The Impact of Population Aging on Green Innovation: An Empirical Analysis Based on Inter-Provincial Data in China

Abstract

China’s green innovation and green transformation is facing the uncertain challenge of an aging population. Based on provincial panel data from 2006 to 2019 in 30 provinces of China, this paper uses the threshold regression approach to test the green innovation effect of population aging. The following important conclusions are obtained through empirical analysis. First, population aging has a significant inhibiting effect on green innovation. This inhibition has shown an overall downward trend. Second, the green innovation effect of population aging has the characteristic of regional heterogeneity. The negative impact of population aging on green innovation in the western region shows a “U” shaped distribution with the deepening of population aging. The inflection point value of the “U” shape is 18.1%. The inhibitory effect of population aging on green innovation in the central region is higher than that in the eastern and western regions, with the degree of inhibitory effect reaching over 41%. The inhibitory effect of population aging on green innovation in the eastern region has obvious marginal decreasing characteristics. Third, the moderating mechanisms of different regional conditions are different. The green innovation effect of population aging is positive when the level of urbanization exceeds 70% and trade openness exceeds 1.1547. Green finance is also a positive moderator. However, population aging can negatively affect green innovation through environmental regulation and human capital channels. Thus, China should adopt supporting measures for green innovation market cultivation and green industry development, and enhance its green innovation capabilities through channels such as trade opening, urbanization and green finance.

1. Introduction

In recent decades, human development has led to increasingly serious environmental problems and instability in political, economic and social development. These problems continue despite the fact that the concept of sustainable development has been proposed for 30 years [1]. Scholars such as Sachs [2] realized that the right combination of economy and technology allows humanity to envision a world with sustainable economic growth [3]. Classical economic growth theories do not include the role of technology and innovation in development in their model analysis, and endogenous growth theories treat R&D investment and human capital as key to economic development [4,5]. However, a limitation of economic theory’s understanding of innovation is that it does not take into account the ecological aspects of innovation, without which it is difficult to consider sustainability [6,7]. Thus, the concept of green innovation, which combines “innovation” and “green”, has become the focus of scholarly attention.

Green innovation is considered an important means to solve environmental problems and achieve environmental sustainability, and is an important component of high-quality development [8]. From the trend of world economic growth, green technology innovation is playing an important role in the new global industrial revolution and technological competition [9].

In particular, for developing countries such as China, facing the daunting challenge of crossing the “middle-income trap”, there is a need to accelerate the process of green innovation in order to take the initiative in the new technological revolution. Thus, the adoption and development of green innovation has become a central part of the policy discussion on China’s economic development and planning [10,11]. The Guiding Opinions on Building a Market-Oriented Green Technology Innovation System issued by Chinese governmental departments has made international green technology innovation a priority for intensity.

However, China’s green innovation and green transformation are facing the uncertain challenge of an aging population. According to UN standards, since 2000 China has officially entered an aging society. China’s population is aging too early relative to its stage of economic development [12]. Even though China has implemented policies such as “separate two-child”, “comprehensive two-child”, and liberalization of “third child” policies, which have contributed to the increase in the fertility rate in the short term, this cannot fundamentally solve the problem of low fertility desire among young people. The seventh census of China in 2021 shows that there are 260 million people aged 60 and above, with a proportion of 18.70%, of which 190 million people are aged 65 and above, with a proportion of 13.50%. Population aging not only causes demographic changes, but also has a particularly important impact on economic–social and technological changes [13]. Currently, China’s “demographic dividend” is disappearing and the economic model must be reformed in order to successfully achieve a green upgrade of the industry, but the rapid aging poses a risk of uncertainty.

Unfortunately, however, the paucity of the existing literature on the direct correlation between population aging and green innovation, coupled with the complexity of the impact of population aging on green innovation, poses difficulties for understanding their relationship. The uncertainty of the impact of population aging on green innovation is manifested in the fact that population aging may have two opposite effects. From the supply side, population aging brings about a decline in the supply of labor resources, which in turn affects the accumulation of physical capital and the enhancement of human capital, leading to a lack of vitality in green innovation. However, the impact of population aging will also strengthen the government’s determination to reform the economy, increase investment in health and human capital, and increase the cultivation of human resources, which will “push” the industrial structure to upgrade and improve the green innovation capacity. From the demand side, rapid aging also means an increase in the burden of the working population, reducing the consumption capacity of the working population, which is not conducive to the promotion of green innovation products and the development of green innovation enterprises, while at the same time, population aging will also increase the demand for investment in human resources, which is beneficial to green innovation. The impact of population aging uncertainty undoubtedly brings challenges to our precise understanding of the relationship between population aging and green innovation. The empirical question of whether population aging affects innovation and through what channels or mechanisms it acts is not fully understood [14]. Additionally, how population aging affects green innovation is rarely addressed in academic studies.

So, what exactly is the impact of population aging on green innovation? This paper attempts to provide an empirical response to this issue in conjunction with a panel dataset of 30 Chinese provinces from 2007 to 2019. Since the degree of population aging in China has huge differences at the provincial level [15], it is difficult to clarify the relationship between population aging and green innovation at the inter-provincial level. The relationship between population aging and green innovation is difficult to dissect from a linear perspective because of the differences in the process of population aging across provinces.

The novel contribution of this paper is to empirically test the relationship between population aging and green innovation from a nonlinear perspective based on a Chinese inter-provincial level panel dataset. In particular, this paper analyzes the regional heterogeneity of the nonlinear relationship between population aging and green innovation, thus providing an empirical basis for the government to formulate targeted policies. More importantly, this paper also examines the moderating effects of urbanization, trade openness, green finance, environmental regulation, human capital and marketization on the relationship between population aging and green innovation, bridging the research gap.

The remainder of this paper is structured as follows. Section 2 provides a literature review. Section 3 discusses the methodology and data. Section 4 provides the empirical results and discussion. In Section 5, we discuss the main conclusions and elaborate on their policy implications.

2. Review of the Literature

Green innovation is defined in various forms, such as conceptual innovations that achieve energy savings and emission reductions [16,17] or green products or process-related innovations [18,19,20,21]. Thus, green technology innovations, including new products, processes, services, and methods, are expected to avoid environmental pollution problems [22,23,24]. Unlike the traditional concept of innovation, green innovation is a combination of “green” and “innovation”, emphasizing the greening of the innovation process and the green output of innovation. In recent decades, there has been a trend of “greening” innovation research. A large number of studies on topics related to green innovation have emerged in the literature, such as new R&D investments, eco-innovation, low-carbon technology innovation, and energy technology innovation.

The literature on the direct association between population aging and green innovation is scarce; however, we can gain insights from studies related to the relationship between aging and technological innovation. The literature is divided into three schools of thought regarding population aging and innovation.

The first school of thought argues that population aging promotes technological innovation. Some scholars argue that the labor scarcity associated with an aging population will instead stimulate technological innovation. For example, studies by Acemoglu [25] and Acemoglu and Restrepo [26] suggest that labor scarcity encourages technological progress if the technology involved is labor-saving. The study by Acemoglu and Restrepo [27] describes how labor scarcity among middle-aged people and young adults triggers robotics (and other automation technologies) adoption and found that, despite a reduction in labor input, this subsequently increased aggregate output. Madsen and Damania [28] argue that firms responding to labor shortages and rising labor costs associated with an aging population increase technological innovation and the adoption of new equipment to become more competitive. Antonelli and Quatraro [29] argue that there is substitutability between capital and technology and labor factors, and when the aging population causes labor shortage, firms adopt technology and capital instead of labor to enhance their market competitiveness. Due to labor scarcity, the model of relying on cheap labor for corporate profits is no longer sustainable, which will force firms to enhance human capital training and capital flow to the R&D input sector (Cutler et al., 1990 [30]) to enhance the innovation capacity and competitiveness of firms. The most recent study conducted by Tan et al. [31] analyzes the relationship between population aging and firm innovation using data from Chinese listed companies from 2004 to 2017, and finds a positive relationship between population aging and firm innovation in China. This facilitative effect is greater in firms with higher labor costs. It has also been argued that older people have certain advantages over younger people in technological innovation. Older employees accumulate more experience, skills and knowledge [32,33,34] and are no less innovative than younger people (Gordo and Skirbekk [35]), thus being more conducive to innovation [36,37]. In addition, an aging population leads to a longer human life expectancy, which will increase the propensity to save, forcing down the interest rate on loans, thus reducing the cost of capital for firms, increasing profits and leaving society with more resources for technological innovation (Bloom et al. [38]).

The second school of thought argues that the aging population inhibits technological innovation. According to the endogenous growth theory, the scarcity of labor and high factor prices due to population aging inhibit technological progress and economic growth. For example, a study by Maestas et al. [39] finds that population aging leads to a reduction in the growth rate of GDP per capita, with one-third of this reduction arising from slower growth of the labor force and two-thirds due to slower growth in the labor productivity of workers across the age distribution. Labor scarcity due to an aging population can also bring about labor market rigidity, which can further discourage firm innovation [40,41,42]. Noda [43] and Gonzales and Niepelt [44] argue that population aging increases the fiscal expenditures of the state for pension security, which has a crowding out effect on technological innovation. An objective disadvantage is that, relative to the young and middle-aged working population, the aging population is less energetic and dynamic in terms of innovation, has a lower learning capacity, and the elderly are less able to absorb new technologies and knowledge, which is not conducive to knowledge dissemination and skills training in firms, and thus is not conducive to the formation of human capital and has a negative impact on innovation [45,46,47]. According to relevant studies in psychology, the cognitive, reasoning, memory, and computational speed abilities of individuals decline significantly after the age of 50 (Nusbaum and Silvia [48]). The peak of innovation ability of Nobel laureates and famous scientists is between 30 and 40 years old, and declines thereafter (Jones [49]). Compared to younger employees, older employees have a rewarding period to innovate, have proven their abilities at work, have less passion and motivation to innovate (Kanfer and Ackerman [50]), and have a lower propensity to adopt new technologies (Lancia and Prarolo [51]). Because the relatively poor learning status of older adults does not match core technology positions (Friedberg [52]), the “learning by doing” theory is not effective in practice in an aging population.

The third school of thought considers the relationship between population aging and innovation to be uncertain. Scholars who hold this view argue that population aging and innovation are not simply linear in a positive or negative direction, but vary with stage and condition. Thus, some scholars argue that the relationship between aging and innovation is nonlinear. For example, Frosch [53] and Jones [49] argue that the effect of aging on innovation may be in the form of an “inverted U”, where population aging first promotes and then inhibits innovation. Some scholars argue that the impact of population aging on innovation is the result of a combination of its positive and negative effects. For example, Xie et al. [54] divided population aging into three stages and argued that the “innovation effect” of population aging would show different characteristics in different periods of population aging. In the early stage of population aging, the negative effect is not yet obvious and the positive effect dominates, so the technological innovation effect of population aging highlights positive characteristics, and this positive effect is increasing as society attaches importance to human capital and technological innovation; in the middle stage of aging, the negative effect continues to be prominent, the positive effect gradually slows down, the power gap between the two is narrowing, and the promotion role of population aging on technological innovation tends to decline. In the later stage of aging, the negative effect dominates, and the technological innovation effect of population aging highlights the negative characteristics.

In conclusion, there is a gap in the literature on population aging and green innovation, and the relevant research focuses on the relationship between population aging and innovation. However, the relationship between population aging and innovation is also debated. We note that a growing number of scholars argue that there are two opposing effects of population aging on innovation. With these two effects waxing and waning, population aging and green innovation are not simply linearly related but vary with different stages or regional conditions. Due to the objective existence of inter-provincial differences in China, the relationship between population aging and green innovation needs to be dissected from a non-linear perspective; however, unfortunately, the literature in this area is inadequate. Thus, this paper dissects the relationship between population aging and green innovation from a nonlinear perspective based on inter-provincial data, which fills the gap of existing studies.

3. Methodology and Data

3.1. Empirical Models

The panel threshold model is an econometric model that can be used as a nonlinear estimator to construct a piecewise function of the regression coefficients of the explanatory variable by estimating thresholds, i.e., analyzing the effects of the explanatory variable on the explanatory variable under different thresholds. Specifically, the threshold is estimated endogenously as an unknown variable to be estimated, and the parameters of the threshold interval are estimated (Yang et al. [55]). Based on Hansen [56], a panel threshold model is developed in this paper to test the threshold effect of population aging on green innovation. The specific econometric model is as follows:

(1)$gi{e}_{it}={\alpha }_0+{\alpha }_{1}ol{d}_{it}\cdot I(tr{s}_{it}\le \gamma )+{\alpha }_{2}ol{d}_{it}\cdot I(tr{s}_{it}>\gamma )+\dots +{\alpha }_{n}ol{d}_{it}\cdot I(tr{s}_{it}\le {\gamma }_n)+\theta x_{it}+{\mu }_i+{\epsilon }_{it}$

where ${\mu }_i$ represents the individual effect of the sample cross-section that does not vary with time, ${\epsilon }_{it}$ denotes the error term, and ${\epsilon }_{it}~iid(0,{\sigma }^2)$. ${\alpha }_0$ is the intercept term, and ${\alpha }_{1,}{\alpha }_{2,}\dots {\alpha }_n$ are the regression coefficients to be estimated. ${gie}_{it}$ is the green innovation efficiency in region i at period $t$, $ol{d}_{it}$ denotes the degree of population aging, and $tr{s}_{it}$ denotes the threshold variable, which can be used either as a regressor of the explanatory variable or as an independent threshold variable (Hansen [57]). The threshold variables in this paper are estimated either as control variables (denoted by $x_{it}$) or as threshold variables for conditional analysis, provided that they all satisfy the independence assumption. $I(·)$ is the indicator function, defined as 1 if the expression in parentheses is true and 0 otherwise. $\gamma$ is the threshold to be estimated, and when a threshold exists the effects of $ol{d}_{it}$ on green innovation can be divided into two threshold intervals, i.e., $old\_1$ and $old\_2$; when two thresholds exist, the effects of $ol{d}_{it}$ on green innovation can be divided into three threshold intervals, i.e., i1, i2 and i3; and when three thresholds exist, the effects of ${old}_{it}$ on green innovation can be divided into four threshold intervals, i.e., $old\_1$, $old\_2$, $old\_3$ and $old\_4$.

For the estimates of $\gamma$, confidence interval analysis was performed using the LR (Likelihood Ratios) statistic with the following equation.

(2)$L{R}_n(\gamma )=n\frac{S_n(\gamma )-S_n(\stackrel{⌢}{\gamma })}{S_n(\stackrel{⌢}{\gamma })}$

The null hypothesis was accepted when $L{R}_n(\gamma )\le c(\alpha )=-2\ln (1-\sqrt{\alpha })$, where $\alpha$ represents the significance level.

For any $\gamma$, the corresponding residual sum of squares is $S_1(\gamma )$. If chosen sequentially from maximum to minimum, it happens that the minimum $S_1(\gamma )$ corresponds to the threshold ${\gamma }^{*}$, which can be expressed as:

(3)${\gamma }^{*}=\arg \min S_1(\gamma )$

When the estimated threshold is obtained, further testing is required to test the reasonableness of the threshold model.

The first is to test the significance of the panel threshold effect. The null hypothesis is $H_0:{\alpha }_1={\alpha }_2$ and the alternative hypothesis is $H_1:{\alpha }_1\ne {\alpha }_2$. The corresponding statistic used as a test can be expressed as:

(4)$F=(S_0(\gamma )-S_1(\widehat{\gamma }))/\widehat{{\sigma }^2}$

In Equation (4), the residual sums of squares $S_0(\gamma )$ and $S_1(\widehat{\gamma })$ are obtained by the parameter estimates in the H₀ and H₁ conditions, respectively, while the residual variance estimated in the H₁ condition is ${\sigma }^2$. The asymptotic distribution of the general F-statistic can be obtained by the self-sampling method to test whether it is significant.

Then, the consistency of the estimates of the thresholds with the true values is tested. The corresponding LR statistic is constructed for the test:

(5)$LR=(S_1(\gamma )-S_1(\widehat{\gamma }))/\widehat{{\sigma }^2}$

The asymptotic distribution of the LR statistic satisfies $c(\alpha )=-2\ln (1-\sqrt{\alpha })$, and the null hypothesis is rejected when $L{R}_1\le c(\alpha )$.

3.2. Sample Selection

In this paper, data from 30 provinces in mainland China from 2006 to 2019 were selected as the sample for the study. Since the data of Tibet, Hong Kong, Macao and Taiwan are missing, they were not included in the sample selection of this paper. The provinces and regions covered by the sample are shown in Figure 1. The main statistics are obtained from the China Statistical Yearbook, China Science and Technology Statistical Yearbook, China Insurance Yearbook, the State Intellectual Property Office and statistical yearbooks of each province.

3.3. Variables

3.3.1. Dependent Variable

Green innovation efficiency (gie) reflects the input–output efficiency in the process of green technology innovation under the conditional constraints of resources and environment [58,59,60,61]. Unlike technological innovation efficiency in the traditional sense, green innovation efficiency reflects the environmental effects of the input–output ratio of green innovation activities [61,62] and represents the degree of greenness of technological innovation efficiency (Li and Du [63]). In the measurement, data envelopment analysis (DEA) [64,65] and stochastic frontier analysis (SFA) [66,67] are two approaches that have been commonly used to measure the efficiency of green innovation. Since DEA models have the drawback of being more sensitive to measurement errors or other noise compared to SFA models (Jacobs [68]), this paper uses the SFA approach to measure green innovation efficiency. For green innovation inputs, this paper follows Qian et al. [69], Song and Han [70] and others, and uses full-time equivalents of R&D personnel as labor input and R&D expenditure as capital input. For green innovation output, this paper selects green innovation patent grants to measure green innovation output considering that green patents are generally used to measure green technology innovation capability [71,72]. Specifically, this paper determines the number of patents based on the correspondence between IPC classification and green technology areas in the “Comparison of Technology Areas and IPC Classification List” published by OECD. There are six types of green technology areas, namely water pollution prevention technology, soil pollution prevention technology, air pollution prevention technology, water saving technology, energy saving technology and recycling technology.

3.3.2. Measuring Population Aging

The UN classification of aging is based on the proportion of the elderly population to the total population, i.e., the proportion of the population aged 60 and over exceeds 10%, or the proportion of the population aged 65 and over exceeds 7%, which means an aging society. This paper follows Tan et al. [31] and uses the percentage of the population aged 65 years and older in the total population to measure the degree of population aging (aging).

3.3.3. Conditioning Variables

To reflect the impact of changes in regional economic conditions on green innovation, this paper selects urbanization, green finance, trade openness, environmental regulation, human capital and marketization variables as threshold variables to examine the moderating effect of regional economic conditions on the relationship between population aging and green innovation from different aspects. The first threshold variable is the level of urbanization (uban). Although aging will bring a shortage in the labor force, the development of urbanization can, to a certain extent, absorb and make full use of the remaining labor resources and take advantage of the scale of the labor force, thus serving as a moderating effect on the aging effect. Therefore, this paper takes urbanization as one of the threshold conditions to regulate the innovation effect of aging. Following Shuai and Fan [73], the level of urbanization is expressed as the proportion of urban population to total population. The second threshold variable is the level of green finance (grf). Green innovation requires a large amount of talent and capital investment, and the development of green finance can provide financial support for green innovation. Following the approach of Song and Han [74], four aspects, i.e., indicators of green investment, green credit, green insurance and government support, are selected to construct a comprehensive evaluation system of green finance, and the entropy value method is used to evaluate the level of green finance. The third threshold variable is the intensity of environmental regulation (hgui). To reduce the negative effects of aging on green innovation, the government needs to implement environmental regulation policies to promote the development of green innovation. Following Sonia Ben [75], the ratio of regional GDP to regional total energy consumption is used to reflect the intensity of environmental regulation. A higher value of this indicator implies lower energy consumption used per unit of output. The fourth threshold variable is the level of human capital (edu). According to endogenous growth theory, an important channel through which aging affects innovation is the accumulation of human capital. Drawing on Zhao and Zhang [76], the average number of years of education in each region is used as a measure of the level of human capital. The fifth threshold variable is the degree of trade openness (trade). The higher the degree of trade openness, the greater the likelihood that a region will acquire advanced green technology through trade. Drawing on Han et al. [77], the ratio of total import and export trade to GDP is used to express the degree of trade openness. The sixth threshold variable is the level of marketization (market). The development of green innovation requires mature market conditions. In this paper, we use the NERI index of marketization based on Fan et al. [78] to measure the level of marketization.

The above threshold variables can also be used as control variables. The choice of control variables is based on the exclusion of endogeneity, covariance, and other measurement problems. In addition to this, we consider the level of total tax burden (tax) as a control variable. For the total tax burden level, we use the share of total tax revenue to GDP to express.

Descriptive statistics for related variables can be found in Table 1. The minimum value of the degree of population aging is 7.40%, which exceeds the UN standard of 7%, indicating that all Chinese provinces have entered the stage of an aging society during the sample period. The mean value of the population aging degree is 13.3% and the maximum value is 22.7%, indicating that China has entered the stage of deep aging.

Figure 2 reveals the degree of population aging in each province. There are significant differences in the degree of population aging among provinces, and most of the provinces have a tendency to deepen the degree of population aging.

4. Empirical Results and Discussion

4.1. Model Tests

Multiple collinear testing was carried out with green innovation performance as the dependent variable, and the test results are shown in Table 2. The result of the VIF statistics shows that the largest VIF value among the variables is uban (VIF = 7.68), the smallest is old (VIF = 1.5), and the average VIF value is 4.39, which is much less than 10, so the model does not have a serious co-linearity problem.

In addition, we consider the possible interrelationship between human capital variables, urbanization variables, and population aging and, thus, they are treated as threshold variables rather than control variables in the later estimation, so as to avoid endogeneity problems as much as possible. Even so, in order to make the findings more objective and rigorous, endogeneity tests are further conducted to exclude possible endogeneity problems that may cause disturbances to the estimation results. Since the variable lags can satisfy both the condition of being highly correlated with the explanatory variable and exogeneity, we use the lags of population aging as the instrumental variables for the endogeneity tests. Specifically, the existence of the endogeneity problem is tested following the method of Hausman [79]. As shown in Table 3, the chi2(9) value of the test is 4.76 and the Prob > chi2 value is 0.8545, indicating that the model satisfies the condition that all explanatory variables are exogenous. In the case of considering heteroskedasticity, the Durbin (score) chi2(1) value of the test is 1.003 (p = 0.317) and the Wu–Hausman F(1,320) is 0.976 (p = 0.3240), which also excludes the endogeneity problem. Thus, the case of interference from endogeneity problems can be excluded.

4.2. Threshold Effects of Population Aging on Green innovation

The Bootstrap method proposed by Hansen [57] is used with 2000 overlapping simulation tests to obtain the F-statistic, p-value and threshold, as shown in Table 4. It can be found that the single threshold, double threshold and triple threshold models are all significant at the 1% significance level when population aging is used as the threshold variable, under both cases of considering control variables and not considering control variables, indicating that there are three thresholds for the effect of population aging on green innovation, and thus it is appropriate to adopt the triple threshold model for threshold effect analysis. On the basis of the three thresholds, the estimated coefficients of the four threshold intervals are obtained recursively. These four threshold intervals are denoted as old_1, old_2, old_3, and old_4, in that order.

To obtain robustness results, we further performed robustness tests. Specifically, the following two robustness tests were adopted: first, changing the time period of the sample and using 2007–2016 as the sample period for the threshold test; second, using the lagged period of old as the core explanatory variable for the threshold model test. The results of both robustness tests indicate that there are three thresholds for the effect of population aging on green innovation, further illustrating the plausibility of using a three-threshold model. By comparing the estimated thresholds, it is not difficult to find that the thresholds for the three scenarios do not differ significantly, except for the case of no control variables. The threshold estimates are different because it is difficult to avoid the estimation error caused by omitted variables in the case of no control variables. Considered together, the estimation using the three-threshold model is reliable and the threshold estimates are robust.

The regression results of the triple threshold model for the effect of population aging on green innovation are reported in Table 5. From model 1 to model 8, we use stepwise regressions to try to avoid estimation errors due to multicollinearity and endogeneity problems. Among them, model 7 is the result of the triple threshold estimation considering the one-period lag of old. Although the estimated coefficients are slightly different, the direction of the effect of the variables on green innovation is consistent and, thus, the test results are robust.

The three thresholds of population aging tested for model 8 are 13.7%, 15.9% and 17.5%, respectively; thus, the green innovation effect of population aging can be divided into the following four threshold intervals: old_1 (the value of old is less than or equal to 13.7%), old_2 (the value of old is greater than 13.7% and less than or equal to 15.9%), old_3 (the value of old is greater than 15.9% and less than or equal to 17.5%), and old_4 (the value of old is greater than 17.5%).

In the estimation results of models 1 to 8, the regression coefficients of population aging on green innovation are all significantly shown to be negative, indicating that population aging has a significant inhibitory effect on green innovation during the sample period. The empirical finding also indicates that this negative effect of population aging decreases with the increase of the threshold, suggesting that as the population aging deepens, the more it will motivate the country to pay attention to adopt green innovation, thus continuously weakening the negative effects of population aging, and thus continuously weakening the negative effects of population aging. Thus, the finding supports the view of Xie et al. [54], that the effect of population aging on green innovation is the result of two opposing forces acting together, and that China is currently in the late stage of population aging, where the technological innovation effect of population aging highlights the negative characteristics. For the effects of the control variables, the direction of the impact coefficients is consistent across the models and the coefficients do not differ significantly, indicating the robustness of the test results. Specifically, green finance, trade liberalization and marketization are positive factors that promote green innovation; environmental regulation and tax burden are negative factors that weaken green innovation.

4.3. Regional Heterogeneity Effects

There are large differences between different regions in China (Clark-Sather et al. [80]), and provincial-level data are usually used to reflect the differences between regions. Heterogeneity effects are analyzed according to the traditional regional division of geospatial China into three regions: east, central, and west.

Based on the Bootstrap method proposed by Hansen [57], overlapping simulation tests were conducted 2000 times and three thresholds for the effect of aging on green innovation were identified in each sample. Thus, the green innovation effect of population aging under each regional sample can be divided into four threshold intervals. Among them, the threshold intervals obtained for the eastern regional sample are: old_1 (the value of old is less than or equal to 16.6%), old_2 (the value of old is greater than 16.6% and less than or equal to 18.0%), old_3 (the value of old is greater than 18.0% and less than or equal to 18.8%), and old_4 (the value of old is greater than 18.8%); the four threshold intervals obtained for the central regional sample are: old_1 (the value of old is less than or equal to 11.1%), old_2 (the value of old is greater than 11.1% and less than or equal to 11.9%), old_3 (the value of old is greater than 11.9% and less than or equal to 13.45%), and old_4 (the value of old is greater than 13.45%); the four threshold intervals obtained for the western region sample are: old_1 (the value of old is less than or equal to 12.4%), old_2 (the value of old is greater than 12.4% and less than or equal to 13.7%), old_3 (the value of old is greater than 13.7% and less than or equal to 18.1%), old_4 (the value of old is greater than 18.1%).

Based on the model estimation, we also conducted two robustness tests, and the results of the tests are shown in Table 6. The results of the robustness tests show that although the impact coefficients differ, the direction of impact is consistent, indicating the robustness of the study findings. In Table 6, population aging has a significant negative effect on green innovation under all three regional samples, further revealing that population aging has an inhibitory effect on green innovation. The spatial distribution of the green innovation effect of population aging shows the following characteristics: First, the negative impact of population aging on green innovation in the western region shows a “U” shaped distribution with the deepening of population aging. The inflection point value is 18.1%, which exceeds the national average population aging level during the sample period. Most of the provinces in the western region have an aging population within 18.1%, which means that the policies adopted by most of the provinces in the western region to address the aging population are effective. Before the inflection point, the inhibitory effect of population aging on green innovation decreases because the labor scarcity and cost increase brought by population aging at a certain stage stimulate regional innovation, thus offsetting the negative effect of population aging to some extent. However, when the degree of population aging exceeds the inflection point, the inhibitory effect of population aging on green innovation becomes increasingly prominent. Second, the inhibitory effect of population aging on green innovation in the central region is the largest among the three regions, with a degree of inhibitory effect above 41%. Thirdly, the inhibitory effect of population aging on green innovation in the eastern region has obvious marginal decreasing characteristics, and the decreasing magnitude of the inhibitory effect is the largest among the three regions. The eastern region is more mature in hedging the negative impact effect of aging because of its higher level of economic development, and thus can absorb the talent resources from the central and western regions.

4.4. Moderating Role of Regional Economic Conditions

In the above study, we found that population aging has a significant inhibitory effect on green innovation. Then, we examine the moderating effect of different regional economic conditions on the green innovation effect of population aging in order to find an effective moderating mechanism. The test results of the model are shown in Table 7.

Through the threshold test, it is found that there are three thresholds for all the selected moderating variables; thus, there are four conditional moderating intervals, namely Condition_Ⅰ, Condition_Ⅱ, Condition_Ⅲ and Condition_Ⅳ, and the corresponding four threshold intervals for the green innovation effect of population aging are old_1, old_2, old _3 and old_4.

Several important conclusions can be drawn, as follows. First, urbanization has a “U”-like dynamic moderating mechanism. The inflection point of the urbanization level is 70%. This suggests that the positive moderating mechanism of urbanization can work only after the 70% inflection point of the urbanization level, and only a few provinces in China can meet this condition. Second, the green innovation effect of population aging is positive when the level of trade openness exceeds 1.1547, suggesting that trade openness, like urbanization, positively moderates the green innovation effect of population aging needs to exceed a certain threshold limit. Overall, the inhibitory effect of population aging on green innovation decreases marginally with increasing trade openness, indicating that higher trade openness is beneficial in hedging the negative effects of population aging. Third, the inhibitory effect of population aging on green innovation basically increases with the increase of environmental regulation intensity and human capital level, which means that population aging will negatively affect green innovation through both environmental regulation and human capital channels, and the higher the environmental regulation intensity and human capital level, the greater the negative impact of population aging on green innovation. For regions with high environmental regulation intensity, the development of industries tends to be restricted due to the priority of protecting the ecological environment, which can lead to the migration of young people to other regions and the high proportion of the remaining old population cannot support the development of local green innovation. For regions with higher human resources, the amount of human capital due to population aging decreases more rapidly, and thus the negative effect on green innovation is greater. This finding supports the view that population aging is detrimental to human capital formation and innovation [46,47]. Fourth, the negative effect of population aging on green innovation decreases as the level of green finance increases, indicating that the increase in the level of green finance is a positive moderating factor, because green finance can help solve problems such as inadequate financing in the green innovation process (Pan et al. [81]). Fifth, the negative effect of population aging on green innovation is similar to an inverted “N”-shaped variation with the change of market level, indicating that the moderating effect of the market level is characterized by unstable and repeated changes, which may be mainly due to the insufficient development of the green innovation market in China at present (Song and Han [70]). The market for green innovation in China faces problems such as low returns to innovation in the private sector (Stucki et al. [82]).

5. Conclusions and Policy Implications

Currently, China is in a high-quality development stage of green transformation, but faces profound challenges brought by population aging. Based on panel data of 30 Chinese provinces from 2006 to 2019, this paper analyzes the nonlinear effects of population aging on green innovation and its regional heterogeneity and moderating mechanisms using the threshold regression technique.

Through the empirical analysis, this paper obtains several important conclusions, as follows. First, population aging has a significant inhibitory effect on green innovation, i.e., population aging has become an important factor limiting the development of green innovation in China. The study also shows that the inhibitory effect of population aging on green innovation is generally decreasing. Second, the green innovation effect of population aging is characterized by regional heterogeneity. The negative impact of population aging on green innovation in the western region shows a “U” shaped distribution with the deepening of population aging. The inflection point value of the “U” shape is 18.1%. The inhibitory effect of population aging on green innovation in the central region is higher than that in the eastern and western regions, with the degree of inhibitory effect reaching over 41%. The inhibitory effect of population aging on green innovation in the eastern region has obvious marginal decreasing characteristics. Third, the moderating mechanisms of different regional conditions are different. The positive moderating mechanism of urbanization on the green innovation effect of population aging can only work after the 70% inflection point of urbanization level, and only a few provinces in China can meet this condition. The green innovation effect of population aging is positive when the trade openness level exceeds 1.1547. The inhibitory effect of population aging on green innovation basically increases with the intensity of environmental regulation and the level of human capital, implying that population aging negatively affects green innovation through both environmental regulation and human capital channels. In addition, the increase in the level of green finance is a positive moderating factor, while the increase in the level of marketization does not play an effective moderating role due to the underdevelopment of the green innovation market in China.

The findings of this paper support the view that population aging has an inhibitory effect on green innovation, and the findings foreshadow that China lacks an effective mechanism to cope with the aging population and urgently needs to adopt corresponding policy measures to hedge against the negative effects of population aging.

The following policy implications can be drawn from the empirical findings of this paper. First, the government should increase the fiscal expenditure on green R&D investment and adopt policies such as tax incentives to stimulate enterprises to green innovation. Second, the knowledge, skills and experience of the older population should be valued, and the innovation capacity of the younger population should be promoted by “bringing the old with the new”. Third, attention should be paid to the cultivation of the green innovation market. The key to the underdevelopment of the green innovation market lies in the insufficient return of private enterprises. Therefore, support for small and medium-sized enterprises should be increased to promote the development of the green innovation market. At the same time, attention must be paid to the involvement and important role of energy companies in creating and promoting green innovation, which is important for energy use efficiency improvement and carbon reduction. Fourth, it should provide talents, capital and other resources for green innovation through urbanization, enhance the international competitiveness of cities, and realize the benign interaction between urbanization and green innovation. Fifth, international advanced green technology, capital and talents should be absorbed through trade opening, and the upgrading of China’s green innovation industry should be promoted through trade opening. Sixth, green finance should be developed to provide financing for corporate green innovation. Seventh, targeted policies suitable for regional differences should be implemented. For the eastern region, the comparative advantages of regional development should be brought into play, regulatory measures should be improved, and the introduction of international green innovation talent resources should be increased to improve international competitiveness. For the central region, the transformation of green industries should be realized as early as possible. For the western region, it should give full play to the advantages of the late-developing regions, seize the opportunity of “One Belt and One Road” and realize the green industrial upgrading.

Although this paper analyzes the impact of population aging on green innovation in China by combining the threshold regression techniques, it still has some limitations, which may also inform the research direction that needs to be expanded in the future. First, due to insufficient data, Tibet, Hong Kong, Macau and Taiwan are not included in the panel dataset of this paper, thus subsequent extension studies are needed. Second, this paper does not focus on the mediating mechanism affecting population aging and green innovation, which requires theoretical and empirical expansion.

Author Contributions

Y.L.: Resources, Data curation, Methodology, Writing—Original draft preparation. M.J.: Conceptualization, Writing—Reviewing and Editing. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. Provinces covered by the sample.

Figure 2. Population aging in each province.

Table 1

Descriptive statistics and definitions of variables.

Variable	Definition	Mean	SD	Min	Max
gie	Green innovation efficiency measured by SFA approach	0.371	0.190	0.116	0.942
old	The percentage of the population aged 65 years and older in the total population	0.133	0.279	0.074	0.227
uban	The proportion of urban population to total population	0.541	0.136	0.275	0.896
grf	The green finance index estimated using the entropy value method	0.155	0.094	0.050	0.759
trade	The ratio of total import and export trade to GDP	0.311	0.367	0.017	1.765
hgui	The ratio of regional GDP to regional total energy consumption	1.270	0.607	0.257	3.928
taxl	The share of total tax revenue to GDP	0.080	0.030	0.041	0.200
market	The NERI index of marketization	6.511	1.916	2.330	11.710
edu	The average number of years of education	8.845	0.987	6.594	12.555

Table 2

Multi-collinearity test.

Variable	VIF	1/VIF
uban	7.68	0.130
grf	5.48	0.182
edu	5.36	0.187
hgui	4.29	0.233
trade	4.22	0.237
market	3.9	0.256
taxl	2.68	0.374
old	1.5	0.667
MeanVIF		4.39

Table 3

Comparing the OLS and 2SLS estimates.

Coefficients
	(b)	(B)	(b − B)	sqrt(diag(V_b − V_B))
Variables	iv	ols	Difference	S.E.
old	0.002	0.000	0.001	0.002
uban	−0.538	−0.512	−0.027	0.064
grf	−0.622	−0.688	0.066	0.048
trade	0.016	0.018	−0.001	0.023
hgui	0.091	0.104	−0.013	0.012
taxl	1.800	1.767	0.033	0.187
market	0.007	0.005	0.002	0.005
edu	−0.053	−0.051	−0.002	0.007
_cons	0.890	0.886	0.004	0.045
chi2(9)	4.76
Prob > chi2	0.8545
Durbin (score) chi2(1)	1.003 (p = 0.3165)
Wu-Hausman F(1,320)	0.976 (p = 0.3240)

Table 4

Results of the threshold model test.

Types of Threshold Model Tests	Threshold Quantity	F Value	p Value	Threshold Value (%)
No Control Variables	Single Threshold	−358.000 ***	0.000	12.200
	Double Threshold	2.466 ***	0.000	18.000
	Triple Threshold	3.2599 ***	0.000	18.700
Control Variables	Single Threshold	−351.000 ***	0.000	13.700
	Double Threshold	8.804 ***	0.000	15.900
	Triple Threshold	5.683 ***	0.000	17.500
Change the sample time period	Single Threshold	8.813 ***	0.000	13.700
	Double Threshold	11.549 ***	0.000	15.900
	Triple Threshold	8.169 ***	0.000	18.000
old lagged one period	Single Threshold	−351.000 ***	0.000	13.400
	Double Threshold	13.602 ***	0.000	15.900
	Triple Threshold	10.688 ***	0.000	18.300

Note: *** indicates significant level of 1%.

Table 5

Regression results of the threshold model.

Variable: old Is the Threshold Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
grf							0.0682 ** (2.050)	0.0731 ** (2.220)
trade						0.0266 *** (4.722)	0.0327 *** (5.151)	0.0338 *** (5.378)
hgui			−0.0327 *** (−15.435)	−0.0275 *** (−11.745)	−0.0284 *** (−12.063)	−0.0198 *** (−6.83)	−0.0262 *** (−6.188)	−0.025 *** (−5.963)
taxl							−0.1651 *** (−4.542)	−0.1683 *** (−3.040)
market					0.0015 ** (2.433)	0.0008 (1.347)	0.0006 (1.038)	0.0007 (1.088)
old_1	−0.013 *** (−7.403)	−0.0038 *** (−5.160)	−0.0042 *** (−6.472)	−0.0041 *** (−6.355)	−0.0043 *** (−6.673)	−0.0036 *** (−5.726)	−0.0036 *** (−5.684)	−0.0039 *** (−6.172)
old_2	−0.012 *** (−7.894)	−0.0034 *** (−5.342)	−0.0036 *** (−6.236)	−0.0035 *** (−6.255)	−0.0037 *** (−6.627)	−0.0032 *** (−5.757)	−0.0031 *** (−5.747)	−0.0034 *** (−6.275)
old_3	−0.013 *** (−10.539)	−0.0025 *** (−3.746)	−0.0032 *** (−6.253)	−0.0032 *** (−6.286)	−0.0034 *** (−6.644)	−0.0028 *** (−5.647)	−0.0028 *** (−5.630)	−0.0031 *** (−6.193)
old_4	−0.012 *** (−12.105)	−0.0031 *** (−5.230)	−0.0025 *** (−5.339)	−0.0025 *** (−5.425)	−0.0027 *** (−5.839)	−0.0024 *** (−5.266)	−0.0024 *** (−5.219)	−0.0027 *** (−5.838)

Note: *** and ** indicate that the coefficients of each variable pass the 1%, and 5% significance levels, respectively. The value of the t-statistic after correction for heteroskedasticity is shown in ().

Table 6

Regional heterogeneity effects of population aging on green innovation.

Variable: old Is the Threshold Variable	Model Estimation Results			Robustness Tests
				Change the Sample Time Period			old Lagged One Period
	Eastern Region	Central Region	Western Region	Eastern Region	Central Region	Western Region	Eastern Region	Central Region	Western Region
grf	0.114 ** (2.442)	0.287 * (1.932)	0.030 (0.299)	0.124 *** (3.072)	0.552 *** (4.795)	0.046 (0.538)	0.092 *** (2.727)	0.224 * (1.946)	−0.025 (−0.343)
trade	0.021 *** (2.863)	−0.0009 (−0.050)	0.081 *** (4.5656)	0.0194 *** (3.378)	−0.0152 (−1.095)	0.059 *** (3.775)	0.010 (1.645)	−0.007 (−0.517)	0.059 *** (4.777)
hgui	−0.039 *** (−4.741)	−0.047 *** (−4.633)	0.0205 ** (2.462)	−0.042 *** (−6.350)	−0.064 *** (−7.978)	0.023 *** (2.694)	−0.040 *** (−7.155)	−0.041 *** (−4.984)	0.021 *** (3.368)
taxl	−0.10 (−1.189)	0.133 (1.162)	−0.007 (−0.086)	−0.098 (−1.541)	0.233 ** (2.459)	−0.005 (−0.062)	−0.120 * (−1.841)	0.237 (1.565)	−0.001 (−0.019)
market	0.0001 (0.1243)	0.0014 (0.936)	0.0006 (0.536)	0.0004 (0.706)	0.001 (−0.033)	0.0002 (0.202)	0.0008 (1.287)	0.0007 (0.681)	0.0007 (0.812)
old_1	−0.0049 ***	−0.0061 ***	−0.0026 **	−0.0044 ***	−0.006 ***	−0.0029 ***	−0.0041 ***	−0.0053 ***	−0.0026 ***
	(−2.652)	(−2.707)	(−2.492)	(−2.961)	(−4.417)	(−3.241)	(−4.970)	(−5.640)	(−3.261)
old_2	−0.0032 **	−0.0049 **	−0.0018 *	−0.031 ***	−0.0046 ***	−0.0023 ***	−0.0032 ***	−0.0048 ***	−0.002 ***
	(−2.462)	(−2.353)	(−1.836)	(−2.537)	(−3.878)	(−2.623)	(−5.008)	(−6.705)	(−2.571)
old_3	−0.002 *** (−4.173)	−0.005 *** (−2.708)	−0.002 *** (−2.634)	−0.0022 *** (−3.869)	−0.006 *** (−4.859)	−0.002 ** (−2.004)	−0.003 *** (−4.9849)	−0.005 *** (−6.170)	−0.001 * (−1.734)
old_4	−0.002 *** (−2.576)	0.004 * (−1.738)	−0.004 *** (−3.628)	−0.002 ** (−2.023)	−0.005 *** (−4.715)	−0.002 *** (−2.900)	−0.0016 *** (−3.307)	−0.0047 *** (−5.531)	−0.002 *** (−2.764)

Note: t values are shown in (); ***, **, and * indicate that the coefficients of each variable pass the 1%, 5%, and 10% significance levels, respectively.

Table 7

Moderating effect of regional economic conditions.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Condition_Ⅰ	The value of uban is less than or equal to 0.421	The value of grf is less than or equal to 0.103	The value of trade is less than or equal to 0.132	The value of hgui is less than or equal to 0.946	The value of edu is less than or equal to 8.054	The value of market is less than or equal to 5.720
old_1	−0.004 *** (−6.082)	−0.0021 *** (−4.768)	−0.003 *** (−6.571)	−0.001 * (−1.859)	−0.0014 *** (−3.405)	−0.002 *** (−5.214)
Condition_Ⅱ	The value of uban is greater than 0.421 and less than or equal to 0.493	The value of grf is greater than 0.103 and less than or equal to 0.148	The value of trade is greater than 0.132 and less than or equal to 0.437	The value of hgui is greater than 0.946 and less than or equal to 1.568	The value of edu is greater than 8.054 and less than or equal to 9.555	The value of market is greater than 5.720 and less than or equal to 5.940
old_2	−0.004 *** (−6.098)	−0.002 *** (−4.572)	−0.0021 *** (−4.898)	−0.0015 *** (−3.778)	−0.0010 ** (−2.417)	−0.0017 *** (−3.941)
Condition_Ⅲ	The value of uban is greater than 0.493 and less than or equal to 0.700	The value of grf is greater than 0.148 and less than or equal to 0.284	The value of trade is greater than 0.437 and less than or equal to 1.155	The value of hgui is greater than 1.568 and less than or equal to 2.883	The value of edu is greater than 9.555 and less than or equal to 10.654	The value of market is greater than 5.940 and less than or equal to 6.180
old_3	−0.005 *** (−7.650)	−0.001 *** (−3.281)	−0.001 ** (−2.043)	−0.002 *** (−4.622)	−0.002 *** (−4.455)	−0.002 *** (−5.546)
Condition_Ⅳ	The value of uban is greater than 0.700	The value of grf is greater than 0.284	The value of trade is greater than 1.155	The value of hgui is greater than 2.883	The value of edu is greater than 10.654	The value of market is greater than 6.180
old_4	0.002 *** (3.497)	−0.001 *** (−4.483)	0.0005 (0.991)	−0.002 *** (−5.666)	−0.003 *** (−6.073)	−0.002 *** (−4.585)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes

Note: ***, **, and * indicate that the coefficients of each variable pass the 1%, 5%, and 10% significance levels, respectively. The value of the t-statistic after correction for heteroskedasticity is shown in ().

References

1. Tomislav, K. The concept of sustainable development: From its beginning to the contemporary issues. Zagreb Int. Rev. Econ.; 2018; 21, pp. 67-94.

2. Sachs, J. The AGE of Sustainable Development; Columbia University Press: New York, NY, USA, 2015.

3. Manioudis, M.; Meramveliotakis, G. Broad strokes towards a grand theory in the analysis of sustainable development: A return to the classical political economy. New Polit. Econ.; 2022; 27, pp. 866-878. [DOI: https://dx.doi.org/10.1080/13563467.2022.2038114]

4. Romer, P. Increasing Returns and Long-Run Growth. J. Polit. Econ.; 1986; 94, pp. 1002-1037. [DOI: https://dx.doi.org/10.1086/261420]

5. Todaro, M.P.; Smith, S.C. Economic Development; 8th ed. Pearson Education Limited: Harlow, UK, 2003.

6. Shiva, V. Resources. The Development Dictionary: A Guide to Knowledge as Power; 2nd ed. Sachs, W. Zed Books: London, UK, New York, NY, USA, 2010; pp. 228-242.

7. Sachs, W. Environment. The Development Dictionary: A Guide to Knowledge as Power; 2nd ed. Sachs, W. Zed Books: London, UK, New York, NY, USA, 2010; pp. 24-37.

8. Fang, Z.; Razzaq, A.; Mohsin, M.; Irfan, M. Spatial spillovers and threshold effects of internet development and entrepreneurship on green innovation efficiency in China. Technol. Soc.; 2022; 68, 101844. [DOI: https://dx.doi.org/10.1016/j.techsoc.2021.101844]

9. Shen, N.; Zhou, J.J. Research on the efficiency of green innovation and the mechanism of key factors in China from the perspective of technological heterogeneity: Based on hybrid DEA and structural equation model. J. Ind. Eng. Manag.; 2018; 32, pp. 46-53.

10. Chen, Z.; Zhang, X.; Chen, F. Do carbon emission trading schemes stimulate green innovation in enterprises? Evidence from China. Technol. Forecast. Soc. Chang.; 2021; 168, 12. [DOI: https://dx.doi.org/10.1016/j.techfore.2021.120744]

11. Yuan, G.; Ye Qin, S.Y. Financial innovation, 0744, information screening and industries’ green innovation—Industry-level evidence from the OECD. Technol. Forecast. Soc. Chang.; 2021; 171, 120998. [DOI: https://dx.doi.org/10.1016/j.techfore.2021.120998]

12. England, R.S. Aging China: The Demographic Challenge to China’s Economic Prospects; Greenwood Publishing Group: Westport, CT, USA, 2005; pp. 1-10.

13. Cuaresma, J.C.; Lábaj, M.; Pružinsky, P. Prospective ageing and economic growth in Europe. J. Econ. Age; 2014; 3, pp. 50-57. [DOI: https://dx.doi.org/10.1016/j.jeoa.2014.05.003]

14. Nishimura, K.G.; Minetaki, K.; Shirai, M.; Kurokawa, F. Effects of Information Technology and Aging Work Force on Labor Demand and Technological Progress in Japanese Industries: 1980–1998; Center for International Research on the Japanese Economy: Tokyo, Japan, 2002.

15. Short, S.; Zhai, F. Looking locally at China’s one-child policy. Stud. Fam. Plan.; 1998; 29, pp. 373-387. [DOI: https://dx.doi.org/10.2307/172250]

16. James, P. The sustainability cycle: A new tool for product development and design. J. Sustain. Prod. Des.; 1997; 2, pp. 52-57.

17. García-Pozo, A.; Sánchez-Ollero, J.L.; Marchante-Lara, M. Eco-innovation and management: An empirical analysis of environmental good practices and labour productivity in the Spanish hotel industry. Innovation; 2015; 17, pp. 58-68. [DOI: https://dx.doi.org/10.1080/14479338.2015.1011057]

18. Chen, Y.S.; Lai, S.B.; Wen, C.T. The influence of green innovation performance on corporate advantage in Taiwan. J. Bus. Ethics; 2006; 67, pp. 331-339. [DOI: https://dx.doi.org/10.1007/s10551-006-9025-5]

19. Chang, C.H.; Chen, Y.S. Green organisational identity and green innovation. Manag. Decis.; 2013; 51, pp. 1056-1070. [DOI: https://dx.doi.org/10.1108/MD-09-2011-0314]

20. Saunila, M.; Ukko, J.; Rantala, T. Sustainability as a driver of green innovation investment and exploitation. J. Clean. Prod.; 2018; 179, pp. 631-641. [DOI: https://dx.doi.org/10.1016/j.jclepro.2017.11.211]

21. Hilkenmeier, F.; Fechtelpeter, C.; Decius, J. How to foster innovation in SMEs: Evidence of the effectiveness of a project-based technology transfer approach. J. Technol. Transf.; 2021; pp. 1-29. [DOI: https://dx.doi.org/10.1007/s10961-021-09913-x]

22. Porter, M.E.; van der Linde, C. Toward a new conception of the environmentcompetiveness relationship. J. Econ. Perspect.; 1995; 9, pp. 97-118. [DOI: https://dx.doi.org/10.1257/jep.9.4.97]

23. Kraus, S.; Breier, M.; Dasí-Rodríguez, S. The art of crafting a systematic literature review in entrepreneurship research. Int. Entrep. Manag. J.; 2020; 16, pp. 1023-1042. [DOI: https://dx.doi.org/10.1007/s11365-020-00635-4]

24. Cheng, Y.; Awan, U.; Ahmad, S.; Tan, Z. How do technological innovation and fiscal decentralization affect the environment? A story of the fourth industrial revolution and sustainable growth. Technol. Forecast. Soc. Chang.; 2021; 162, 120398. [DOI: https://dx.doi.org/10.1016/j.techfore.2020.120398]

25. Acemoglu, D. When does labor scarcity encourage innovation?. J. Polit. Econ.; 2010; 118, pp. 1037-1078. [DOI: https://dx.doi.org/10.1086/658160]

26. Acemoglu, D.; Restrepo, P. Secular stagnation? The effect of aging on economic growth in the age of automation. Am. Econ. Rev.; 2017; 107, pp. 174-179. [DOI: https://dx.doi.org/10.1257/aer.p20171101]

27. Acemoglu, D.; Restrepo, P. The race between man and machine: Implications of technology for growth, factor shares, and employment. Am. Econ. Rev.; 2018; 108, pp. 1488-1542. [DOI: https://dx.doi.org/10.1257/aer.20160696]

28. Madsen, J.B.; Damania, R. Labour Demand and Wage-induced Innovations: Evidence from the OECD countries. Int. Rev. Appl. Econ.; 2001; 3, pp. 323-334. [DOI: https://dx.doi.org/10.1080/02692170110052365]

29. Antonelli, C.; Quatraro, F. The Effects of Biased Technological Changes on Total Factor Productivity: A Rejoinder and New Empirical Evidence. J. Technol. Transf.; 2014; 2, pp. 281-299. [DOI: https://dx.doi.org/10.1007/s10961-013-9328-5]

30. Cutler, D.; Poterba, J.; Sheiner, L.; Summers, L. An Aging Society: Opportunity or Challenge. Brook. Pap. Econ. Act.; 1990; 21, pp. 1-73. [DOI: https://dx.doi.org/10.2307/2534525]

31. Tan, Y.; Liu, X.; Sun, H.; Zeng, C. Population ageing, labour market rigidity and corporate innovation: Evidence from China. Res. Policy; 2022; 51, 104428. [DOI: https://dx.doi.org/10.1016/j.respol.2021.104428]

32. Wachsen, E.; Blind, K. More labor market flexibility for more innovation? Evidence from employer-employee linked micro data. Res. Policy; 2016; 5, pp. 941-950. [DOI: https://dx.doi.org/10.1016/j.respol.2016.01.020]

33. Hoxha, S.; Kleinknecht, A. When labour market rigidities are useful for innovation. Evidence from German IAB firm-level data. Res. Policy; 2020; 49, 104066. [DOI: https://dx.doi.org/10.1016/j.respol.2020.104066]

34. Kleinknecht, A. The (Negative) impact of supply-side labor market reforms on productivity: An overview of the evidence. Camb. J. Econ.; 2021; 44, pp. 445-464. [DOI: https://dx.doi.org/10.1093/cje/bez068]

35. Gordo, L.R.; Skirbekk, V. Skill demand and the comparative advantage of age: Jobs tasks and earnings from the 1980s to the 2000s in Germany. Labour Econ.; 2013; 22, pp. 61-69. [DOI: https://dx.doi.org/10.1016/j.labeco.2012.09.003]

36. Cai, J.; Stoyanov, A. Population Ageing and Comparative Advantage. J. Int. Econ.; 2016; 4, pp. 1-21. [DOI: https://dx.doi.org/10.1016/j.jinteco.2016.04.006]

37. Prettner, K.; Bloom, D.E.; Strulik, H. Declining fertility and economic well-being: Do education and health ride to the rescue?. Labour Econ.; 2013; 22, pp. 70-79. [DOI: https://dx.doi.org/10.1016/j.labeco.2012.07.001] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26388677]

38. Bloom, D.E.; Canning, D.; Graham, B. Longevity and life cycle savings. Scand. J. Econ.; 2003; 105, pp. 319-338. [DOI: https://dx.doi.org/10.1111/1467-9442.t01-1-00001]

39. Maestas, N.; Mullen, K.J.; Powell, D. The effect of population aging on economic growth, the labor force and productivity. NBER Working Paper; National Bureau of Economic Research, Inc.: Cambridge, MA, USA, 2016; 22452.

40. Tressel, T.; Scarpetta, S. Boosting Productivity via Innovation and Adoption of New Technologies: Any Role for Labor Market Institutions?; World Bank Publications: Washington, DC, USA, 2004.

41. Bassanini, A.; Nunziata, L.; Venn, D. Job protection legislation and productivity growth in OECD countries. Econ. Policy; 2009; 24, pp. 349-402. [DOI: https://dx.doi.org/10.1111/j.1468-0327.2009.00221.x]

42. Bartelsman, E.J.; Gautier, P.A.; De Wind, J. Employment protection, technology choice, and worker allocation. Int. Econ. Rev.; 2016; 57, pp. 787-826. [DOI: https://dx.doi.org/10.1111/iere.12176]

43. Noda, H. Population Ageing and Creative Destruction. J. Econ. Res.; 2011; 1, pp. 29-58.

44. Gonzales, E.M.; Niepelt, D. Ageing, Government Budgets, Retirement and Growth. Eur. Econ. Rev.; 2012; 1, pp. 97-115. [DOI: https://dx.doi.org/10.1016/j.euroecorev.2011.05.007]

45. Verhaeghen, P.; Salthouse, T.A. Meta-analyses of age-cognition relations in adulthood: Estimates of linear and nonlinear age effects and structural models. Psychol. Bull.; 1997; 122, pp. 231-249. [DOI: https://dx.doi.org/10.1037/0033-2909.122.3.231]

46. Behaghel, L.; Greenan, N. Training and Age—Biased Technical Change. Ann. Stat.; 2010; 99/100, pp. 317-342. [DOI: https://dx.doi.org/10.2307/41219169]

47. Ashworth, M.J. Preserving Knowledge Legacies: Workforce Aging, Turnover and Human Resource Issues in the US Electric Power Industry. Int. J. Hum. Resour. Manag.; 2006; 9, pp. 1659-1688. [DOI: https://dx.doi.org/10.1080/09585190600878600]

48. Nusbaum, E.C.; Silvia, P.J. Are Intelligence and Creativity Really so Different? Fluid Intelligence, Executive Processes, and Strategy Use in Divergent Thinking. Intelligence; 2011; 1, pp. 36-45. [DOI: https://dx.doi.org/10.1016/j.intell.2010.11.002]

49. Jones, B.F. Age and Great Invention. Rev. Econ. Stat.; 2010; 1, pp. 1-14. [DOI: https://dx.doi.org/10.1162/rest.2009.11724]

50. Kanfer, R.; Ackerman, P. Individual Differences in Work Motivation: Further Exploration of a Trait Framework. Appl. Psychol.; 2000; 3, pp. 470-482. [DOI: https://dx.doi.org/10.1111/1464-0597.00026]

51. Lancia, F.; Prarolo, G. A Politico-economic Model of Ageing, Technology Adoption, and Growth. J. Popul. Econ.; 2012; 3, pp. 989-1018. [DOI: https://dx.doi.org/10.1007/s00148-011-0364-x]

52. Friedberg, R.M. The impact of mass migration on the Israeli labor market. Q. J. Econ.; 2001; 116, pp. 1373-1408. [DOI: https://dx.doi.org/10.1162/003355301753265606]

53. Frosch, K.; Tivig, T. Age, Human Capital and the Geography of Innovation. Labour Mark. Demogr. Chang.; 2009; 6, pp. 137-146.

54. Xie, X.; Zhu, X. Population Aging, Technological Innovation and Economic Growth. China Soft Sci.; 2020; 6, pp. 42-53. (In Chinese)

55. Yang, X.; He, L.; Xia, Y.; Chen, Y. Effect of government subsidies on renewable energy investments: The threshold effect. Energy Policy; 2019; 132, pp. 156-166. [DOI: https://dx.doi.org/10.1016/j.enpol.2019.05.039]

56. Hansen, B.E. Threshold Effect in Non-dynamic Panels: Estimation, Testing, and Inference. J. Econom.; 1999; 93, pp. 345-368. [DOI: https://dx.doi.org/10.1016/S0304-4076(99)00025-1]

57. Hansen, B.E. Sample Splitting and threshold estimation. Econometrica; 2000; 68, pp. 575-603. [DOI: https://dx.doi.org/10.1111/1468-0262.00124]

58. Lin, S.; Sun, J.; Marinova, D.; Zhao, D. Evaluation of the green technology innovation efficiency of China’s manufacturing industries: DEA window analysis with ideal window width. Technol. Anal. Strat. Manag.; 2018; 30, pp. 1166-1181. [DOI: https://dx.doi.org/10.1080/09537325.2018.1457784]

59. Zhang, J.; Kang, L.; Li, H.; Ballesteros-Pérez, P.; Skitmore, M.; Zuo, J. The impact of environmental regulations on urban Green innovation efficiency: The case of Xi’an. Sustain. Cities Soc.; 2020; 57, 102123. [DOI: https://dx.doi.org/10.1016/j.scs.2020.102123]

60. Zhu, L.; Luo, J.; Dong, Q.; Zhao, Y.; Wang, Y.; Wang, Y. Green technology innovation efficiency of energy-intensive industries in China from the perspective of shared resources: Dynamic change and improvement path. Technol. Forecast. Soc. Chang.; 2021; 170, 120890. [DOI: https://dx.doi.org/10.1016/j.techfore.2021.120890]

61. Zhao, N.; Liu, X.; Pan, C.; Wang, C. The performance of green innovation: From an efficiency perspective. Socio-Econ. Plan. Sci.; 2021; 78, 101062. [DOI: https://dx.doi.org/10.1016/j.seps.2021.101062]

62. Fan, F.; Lian, H.; Liu, X.; Wang, X. Can environmental regulation promote urban green innovation efficiency? An empirical study based on Chinese cities. J. Clean. Prod.; 2021; 287, 125060. [DOI: https://dx.doi.org/10.1016/j.jclepro.2020.125060]

63. Li, J.; Du, Y. Spatial effect of environmental regulation on green innovation efficiency: Evidence from prefectural-level cities in China. J. Clean. Prod.; 2021; 286, 125032. [DOI: https://dx.doi.org/10.1016/j.jclepro.2020.125032]

64. Miao, C.-L.; Duan, M.-M.; Zuo, Y.; Wu, X.-Y. Spatial heterogeneity and evolution trend of regional green innovation efficiency–an empirical study based on panel data of industrial enterprises in China’s provinces. Energy Policy; 2021; 156, 112370. [DOI: https://dx.doi.org/10.1016/j.enpol.2021.112370]

65. Chen, X.; Liu, X.; Gong, Z.; Xie, J. Three-stage super-efficiency DEA models based on the cooperative game and its application on the R&D green innovation of the Chinese high-tech industry. Comput. Ind. Eng.; 2021; 156, 107234.

66. Lin, B.; Luan, R. Are government subsidies effective in improving innovation efficiency? Based on the research of China’s wind power industry. Sci. Total Environ.; 2020; 710, 136339. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2019.136339]

67. Sun, H.; Edziah, B.K.; Kporsu, A.K.; Sarkodie, S.A.; Taghizadeh-Hesary, F. Energy efficiency: The role of technological innovation and knowledge spillover. Technol. Forecast. Soc. Chang.; 2021; 167, 120659. [DOI: https://dx.doi.org/10.1016/j.techfore.2021.120659]

68. Jacobs, R. Alternative Methods to Examine Hospital Efficiency: Data Envelopment Analysis and Stochastic Frontier Analysis. Health Care Manag. Sci.; 2001; 4, pp. 103-115. [DOI: https://dx.doi.org/10.1023/A:1011453526849]

69. Qian, L.; Xiao, R.; Chen, Z. Study of Industrial Enterprises’ Technology Innovation Efficiency and Regional Disparities in China—Based on the Theory of Meta-Frontier and DEA Model. Econ. Theory Bus. Manag.; 2015; 1, pp. 26-43. (In Chinese)

70. Song, W.; Han, X. The bilateral effects of foreign direct investment on green innovation efficiency: Evidence from 30 Chinese provinces. Energy; 2022; 261, 125332. [DOI: https://dx.doi.org/10.1016/j.energy.2022.125332]

71. Earnhart, D. Regulatory factors shaping environmental performance at publiclyowned treatment plants. J. Environ. Econ. Manag.; 2004; 48, pp. 655-681. [DOI: https://dx.doi.org/10.1016/j.jeem.2003.10.004]

72. Kammerer, D. The effects of customer benefit and regulation on environmental product innovation.: Empirical evidence from appliance manufacturers in Germany. Ecol. Econ.; 2009; 68, pp. 2285-2295. [DOI: https://dx.doi.org/10.1016/j.ecolecon.2009.02.016]

73. Shuai, S.; Fan, Z. Modeling the role of environmental regulations in regional green economy efficiency of China: Empirical evidence from super efficiency DEA-Tobit model. J. Environ. Manag; 2020; 261, 110227. [DOI: https://dx.doi.org/10.1016/j.jenvman.2020.110227] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32148297]

74. Song, W.; Han, X. A bilateral decomposition analysis of the impacts of environmental regulation on energy efficiency in China from 2006 to 2018. Energy Strategy Rev.; 2022; 43, 100931. [DOI: https://dx.doi.org/10.1016/j.esr.2022.100931]

75. Sonia Ben, K.; Natalia, Z.-S. The pollution haven hypothesis: A geographic economy model in a comparative study. Fond. Eni Enrico Mattei; 2008; 44223.

76. Zhao, Q.W.; Zhang, C. Foreign direct investment and China’s technical efficiency improvement: An empirical analysis based on stochastic frontiers production model. World Econ. Stud.; 2009; 6, pp. 61-67.

77. Han, J.; Miao, J.; Shi, Y.; Miao, Z. Can the semi-urbanization of population promote or inhibit the improvement of energy efficiency in China?. Sustain. Prod. Consump.; 2021; 26, pp. 921-932. [DOI: https://dx.doi.org/10.1016/j.spc.2021.01.008]

78. Fan, G.; Wang, X.; Zhu, H. Neri Index of Marketization of China’s Provinces 2011 Report; Economic Science Press: Beijing, China, 2011.

79. Hausman, J.A. Specification tests in econometrics. Econometrica; 1978; 46, pp. 1251-1271. [DOI: https://dx.doi.org/10.2307/1913827]

80. Clarke-Sather, A.; Qu, J.; Wang, Q.; Zeng, J.; Li, Y. Carbon inequality at the sub-national scale: A case study of provincial-level inequality in CO₂ emissions in China 1997–2007. Energy Policy; 2011; 39, pp. 5420-5428. [DOI: https://dx.doi.org/10.1016/j.enpol.2011.05.021]

81. Pan, X.F.; Zhang, J.; Li, C.Y.; Quan, R.; Li, B. Exploring dynamic impact of foreign direct investment on China’s CO2 emissions using Markov-switching vector error correction model. Comput. Econ.; 2018; 52, pp. 1139-1151. [DOI: https://dx.doi.org/10.1007/s10614-017-9745-x]

82. Stucki, T.; Woerter, M.; Arvanitis, S.; Peneder, M.; Rammer, C. How different policy instruments affect green product innovation: A differentiated perspective. Energy Policy; 2018; 114, pp. 245-261. [DOI: https://dx.doi.org/10.1016/j.enpol.2017.11.049]