is not contradictory to economic growth, however, and that it has importance in maintaining medium-high growth for China's economy. Referencing the research on the impact of service-oriented industrial structure on economic growth, we study the impact of service-oriented industrial structure on the quality of economic development. What is the impact of service-oriented industrial structure on the quality of economic development? Has service-oriented industrial structure have a long-term nonlinear impact on the quality of economic development? Will the initial level of service-oriented industrial structure play a moderating role in their relationship? Based on the research on the impact of service-oriented industrial structure on economic growth, the following three hypotheses are proposed:

H1

The service-oriented trend of industrial structure has a significant impact on the quality of economic development.

H2

The initial level of service-oriented industrial structure has a moderating effect on this impact.

H3

This impact is a U-shaped one, that is, when the service-oriented level of industrial structure is low, it will have a negative effect on the quality of economic development. However, when the service-oriented level exceeds a certain threshold, it will have a possible positive effect on the quality of economic development.

4. Descriptions of data and methodology

4.1. Descriptions of data

This article uses 287 cities in China and a total of 17 years, ranging from 2003 to 2019, to carry out empirical research and robustness test. Observation samples are 4758, where 7 cities are removed, because of a lack of a large amount of data: Chaohu (Anhui), Lhasa (Tibet), Longnan (Gansu), Zhongwei (Ningxia), Shizuishan (Ningxia), Wuzhong (Ningxia), Guyuan (Ningxia).2 The data in this paper is mainly from CNKI China Economic and Social Development Statistical Database and CEIC China Economic Database. In addition, China's urban innovation index comes from "China's Urban and Industrial Innovation Capacity Report ". Variables are defined as follows:

Explained variables: Economic development quality (QGI). Many researchers use economic total factor productivity, real GDP per capita, pollution emissions of the unit output, and other indicators to measure quality of economic development. However, the single index has difficulty reflecting the quality of economic development; it cannot reflect the social needs and the needs of sustainable development. Therefore, this paper measures the quality level of economic development in different regions by constructing a multi-dimensional comprehensive evaluation system and by using a two-step principal component analysis to measure the economic development quality index. Indicators selected by the economic development quality are shown in Table A1, and descriptive statistics are shown in Table A2.

Referencing [28], this paper uses the entropy weight method to recalculate the economic development quality index in Section 5.2 Robustness test. Firstly, each basic index $X_{ij}$ is standardized by employing the range method and then calculating the information entropy $E_j$， $E_{\mathrm{j}}=\ln \left\{\frac{-1}{n}\sum _{i=1}^n\left(X_{ij}/\sum _{i=1}^n{X}_{ij}\right)\ln \left(X_{ij}/\sum _{i=1}^n{X}_{ij}\right)\right\}$, where $n$ is the observed sample. Using these variables, the weight of each basic index could be calculated as $W_j=\frac{\left(1-E_j\right)}{\sum _j\left(1-E_j\right)}$ . The economic development quality index of each prefecture-level city, which we denote as QGIS, can be calculated through the weighted average. The 15 core indicators to recalculate QGIS are consistent with those of the principal component calculation.

The core explanatory variable: Servitization of industrial structure (TS). Industrial structure changes are usually measured by the proportion of non-agricultural output value according to Clark's law. However, under the promotion of information technology, the economic structure service-oriented trend has become an important feature in industrial structure upgrading. Therefore, this paper uses the ratio of tertiary industry output value to secondary industry output value as well as labor productivity value of tertiary industry to secondary industry to express the level of industrial structure service-oriented trend [29]. Therefore, the larger TS indicates the higher the degree of industrial structure service-oriented trend; the initial value of industrial structure service-oriented trend (INIT) is expressed by the industrial structure service-oriented trend level of each prefecture-level city in the base year of 2003, and INIT is a dummy variable, 1 above the median, 0 otherwise.

Mediating variables: Based on the shift-share model of [30], the perspective of labor income ratio of [31], and the theoretical research of [32], three mediating variables are used in this paper: Labour productivity (LP), expressed as the logarithm of output per employed people; Wage level (GZ), which is expressed as the logarithm of the average wage of employees on the job; Environmental protection (ENV) is expressed as the logarithm of solid waste per output.

Control variables: Refers to the research of [33] along with [34], while trying to avoid the underlying indicators used in measuring the explanatory variable QGI, the following control variables are used in this paper: Openness (OPEN), measured by total imports and exports as a share of GDP; level of government control (GOV), measured by government fiscal expenditure as a share of GDP; urbanization rate (URBAN), expressed by dividing urban population by total population; innovation level (INOV), expressed by the logarithm of China's urban innovation index; urban area (LAND), expressed by the logarithm of total land area in administrative areas; urban population (POP), expressed by the logarithm of household registration population; and net urban inflow (FLOW), expressed by the logarithm of the ratio of resident population to household registration population. Descriptions of variables are shown in Table 1; and descriptive statistic of variables are shown in Table 2.

Table 1.

Description of variable summary.

	Variables	Symbol	Calculation method
Explained variables	Economic development quality	QGI	ln (Principal component analysis method)
		QGIS	ln (Entropy weight method)
Explanatory Variables	Servitization of industrial structure	TS	ln (Output value of tertiary industry/secondary industry)
	Initial value of servitization of industrial structure	INIT	INIT is a dummy variable, 1 if service level of industrial structure in the base year is greater than the median, 0 otherwise
	The square of servitization of industrial structure	TS2	The square of servitization of industrial structure
	Servitization of industrial structure	TSL	ln (Labor productivity value of tertiary industry/secondary industry)
Mediating variables	Labour productivity	LP	ln (Total output/employed population)
	Average wage	GZ	ln (Average wage of on-the-job workers)
	Environmental pollution	ENV	ln (Solid waste output per unit output)
Control variables	Degree of openess	OPEN	ln (Total import and export/GDP)
	Government control	GOV	ln (Government expenditure/GDP)
	Urbanization rate	URBAN	ln (Urban population/total population)
	Innovation level	INOV	ln (China Urban innovation index)
	Urban area	LAND	ln (Land area of administrative region)
	Urban population	POP	ln (Registered residence population)
	Net urban inflow	FLOW	ln (Resident population/registered residence population)

Notes: Urbanization rate is calculated from CNKI China Economic and Social Development Statistical Database; Innovation level is come from China's urban and industrial innovation report; and all the other variables come from CEIC China Economic Database.

Table 2.

Descriptive statistics of variables.

Variables	Observations	Mean	Standard deviation	Minimum value	Maximum value
QGI	4758	2.862	0.407	1.538	4.504
QGIS	4758	1.825	0.499	0.793	3.662
TS	4758	4.429	0.459	2.244	6.320
INIT	4758	0.500	0.500	0.000	1.000
INIT*TS	4758	2.331	2.348	0.000	6.320
TS2	4758	19.822	4.086	5.033	39.940
TSL	4758	4.210	0.540	0.335	5.812
OPEN	4758	11.997	1.677	1.001	17.257
GOV	4758	2.691	0.471	1.141	4.255
URBAN	4758	3.876	0.357	0.339	4.605
INOV	4758	1.112	1.221	0.000	7.438
LAND	4758	9.342	0.821	7.015	12.443
POP	4758	8.177	0.689	5.098	10.439
FLOW	4758	−0.007	0.221	−1.539	2.376
LP	4758	12.295	0.808	9.251	14.310
GZ	4758	10.384	0.627	8.686	12.062
ENV	4758	6.345	1.226	1.055	9.963

Note: Samples with missing data of variables have been excluded. Moving average method is used to make up the missing data.

4.2. Methodology

Economic development is a dynamic process. As a result, the current quality of economic development depends on its past level. Furthermore, when the sample size is large, $N$, and small, $T$, the GMM based instrumental variable estimation method is often more effective. Therefore, this paper uses the systematic and differential GMM estimation to study the impact of the industrial structure service-oriented trend on the quality of economic development. The dynamic model of economic development quality is obtained by introducing the lag of the first and second phases of the dependent variables into this model:

$$QG{I}_{\mathrm{i}\mathrm{t}}={\beta }_0+{\beta }_{1}QG{I}_{\mathrm{i}\mathrm{t}-1}+{\beta }_{2}QG{I}_{\mathrm{i}\mathrm{t}-2}+{\beta }_{3}T{S}_{it}+\sum _m{\theta }_m{X}_{it}+{\mu }_{it}$$ (1)

Where $i$ is the region， $t$ is the time， $QG{I}_{\mathrm{i}\mathrm{t}}$ is the explained variable，that is, the quality of economic development， $QG{I}_{\mathrm{i}\mathrm{t}-1}$、 $QG{I}_{\mathrm{i}\mathrm{t}-2}$ are the first and second lags of the explained variable， $T{S}_{it}$ is the level of industrial structure service-oriented trend， $X_{it}$ is the vector of other control variables, including degree of openness, level of government control, urbanization rate, level of innovation, urban area, urban population, and net urban inflow，and ${\mu }_{it}$ is the random error term.

To study the nonlinear relationship that may exist between industrial structure service-oriented trend and the quality of economic development, the interaction term of the level of industrial structure service-oriented trend in the base year $TS*INI{T}_{\mathrm{i}\mathrm{t}}$, and the square term of industrial structure service-oriented trend $T{S}_{\mathrm{i}\mathrm{t}}^2$ are introduced on the basis of the basic model, respectively.

$$QG{I}_{\mathrm{i}\mathrm{t}}={\beta }_0+{\beta }_{1}QG{I}_{\mathrm{i}\mathrm{t}-1}+{\beta }_{2}QG{I}_{\mathrm{i}\mathrm{t}-2}+{\beta }_{3}T{S}_{it}+{\beta }_{4}T{S}_{it}*INI{T}_{\mathrm{i}\mathrm{t}}+\sum _m{\theta }_m{X}_{it}+{\mu }_{it}$$ (2)

$$QG{I}_{\mathrm{i}\mathrm{t}}={\beta }_0+{\beta }_{1}QG{I}_{\mathrm{i}\mathrm{t}-1}+{\beta }_{2}QG{I}_{\mathrm{i}\mathrm{t}-2}+{\beta }_{3}T{S}_{it}+{\beta }_{4}T{S}_{\mathrm{i}\mathrm{t}}^2+\sum _m{\theta }_m{X}_{it}+{\mu }_{it}$$ (3)

In order to test whether labor productivity, average wage and environmental pollution are the three channels through which servitization of industrial structure affects the quality of economic development, this study uses the mediating effect model of [35,36] to test the effect mechanism. The above models (1) and (2) correspond to (4)–(5) and (6)–(7) mechanism models, respectively:

$$E_{\mathrm{i}\mathrm{t}}={\alpha }_0+{\alpha }_1{E}_{\mathrm{i}\mathrm{t}-1}+{\alpha }_{2}T{S}_{it}+\sum _k{\alpha }_k{X}_{it}+{\mu }_{it}$$ (4)

$$QG{I}_{\mathrm{i}\mathrm{t}}={\theta }_0+{\theta }_{1}QG{I}_{\mathrm{i}\mathrm{t}-1}+{\theta }_{2}QG{I}_{\mathrm{i}\mathrm{t}-2}+{\theta }_3{E}_{it}+{\theta }_{4}T{S}_{it}+\sum _k{\theta }_k{X}_{it}+{\mu }_{it}$$ (5)

$$E_{\mathrm{i}\mathrm{t}}={\alpha }_0+{\alpha }_1{E}_{\mathrm{i}\mathrm{t}-1}+{\alpha }_{2}T{S}_{it}+{\alpha }_{3}T{S}_{it}*INI{T}_{\mathrm{i}\mathrm{t}}+\sum _k{\alpha }_k{X}_{it}+{\mu }_{it}$$ (6)

$$QG{I}_{\mathrm{i}\mathrm{t}}={\theta }_0+{\theta }_{1}QG{I}_{\mathrm{i}\mathrm{t}-1}+{\theta }_{2}QG{I}_{\mathrm{i}\mathrm{t}-2}+{\theta }_3{E}_{it}+{\theta }_{4}T{S}_{it}+{\theta }_{5}T{S}_{it}*INI{T}_{\mathrm{i}\mathrm{t}}+\sum _k{\theta }_k{X}_{it}+{\mu }_{it}$$ (7)

where $E_{\mathrm{i}\mathrm{t}}$ is the mediating variable, which refers to labor productivity, average wage, and environmental pollution. If ${\alpha }_2$, ${\theta }_3$, and ${\theta }_4$ are all significant，it means that servitization of industrial structure affects the quality of economic development through partial mediating effect; if ${\alpha }_2$ and ${\theta }_3$ are significant, however, ${\theta }_4$ is not significant，it means that servitization of industrial structure affects the quality of economic development through complete mediating effect.

5. Empirical results

5.1. Baseline empirical results

We apply the Im-Pesaran-Shin unit-root test and the Hadri LM test to test two time series: economic development quality, (QGI), and servitization of industrial structure, (TS), and find that these two time series are nonstationary (i.e., there is a unit root).3 However, when the Kao test and Westerlund test for cointegration are applied, it is found that these two time series are cointegrated, or in other words, there seems to be a long-run or equilibrium relationship between the two variables.4

Tables 3 and 4 report the results of the systematic and differential GMM estimation of the impact of industrial structure service-oriented trend on the quality of economic development, expressed as the ratio of output value of tertiary to secondary industries, and the ratio of labor productivity value of tertiary to that of secondary industries respectively. To enhance the reliability of the regression results, the Arellano-Bond test is conducted for the rationality of the model setting; the Arellano-Bond test indicates that there is no second-order autocorrelation. Model (1) in the table is the basic model, and the estimation results are shown in models (2) and (3) in the table after considering the initial level of industrial structure service-oriented trend $INI{T}_{\mathrm{i}\mathrm{t}}$ and the squared term of industrial structure service-oriented trend in the base year$T{S}_{\mathrm{i}\mathrm{t}}^2$.

Table 3.

The influence of servitization of industrial structure (output value ratio) on the quality of economic devetlopment.

VARIABLES	System GMM			DIF-GMM
	Model (1)	Model (2)	Model (3)	Model (1)	Model (2)	Model (3)
	QGI	QGI	QGI	QGI	QGI	QGI
TS	−0.0535***	−0.0682***	−0.2068***	−0.0796***	−0.0787***	−0.1801**
	(-5.8602)	(-5.8087)	(-2.6426)	(-5.3442)	(-6.0275)	(-2.2248)
INIT *TS		0.0077***			0.0268*
		(2.8934)			(1.7056)
TS2			0.0175**			0.0159*
			(2.0216)			(1.6652)
OPEN	−0.0054	−0.0023	−0.0029	−0.0171***	−0.0162***	−0.0199***
	(-1.2762)	(-0.4891)	(-0.7277)	(-3.0168)	(-4.5328)	(-4.2470)
GOV	0.0618***	0.0604***	0.0595***	0.0599**	0.0781***	0.0553***
	(3.5762)	(3.5801)	(3.9698)	(2.5359)	(4.4776)	(2.8169)
URBAN	0.0927***	0.0799**	0.0722**	0.1389*	0.0855*	0.1189
	(3.0616)	(2.2260)	(2.1245)	(1.6768)	(1.7881)	(1.6041)
INOV	0.0175***	0.0294***	0.0132**	0.0561***	0.0484***	0.0278***
	(2.8806)	(4.1610)	(2.1193)	(5.1793)	(6.1753)	(2.8697)
LAND	−0.0567***	−0.0755***	−0.0517***	−0.1462	−0.0539	−0.0607
	(-3.8679)	(-4.5595)	(-3.6584)	(-1.6090)	(-1.2346)	(-1.0980)
POP	0.1037***	0.0755***	0.0935***	0.0895	−0.0217	0.0712
	(5.9235)	(4.6220)	(5.4426)	(0.7044)	(-0.4116)	(0.9250)
FLOW	0.1124***	0.0924***	0.0764**	0.0071	−0.0137	−0.0139
	(3.1467)	(2.7578)	(2.4613)	(0.2094)	(-0.6328)	(-0.5182)
L.QGI	0.6721***	0.6789***	0.7096***	0.4994***	0.5551***	0.5846***
	(11.6581)	(11.9979)	(12.4166)	(7.0825)	(8.0089)	(9.7915)
L2.QGI	0.1192**	0.1146**	0.1235**	0.1137***	0.1002**	0.0996**
	(2.4494)	(2.2180)	(2.3249)	(2.7432)	(2.3761)	(2.2656)
Constant	0.0698	0.5213*	0.3807	1.5698	1.5814**	1.0004
	(0.3118)	(1.8855)	(1.2085)	(1.5460)	(2.3934)	(1.1124)
Observations	4194	4194	4194	3912	3912	3912
AR1	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000
AR2	0.4736	0.5550	0.4992	0.2996	0.5282	0.5233
${\chi }^2$	22714	15920	21985	6974	12800	11466

Notes: all GMM estimates use the lag order of dependent variable and independent variable TS、INIT*TS、TS2 as instrumental variables. Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. *, **, and *** denote variables significant at the 10%, 5%, and 1% level, respectively.

Table 4.

The influence of servitization of industrial structure (labor productivity ratio) on the quality of economic devetlopment.

VARIABLES	System GMM			DIF-GMM
	Model (1)	Model (2)	Model (3)	Model (1)	Model (2)	Model (3)
	QGI	QGI	QGI	QGI	QGI	QGI
TSL	−0.0400***	−0.0259**	0.0249	−0.0649***	−0.0113	0.0311
	(-3.8432)	(-2.4469)	(0.8199)	(-2.9358)	(-0.4067)	(0.3148)
INITL*TSL		−0.0005			−0.1181**
		(-0.1370)			(-2.4441)
TSL2			−0.0062			−0.0075
			(-1.5489)			(-0.6201)
Control	YES	YES	YES	YES	YES	YES
Observations	4194	4194	4194	3912	3912	3912
AR1	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000
AR2	0.2881	0.4136	0.5399	0.1747	0.2964	0.3085
${\chi }^2$	19209	15852	20880	4865	4344	8709

Notes: all GMM estimates use the lag order of dependent variable and independent variable TSL、INITL*TSL、TSL2 as instrumental variables. Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. **, and *** denote variables significant at the 5% and 1% level, respectively.

Model (1) in Table 3 shows that the service-oriented level of industrial structure has a significant negative effect on the quality of economic development, with coefficients of −0.05 in system GMM, and −0.08 in Diff-GMM; both of which are significant at the 1% level. The results support the idea of structural deceleration, that is, the improvement of the service-oriented level of industrial structure has a restraining effect on the quality of economic development. This result is similar to Ref. [24], which expresses the economic development as the total factor production of tertiary industry by applying the panel data of 30 provinces between the years 1995 and 2018 and applies the ratio of production service to tertiary industry to measure industrial structure transition index. On the other hand, this result is contrary to Ref. [25], which measures the economic development quality index by applying the panel data of 30 provinces between the years 2000 and 2017, and applies Norm of Absolute Values to measure industrial structure transition index.

Model (2) contains the industrial structure and its initial level of service interaction, coefficient of 0.01 in system GMM and 0.03 in Diff-GMM, and is significant at the 1% and 10% level, respectively. It shows that the initial level of industrial structure servitization has a moderating effect on the relationship between servitization of industrial structure and the quality of economic development. The higher the initial service level of the industrial structure, the weaker the inhibitory effect on the quality of economic development. When the initial service level of the industrial structure is high enough, servitization of industrial structure has a promoting effect on quality of economic growth. Considering the quadratic terms of service-oriented industrial structure. The regression results of model (3) shows that the coefficient of the squared term is 0.02 both in system GMM and Diff-GMM, and is significant at the 5% and 10% level, respectively. It means that there may be U-shaped relationship between service-oriented industrial structure and quality of economic development. That is to say, when the degree of servitization of industrial structure is low, it will hinder the quality of economic development. However, when the level of servitization of industrial structure exceeds a certain threshold, it will have a possible positive effect on the quality of economic development. Thus, hypothesis 1-3 is verified.

Model (1) in Table 4, considering the ratio of labor productivity of tertiary industry to secondary industry (TSL), shows that the service-oriented level of industrial structure has a significantly negative effect on the quality of economic development. The coefficient is −0.04 in system GMM and −0.06 in Diff-GMM. The coefficients are significant at 1%. INITL is a dummy variable, 1 if service level of industrial structure (TSL) in the base year is greater than the median, 0 otherwise. TSL2 is the square of servitization of industrial structure (TSL). However, regression results of initial level and square terms are not similar to that of Table 4.

Overall, the ratio of the output value, whether it be a difference of GMM or system GMM regression, lag phase one and lag phase two of the quality of economic development model have a significant promoting effect on current economic development. The degree of openness has an inhibitory effect on quality of economic development in difference of GMM; Government control has a positive effect on the quality of economic development, which is contrary to the research conclusion of [34]; urbanization level has significant positive effect in almost all cases; Innovation has a significant positive role in regional economic development. The results of the system GMM estimation demonstrates that the urban area significantly inhibits the quality of economic development, while urban population and urban net inflow significantly improves the quality of economic development.

5.2. Robustness test

The above empirical results have validated three proposed research hypotheses, and to further test the accuracy of these findings, this study remeasures the economic development quality index using the entropy weight method. The regression results in Table 5 show that the coefficients of TS are −0.03 in system GMM and −0.06 in Diff-GMM, and are significant at the 1% level; The coefficients of the interaction term of industrial structure service-oriented trend and its initial value are 0.01 and 0.02, and are significant at the 1% and 10% level, respectively; The coefficients of the quadratic term of industrial structure service-oriented trend are both 0.02, and are significant at the 1% and 5% level significant, respectively. All these results are consistent with the previous findings. As for other control variables, results are similar to the previous ones.

Table 5.

The robustness test.

VARIABLES	System GMM			DIF- GMM
	Model (1)	Model (2)	Model (3)	Model (1)	Model (2)	Model (3)
	QGIS	QGIS	QGIS	QGIS	QGIS	QGIS
TS	−0.0273***	−0.0288***	−0.1904***	−0.0564***	−0.2048***	−0.0654***
	(-3.0258)	(-2.6643)	(-3.8437)	(-4.9493)	(-2.9517)	(-4.3667)
INIT*TS		0.0067***				0.0247*
		(2.8381)				(1.7490)
TS2			0.0193***		0.0197**
			(3.4848)		(2.4477)
Control	YES	YES	YES	YES	YES	YES
Observations	4194	4194	4194	3912	3912	3912
AR1	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000
AR2	0.7733	0.4026	0.6375	0.4606	0.6765	0.4520
${\chi }^2$	62149	45808	63050	44080	48365	40506

Notes: all GMM estimates use the lag order of dependent variable and independent variable TS、INIT*TS、TS2 as instrumental variables. Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. *, **, and *** denote variables significant at the 10%, 5%, and 1% level, respectively.

5.3. Heterogeneity analysis

Considering the imbalance of regional economic development in China, there are great differences between regions in industrial structure and economic development level. This paper likewise analyzes the effect of industrial structure on quality of economic development by regions. According to Ref. [37], four regions are divided: the eastern, central, western and northeastern regions. The eastern region includes 10 provinces, including Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong and Hainan. The central region comprises 6 provinces, including Shanxi, Anhui, Jiangxi, Henan, Hunan and Hubei. The western region includes 12 provinces, including Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia and Xinjiang. The northeastern region includes Jilin, Heilongjiang and Liaoning provinces.

The empirical results in Table 6 show that, in the eastern and central cities, the service-oriented level of industrial structure has a significant negative effect on the quality of economic development. The inhibit effect of the eastern city is notably stronger than that of the central city. However, in western and northeastern city, servitization of industrial structure has not a significant effect on quality of economic growth. The above results indicate that the inhibiting effect of service-oriented industrial structure on the quality of economic development in China is dominated by the eastern and central regions.

Table 6.

Heterogeneity test.

VARIABLES	Model (1)	Model (1)	Model (1)	Model (1)
	EAST	MIDDLE	WEST	NORTHEAST
	QGI	QGI	QGI	QGI
TS	−0.1077***	−0.0659***	−0.0806	−0.1750
	(0.0301)	(0.0173)	(0.0930)	(0.5618)
Control	YES	YES	YES	YES
Observations	1414	1238	784	476
AR1	0.0000	0.0001	0.4554	0.8589
AR2	0.3173	0.8739	0.7857	0.9225
${\chi }^2$	2072	7179	3915	440.5

Notes: Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. *** denotes variables significant at the 1% level.

5.4. Analysis of influence mechanism

The robustness test and heterogeneity test show that our hypothesis is still valid. Then how does the service-oriented industrial structure affect the quality of economic development? Three effect mechanism are studied. The first one is the mediating effect of labor productivity. The empirical results of models (4) and (6) show that servitization of industrial structure has significantly positive impacts on the labor productivity in Table 7. Furthermore, intermediary effect is tested in models (5) and (7). The empirical results show that the labor productivity has a significantly positive effect on quality of economic growth, and coefficients of industrial structure are significant. According to the condition of mediating effect, labor productivity has a partial mediating effect on the impact of industrial structure on the quality of economic development. From the empirical results, the service-oriented trend of industrial structure has an improving effect on labor productivity and the improvement of labor productivity can promote the quality of economic development. Therefore, negative effect of the service-oriented trend of industrial structure on economic development quality is weakened by improving the overall labor productivity. The second one is the mediating effect of average wage. The regression results show that the average wage plays a partial mediating role in the effect of service-oriented industrial structure on the quality of economic development in Table 8. The third one is the mediating effect of environmental pollution. The empirical results in Table 9 show that service-oriented trend of industrial structure has a restraining effect on environmental pollution, and the effect of environmental pollution on the quality of economic development is significant, that is, environmental pollution plays a partial mediating role in the effect of service-oriented industrial structure on the quality of economic development.

Table 7.

The test of mediating effect based on labor productivity.

VARIABLES	Model (1)	Model (2)	Model (4)	Model (5)	Model (6)	Model (7)
	QGI	QGI	LP	QGI	LP	QGI
LP				0.0779***		0.0803***
				(3.4800)		(3.9149)
TS	−0.0796***	−0.0787***	0.0346*	−0.0603***	0.0414*	−0.0526***
	(-5.3442)	(-6.0275)	(1.7727)	(-3.3123)	(1.6510)	(-2.6563)
INIT*TS		0.0268*			−0.0084	0.0091
		(1.7056)			(-0.3530)	(0.4409)
L.LP			1.2220***		1.2042***
			(26.3797)		(32.5348)
L2.LP			−0.2601***		−0.2417***
			(-5.1893)		(-6.2581)
L.QGI	0.4994***	0.5551***		0.3728***		0.3837***
	(7.0825)	(8.0089)		(5.2598)		(5.6222)
L2.QGI	0.1137***	0.1002**		0.0993***		0.1099**
	(2.7432)	(2.3761)		(2.6604)		(2.1232)
Control	YES	YES	YES	YES	YES	YES
Observations	3912	3912	3912	3912	3912	3912
AR1	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000
AR2	0.2996	0.5282	0.1707	0.3390	0.2306	0.3396
${\chi }^2$	6974	12800	35523	4212	40921	5029

Notes: Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. *, **, and *** denote variables significant at the 10%, 5%, and 1% level, respectively.

Table 8.

The test of mediating effect based on average wage.

VARIABLES	Model (1)	Model (2)	Model (4)	Model (5)	Model (6)	Model (7)
	QGI	QGI	GZ	QGI	GZ	QGI
GZ				0.1738***		0.2127***
				(9.5545)		(8.6077)
TS	−0.0700***	−0.0787***	−0.0361***	−0.0869***	−0.0352***	−0.0735***
	(-6.1752)	(-6.0275)	(-4.8933)	(-6.3326)	(-2.9868)	(-3.1468)
INIT*TS		0.0268*			0.0204*	−0.0247
		(1.7056)			(1.8509)	(-0.8633)
L.GZ			0.9327***		0.9446***
			(29.4465)		(17.0723)
L2.GZ			0.0153		0.0001
			(0.5366)		(0.0019)
L.QGI	0.5507***	0.5551***		0.2875***		0.1289**
	(10.2720)	(8.0089)		(5.0322)		(2.0385)
L2.QGIi	0.1248***	0.1002**		0.0931***		0.0759
	(3.1480)	(2.3761)		(3.0557)		(1.4754)
Control	YES	YES	YES	YES	YES	YES
Observations	3912	3912	3912	3912	3912	3912
AR1	0.0000	0.0000	0.0000	0.0000	0.0000	0.0079
AR2	0.2562	0.5282	0.0198	0.4227	0.1914	0.6046
${\chi }^2$	12084	12800	270607	5613	266009	2814

Notes: Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. *, **, and *** denote variables significant at the 10%, 5%, and 1% level, respectively.

Table 9.

The test of mediating effect based on environmental pollution.

VARIABLES	Model (1)	Model (2)	Model (4)	Model (5)	Model (6)	Model (7)
	QGI	QGI	ENV	QGI	ENV	QGI
ENV				−0.0207**		−0.0139*
				(-2.5718)		(-1.6486)
TS	−0.0730***	−0.0787***	−0.1361**	−0.0819***	0.0858	−0.0834***
	(-5.5889)	(-6.0275)	(-2.2553)	(-5.9430)	(1.5588)	(-4.1554)
INIT*TS		0.0268*			−0.1320*	0.0120
		(1.7056)			(-1.9478)	(0.6178)
L.ENV			0.7194***		0.8761***
			(14.3298)		(19.2453)
L2.ENV			−0.0579***		−0.0683**
			(-2.7347)		(-2.0062)
L.QGI	0.5257***	0.5551***		0.4778***		0.4800***
	(8.6961)	(8.0089)		(7.8974)		(6.3117)
L2.QGI	0.1128***	0.1002**		0.1072***		0.1085**
	(2.9351)	(2.3761)		(2.8670)		(2.0171)
Control	YES	YES	YES	YES	YES	YES
Observations	3912	3912	3912	3912	3912	3912
AR1	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000
AR2	0.3174	0.5282	0.0000	0.3244	0.0001	0.3814
${\chi }^2$	8837	12800	4217	7120	9071	6587

Notes: Samples with missing data of variables have been excluded. The numbers in the parentheses denote z values. *, **, and *** denote variables significant at the 10%, 5%, and 1% level, respectively.

6. Conclusions and policy implications

This study for the first time uses the panel data of 280 prefecture-level cities ranging from 2003 to 2019 in China, applies the dynamic panel GMM estimation method, employs labor productivity, average wage, and environmental pollution as intermediary variables, and analyzes for the first time how servitization of industrial structure affects the quality of economic development. The results show that there is a U-shaped effect for servitization of industrial structure on quality of economic development. When servitization of industrial structure is low, it has a negative effect on the quality of economic development. However, when the former exceeds a certain threshold, it may have a positive effect on the latter. The higher the initial level of servitization of industrial structure, the weaker the negative effect of servitization of industrial structure on quality of economic development. When the initial level of the former is high enough, the former can promote the latter. Moreover, the effect differs by region.

Furthermore, it is found that labor productivity, average wage, and environmental pollution are the key intermediary variables of the effect of industrial structure on the quality of economic development through the analysis of the mechanism of the effect of industrial structure on the quality of economic development. Negative effect of the service-oriented trend of industrial structure on economic development quality is weakened by improving the overall labor productivity and reducing environmental pollution. Negative effect of the service-oriented trend of industrial structure on economic development quality is strengthened by reducing the average wage level.

Under the background of further development of a service-oriented industrial structure, how can we achieve sustainable and high-quality economic development? This study makes the following policy recommendations. First, it is necessary to accelerate the transformation and upgrading of traditional industries to foster and develop emerging industries vigorously. Second, it is important to upgrade the internal structure of the service industry to increase labor productivity and improve policies to raise wages in line with productivity improvements, recognizing the transition and transformation from labor-intensive to technology-and-knowledge-intensive. Third, it is necessary to continue to improve industrial policies with regional characteristics and achieve balanced and coordinated economic and industrial development in all regions. Fourth, it is essential to develop a low-carbon economy, save resources and energy, reduce emissions, protect the environment, and promote the sustainable development of society.

The effect of servitization on the industrial structure on economic quality is a subject that needs long-term research. The time period of our paper is shorter, so we need to use longer data to further verify the reliability of our research conclusions in the future. In addition, industry servitization has become a common trend in many countries. How does industry servitization affect the quality of economic development? It is necessary to carry out a comparative study to reveal the relevant common characteristics in the future.

Funding

This work was supported by Fujian Undergraduate Education and Teaching Reform Research Project (FBJG20210010), the Key Research Institutes of Humanities and Social of the Ministry of Education of China (17JJD790014). We thank Jiajun Yuan for providing lab support. The remaining errors are ours.

Author contribution statement

Ruoran Zhu: Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Juan Jiang: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper. Yan Cong: Analyzed and interpreted the data; Wrote the paper. Guifu Chen: Conceived and designed the experiments; Wrote the paper.

Does Servitization of Industrial Structure Adversely Affect the Quality of Economic Development? Evidence from China

Appendix.

Table A1.

Index system of economic development quality

Comprehensive indicators	First level indicators	Second level indicators	Third level indicators	Computing method	Unit
The Index of the quality of economic growth (QGI)	Economic development	Validity	GDP per capita	GDP/resident population	Yuan
			Fiscal revenue per capita	Fiscal revenue/resident population	Yuan
			Investment in fixed assets	–	100 million Yuan
		Coordination	Dual contrast coefficient	Agricultural/non-agricultural labor productivity	–
			Binary contrast factor	Proportion of non-agricultural output value - proportion of non-agricultural employment	–
			Balance of deposits and loans/GDP	–	–
			FDI/GDP	–	–
		Innovation	College students/ten thousand people	–	person
			Science and technology expenditure/fiscal expenditure	–	–
			Education expenditure/fiscal expenditure	–	–
		Sharing	Urban per capita disposable income	–	Yuan
			Urban per capita consumption expenditure	–	Yuan
			Number of internet users	–	Household
		Stability	Coefficient of variation	Standard deviation/mean of GDP growth rate	–
			Inflation rate	–	%
			Unemployment rate	Unemployed population/labor force population	%
	Economic quality	Basic quality	Highway mileage/population	–	km/10 thousand people
		Coordination quality	Green space	–	hectare
	Social achievements	Education	Teacher- student ratio	–	–
		Health	Life expectancy	–	Year
		Social security	Number of beds in Medical Institutions	–	bed
		Social equity	Multiple of income gap between urban and rural residents	Per capita urban disposable income/rural net income	–
	Resources and environment	Resource consumption	Energy intensity per unit GDP	Total energy consumption/GDP	10000 tons of standard coal/100 million yuan
			Water consumption per unit GDP	Domestic water consumption/GDP	Million cubic meters/100 million yuan
		Environmental pollution	Air pollution level per unit output	Industrial emissions/GDP	100 million cubic meters/100 million yuan
			Sewage discharge per unit output	Industrial waste water discharge/GDP	10000 tons/100 million yuan

Tabla A2.

Statistical description of indicator of China's economic development quality

Variables	Mean	SD	Minimum	Maximum
Energy intensity per unit GDP	0.830	0.145	0.000	1.000
Water consumption per unit GDP	0.948	0.057	0.000	1.000
Air pollution level per unit output	0.972	0.049	0.000	1.000
Sewage discharge per unit output	0.978	0.037	0.000	1.000
Life expectancy	0.342	0.149	0.000	1.000
Number of beds in medical institutions	0.079	0.080	0.000	1.000
Income gap between urban and rural residents	0.961	0.037	0.000	1.000
Teacher- student ratio	0.289	0.132	0.000	1.000
Green space	0.039	0.086	0.000	1.000
Highway mileage/population	0.162	0.104	0.000	1.000
Inflation rate	0.858	0.094	0.000	1.000
Unemployment rate	0.692	0.052	0.000	1.000
Coefficient of variation	0.740	0.184	0.000	1.000
Number of internet users	0.050	0.072	0.000	1.000
Urban per capita consumption expenditure	0.224	0.120	0.000	1.000
Urban per capita disposable income	0.277	0.152	0.000	1.000
Science and technology expenditure/fiscal expenditure	0.081	0.091	0.000	1.000
Education expenditure/fiscal expenditure	0.411	0.112	0.000	1.000
College students/ten thousand people	0.108	0.145	0.000	1.000
FDI/GDP	0.060	0.067	0.000	1.000
Balance of deposits and loans/GDP	0.448	0.075	0.000	1.000
Binary contrast factor	0.570	0.148	0.000	1.000
Dual contrast coefficient	0.003	0.020	0.000	1.000
GDP per capita	0.077	0.091	0.000	1.000
Fiscal revenue per capita	0.044	0.071	0.000	1.000
Investment in fixed assets	0.058	0.080	0.000	1.000

Notes: Observations are 5318; All data are standardized.

Tabla A3.

Unit-root and cointegration test

Panel A: Unit-root test	Im-Pesaran-Shin unit-root test	Hadri LM test
	H0: All panels contain unit roots	H0: All panels are stationary
	p-value of Z-t-tilde-bar	p-value of z
QGI	0.474	0.000
TS	1.000	0.000
Panel B: Cointegration test	Kao test	Westerlund test
	H0: No cointegration	H0: No cointegration
	p-value of Augmented Dickey-Fuller t	p-value of Variance ratio
QGI and TS	0.000	0.000