Can industrial agglomeration affect biodiversity loss?

Abstract

Reducing the rate of species extinction and protecting biodiversity are major international concerns. The loss of biodiversity is closely related to international trade as an inevitable result of industrial agglomeration and improvements in green economy efficiency. In contrast to previous studies on biodiversity loss from the perspectives of deforestation, hunting, and fire, this study examines biodiversity loss from an international trade perspective, calculates the biodiversity footprint of each country as an indicator of biodiversity loss, and innovatively elaborates on the theoretical mechanisms of industrial agglomeration, green economy efficiency, and biodiversity loss. An empirical analysis used panel data from 148 countries from 2006 to 2020. This study identifies that industrial agglomeration directly and indirectly aggravates biodiversity loss through green economy efficiency, which mediates the relationship between industrial agglomeration and biodiversity loss. The effects of industrial agglomeration and green economy efficiency on biodiversity loss driven by export trade in developed and developing countries are consistent with the benchmark test results. The effect of industrial agglomeration on green economy efficiency is positive in developed countries and vice versa in developing countries. By region, green economy efficiency can significantly mitigate the embodied biodiversity loss in the export trade in Asia, Africa, and North America, whereas its influence in Europe and North America is insignificant. This study extends the perspective of biodiversity research from the natural to economic fields, delves into the underlying economic causes of the current state of trade-driven biodiversity loss, and provides important evidence for reducing biodiversity losses caused by international trade.

Keywords

Industrial agglomeration green economy efficiency biodiversity loss mediating effects trade

Introduction

The global ecological crisis has worsened in recent years primarily because of widespread forest loss, increased land desertification, continuous wetland degradation, accelerated species extinction, and severe soil and water erosion. Among these, species extinction is the most prominent. Humans have already occupied 90% of the world's biologically active area with 0.01% biomass and have destroyed 83% of wild mammals and 50% of plants.¹ According to the 2021 IUCN Red List of Threatened Species, endangered species account for up to 28% of the assessed species. If no measures are taken to prevent this trend, the risk of species extinction will further increase, potentially hastening the sixth mass extinction.² Therefore, there is an urgent need to conserve biodiversity. Biodiversity is directly related to the stability and sustainability of ecological and social systems. Their self-regulatory capacity, ecological services, and material resources are crucial for the sustainable development of human societies, particularly the economy, and have gained considerable attention from the international community.

In 1992, the United Nations Conference on Environment and Development adopted the Convention on Biological Diversity, which aims to protect endangered plants and animals and requires developed countries to provide financial resources or transfer technology to developing countries for facilitating the conservation of biological resources. In 2010, the United Nations set 20 “Aichi Targets” to focus on the short-term effects of biodiversity conservation over a decade. In 2015, the United Nations adopted 17 sustainable development goals. Goal 15 aimed to protect, restore, and promote the sustainable use of terrestrial ecosystems, sustainable forest management, combat desertification, and halt land degradation and biodiversity loss. However, biodiversity governance has been ineffective. Only 6 of the 20 “Aichi Targets” were partially achieved, and several sub-goals worsened. Since 1996, The Red List Index, which measures biodiversity, has decreased from 0.82 to approximately 0.75. The number of critically endangered species increased from 1820 to 8404, endangered species increased from 2375 to 14,647, and vulnerable species increased from 6337 to 15,492.

In recent years, approximately 30% of threats to species have been caused by international trade,³ and this proportion is increasing, making trade a major cause of the current threat to biodiversity in all countries. Relying on complex global value chains, international trade enables a country's consumption demand to be met by producing and exporting products to other countries. The existence of “pollution havens” accelerates the international transfer of pollution-intensive industries such as mining.⁴ These industries have a greater negative impact on the local environment and tend to accelerate habitat degradation and fragmentation of regional habitats, leading to an increase in endangered species and a significant loss of biodiversity.⁵ Pollution havens always exist in countries with weakly enforced legislation and poverty, such as metal minerals mining in Kazakhstan and oil extraction in the Amazon Basin. As a result, the loss of biodiversity caused by international trade is mostly in developing countries. Analogous to the concept of embodied carbon, endangered species resulting from products produced and processed at home and ultimately consumed abroad are embodied in biodiversity loss. Calculating embodied biodiversity loss can be useful for measuring the complex threats to species resulting from international trade.³

Meanwhile, trade brings larger overseas markets and higher profits, improves regional production technology through technology spillover effects, reduces costs, and improves production efficiency, thus gradually concentrating industrial activities and increasing the degree of agglomeration. With the increase in trade activities, transaction efficiency improves and trade costs fall.⁶ Profit drives enterprises to continue concentrating in areas with lower trade costs, promoting industrial agglomeration. In addition, trade encourages countries to continuously deepen the division of production to specialize in different production links for the same product. As a result, vertical linkages are formed, and production costs are reduced,⁷ promoting industries’ spatial agglomeration.

Industrial agglomeration, which causes spatial diffusion and the transfer of environmental problems, is also a significant factor affecting biodiversity.⁸ Increased resource demand accompanies industrial agglomeration, and the over-exploitation of natural resources has caused significant biological habitat loss. The gradual increase in the use of fossil energy sources also directly affects carbon dioxide emissions,⁹ which are likely to produce the greenhouse effect and exacerbate climate change, thus threatening the survival of species. Additionally, agglomeration can aggravate environmental pollution to a certain extent. This will have a catastrophic impact on the atmosphere, soil, and ocean, damaging ecosystem functions and natural habitats, increasing habitat species density, and accelerating species competition and extinction. Thus, industrial agglomeration can have a direct negative impact on species, further amplifying biodiversity losses.

Moderate industrial agglomeration relies on conventional mechanisms such as resource sharing, cost savings, human capital, scale effects, technological spillovers, enhancing competition, collaborative innovation, and reducing information asymmetry to promote green economy efficiency improvements.¹⁰ With the global emphasis on harmonious economic and environmental development, countries can rely on the improvement of green total factor productivity to introduce green factors into economic production to achieve sustainable production and consumption, as well as the transformation from an extensive to an intensive economy, thus avoiding the uncontrolled exploitation of resources, excessive use of pesticides and fertilizers, and irrational industrial production activities, and providing a better living space for species. It can also promote the upgrading of the country's position in the global value chain by improving green total factor productivity, improving the structure of product trade, reducing the negative impact of industrial activities on natural ecosystems,¹¹ and protecting biological reproduction and biodiversity, thus solving the dilemma of the continuous expansion of biodiversity loss.

This study incorporates industrial agglomeration, green economic efficiency, and biodiversity loss into the same analytical framework, with marginal contributions in the following aspects. When discussing the relationship between international trade and the environment, almost all studies have focused on the impact of trade on embodied carbon emissions¹² and energy.¹³ However, few scholars have studied the impact of trade on biodiversity. Biodiversity research has focused on measuring the loss of species richness. This study uses the carbon footprint concept to calculate the biodiversity footprint as a measure of biodiversity loss, providing a new concept for accurately measuring biodiversity loss in countries driven by international trade and achieving biodiversity conservation. Second, although Lenzen et al.³ linked international trade to biodiversity loss to quantify the important role of trade as a driver of species threats, subsequent relevant studies lacked a systematic analysis of the impact mechanism. Economic efficiency and industrial agglomeration are not considered in studies on biodiversity loss driven by international trade. Based on the measurement of biodiversity loss, this study presents an in-depth analysis of the relationship between industrial agglomeration, green economy efficiency, and biodiversity loss, focusing on how industrial agglomeration and green economy efficiency jointly contribute to biodiversity loss and identifying the interaction mechanism between the three factors, which is of practical significance for countries to slow down the rate of species extinction. Third, some scholars have revealed the different impacts of international trade on biodiversity in developed and developing countries. However, they studied this from the perspective of their own country. However, the heterogeneous effects of regions have not yet been mentioned. Therefore, this study examines the relationship among industrial agglomeration, green economy efficiency, and biodiversity loss by country and region, answers the question of whether there are differences in action paths between developing and developed countries and between different regions, and provides empirical support for effective local adaptation to curb biodiversity loss.

The remainder of this paper is structured as follows: the second section contains the literature review and theoretical hypotheses; the third section contains the modeling and description of indicators; the fourth section contains the empirical analysis; the fifth section contains the test of heterogeneity; and the section contains the conclusion and policy recommendations.

Literature review and theoretical hypotheses

Biodiversity loss is a serious global issue, as evidenced by the IUCN for Conservation of Nature's Red List of Threatened Species. To mitigate the species extinction crisis, scholars have primarily researched agricultural production,¹⁴ land use,¹⁵ climate change,¹⁶ illegal hunting,¹⁷ international trade,¹⁸ and other influencing factors. As an academic research hotspot, the environmental effects of trade are complex. Li et al.¹⁹ found that trade openness helps reduce global carbon emissions and increases carbon emissions in the lower middle-income group. Wang et al.²⁰ believed that trade openness increases the ecological footprint and environmental pressure. Wang et al.²¹ thought that income inequality had changed the relationship between trade and carbon emissions from an inverted U-shape to an N-shape and has increased the complexity of trade and carbon emissions decoupling. Wang et al.²² investigated the decoupling impact of trade on carbon emissions, pointing out that trade openness favors carbon neutrality in rich countries but not in less developed countries. Wang et al.²³ investigated the effect of trade openness on carbon emissions within the framework of the environmental Kuznets curve (EKC) hypothesis, and the results showed that trade openness has a mitigating effect on carbon emissions in only a few countries. Although many studies related to the impact of trade on the environmental footprint, especially the carbon footprint, only a few scholars have used international trade^3,18 as an entry point for calculating biodiversity loss in each country by measuring biodiversity footprints in trade. Previous studies have focused on the characterization of biodiversity loss and the allocation of responsibility without an in-depth analysis of the specific influencing factors and underlying mechanisms of action. However, there is a significant research gap remains in this area.

The implied export of species in trade, that is, the biodiversity loss, depends on the impact of externalities of industrial activities on biodiversity on the one hand and is determined by the product mix of export trade on the other. The development of international trade accelerates industrial agglomeration and affects biodiversity through various externalities, affecting biodiversity loss. Enterprises typically select factor-intensive areas during agglomeration to reduce their production transaction costs. In particular, resource-intensive industries are mostly located in raw material production areas, thus inducing the overexploitation of biological resources and directly reducing the number of species. Meanwhile, the layout can lead to the blind use of other resources, thus destroying the balance between ecosystems and biological habitats, causing the degradation of biological habitats and indirectly affecting the species’ long-term survival.²⁴ Industrial agglomeration also contributes to infrastructure development, particularly the transport infrastructure.²⁵ The higher the degree of agglomeration, the better the transport infrastructure and the more pronounced the impact on biodiversity. The developed transport networks restrict the space for biological activities and block genetic exchange. Fragmented distribution patterns of species have led to shrinkage and extinction. At the same time, the construction of transport infrastructure occupies large amounts of land and compresses habitats, increasing the density of species in the habitat and intensifying competition and pressure for survival. Developed transportation also overcomes geographical barriers to some extent and creates conditions for hunting, logging, and other forms of exploitation that are not conducive to species conservation.

Furthermore, industrial agglomeration can harm the environmental quality and threaten biodiversity. Specialized agglomerations tend to generate path dependence and technology locking, which can reduce economic efficiency and aggravate environmental pollution.²⁶ The crowding and intensification effects of diversified agglomerations can also increase pollution,²⁷ ultimately increasing the rate of species extinction and biodiversity loss. Industrial agglomeration also affects biodiversity by influencing climate change. Climate change has been a recurring factor in historical mass extinction crises.²⁸ Global warming has become evident in recent years, and greenhouse gas (GHG) emissions, such as carbon dioxide, have increased. Industrial agglomeration significantly impacts carbon emissions: the higher the degree of industrial agglomeration, the greater the carbon dioxide emissions²⁹ and the clearer the greenhouse effect. Ensuing extreme climates, such as high temperatures and droughts, cause changes in biological habitats, and biological organisms die in large numbers because they cannot adapt to new environments, leading to biodiversity loss. Therefore, we propose Hypothesis 1:

Hypothesis 1 (H1): Industrial agglomeration aggravates biodiversity loss, and these two factors have a significant positive relationship.

Green economy efficiency, which reflects technological progress, efficiency improvements, and the green environmental protection of production, directly affects biodiversity through positive environmental externalities. Specifically, advances in production technology can improve productivity and resource use efficiency, reduce factor inputs and energy consumption per unit of output,³⁰ and avoid overexploitation of natural resources, thus reducing direct species loss through decreased logging and hunting, as well as maintaining ecological balance,³¹ mitigating external damage to habitats, and reducing indirect species loss. Advances in environmental technologies, including using green and clean energy, pollution treatment equipment, and clean and environment-friendly production processes, can reduce pollutant emissions and environmental pollution.³² In particular, reductions in toxic pollutant emissions and eutrophication can help improve atmospheric, marine, freshwater, and soil quality, thereby creating healthy biological habitats for species survival. Additionally, green economy efficiency can contribute to upgrading industrial structures by influencing the efficiency of production factors.³³ Modern and high-tech industries consume less energy than mining and manufacturing industries. The structural optimization effect of green economy efficiency can contribute to the upgrading of industrial structures by influencing the efficiency of the allocation of production factors, thereby reducing the demand for fossil fuels, such as coal and oil, effectively reducing carbon dioxide emissions^11,34 and slowing down species extinction and loss caused by global warming.

Green economy efficiency can also change the structure of trade products, thus promoting the upgrading of global value chains to reduce biodiversity losses in the export trade. Improvements in green economy efficiency can lead to factor integration, optimizing resource allocation, strengthening the market power of the corresponding segments, enhancing the value-added capacity of domestic factors, upgrading products and functions, breaking the “low-end lock” dilemma of developing countries, and moving from low-end to high-end segments.³⁵ The rise in the status of value chains will change the structure of trade products to some extent, with exports gradually transitioning from resource-and labor-intensive products to technology-intensive products. As technology-intensive industries have fewer negative environmental externalities than resource-and labor-intensive industries and the negative impacts of industrial activities on ecology and species are weaker, the embodied biodiversity loss in export products gradually decreases along with the improvement of green economy efficiency and rise in the status of the value chain. Therefore, we propose Hypothesis 2.

Hypothesis 2 (H2): Green economy efficiency can mitigate biodiversity loss, and a significant negative relationship exists between these two factors.

Industrial agglomeration is closely related to green economic efficiency. Some researchers agree that industrial agglomeration can promote green technological progress. Specialized agglomeration positively impacts green technological progress³⁶ through Martian externalities, mainly economies of scale and knowledge spillovers. The scale economy effect can significantly reduce management and information search costs, and enterprises can invest more in technological innovation and achieve green technological progress. Specialization agglomeration enables the smooth flow of talent among enterprises, promotes the dissemination and sharing of knowledge, technology, and information among similar industries, and causes technological knowledge spillover, which promotes green technological progress.³⁷ Diversification agglomeration positively affects green technological progress through Jacobs' externalities, mainly through innovation links and cooperative effects. On the one hand, diversification agglomeration can promote a bidirectional flow of knowledge and technology between industries and lead to complementary technological innovation in other related industries through technological innovation in one industry, thus forming a mutually beneficial industrial chain and accelerating green technological progress through the innovation linkage effect.³⁸ Diversification agglomeration is also conducive to strengthening synergistic cooperation among enterprises, reducing the exclusivity of knowledge among different industries and enterprises, and realizing scientific research cooperation and green technological progress through technology and cost sharing.³⁹

However, the objective of this positive influence is green technological progress, and the premise is that industrial agglomeration is reasonable and moderate. Excessive industrial agglomeration is likely to cause diseconomies of scale, traffic congestion, inadequate infrastructure supply, increased energy consumption, severe environmental pollution, and ecological degradation. As a result, the cost of production factors, such as land and capital, increases, leading to wasted resources and lower productivity.⁴⁰ Eventually, a congestion effect forms, restricting the improvement in green economy efficiency.⁴¹ In other words, when the congestion effect is sufficiently large, industrial agglomeration hinders green economy efficiency.⁴² Considering that the degree of industrial agglomeration in various countries is generally high, the impact of industrial agglomeration on green economic efficiency is negative. Combined with Hypothesis 2, it is clear that industrial agglomeration can affect biodiversity directly or indirectly by influencing the structure of trade products by improving green economic efficiency, thereby affecting biodiversity loss. Therefore, we propose Hypothesis 3.

Hypothesis 3 (H3): Industrial agglomeration negatively contributes to green economic efficiency, mediating biodiversity loss.

Methodology and data

Regression model

A mediating effect model was developed based on the theoretical hypotheses of this study.

L n b f_{i t} = α_{0} + α_{1} L n a g g l_{i t} + \sum C o n t r o l + ε_{i t} + μ_{i} + w_{t}

(1)

L n g t f p_{i t} = β_{0} + β_{1} L n a g g l_{i t} + \sum C o n t r o l + ε_{i t} + μ_{i} + w_{t}

(2)

L n b f_{i t} = γ_{0} + γ_{1} L n a g g l_{i t} + γ_{2} L n g t f p_{i t} + \sum C o n t r o l + ε_{i t} + μ_{i} + w_{t}

(3)

where bf denotes embodied biodiversity loss in export trade, aggl denotes the degree of industrial agglomeration, gtfp denotes green economy efficiency, i and t denote country and year, respectively, and Control represents a series of control variables including aging, gdp, adv, and open, which denote aging population, real GDP per capita, value-added rate, and degree of openness to the outside world, respectively. Additionally,

ε_{i t}

is a random disturbance term,

u_{i}

is the individual effect, and

w_{t}

is the time effect. According to Baron and Kenny's⁴³ study on mediating effects between variables,

α_{1}

reflects the total effect of industrial agglomeration on the embodied biodiversity loss in export trade,

γ_{1}

represents the direct effect of industrial agglomeration on the embodied biodiversity loss in export trade, and the size of the mediating effect is measured by

β_{1} γ_{2}

Biodiversity footprint (bf) measurement model

This study refers to Lenzen's³ method of measuring the biodiversity footprint (bf) to represent biodiversity loss in the export trade. First, the Red List of Threatened Species was collected, and the statistical range of species was classified as endangered, critically endangered, or vulnerable according to the IUCN criteria. Simultaneously, three broad categories of threats were excluded: natural system modification, invasive and other problematic species and genes, and geological events unrelated to international trade. Ten broad categories and 82 sub-causes of threats related to international trade were identified and matched to the species. Species resources were then incorporated into the environmental satellite accounts in the input–output analysis, and the Red List of Threatened Species and multiregional input–output tables were effectively combined through multiple matching to explore biodiversity loss for each country in the following manner.

First, we match the production activities involved in the threat with the product using CPC V1.0, as the common product classification (CPC) and construct a matrix with the threat as the row and CPC as the column, with one of the threats corresponding to the product and zero otherwise, to obtain matrix M1. Second, according to the international standard industry classification, we matched the common products in CPC V1.0 with the industry sectors in the input–output table. If it corresponds to an industry, it is one; otherwise, it is zero. Then, we obtain the matrix M2. Finally, threats to species were matched to industry sectors in international trade as follows:

M_{C} = M_{1} \times M_{2}

(4)

The M_C matrix was standardized to obtain the N_C matrix to ensure that all threats were not duplicated in accounting.

N_{C} = (\hat{M_{C} X})^{- 1} M_{C} X

(5)

After database matching, this study assumes that the endangered species in a country are only related to the production activities and threat exists only in the country. On that basis, the species is introduced by constructing a matrix of rows with species/country/threat and columns with country/threat. If the country is the same and the threat corresponds, the element is recorded as one; otherwise, it is recorded as zero. A Z-matrix was obtained. The Q matrix was calculated based on the Z matrix, as follows:

Q = Z_{C} \times N_{C}

(6)

To eliminate threats in the matrix, we further processed the Q matrix by summing the statistics of the same rows of species/countries to obtain the Q_ag matrix and then normalizing it to obtain the W matrix as follows:

W = (\hat{Q_{a g} 1})^{- 1} Q_{a g}

(7)

To remove national boundaries, we further processed the W matrix by summing the rows with the same species to obtain the W_ag matrix, which showed the threats imposed on biological species by the production activities of the industry in each country and conformed to the standard format of satellite accounts required for integrated environmental economic analysis.

Compared with the single-region model, the multiregion input–output model considers the variability of intermediate inputs and technology levels in different countries, and the complete consumption coefficient is somewhat modified, resulting in a more accurate measurement. Therefore, the multiregional input–output (MRIO) model was used to calculate the biodiversity footprint. Depending on where the product is used, the MRIO model can be expressed as

X_{S} = \sum (A_{s r} X_{r} + Y_{s r}), r, s = 1, 2, \dots m

(8)

where X_s represents the total output of country s, Y_sr represents the final product used to meet the demand of country r of the total output of country s, A_sr represents the direct consumption coefficient, which represents the intermediate product of country s consumed per unit of total output of country r, and A_sr X_sr represents the intermediate product exported to country r of the total output of country s. The deformation of Equation (8) yields Equation (9):

X = (I - T {\hat{X}}^{- 1})^{- 1} Y = (I - A)^{- 1} Y = L Y

(9)

Leontief demonstrated that a pollution intensity matrix can be added to the traditional input–output matrix to calculate the carbon footprint, regardless of whether the carbon intensity of imports is consistent with that of domestic products for both unilateral and multilateral trade.⁴⁴ Biological species belong to the environmental resource account in the input–output analysis. Therefore, the carbon footprint calculation method is equally applicable when measuring the biodiversity footprint. Compared to the previous method proposed by Lenzen,³ the method proposed in this study is different. The steps were identical before calculating the biodiversity footprint. The methodological improvement and innovation point is the calculation of the biodiversity consumption coefficient from the species and country/industry matrices. The biodiversity consumption coefficient matrix reflects each country's current status of biodiversity depletion, owing to its economic and industrial status. If E_s denotes the biodiversity consumption coefficient vector for country s, the embodied species loss in the export trade from country s to country r can be expressed by Equation (10) or Equation (11).

F_{s r} = E X_{s r} = \hat{E_{s}} X_{s r} = \hat{E_{s}} \sum L_{s t} Y_{t r}, r, s = 1, 2, \dots m

(10)

F = E X = E L Y

(11)

(\begin{matrix} f_{1, 1} & f_{1, 2} & \dots & f_{1, 188} \\ f_{2, 1} & f_{2, 2} & \dots & f_{2, 188} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ f_{188, 1} & f_{188, 2} & \dots & f_{188, 188} \end{matrix}) = (\begin{matrix} E_{1} & 0 & \dots & 0 \\ 0 & E_{2} & \dots & 0 \\ ⋮ & ⋮ & ⋱ & ⋮ \\ 0 & 0 & \dots & E_{188} \end{matrix}) (\begin{matrix} L_{1, 1} & L_{1, 2} & \dots & L_{1, 188} \\ L_{2, 1} & L_{2, 2} & \dots & L_{2, 188} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ L_{188, 1} & L_{188, 2} & \dots & L_{188, 188} \end{matrix}) (\begin{matrix} y_{1, 1} & y_{1, 2} & \dots & y_{1, 188} \\ y_{2, 1} & y_{2, 2} & \dots & y_{2, 188} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ y_{188, 1} & y_{188, 2} & \dots & y_{188, 188} \end{matrix})

The diagonal elements of E_s represent the species consumed per unit of total output in each sector of country s. The corresponding elements of the row vector obtained by summing the rows of matrix W_ag are divided by the corresponding national sector output, resulting in biodiversity consumption coefficients placed on the matrix's diagonal, that is, the biodiversity consumption coefficient matrix E. The biodiversity footprint matrix F quantified the threats to species indirectly caused by the final consumer, with the biodiversity footprint extending from domestic to foreign countries. Based on Lyontief input–output analysis, we observed that summing the row elements of the F matrix yields the biodiversity loss for each country, that is, the loss of biodiversity in a country driven by international trade.

Variable descriptions

Independent variables

Area entropy index (aggl). As the location entropy index can eliminate differences at the regional scale and truly reflect the spatial distribution of geographical factors, this study chose the location entropy index to characterize the level of industrial agglomeration, which is calculated as follows:

a g g l = \frac{E_{i j} / \sum_{j = 1}^{m} E_{i j}}{\sum_{i = 1}^{n} E_{i j} / \sum_{j = 1}^{n} \sum_{i = 1}^{m} E_{i j}}

(12)

where j denotes the industry, i denotes the country, E_ij denotes the number of people employed in industry j in country i, and n and m represent the number of industries and countries, respectively. The location entropy index reflects the relative level of industry agglomeration within a certain region; the larger the location entropy index, the higher is the degree of industry agglomeration.

Green Economy Efficiency (GTFP). This study applies the Malmquist-Lenberger (ML) index method based on the Slacks-Based Measure (SBM) directional distance function to measure total factor productivity, which not only considers the impact of input and output slack on productivity but also eliminates the need to choose the measurement angle and form of the production function. Given data availability, the following input–output indicators are defined: (a) output indicators. In this study, we use the total output value obtained by summing the output value of the sectors at current prices in each country as the desired output indicator, which includes both the transfer value of intermediate goods and the value newly created in the production process and is more reasonable than the value-added and net output indicators. (b) Non-desired output indicators. Greenhouse gas emissions, which exacerbate global climate change and air pollution, are among the most significant manifestations of environmental pollution worldwide. Considering data availability, this study selected the indicator of greenhouse gas emissions to reflect unexpected outputs in the national economic development process. (c) Input indicators. Capital input: This study considers the capital stock of current PPPs as a capital input indicator. Labor input: Owing to the lack of relevant data on labor time and the education level of workers, indicators of labor quality and efficiency are not available. Therefore, this study summed up the labor force in every industry as an indicator of labor input. Energy input: Because the energy-use data in the Eora database for some years were identical, this study chose the energy footprint as an indicator of energy input for data accuracy. Land input: Cropland is an important land use type. Therefore, this study used the total cropland area to indicate land input.

Control variables

Aging population (aging). An aging population increases the cost of labor and capital factors. However, aging positively promotes the export of capital-intensive goods and causes capital-intensive industries to gradually replace labor-intensive industries as a comparative advantage of exports only when aging exceeds the corresponding inflection point. The export competitiveness of technology-intensive products in countries with a greater degree continues to decline.⁴⁵ This reflects the complex impact of population aging on upgrading the trade structure. Therefore, the influence of an aging population on the implied loss of biodiversity in export trade and the expected sign is also uncertain.

Economic development level (GDP). Generally, the higher the level of economic development, the higher the country's demand for sustainable and green development guided by government policies. The greater the importance of protecting the ecological environment, including biological species, the smaller the negative externalities of industrial activities,⁴⁶ and the smaller the diversity of species consumed by export trade products. We used real GDP per capita to measure the level of economic development with an expected negative sign.

Rate of value-added (value). The value-added rate reflects the country's position in the global industrial chain. A low value-added rate indicates the country is at the lower end of the international division of labor with low-technology products. Currently, most industries are low value-added, such as the primary processing of agricultural and mineral products. These industries are less environmentally friendly, and their industrial activities add to the environmental burden⁴⁷ and amplify the embodied loss of species diversity in trade. We measure the value-added rate in terms of value-added/total output, with an expected negative sign

Degree of openness to the outside world (open). In the present study, we examined the loss of species diversity in international trade. Embodied biodiversity loss in the export trade, as characterized by the biodiversity footprint, reflects the impact of foreign markets on species diversity. The greater the degree of openness to the outside world, the greater the final consumption demand from abroad⁴⁸ and the greater the loss of species diversity transferred from other countries. We measured the degree of openness by the proportion of total exports and imports to GDP, with the expected positive sign.

Data

This study selected the data for 148 countries between 2006 and 2020 from the IUCN Red List of Threatened Species, Eora database, WIOD database, World Trade Organization, and World Bank. Descriptive statistics for the explanatory, core explanatory, and control variables are presented in Table 1.

Table 1.

Descriptive statistics of variables.

Variables	Unit	Observations	Average value	s.d.	Maximum value	Minimum value	Prediction symbol for the impact on bf (gtfp)
bf	species/$1	2220	3.667	23.652	419.639	0.024	/
aggl	proportion	2220	1.023	0.173	1.466	0.551	+ ( − )
gtfp	proportion	2220	0.293	0.257	1.924	0.002	−
aging	percentage	2220	0.083	0.058	0.273	0.002	?
gdp	ten thousand dollars/per person	2220	1.531	2.073	12.368	0.017	−
value	percentage	2220	0.46	0.06	0.58	0.28	−
open	percentage	2220	0.516	0.338	8.362	0.015	+

Analysis and results

Benchmark test

Before performing the benchmark tests, we conducted a variance inflation factor (VIF) test to prove the absence of multicollinearity. Table 2 presents the results of the study.

Table 2.

VIF test results

Variables	VIF	1/VIF
Ln(aggl)	1.09	0.92
Ln(gtfp)	1.62	0.62
Ln(aging)	1.65	0.61
Ln(gdp)	2.42	0.41
Ln(value)	1.14	0.87
Ln(open)	1.21	0.82
Mean VIF	1.52

VIF was used to test whether the model had multicollinearity. The larger the VIF, the more serious the multicollinearity issue. When the maximum VIF does not exceed 10, the empirical rule does not consider multicollinearity issues in the model. The results in Table 2 show that the VIF of each variable in the model was less than 10, and the maximum VIF was 2.42. This indicates that there was no multicollinearity.

To minimize model endogeneity and make the regression results realistic and reliable, this study used the two-stage least squares (2SLS) model⁴⁹ for benchmark hypothesis testing to verify the impact of industrial agglomeration on the biodiversity footprint in export trade. The results are shown in Table 3.

Table 3.

Benchmark tests.

Variables	1 stageLn(aggl)(1)	2 stageLn(bf)(2)	1 stageLn(aggl)(3)	2 stageLn(bf)(4)	1 stageLn(aggl)(5)	2 stageLn(bf)(6)	1 stageLn(aggl)(7)	2 stageLn(bf)(8)
Ln(aggl)		1.259***(0.3756)		1.362***(0.3762)		1.957***(0.1896)		1.860***(0.2024)
Ln(aggl_t−1)	0.994***(0.0365)		0.998***(0.0373)		0.983***(0.0362)		0.947***(0.0401)
Ln(gdp)	0.010***(0.0024)	−0.297***(0.0339)	0.011***(0.0024)	−0.291***(0.0321)	0.010***(0.0024)	−0.289***(0.0270)	0.010***(0.0023)	−0.289***(0.0276)
Ln(open)			0.006(0.0042)	0.158***(0.0309)	0.007***(0.0040)	0.118***(0.0278)	0.005(0.004)	0.113***(0.0281)
Ln(addv)					0.006***(0.0043)	−0.234***(0.0190)	0.006(0.004)	−0.232***(0.0196)
Ln(aging)							0.036(0.009)	0.095*(0.0540)
N	2220	2220	2220	2220	2220	2220	2220	2220
F	739.75	44.66	714.08	49.28	736.33	85.70	558.55	67.71
Kleibergen–Paap rk LM	67.230		68.254		108.017		85.767
Cragg–Donald Wald F	3143.151		3125.109		2905.447		2442.359

In the regression of industrial agglomeration and biodiversity footprint in export trade, control variables such as level of economic development, degree of openness to the outside world, value-added rate and aging population were gradually introduced. From the regression results, the coefficient of industrial agglomeration is significantly positive, indicating that industrial agglomeration aggravates the embodied biodiversity loss in export trade, and Hypothesis 1 is confirmed. It is generally believed that there is a nonlinear relationship between industrial agglomeration and environmental efficiency in past works,⁵⁰ while the conclusion of this paper is different.

The degree of openness to the outside world has a significantly positive coefficient for embodied biodiversity loss in exports, which is consistent with expectations. The economic development level and rate of value-added all have a significantly negative effect on the embodied biodiversity loss in export trade, which is consistent with expectations. The aging population has a significant positive coefficient on embodied biodiversity loss in export trade. This means that with the increase in the aging population, biodiversity loss driven by international trade intensifies. This could be because an aging population leads to a decline in the labor participation rate and productivity. At the same time, the reversal effect brought about by rising labor costs is small, making it difficult to realize the transformation of the trade structure and upgrade the global value chain status. The results show that an aging population exacerbates trade-driven biodiversity loss.

Test for mediating effects

To determine the relationship between industrial agglomeration, green economy efficiency, and embodied biodiversity loss in export trade and to verify the mediating effect of green economy efficiency in the process of industrial agglomeration affecting embodied biodiversity loss in export trade, we continued to use the two-stage least squares method to conduct an empirical test, the results of which are shown in Table 4.

Table 4.

Intermediary effects test.

Variables	Ln(bf)(1)	Ln(gtfp)(2)	Ln(bf)(3)
Ln(aggl)	1.860***(0.2024)	−0.757*(0.4454)	1.761***(0.1892)
Ln(gtfp)			−0.130***(0.0246)
Ln(gdp)	−0.289***(0.0276)	0.292***(0.0470)	−0.251***(0.0259)
Ln(open)	0.113***(0.0281)	0.062(0.0561)	0.121***(0.0267)
Ln(value)	−0.232***(0.0196)	0.084(0.0750)	−0.221***(0.0146)
Ln(aging)	0.095*(0.0540)	0.365***(0.0998)	0.143***(0.0544)
N	2220	2220	2220
F	67.71	19.44	90.51
Kleibergen–Paap rk LM	85.767	85.767	88.406
Cragg–Donald Wald F	2442.359	2442.359	2460.959

In Model (1) of Table 4, the coefficient of industrial agglomeration is 1.860, which is significantly positive at the 1% level, indicating that industrial agglomeration exacerbates embodied biodiversity loss in export trade, where 1.860 represents the total effect of industrial agglomeration on embodied biodiversity loss in export trade. Thus, Hypothesis 1 is confirmed. In Model (2), the coefficient of industrial agglomeration is −0.757, which is significantly negative at 10% confidence level, indicating that industrial agglomeration negatively contributes to green economy efficiency. Hypothesis 2 is confirmed at a 90% confidence interval.

In Model (3), industrial agglomeration and green economy efficiency jointly contribute to embodied biodiversity loss in export trade. At this point, the coefficient of industrial agglomeration was 1.761, which was significantly positive at the 1% level, indicating a direct effect of industrial agglomeration on embodied biodiversity loss in export trade. The coefficient of green economic efficiency is −0.130, which is significantly negative at the 1% level. Calculations based on the regression results of Models (2) and (3) show that the mediating effect of industrial agglomeration on embodied biodiversity loss in export trade is 0.098, implying that green economic efficiency plays a partial mediating role in examining the impact of industrial agglomeration on embodied biodiversity loss in export trade. Thus, Hypothesis 3 was confirmed. This result differs from past works, which believe that agglomeration will indirectly reduce environmental costs through the mediating effect of green economy efficiency.⁵¹ This may be primarily due to the high degree of industrial agglomeration, which leads to a negative externality of scale exceeding the Marshallian and Jacobs positive externalities.

Robustness test

This study conducts a robustness test by replacing the independent variable with the relative specialization agglomeration index (MAR) and the relative diversity agglomeration index (JAC). The indices were calculated as follows:

M A R_{i} = m a x_{j} (\frac{f_{i j}}{f_{j}})

J A C_{i} = \frac{1}{\sum_{j} | f_{i j} - f_{j} |}

The agglomeration indices obtained from the above calculations were introduced into Equations (1)–(3) for the regression analysis, and the results are shown in Table 5.

Table 5.

Robustness tests.

Variables	Ln(bf)(1)	Ln(gtfp)(2)	Ln(bf)(3)	Ln(bf)(4)	Ln(gtfp)(5)	Ln(bf)(6)
Ln(MAR)	3.287***(0.7057)	−4.647***(1.3272)	2.706***(0.6667)
Ln(JAC)				1.695***(0.3156)	−1.748***(0.6059)	1.432***(0.2844)
Ln(gtfp)			−0.125***(0.0280)			−0.150***(0.0281)
Ln(gdp)	−0.315***(0.0445)	0.391***(0.0761)	−0.266***(0.0410)	−0.253***(0.0348)	0.288***(0.0570)	−0.210***(0.0303)
Ln(open)	0.127***(0.0360)	0.022(0.0801)	0.130***(0.0330)	0.053*(0.0322)	0.125*(0.0642)	0.072**(0.0302)
Ln(value)	−0.267***(0.0258)	0.170**(0.0672)	−0.246***(0.0231)	−0.243***(0.0357)	0.117(0.0890)	−0.226***(0.0248)
Ln(aging)	−0.172(0.1138)	1.054***(0.3166)	−0.040(0.1061)	−0.058(0.0772)	0.740**(0.2167)	0.053(00.0730)
N	2220	2220	2220	2220	2220	2220
F	38.34	6.85	49.85	29.48	8.77	41.77
Kleibergen–Paap rk LM	36.748	36.748	35.819	67.118	67.118	69.417
Cragg–Donald Wald F	48.522	48.522	48.754	133.378	133.378	144.283

Models (1) and (4) in Table 5 show that industrial agglomeration is positively related to embodied biodiversity loss in export trade, which is consistent with the results of the benchmark regression, and Hypothesis 1 is again confirmed. Models (2) and (5) show that industrial agglomeration plays a significantly negative role in green economy efficiency, with no change in the sign of the coefficients, again confirming Hypothesis 2. Models (3) and (6) examine the joint effect of industrial agglomeration and green economy efficiency on the embodied biodiversity loss in export trade and show that industrial agglomeration exacerbates the net embodied biodiversity loss in export trade, while green economy efficiency mitigates the embodied biodiversity loss in export trade. The observations remain robust, and the mediating effect of green economy efficiency in the effect of industrial agglomeration on the implied biodiversity loss in export trade is significant, and Hypothesis 3 is again confirmed. The regression results for the control variables are generally consistent with those above. The regression results are robust and reliable.

Heterogeneity test

Country heterogeneity

Different countries have different levels of economic development and, correspondingly, different degrees of industrial agglomeration and green economy efficiency, which may affect the complex relationship among industrial agglomeration, green economy efficiency, and embodied biodiversity loss in export trade. To better explore the country heterogeneity of the relationship between industrial agglomeration, green economy efficiency, and embodied biodiversity loss in export trade, we divided 148 countries into developed and developing countries for empirical analysis based on the list of developed countries released by the United Nations at the end of 2023, of which 36 were developed countries, and 112 were developing countries. The regression results are presented in Table 6.

Table 6.

Tests of mediating effects by country.

Variables	Developed countries			Developing countries
Variables	Ln(bf)(1)	Ln(gtfp)(2)	Ln(bf)(3)	Ln(bf)(4)	Ln(gtfp)(5)	Ln(bf)(6)
Ln(aggl)	3.118***(0.5956)	1.584**(0.6854)	3.310***(0.5636)	1.461***(0.2166)	−0.978*(0.5065)	1.317***(0.1970)
Ln(gtfp)			−0.121***(0.0301)			−0.147***(0.0303)
Ln(gdp)	−0.408***(0.0437)	0.931***(0.0850)	−0.295***(0.0450)	−0.260***(0.0273)	0.261***(0.0498)	−0.222***(0.0251)
Ln(open)	0.262***(0.0582)	0.283***(0.1130)	0.296***(0.0568)	0.065**(0.0290)	0.008(0.0626)	0.066**(0.0277)
Ln(value)	0.060(0.1336)	−0.378(0.2454)	0.014(0.1354)	−0.227***(0.0213)	0.091(0.0740)	−0.214***(0.0143)
Ln(aging)	0.182(0.2103)	−0.018(0.2746)	0.180(0.2049)	0.001(0.0443)	0.285***(0.0919)	0.043(0.0412)
N	540	540	540	1680	1680	1680
F	55.70	30.21	49.203	47.72	13.71	76.81
Kleibergen–Paap rk LM	47.132	47.132	49.203	68.015	68.015	71.083
Cragg–Donald Wald F	410.112	410.112	399.869	1736.116	1736.116	1744.549

As shown in Table 6, the coefficient of industrial agglomeration is significantly positive in both developed and developing countries, indicating that industrial agglomeration exacerbates embodied biodiversity loss in export trade. Thus, Hypothesis 1 was verified. The effect of green economy efficiency on embodied biodiversity loss in export trade is negative for both developed and developing countries, indicating that green economy efficiency mitigates embodied biodiversity loss in export trade. Thus, Hypothesis 2 was verified.

In contrast, the dampening effect of green economy efficiency on embodied biodiversity loss in exports is greater in developing countries than in developed countries. The high position of developed countries in the international division of the labor value chain determines that developing countries will bear a large part of the material production of goods consumed by developed countries and become the destination for transferring energy consumption and species diversity losses from developed countries. According to the measurement results, the embodied biodiversity loss in the export trade of developing countries was much higher than that of developed countries. Their green economy efficiency started at a lower level. Therefore, there is room for improvement in green economic efficiency and a reduction in biodiversity loss. The marginal effects of an improvement in green economy efficiency are more significant. With the same degree of improvement in green economy efficiency, developing countries can save more resources with greater production efficiency, energy savings, emissions reductions, and environmental improvement effects. Meanwhile, they can enhance their embedded position in global value chain to a greater extent and improve the structure of trade products more effectively to reduce the direct or indirect negative impact of export industrial activities on species and better suppress the embodied biodiversity loss in export trade.

The only difference between the empirical results of country heterogeneity and the analysis above is the influence of industrial agglomeration on green economy efficiency. The coefficient of industrial agglomeration is significantly positive in Model (2) of Table 6, whereas the opposite is true in Model (5). This means that industrial agglomeration in developed countries can promote the improvement of green economy efficiency, whereas industrial agglomeration in developing countries has a negative impact on green economy efficiency. This result can be explained as follows: Industries in developed countries are primarily capital- and technology-intensive. Compared to other industries, they have higher production efficiency and smaller negative environmental externalities. The green technological progress brought about by economies of scale, knowledge spillover, and innovation link effects in agglomeration is sufficient to offset the reduction in green technology efficiency caused by the congestion effect. Industries in developing countries are predominantly labor- and resource-intensive. Their industrial agglomeration is more likely to reduce green technological efficiency and less likely to promote green technological progress. Consequently, industrial agglomeration negatively affects green economic efficiency. The other control variables were generally consistent with the results of the above tests.

Region heterogeneity

Industrial agglomerations must have specific locations. Regional differences imply that heterogeneous locations attract different types of industrial agglomerations. This means that the externalities of industrial agglomeration may vary among regions, leading to different effects of industrial agglomeration on green economy efficiency and species. This study divides the world into five major regions—Asia, Europe, Africa, North America, and South America—to determine the regional characteristics of the relationship between industrial agglomeration, green economy efficiency, and embodied biodiversity losses in export trade. Owing to data limitations, there were only three Oceanian countries. However, their sample size was small. Consequently, data from Australia, Fiji, and New Zealand were excluded, leaving 145 countries, as shown in Table 7. The regression results are presented in Table 8.

Table 7.

Region classification.

Regions	Countries
Asia	Armenia, Azerbaijan, Bahrain, Bangladesh, Bhutan, Brunei Darussalam, Cambodia, China, Cyprus, Georgia, Hong Kong, India, Indonesia, Iran (Islamic Republic of), Iraq, Israel, Japan, Jordan, Kazakhstan, Korea, Kuwait, Kyrgyzstan, Lao People's DR, Lebanon, Malaysia, Macao, Mongolia, Nepal, Oman, Pakistan, Philippines, Qatar, Saudi Arabia, Singapore, Syrian Arab Republic, Tajikistan, Thailand, Turkey, Turkmenistan, UAE, Uzbekistan, Viet Nam
Africa	Algeria, Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cape Verde, Central, African Republic, Chad, Congo, Côte d’Ivoire, DR Congo, Egypt, Eswatini, Gabon, Gambia, Ghana, Guinea, Kenya, Lesotho, Madagascar, Mali, Mauritania, Mauritius, Morocco, Mozambique, Namibia, Niger, Nigeria, Rwanda, Senegal, Sierra Leone, South Africa, Tanzania, Togo, Tunisia, Uganda, Zambia
Europe	Albania, Austria, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Moldova, Montenegro, Netherlands, North Macedonia, Norway, Poland, Portugal, Romania, Russian Federation, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, UK, Ukraine
North America	Bahamas, Barbados, Belize, Canada, Costa Rica, Dominican Republic, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, USA
South America	Argentina, Bolivia (Plurinational State of), Brazil, Chile, Colombia, Ecuador, Paraguay, Peru, Uruguay, Venezuela (Bolivarian Republic of)

Table 8.

Test results for industrial agglomeration, green economy efficiency and embodied biodiversity loss in export trade by region.

Variables	Asia	Africa	Europe	North America	South America
Ln(aggl)	0.941**(0.2737)	0.664***(0.2303)	3.148***(0.4587)	3.223***(0.7799)	2.065***(0.6067)
Ln(gtfp)	−0.173***(0.0304)	−0.070*(0.0410)	−0.078(0.0519)	−0.013(0.0193)	−0.333***(0.0893)
Ln(gdp)	−0.174***(0.0191)	−0.319***(0.0371)	−0.266***(0.0635)	−0.398***(0.0641)	−0.287***(0.0571)
Ln(open)	0.197***(0.0390)	−0.048(0.0360)	0.077(0.0755)	0.042(0.0796)	0.351***(0.0794)
Ln(value)	0.411**(0.1757)	−0.256***(0.0902)	−0.270***(0.0219)	−0.931***(0.2172)	−0.213(0.1612)
Ln(aging)	−0.003(0.0344)	0.452***(0.1326)	0.411**(0.1846)	0.321(0.3580)	0.314(0.1978)
N	630	585	585	225	150
F	36.98	26.69	76.02	19.57	31.12
Kleibergen–Paap rk LM	63.705	27.016	18.426	16.891	13.779
Cragg–Donald Wald F	699.601	849.829	234.621	120.960	152.312

Based on the regression results, the coefficient of industrial agglomeration is significantly positive, regardless of region. In other words, industrial agglomeration always exacerbates the loss of embodied biodiversity in export trade. The coefficients of green economy efficiency in Asia and Africa are significantly negative, consistent with the above results. The coefficients of green economy efficiency in Europe and America are negative but insignificant. Countries in Europe are generally at higher latitudes and have a relative lack of biodiversity compared with countries at lower latitudes. A small biodiversity base results in a less sensitive response to the indirect effects of green economic efficiency. Countries in Europe and North America are usually economically developed and at the high end of the global value chain. They mainly choose to transfer high-consumption, high-pollution industries to other countries to pass on the environmental costs. As a result, their embodied biodiversity loss in export trade is already much smaller than that of other regions. Even if improving green economy efficiency can reduce negative environmental externalities and upgrade the global value chain, its indirect impact on the embodied biodiversity loss in the export trade is still insignificant.

The regression results for the control variables were generally consistent with those above, except for the value-added rate variable. Interestingly, the coefficient of the value-added rate is significantly positive in Asia, which differs from other regions. This finding was only observed in China in Liu's⁵² study, which extends its application to Asia. This may be caused by Asia's special international division of labor status. Asian countries have long participated in the global value chain by transferring foreign-polluting industries. Although they gain economic benefits and increase the value-added rate, they also bear the environmental costs of unclean growth, including biodiversity loss.

Conclusions and discussion

The scientific value of this study is as follows: First, it referred to the carbon footprint method to innovatively calculate the biodiversity footprint and uses it to measure the embodied biodiversity loss in export trade. Second, this study used economic principles to explore the mechanisms of trade-driven biodiversity loss. A panel of 148 countries from 2006 to 2020 was selected to explore the effects of industrial agglomeration and green economy efficiency on embodied biodiversity loss in export trade using the 2SLS method to examine whether green economy efficiency can act as a mediating variable of industrial agglomeration on embodied biodiversity loss in export trade and to further investigate the mechanisms of their effects. Third, the study examined the effects of industrial agglomeration and green economy efficiency on biodiversity loss by country and region, which may provide targeted recommendations for countries to conserve biodiversity from an economic perspective.

Conclusions

We draw four main conclusions. First, industrial agglomeration can aggravate embodied biodiversity loss in exports, and there is a significant positive relationship between the two factors. Second, green economy efficiency can mitigate the embodied biodiversity loss in export trade, and there is a significant negative relationship between the two factors. Third, industrial agglomeration plays a negative role in promoting green economic efficiency, and there is a significant negative relationship between these two factors. Green economy efficiency can act as a mediating variable for industrial agglomeration, affecting the loss of embodied biodiversity in the export trade. Fourth, there is some country and regional heterogeneity in the relationship between industrial agglomeration, green economic efficiency, and embodied biodiversity loss in the export trade. Industrial agglomeration in developed countries promotes green economic efficiency, whereas the opposite is true in developing countries. Green economy efficiency can significantly mitigate biodiversity losses from trade in Asia, Africa, and South America. However, it is not effective in Europe or North America.

Implications

Based on the above observations, this study makes several recommendations. First, countries should avoid excessive industrial agglomeration and select an appropriate scale of industrial agglomeration based on the actual situation. Governments must play a guiding role for countries that receive industrial transfers and not overly pursue industrial agglomeration, which should be limited by local resources and environmental carrying capacities. They should selectively guide industrial agglomerations through screening, strengthen the planning and management of industrial agglomeration areas, and encourage the development of diversified industries. Moreover, investment in environmental protection can be increased and strict environmental protection standards can be enforced to limit pollutant emissions by enterprises in areas with serious agglomeration. Strengthening social supervision, guiding enterprises to operate in a compliant manner, and implementing corporate social responsibility are also effect measures to reduce the negative impact of excessive industrial agglomeration on biodiversity

Second, the government should increase preferential policy, improve investment in green technology innovation, provide policy support for enterprises or scientific research institutions that carry out basic technological research, and provide corresponding rewards to enterprises that make key technological breakthroughs to encourage the transformation of technological progress paths to green and improve green economy efficiency. Simultaneously, laws and regulations should be introduced to improve infrastructure, increase investment in higher education, and provide talent reserves and support for technological innovation to achieve green technological progress. As the main carriers of improving green economy efficiency, enterprises should take the initiative to eliminate backward production methods with high energy consumption and pollution and consciously shift some enterprise profits from expanding production scale to technological research. The leading role of model enterprises is to encourage enterprises with strong technological progress capabilities to provide technological transformation programs to other enterprises. These measures can promote green production, energy conservation, and emission reduction in enterprises with significant negative environmental externalities and improve green economy efficiency to reduce embodied biodiversity loss in export trade.

Finally, countries could adopt different policies for industrial agglomeration and green economic efficiency. Thus, reducing the scale of industrial agglomeration is the best strategy for developing countries. Considering the positive effect of industrial agglomeration on green economy efficiency, developed countries should maintain an appropriate scale of industrial agglomeration to effectively exert the effects of economies of scale, knowledge spillovers, innovation links, and shared facilities. In this way, they can stimulate technological innovation, improve production efficiency, and reduce pollution, thereby promoting green economic efficiency and protecting biological habitats to protect biodiversity. Compared with countries in Europe and North America, countries in Asia, Africa, and South America should focus more on improving green economy efficiency and fully play its significantly positive effect on reducing biodiversity loss.

Limitations and discussion

Our study has several limitations. First, because the characteristics of industries differ, the impact of different factors on biodiversity loss may have industry heterogeneity. However, industrial data were lacking. Therefore, we do not conduct an industry heterogeneity study. Relevant industry data searches must be developed further. Second, research on biodiversity from a trade perspective is still in its infancy. Since the Economics of Ecosystems and Biodiversity are different from Environmental Economics, the method that adopted the theoretical knowledge of environmental costs, such as energy footprint, water footprint, and carbon footprint, when expounding the theoretical mechanism lacks sufficient scientific knowledge. Further refinement of this theory will be performed in future studies. Finally, the position and degree of embedding of different industries in the value chain are closely related to biodiversity loss. Further incorporation of the GVC into the model and discussion of its effect on biodiversity loss are urgent issues to be studied in-depth in biodiversity exploration.

Footnotes

Acknowledgment

All authors read and approved the final manuscript.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Humanities and Social Science Planning Projects of the Ministry of Education in 2022 China Grant (22YJA790058).

ORCID iD

Shuhong Wang

Lu Xing is pursuing her Doctor degree of Economics from Ocean University of China. Her current research interests include biodiversity and environmental management. Her analysis relies on the development and use of input output model, econometrics model and methods on compositional data.

Shuhong Wang is a professor in Shandong University of Finance and Economics. He obtained PH.D. degree of Applied Economics from Nankai University in 2014. His current research interests are Environmental Economics and Environmental Management. He has published about 80 refereed papers in high-ranked journals such as Omega, Journal of the Operational Research Society, and Journal of Environmental Management.

Appendix

See Table A1.

References

Bar-on

Phillips

Milo

. The biomass distribution on earth. Proc Natl Acad Sci USA 2018; 115: 6506–6511.

Ceballos

Ehrlich

Raven

. Vertebrates on the brink as indicators of biological annihilation and the sixth mass extinction. Proc Natl Acad Sci USA 2020; 117: 13596–13602.

Lenzen

Moran

Kanemoto

, et al. International trade drives biodiversity threats in developing nations. Nature 2012; 486: 109–112.

Sun

Chen

Zhang

. Exploring the trading embodied CO2 effect and low-carbon globalization from the international division perspective. Environ Impact Asses 2020; 83: 106414.

Kok

Oostvogels

de Olde

, et al. Balancing biodiversity and agriculture: conservation scenarios for the Dutch dairy sector. Agric Ecosyst Environ 2020; 302: 107103.

Greenaway

Kneller

. Exporting, productivity and agglomeration. Eur Econ Rev 2008; 52: 919–939.

Melitz

. The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 2003; 71: 1695–1725.

Grazi

van den Bergh

JCJM

Rietveld

. Spatial welfare economics versus ecological footprint: modeling agglomeration, externalities and trade. Environ and Resour Econ 2007; 38: 135–153.

Wang

Chen

. Global value chains and carbon emission reduction in developing countries: does industrial upgrading matter? Environ Impact Asses 2022; 97: 106895.

10.

Cheng

Jin

. Agglomeration economy and the growth of green total-factor productivity in Chinese industry. Socio Econ Plann Sci 2020; 83: 101003.

11.

Wang

Tang

, et al. Export trade, embodied carbon emissions, and environmental pollution: an empirical analysis of China’s high-and new-technology industries. J Environ Manage 2020; 276: 111371.

12.

Hou

Wang

Yuan

, et al. Estimating the mitigation potential of the Chinese service sector using embodied carbon emissions accounting. Environ Impact Asses 2021; 86: 106510.

13.

Chen

Tan

Liu

. Direct and embodied energy-water-carbon nexus at an inter-regional scale. Appl Energy 2019; 251: 113401.

14.

Zabel

Delzeit

Schneider

, et al. Global impacts of future cropland expansion and intensification on agricultural markets and biodiversity. Nat Commun 2019; 10: 2844.

15.

Chaudhary

Brooks

. National consumption and global trade impacts on biodiversity. World Dev 2019; 121: 178–187.

16.

Arneth

Shin

Leadley

, et al. Post-2020 biodiversity targets need to embrace climate change. Proc Natl Acad Sci USA 2020; 117: 30882–30891.

17.

Gray

TNE

Hughes

Laurance

, et al. The wildlife snaring crisis: an insidious and pervasive threat to biodiversity in Southeast Asia. Biodivers Conserv 2018; 27: 1031–1037.

18.

Moran

Kanemoto

. Identifying species threat hotspots from global supply chains. Nat Ecol Evol 2017; 1: 0023.

19.

Wang

Liu

, et al. Per-capita carbon emissions in 147 countries: the effect of economic, energy, social, and trade structural changes. Sustain Prod Consump 2021; 27: 1149–1164.

20.

Wang

. Does urbanization redefine the environmental Kuznets curve? An empirical analysis of 134 countries. Sust Cities Soc 2021; 76: 103382.

21.

Wang

Yang

. Does income inequality reshape the environmental Kuznets curve (EKC) hypothesis? A nonlinear panel data analysis. Environ Res 2022; 216: 114575.

22.

Wang

. Trade protectionism jeopardizes carbon neutrality-decoupling and breakpoints roles of trade openness. Sustain Prod Consump 2023; 35: 201–215.

23.

Wang

Zhang

. Revisiting the environmental kuznets curve hypothesis in 208 counties: the roles of trade openness, human capital, renewable energy and natural resource rent. Environ Res 2023; 216: 114637.

24.

Rybicki

Hanski

. Species-area relationships and extinctions caused by habitat loss and fragmentation. Ecol Lett 2013; 16: 27–38.

25.

Martin

Rogers

. Industrial location and public infrastructure. J Int Econ 1995; 39: 335–351.

26.

Xie

Zhang

. The spatial technology spillovers effect of Chinese manufacturing industries’ agglomeration: the difference of institutional environment. Geogr Res 2016; 35: 909–928.

27.

You

Hong

, et al. The impact of industrial agglomeration on ecological efficiency: an empirical analysis based on 244 Chinese cities. Environ Impact Asses 2022; 96: 106841.

28.

Urban

. Accelerating extinction risk from climate change. Science 2015; 348: 571–573.

29.

Liu

Zhu

. The impact of industrial agglomeration on industrial pollutant emission: evidence from China under new normal. Clean Technol Environ Policy 2017; 19: 2327–2334.

30.

Cheng

Zhu

, et al. Measuring CO2 emissions performance of China’s construction industry: a global Malmquist index analysis. Environ Impact Asses 2022; 92: 106673.

31.

Wang

Zhou

. Spatial agglomeration and driving factors of environmental pollution: a spatial analysis. J Cleaner Prod 2021; 279: 123839.

32.

Tokimatsu

Konishi

Ishihara

, et al. Role of innovative technologies under the global zero emissions scenarios. Appl Energy 2016; 162: 1483–1493.

33.

Wang

Feng

. Revealing the pattern and evolution of global green development between different income groups: a global meta-frontier by-production technology approach. Environ Impact Asses 2021; 89: 106600.

34.

Zhu

Zhang

. The impact of cross-region industrial structure optimization on economy, carbon emissions and energy consumption: a case of the Yangtze river Delta. Sci Total Environ 2021; 778: 146089.

35.

Ryan

Buciuni

Giblin

, et al. Subsidiary upgrading and global value chain governance in the multinational enterprise. Glob Strateg J 2020; 10: 496–519.

36.

Zeng

Huang

. Industrial collaborative agglomeration, marketization, and green innovation: evidence from China’s provincial panel data. J Cleaner Prod 2021; 279: 123598.

37.

Soltan

van Dijk

. Palestinian Clusters: from agglomeration to innovation. Eur Sc J 2017; 13: 323–336.

38.

Tao

Luo

, et al. Agglomeration economies in creative industries. Reg Sci Urban Econ 2019; 77: 141–154.

39.

Glaeser

Kallal

Scheinkman

, et al. Growth in cities. J Pol Econ 1992; 100: 1126–1152.

40.

Lin

Yang

. Agglomeration and productivity: firm-level evidence from China’s textile industry. China Econ Rev 2011; 22: 313–329.

41.

. The impact of eco-environmental regulation on green energy efficiency in China - Based on spatial economic analysis. Energy Environ 2022. doi:10.1177/0958305X211072435.

42.

Liu

Wang

, et al. Can industrial agglomeration improve regional green total factor productivity in China? An empirical analysis based on spatial econometrics. Growth Change 2021; 52: 1011–1039.

43.

Baron

Kenny

. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986; 51: 1173–1182.

44.

Leontief

. Environmental repercussions and the economic structure: an input-output approach. Rev Econ Stat 1970; 52: 262–271.

45.

Cai

Stoyanov

. Population aging and comparative advantage. J Int Econ 2016; 102: 1–21.

46.

Zhao

Liu

. Carbon emission intensity embodied in trade and its driving factors from the perspective of global value chain. Environ Sci Pollut Res 2020; 27: 32062–32075.

47.

Liu

Lackner

Fan

. Value-added involved in CO2 emissions embodied in global demand-supply chains. J Environ Plan Manag 2020; 64: 76–100.

48.

Islam

Kanemoto

Managi

. Impact of trade openness and sector trade on embodied greenhouse gases emissions and air pollutants. J Ind Ecol 2016; 20: 494–505.

49.

Oczkowski

. Discrimination between measurement scales using non-nested tests and 2SLS: Monte Carlo evidence. Struct Equ Modeling 2002; 9: 103–125.

50.

Shen

Peng

. Can industrial agglomeration achieve the emission-reduction effect? Socio-Econ Plan Sci 2021; 75: 100867.

51.

Wang

Dong

. How does ICT agglomeration affect carbon emissions? The case of Yangtze river Delta urban agglomeration in China. Energy Econ 2022; 111: 106107.

52.

Liu

Zong

Hynes

, et al.

Can China reduce the carbon emissions of its manufacturing exports by moving up the global value chain?

Res Int Bus Financ 2019; 51: 101101.