Asymmetric Effects of Economic Development, Agroforestry Development, Energy Consumption, and Population Size on CO₂ Emissions in China

Abstract

The COVID-19 epidemic and the Russian–Ukrainian conflict have led to a global food and energy crisis, making the world aware of the importance of agroforestry development for a country. Modern agriculture mechanization leads to massive energy consumption and increased CO${}_2$ emissions. At the same time, China is facing serious demographic problems and a lack of consumption in the domestic market. The Chinese government is faced with the dilemma of balancing environmental protection with economic development in the context of the “double carbon” strategy. This article uses annual World Bank statistics from 1990 to 2020 to study the asymmetric relationships between agroforestry development, energy consumption, population size, and economic development on CO${}_2$ emissions in China using the partial least squares path model (PLS-PM), the autoregressive VAR vector time series model, and the Granger causality test. The results are as follows: (1) The relationship between economic development and carbon dioxide emissions, agroforestry development and carbon dioxide emissions, energy consumption and carbon dioxide emissions, and population size and carbon dioxide emissions are both direct and indirect, with an overall significant positive effect. There is a direct negative relationship between population size and carbon dioxide emissions. (2) The results of the Granger causality test show that economic development, energy consumption, and CO${}_2$ emissions are the causes of the development of agroforestry; economic development, agroforestry development, population size, and CO${}_2$ emissions are the causes of energy consumption; energy consumption is the cause of economic development and CO${}_2$ emissions; and agroforestry development is the cause of population size and energy consumption. (3) In the next three years, China’s agroforestry development will be influenced by the impulse response of economic development, energy consumption, and CO${}_2$ emission factors, showing a decreasing development trend. China’s energy consumption will be influenced by the impulse response of economic development, agroforestry development, population size, and CO${}_2$ emission factors, showing a decreasing development trend, followed by an increasing development trend. China’s CO${}_2$ emission will be influenced by the impulse response of energy consumption and agroforestry development. China’s CO${}_2$ emissions will be influenced by the impulse response of energy consumption and agroforestry development factors, showing a downward and then an upward development trend.

1. Introduction

The United Nations Declaration on Sustainable Development Goals (SDGs) [1] recognizes the importance of improving the quality of the global environment. These goals aim to achieve environmental sustainability, alongside global economic and social development, by 2030 [2,3]. As a result, the implementation of energy efficiency and emission reduction policies has received attention from ecologists, policymakers, and economists around the world. Climate change and the energy crisis are major environmental issues of great concern worldwide [4,5]. The State of Food Security and Nutrition in the World 2021 Report [6] states that an estimated 720–811 million people worldwide will face hunger in 2020 and that the world will not meet the goal of eradicating hunger and all forms of malnutrition by 2030 due to a combination of major factors. The challenges we face are becoming more acute. These include conflicts, climate change, and extreme events [6], as well as economic slowdowns and recessions, all exacerbated by the underlying causes of poverty and by very high and persistent levels of inequality. Furthermore, the Russian–Ukrainian conflict has led to a global food and energy crisis, making the world aware of the importance of agroforestry development. China is striving to reach peak carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060 [7,8]. At the same time, China faces serious population structure problems, and domestic market consumption is weak. Under the strategic goal of “double carbon”, the Chinese government has to give attention to environmental protection and economic development and is caught in a dilemma. Therefore, it is very important to study the impact of agricultural and forestry development, energy consumption, population, and economic development on China’s carbon dioxide emissions, as well as the relationships between these factors.

The innovation of this paper is that it integrates the asymmetric effects of economic development, agroforestry development, population size, and energy consumption on CO${}_2$ emissions based on previous studies.

In this paper, the size of the population is added to the list of variables that have rarely been considered in previous studies on CO${}_2$ emissions. In addition, a total of 103 secondary indicators are selected in this paper, which is more comprehensive than previous studies.

There are two main points of innovation in this paper.

China’s economic development is facing the triple pressure of shrinking demand, supply shocks, and weakening expectations. At present, China’s economic policy emphasizes stability. In order to provide suggestions for China’s future carbon-neutral efforts, this paper attempts to examine the causal relationship and influence of consumption and population size on carbon emissions in terms of economic development, agroforestry development, energy, and the environment. In previous studies of CO${}_2$ emissions, few scholars have considered the impact of CO${}_2$ emissions on China from these five dimensions in an integrated manner. Existing studies mostly use a few variables to describe a country’s economic development, energy consumption, CO${}_2$ emissions, and so on [9,10,11,12,13]. Due to the complexity of economic systems and ecosystems, the same variable may have opposite effects on carbon dioxide emissions under different circumstances [14,15], so using a small number of variables to characterize a system may not be enough. A total of 103 secondary indicators have been selected in this paper, which is a more comprehensive study compared to previous studies, and the scope of the study is more extensive. Previous studies on CO${}_2$ emissions have only included 10 to 30 indicators, so our study is more credible.

Section 1 is the introduction, which introduces the background of the study, the significance of the study, the contribution of this paper, and the innovation points. Section 2 is the literature review, which reviews the research results of domestic and international scholars. Section 3 and Section 4 are empirical analyses. Section 3 uses the PLS-PM model to analyze the original data and perform dimensionality reduction to make the data easier to handle and prepare for the analysis in the next section. Section 4 uses the VAR autoregressive model for the empirical analysis. Section 5 is the discussion section, which presents the shortcomings of this paper’s research and the outlook for future research. Section 6 is the conclusion, recommendations, and practical implications section, which gives a brief statement on the value of this paper’s research.

2. Literature Review

The Temperate Agroforestry Association (AFTA) defines agroforestry as an intensive and sustainable land management system that meets the production needs of farmers and landowners while providing environmental benefits to society [16,17]. Since agroforestry practices can enhance food, nutrition, environmental, and energy security at the same time, agroforestry can further promote the achievement of other sustainable development goals by identifying patterns that reflect local characteristics and guiding the participation of the population by rigorously evaluating their effectiveness and using “slope” management [18]. Since 2000, China has implemented a large-scale natural forest protection program (NFPP) for the conservation of natural vegetation for sustainable development [19]. Since the Thirteenth World Forestry Congress, it has been widely recognized that prosperity in the forestry sector is important for economic and social benefits and essential for achieving sustainable development goals [20]. Agroforestry is a land management practice in which trees are planted around or between crops or pastures. This integration of agriculture and forest is often seen as an option that ensures food security and brings together a variety of environmental benefits. However, quantitative research that integrates multiple aspects of agroforestry is rare [21]. Declining biodiversity is affecting food security, agricultural sustainability, and environmental quality. Agroforestry is considered a possible partial solution for biodiversity conservation and improvement [22,23]. As climate change poses a global threat and can jeopardize the sustainability of agriculture, forestry, and other surface systems [24], it is important to study the impact relations between the development of agroforestry and CO${}_2$ emissions.

Balancing the nutritional needs of a growing population with the sustainability of agriculture is a fundamental challenge for the 21st century [25]. Productive and sustainable agricultural systems are essential for human survival [26]. By 2050, the global population is expected to reach 9.8 billion [27,28]. Food production is growing exponentially to meet the needs of a growing population. However, limited land and water resources, climate change, and an increase in extreme events may pose significant threats to the achievement of sustainable agriculture goals [29].

Agriculture makes a significant contribution to the economy of a country, and China is no exception. Over the years, the accumulation of carbon dioxide in the atmosphere has increased dramatically due to various human activities such as deforestation and agriculture. Rapidly developing agriculture and agricultural mechanization have led to a significant increase in global energy use and CO${}_2$ emissions [30]. The studies of Platis and Sharma show agroforestry development activities also contribute to carbon emissions. Agriculture accounts for 5% of all energy used worldwide. Most of it is not in a renewable form, so it can be linked to greenhouse gas emissions [14,15]. On the other hand, agriculture provides food, fuel, and raw materials that are vital to human livelihoods. With an unprecedented and still growing population, it is crucial to focus on agroforestry development [31].

The expansion of renewable energy aims to meet global energy demand while replacing fossil fuels. However, it requires large areas of land. At the same time, food security is affected by climate change and the growth of the world’s population. This has led to increased competition for limited land resources [32]. Increased renewable energy and agricultural development will reduce CO${}_2$ emissions, while non-renewable energy is positively correlated with emissions [33]. The expansion of economic activities has led to an unprecedented rise in global energy consumption, which in turn has led to serious environmental problems, such as global warming [34]. As one of the pillars of modern economic and social development, fossil energy consumption can significantly promote regional economic growth [35].

Kongkuah et al. investigated the relationship between carbon dioxide emissions, economic growth, and energy consumption in China during a period of increasing urbanization and international trade [36]. It also studied the future paths of these related variables and tested the EKC hypothesis in China. According to this study, the EKC hypothesis is not valid in China. The study also demonstrates that energy consumption has a positive and significant effect on carbon dioxide emissions in the long run, so pollution from carbon dioxide emissions is worsened.

In addition, Azam investigates the impact of renewable electricity consumption on economic growth for a panel of 10 newly industrialized countries from 1990 to 2015 [37]. The panel Granger causality findings indicate that a bidirectional causality is found between renewable electricity consumption and GDP, total labor force and GDP, and non-renewable electricity consumption and labor force in the short run. There is a short-run unidirectional causal relationship running from renewable electricity consumption to the labor force, from non-renewable electricity consumption to economic growth, and from non-renewable electricity consumption to renewable electricity consumption.

A major contributor to climate change and global warming is the increase in global carbon emissions from human activities, such as deforestation and the burning of fossil fuels [38]. Deforestation, land clearing, fertilization with highly environmentally damaging chemicals, neglected integration of agroforestry, and social forest practices are all likely to increase CO${}_2$ emissions [39]. In complex agroforestry systems, forest soils have a higher GWP than agricultural soils, and hedgerow woodland soils have a higher GWP than shelterbelts [40].

Forestry and agroforestry can reduce CO${}_2$ emissions, so extensive deforestation must be reduced [41]. Trees in traditional agricultural systems form a better solution to climate change and support agriculture by providing mutual synergies for ecology and food security [42,43,44]. Climate change caused by anthropogenic greenhouse gas emissions (mainly CO${}_2$) is one of the three major threats of modern times. The main cause is the high dependence on fossil fuels for electricity generation, transport, manufacturing, and intensive land use (deforestation) [45]. In China, since the 1960s, large areas of reclaimed coastal wetlands have been converted to other uses, including salt farms, agricultural land, woodlands, and aquaculture ponds, which may affect vegetation photosynthesis [46]. Such increases in yield and efficiency often come at the expense of the environment, including water pollution (leaching and runoff of nitrogen, phosphorus, and pesticides), soil degradation (e.g., erosion, compaction, and loss of soil organic matter), loss of biodiversity, and increased emissions of greenhouse gases such as CO${}_2$ and CH${}_4$ [47,48].

3. PLS-PM Model

3.1. Data Sources and Variable Descriptions

3.1.1. Selection of Indicators

The main variables studied in this paper are economic development (GDP), agroforestry development (AFD), energy consumption (EC), population size (PS), and CO${}_2$ emissions (CO${}_2$), involving 103 relevant secondary indicators, shown in Appendix A 

Table A1. Assignment of indicator weights in PLS-PM model results.

Level 1 Indicator	Indicator Name	Symbol	Weight
Economic Development	Domestic general government health expenditure (% of GDP)	X${}_1$	0.058601
(GDP)	Manufacturing, value added (% of GDP)	X${}_2$	−0.055852
	Market capitalization of listed domestic companies (% of GDP)	X${}_3$	0.045206
	Research and development expenditure (% of GDP)	X${}_4$	0.059284
	GDP per capita (current LCU)	X${}_5$	0.056813
	GDP per capita, PPP (constant 2017 international $)	X${}_6$	0.058212
	Industry (including construction), value added (% of GDP)	X${}_7$	−0.028804
	GDP per capita growth (annual %)	X${}_8$	−0.022887
	GDP per capita (current USD)	X${}_9$	0.056472
	Foreign direct investment, net inflows (% of GDP)	X${}_{10}$	−0.027652
	GDP (current USD)	X${}_{11}$	0.056251
	Gross national expenditure (% of GDP)	X${}_{12}$	−0.000448
	GDP (current LCU)	X${}_{13}$	0.056500
	Trade (% of GDP)	X${}_{14}$	0.018994
	Gross domestic savings (% of GDP)	X${}_{15}$	0.039865
	GDP growth (annual %)	X${}_{16}$	−0.028592
	Broad money (% of GDP)	X${}_{17}$	0.059169
	Merchandise trade (% of GDP)	X${}_{18}$	−0.000515
	Inflation, GDP deflator: linked series (annual %)	X${}_{19}$	−0.028316
	Services, value added (% of GDP)	X${}_{20}$	0.058527
	General government final consumption expenditure (% of GDP)	X${}_{21}$	0.042862
	Claims on central government, etc. (% GDP)	X${}_{22}$	0.049255
	GDP per capita, PPP (current international $)	X${}_{23}$	0.058145
	PPP conversion factor, GDP (LCU per international $)	X${}_{24}$	0.057993
	GDP, PPP (current international $)	X${}_{25}$	0.057935
	GDP deflator (base year varies by country)	X${}_{26}$	0.059344
Agroforestry Development	Agriculture, forestry, and fishing, value added (constant LCU)	X${}_{27}$	0.151855
(AFD)	Agriculture, forestry, and fishing, value added (% of GDP)	X${}_{28}$	−0.145671
	Agriculture, forestry, and fishing, value added (current USD)	X${}_{29}$	0.146749
	Agriculture, forestry, and fishing, value added (annual % growth)	X${}_{30}$	−0.031939
	Agriculture, forestry, and fishing, value added per worker (constant 2015 USD)	X${}_{31}$	0.145551
	Agriculture, forestry, and fishing, value added (current LCU)	X${}_{32}$	0.148642
	Agriculture, forestry, and fishing, value added (constant 2015 USD)	X${}_{33}$	0.151895
	Forest rents (% of GDP)	X${}_{34}$	−0.131377
	Total natural resources rents (% of GDP)	X${}_{35}$	−0.045632
Energy consumption	Energy-related methane emissions (% of total)	X${}_{36}$	0.067541
(EC)	Renewable energy consumption (% of total final energy consumption)	X${}_{37}$	−0.065592
	Nitrous oxide emissions in energy sector (thousand metric tons of CO${}_2$ equivalent)	X${}_{38}$	0.064501
	Energy imports, net (% of energy use)	X${}_{39}$	0.066981
	Investment in energy with private participation (current USD)	X${}_{40}$	−0.007817
	Fossil fuel energy consumption (% of total)	X${}_{41}$	0.066547
	Public private partnerships investment in energy (current USD)	X${}_{42}$	−0.001304
	Energy use (kg of oil equivalent per capita)	X${}_{43}$	0.066538
	Adjusted savings: energy depletion (current USD)	X${}_{44}$	0.047159
	Nitrous oxide emissions in energy sector (% of total)	X${}_{45}$	−0.047572
	Alternative and nuclear energy (% of total energy use)	X${}_{46}$	0.066999
	Adjusted savings: energy depletion (% of GNI)	X${}_{47}$	−0.026126
	GDP per unit of energy use (PPP $ per kg of oil equivalent)	X${}_{48}$	0.068223
	Combustible renewables and waste (% of total energy)	X${}_{49}$	−0.067900
	Methane emissions in energy sector (thousand metric tons of CO${}_2$ equivalent)	X${}_{50}$	0.066770
	GDP per unit of energy use (constant 2017 PPP $ per kg of oil equivalent)	X${}_{51}$	0.067174
	Energy intensity level of primary energy (MJ/$2011 PPP GDP)	X${}_{52}$	−0.063757
	Energy use (kg of oil equivalent) per $1,000 GDP (constant 2017 PPP)	X${}_{53}$	−0.063351
	Coal rents (% of GDP)	X${}_{54}$	0.014803
	Mineral rents (% of GDP)	X${}_{55}$	0.009687
	Natural gas rents (% of GDP)	X${}_{56}$	0.055533
	Oil rents (% of GDP)	X${}_{57}$	−0.049966
Population size	Population, female (% of total population)	X${}_{58}$	0.036775
(PS)	Population ages 40–44, male (% of male population)	X${}_{59}$	−0.048668
	Population ages 20–24, female (% of female population)	X${}_{60}$	0.053098
	Population ages 0–14, male	X${}_{61}$	0.065278
	Age dependency ratio (% of working-age population)	X${}_{62}$	0.060491
	Population ages 20–24, male (% of male population)	X${}_{63}$	0.050611
	Rural population	X${}_{64}$	0.066585
	Age dependency ratio, old (% of working-age population)	X${}_{65}$	−0.060103
	Population ages 25–29, female (% of female population)	X${}_{66}$	0.045971
	Population ages 0–14, total	X${}_{67}$	0.065667
	Population, male (% of total population)	X${}_{68}$	−0.036775
	Population, total	X${}_{69}$	−0.069061
	Age dependency ratio, young (% of working-age population)	X${}_{70}$	0.067056
	Population ages 0–14 (% of total population)	X${}_{71}$	0.067810
	Population ages 15–64, total	X${}_{72}$	−0.068139
	Population ages 0–14, female	X${}_{73}$	0.065971
	Urban population	X${}_{74}$	−0.068701
	Population ages 15–64, female (% of female population)	X${}_{75}$	−0.058857
	Population ages 30–34, female (% of female population)	X${}_{76}$	0.019239
	Population ages 35–39, male (% of male population)	X${}_{77}$	0.003181
	Population ages 30–34, male (% of male population)	X${}_{78}$	0.018523
	Population ages 35–39, female (% of female population)	X${}_{79}$	0.002284
	Population ages 10–14, female (% of female population)	X${}_{80}$	0.056126
CO${}_2$ emissions	CO${}_2$ emissions (kt)	X${}_{81}$	0.054550
(CO${}_2$)	CO${}_2$ emissions from liquid fuel consumption (% of total)	X${}_{82}$	−0.037179
	CO${}_2$ emissions from electricity and heat production, total (% of total fuel combustion)	X${}_{83}$	0.047387
	CO${}_2$ emissions (kg per 2017 PPP $ of GDP)	X${}_{84}$	−0.051396
	CO${}_2$ emissions from other sectors, excluding residential buildings and commercial and public services (% of total fuel combustion)	X${}_{85}$	−0.049226
	Nitrous oxide emissions (thousand metric tons of CO${}_2$ equivalent)	X${}_{86}$	0.055019
	CO${}_2$ emissions from solid fuel consumption (% of total)	X${}_{87}$	−0.039872
	CO${}_2$ emissions from manufacturing industries and construction (% of total fuel combustion)	X${}_{88}$	-0.027487
	CO${}_2$ emissions from solid fuel consumption (kt)	X${}_{89}$	0.053315
	Methane emissions (kt of CO${}_2$ equivalent)	X${}_{90}$	0.053228
	CO${}_2$ emissions from liquid fuel consumption (kt)	X${}_{91}$	0.055332
	Agricultural nitrous oxide emissions (thousand metric tons of CO${}_2$ equivalent)	X${}_{92}$	0.050109
	CO${}_2$ emissions (metric tons per capita)	X${}_{93}$	0.054258
	CO${}_2$ emissions from gaseous fuel consumption (% of total)	X${}_{94}$	0.050284
	Agricultural methane emissions (thousand metric tons of CO${}_2$ equivalent)	X${}_{95}$	−0.028583
	CO${}_2$ emissions from gaseous fuel consumption (kt)	X${}_{96}$	0.052104
	CO${}_2$ emissions (kg per 2015 USD of GDP)	X${}_{97}$	−0.051440
	Total greenhouse gas emissions (kt of CO${}_2$ equivalent)	X${}_{98}$	0.054590
	CO${}_2$ emissions from residential buildings and commercial and public services (% of total fuel combustion)	X${}_{99}$	−0.051083
	CO${}_2$ emissions (kg per PPP $ of GDP)	X${}_{100}$	−0.051347
	CO${}_2$ emissions from transport (% of total fuel combustion)	X${}_{101}$	0.046346
	Other greenhouse gas emissions, HFC, PFC and SF6 (thousand metric tons of CO${}_2$ equivalent)	X${}_{102}$	−0.047617
	CO${}_2$ intensity (kg per kg of oil equivalent energy use)	X${}_{103}$	0.051875

.

(1)

Economic development (GDP)

At a macro level, the inconsistency between population distribution and economic development reflects, to some extent, the lag in population mobility, and also widens regional economic disparities [49]. Sustained improvements in the reduction of CO${}_2$ emissions, carbon intensity, and energy intensity are difficult to achieve, and most countries are experiencing a weak decoupling between CO${}_2$ emissions and GDP [50]. Indicators related to economic development are, of course, not limited to GDP. We select 26 indicators, X${}_1$–X${}_{26}$, to describe China’s economic development as comprehensively as possible. China has implemented the Healthy China strategy in recent years, so X${}_1$ is selected. China is in a stage of rapid industrialization, with a large amount of energy consumption and CO${}_2$ emissions. The growth of GDP has mainly benefited from the prosperity of the secondary industry over the past three decades, so X${}_2$ and X${}_7$ are selected. X${}_4$–X${}_6$, X${}_{20}$, and X${}_{23}$ are also important variables to measure the level of economic development [9,10]. Other variables describe China’s economic development from the perspective of trade and GDP. Among them, studies have shown that GDP variables (X${}_5$, X${}_6$, X${}_{11}$, etc.) [11] and X${}_7$ [11] should not be ignored in contributing to CO${}_2$ emissions. X${}_1$ [12] can reduce a country’s domestic carbon emissions. X${}_{18}$ [13,36] is also an important factor affecting carbon emissions.

(2)

Agroforestry development (AFD)

Agroforestry is considered as a way to support the transition of agriculture from a system with sustainability issues to a system that includes renewable activities and thus provides solutions for sustainable development [51]. The main objective of agroforestry systems is to minimize competition for resources and maximize ecological and economic benefits [52]. In temperate agro-environments, agroforestry offers great potential to conserve and restore biodiversity, reduce non-point-source pollution, improve soil microbial resilience to water stress, and cope with climate change [24,53,54]. Therefore, in this paper, based on previous studies in the literature, X${}_{27}$–X${}_{35}$ are used to describe AFD. Agriculture accounts for 5% of all energy used worldwide. Most of it is not in a renewable form, so it can be linked to greenhouse gas emissions [14,15]. Limited by data sources, and by depicting the development level and structure of AFD, it can reflect the frequency of economic behavior of AFD.

(3)

Energy consumption (EC)

Agriculture accounts for 5% of total global energy consumption. Most of this is not renewable energy and therefore may be associated with greenhouse gas emissions [15]. CO${}_2$ emissions from direct energy consumption vary considerably between cities and are therefore susceptible to regional conditions [55]. As one of the pillars of modern economic and social development, energy plays an important role in promoting regional economic growth [35]. Energy consumption includes: primary energy such as raw coal, crude oil, and natural gas; secondary energy such as coke, gas, refined oil, and other products generated at the same time; other fossil energy sources; renewable energy; and new energy sources. Therefore, X${}_{36}$–X${}_{57}$ are selected with regard to energy investment, renewable energy, and energy prices. EC increases carbon emissions [36], such as X${}_{36}$, X${}_{41}$, X${}_{43}$, and X${}_{54}$–X${}_{57}$. However, X${}_{37}$ and X${}_{49}$ reduce carbon emissions environmentally [56]. The other variables reflect China’s energy mix and level of development.

(4)

Population size (PS)

Population factors are one of the main drivers of CO${}_2$ emission projections, and there is a significant positive relationship between GDP and CO${}_2$ emissions [57]. At the micro (household) level, research on household carbon emissions has addressed the effects of population size and socio-economic factors [58]. Household income and population size can also influence the level of transport-related CO${}_2$ emissions [59]. Factors such as population density and intensity of population activity have a stronger impact on urban CO${}_2$ emissions [60]. The contribution of population to the increase in energy-related CO${}_2$ emissions of residents in urban and rural areas is small and the most stable [61]. Pan et al [62] analyzed the relationship between demographic change and carbon emissions using ridge regression in four dimensions: population size, population age structure, population consumption structure, and population employment structure. In this paper, it is considered that these indicators alone are not enough to study the demographic factors; therefore, X${}_{58}$–X${}_{80}$ are used to describe PS. The change in population age structure diminishes the growth of carbon emissions in China, but the effects of changing urban and rural population and population size generate increases in carbon emissions [63].

(5)

CO${}_2$ emissions (CO${}_2$)

Over-reliance on fossil fuels such as coal and oil not only places a burden on economic development, but also directly contributes to a series of environmental problems, such as global warming and CO${}_2$ emissions [64]. As economic output and the industrial sector grows, the release of carbon dioxide tends to rise, which makes GDP and carbon dioxide emissions increasingly relevant indicators as drivers of carbon dioxide emissions [39]. Therefore, in this paper, 23 secondary indicators related to CO${}_2$ emissions are selected based on previous scholarly research.

3.1.2. Data Sources

The main variables studied in this paper are economic development (GDP), agroforestry development (AFD), energy consumption (EC), population size (PS), and CO${}_2$ emissions (CO${}_2$), involving 103 relevant secondary indicators. The table of indicator weights assigned to the results of the PLS-PM model is detailed in Appendix Table A1.

The study uses annual time series data from 1990 to 2020, and all data used in this study are from the World Bank (https://data.worldbank.org/country/CN, accessed on 6 May 2022). Due to the presence of missing values in some of the data, the ARIMA model was used to fill in the missing values, and then the partial least squares path model (PLS-PM) was used to complete the data processing using R4.1.1 software. Descriptive statistics for the variables are shown in

Table 1. Descriptive analysis of variables.

Variable	Year	GDP	AFD	EC	PS	CO${}_2$
Obs	31	31	31	31	31	31
Mean	2005	2710 × 10${}^9$	1250 × 10${}^9$	3100 × 10${}^6$	−1050 × 10${}^5$	1,150,694
Std. dev.	9.092121	2510 × 10${}^9$	631 × 10${}^9$	2950 × 10${}^6$	341 × 10${}^5$	524,396.7
Min	1990	191 × 10${}^9$	477 × 10${}^9$	284 × 10${}^6$	−1560 × 10${}^5$	484,923.7
Max	2020	8000 × 10${}^9$	2520 × 10${}^9$	11,100 × 10${}^6$	−517 × 10${}^5$	1,869,574

.

3.2. Principles of the PLS Model

Partial least squares structural equation modeling (PLS-SEM) is a new generation of statistical data analysis that, despite its infancy, is rapidly gaining traction in the academic community: it has stimulated academic interest in a variety of methods, leading to a dynamic and evolving tool. Managers in organizations and government agencies, as well as academics and researchers, now have access to large amounts of data to evaluate decisions and new findings [65]. The goal of PLS-SEM is to maximize the explained variance of the endogenous latent components (dependent variables). PLS-SEM was chosen because of its ability to detect significant driving structures and deal with non-normal data sets, its minimal sample size requirements, and its precise computational and modern approach to developing models and testing the validity of hypotheses [66,67]. In addition, the PLS-PM has low dependence on data and does not need to restrict parameters, and the model setting is simple and easy to operate. Therefore, the extended model is not used. Similarly, this paper uses the VAR model instead of the VAR extended model.

PLS-PM is a partial least squares estimation of structural equation models. Similar to SEM, PLS-PM also calculates regression analysis metrics such as standard path coefficients and R${}^2$ values.

PLS-PM uses the GOF (goodness-of-fit) index to determine the overall fit of the model, with larger GOF values indicating a better fit. The higher the GOF value, the better the fit of the model. The PLS-PM model constructed in this paper is shown in Figure 1.

Assuming that there are i groups of explanatory variables and that each group contains $P_i$ variables, each group of variables can be expressed as in Equation (1).

$$X_i=\left(x_{i1},x_{i2},\dots ,x_{ipi}\right)(i=1,2,\dots ,i)$$

(1)

In general, it is assumed that each of the variables in the explicit variables $x_{ih}$ (i = 1, 2, …, ih = 1, 2, …, $P_i$) is centralized and that they are all based on n common observations. In addition, the hidden variable $\delta$${}_i$, corresponding to each set of explicit variables $x_i$, is assumed to be standardized. In the PLS-PM model, the measurement model consists of the explicit variables $x_i$ (i = 1, 2, …, i) with their corresponding hidden variables $\delta$${}_i$. The structural model consists of different groups of hidden variables together $\delta$${}_i$. The PLS-PM model estimates the hidden variables $\delta$${}_i$ by means of linear regression with multiple iterations. The relationship equation between the explicit and implicit variables is then estimated through the set network model. There are two main methods for estimating the explicit variable $x_i$ (i = 1, 2, …, i) and its corresponding hidden variable $\delta$${}_i$: one is external estimation, where the hidden variable is calculated through the relationship between the explicit variable and the explanatory variable; the other method is internal estimation, where it is calculated through the correlation between the hidden variables. Theoretically, the larger the sample size, the better, and there is no point in estimating on a small sample.

In this paper, an iterative algorithm is used to calculate the PLS-PM model hidden variables, and the measurement model and structural equation model are then calculated based on the estimated values of the hidden variables. The detailed calculation steps are as follows:

Step 1. The initial value of the orientation quantity $Y_i$ is equal to $x_{i1}$.

Step 2. The estimated value of $Z_i$ is calculated, as shown in Equations (2) and (3). That is,

$$Z_i={\left(\sum _{i:\beta ij\ne 0}{e}_{ij}{Y}_i\right)}^{*}$$

(2)

In the above equation,

$$e_{ij}=sign\left(r\left(Y_j,Y_i\right)\right)$$

(3)

Step 3. According to the estimated value of $Z_i$, through Equation (4), we obtain

$$w_i={\left(X_i^T{X}_i\right)}^{-1}{X}_i^T{Z}_i$$

(4)

Calculate the weight vector wi as in Equation (5).

$$w_i=\frac{1}{n}{X}_i^T{Z}_i$$

(5)

Step 4. Using the calculated wi, by Equation (6), we obtain

$$Z_i={\left(\sum _{j:\beta ji\ne 0}{e}_{ij}{Y}_i\right)}^{*}$$

(6)

Calculate the new $Y_i$ as in Equation (7).

$$Y_i={\left(\sum _{b=1}^{Pi}{w}_{ih}{x}_{ih}\right)}^{*}={\left(X_i{W}_i\right)}^{*}$$

(7)

Step 5. Return to the second step until the computation converges to the position for the final obtained $Y_i$ as an estimate of the hidden variable $\delta$${}_i$.

Many studies have rigorously demonstrated the good convergence of this computational method in path analysis for two sets of variable sets, and a large body of literature has demonstrated that the method is equally applicable to the multi-variable set case After obtaining an estimate of the hidden variable $\delta$${}_i$, we can estimate the measurement model using a one-variable linear regression method, ${\widehat{x}}_{jh}={\lambda }_{jh}{\widehat{\delta }}_j$. The path coefficients in the structural model are then estimated using multiple regression, and the hidden variable $\delta$${}_j$ is calculated as shown in Equation (8).

$${\widehat{\delta }}_j=\sum _{i;cij\ne 0}{\beta }_{ji}{\widehat{\delta }}_i$$

(8)

where $c_{ij}$ denotes the element in the design matrix, and $c_{ij}$ not being equal to 0 indicates that there is a correlation between $\delta$${}_j$ and $\delta$${}_i$.

3.3. Analysis of PLS Results

The correlations between the variables of the PLS-PM model constructed in this paper and the indicators of the goodness of fit of the model are shown in

Table 2. PLS-PM model variable correlations and model goodness-of-fit.

Variables	GDP	AFD	EC	PS	CO${}_2$
GDP	1.00	0.99	0.95	−0.90	0.96
AFD	0.99	1.00	0.97	−0.92	0.98
EC	0.95	0.97	1.00	−0.98	0.99
PS	−0.90	−0.92	−0.98	1.00	−0.96
CO${}_2$	0.96	0.98	0.99	−0.96	1.00
Estimate	−0.31 ***	0.77 ***	0.37 **	−0.16 *
Std. Error	0.09	0.14	0.14	0.08
t value	−3.30	5.26	2.57	−1.96
Pr(>|t|)	0.00	0.00	0.01	0.06
R${}^2$	0.00	0.98	0.94	0.96	0.99
Type	Exogenous	Endogenous	Endogenous	Endogenous	Endogenous

*** indicates significance at the 1% level. ** indicates significance at the 5% level. * indicates significance at the 10% level. Similarly hereinafter.

. From Table 2, it can be concluded that the maximum p-value of all path coefficients of the model = 0.0611 are significant at the 10% level; the minimum R${}^2$ = 0.9460, and the overall goodness of fit GOF = 0.8203, indicating a good fit of the model.

3.4. Result of PLS-PM Model

As shown in

Table 3. Influence relationships between variables in PLS-PM model results.

	Relationships	Direct	Indirect	Total
1	GDP ->AFD	0.9900	0.0000	0.9900
2	GDP ->EC	0.0000	0.9630	0.9630
3	GDP ->PS	0.0000	−0.9440	−0.9440
4	GDP ->CO${}_2$	−0.3160	1.2880	0.9720
5	AFD ->EC	0.9730	0.0000	0.9730
6	AFD ->PS	0.0000	−0.9540	−0.9540
7	AFD ->CO${}_2$	0.7740	0.5280	1.3010
8	EC ->PS	−0.9800	0.0000	−0.9800
9	EC ->CO${}_2$	0.3790	0.1630	0.5420
10	PS ->CO${}_2$	−0.1670	0.0000	−0.1670

, the results of the PLS-PM model show that there is a direct and significant positive effect between economic development (GDP) and agroforestry development (AFD); an indirect and significant positive effect between economic development (GDP) and energy consumption (EC); an indirect and significant negative effect between economic development (GDP) and energy consumption (PS); and both a direct and an indirect effect between economic development (GDP) and carbon dioxide emissions (CO${}_2$). The relationship between economic development (GDP) and energy consumption (PS) is indirect and significantly negative, while the relationship between economic development (GDP) and carbon dioxide emissions (CO${}_2$) is both direct and indirect, with an overall significant positive effect. The relationship between agroforestry development (AFD) and energy consumption (EC) is direct and significantly positive; the relationship between agroforestry development (AFD) and population size (PS) is indirect and significantly negative; the relationship between agroforestry development (AFD) and carbon dioxide emissions (CO${}_2$) is both direct and indirect, with an overall positive effect. The relationship between AFD and CO${}_2$ emissions is both direct and indirect, with an overall significant positive effect. There is a direct and negative relationship between energy consumption (EC) and population (PS), and a direct and indirect relationship between energy consumption (EC) and carbon dioxide emissions (CO${}_2$), with an overall positive relationship. The relationship between population size (PS) and carbon dioxide emissions (CO${}_2$) is direct and negative.

The following conclusions can be drawn from this:

(1)

Agroforestry development is directly influenced by economic development and is significantly positive.

(2)

Energy consumption is significantly and positively influenced by economic development and agroforestry development. In other words, economic development and agroforestry development still depend on a large amount of energy consumption.

(3)

Population size is significantly negatively influenced by economic development, agroforestry development, and energy consumption, that is, population growth is significantly constrained and limited by economic development, agroforestry development, and energy consumption.

(4)

CO${}_2$ emissions are significantly positively influenced by economic development, agroforestry development, and energy consumption, but slightly negatively and directly influenced by population size. In other words, population growth does not have a significant impact on CO${}_2$ emissions, but rather economic development, agroforestry development, and energy consumption. It is economic development, agroforestry development, and energy consumption that really affect CO${}_2$ emissions, and the government should focus on these three areas in order to achieve the goal of carbon neutrality.

4. VAR Model

4.1. Principle of the VAR Model

The VAR model was first proposed by Sims [68,69] to study the dynamic relationships between multivariates. Based on research using the VAR model, there are three main applications: (1) the causal relationship between variables is found by the Granger causation test; (2) quantitative analysis of the relationship between variables through variance decomposition and pulse response; (3) short-term projections are made for each variable. The basic form of the VAR model is shown in Equation (9).

$$Y_{\mathrm{t}}={\mu }_0+{\mu }_1{Y}_{t-1}+{\mu }_2{Y}_{t-2}+\dots +{\mu }_m{Y}_{t-m}+{\epsilon }_t$$

(9)

where Y represents the variable studied, $Y_t$ represents the variable of the current period, and Y${}_{t-m}$ represents the variable of the lag m period (m≥ 1), as shown in Equation (10).

$$Y_t=\left[\begin{array}{c}{y}_{1t}\\ y_{2t}\\ ·\\ ·\\ ·\\ y_{nt}\end{array}\right],{\epsilon }_t=\left[\begin{array}{c}{\epsilon }_{1t}\\ {\epsilon }_{2t}\\ ·\\ ·\\ ·\\ {\epsilon }_{nt}\end{array}\right],{\mu }_0=\left[\begin{array}{c}{\mu }_{10}\\ {\mu }_{20}\\ ·\\ ·\\ ·\\ {\mu }_{n0}\end{array}\right],{\mu }_i=\left[\begin{array}{cccc}{\mu }_{11}^i& {\mu }_{12}^i& \dots & {\mu }_{1n}^i\\ {\mu }_{21}^i& {\mu }_{22}^i& \dots & {\mu }_{2n}^i\\ ·& ·& ·& ·\\ ·& ·& ·& ·\\ ·& ·& ·& ·\\ {\mu }_{n1}^i& {\mu }_{n2}^i& \dots & {\mu }_{nm}^i\end{array}\right],(i=1,2,\dots ,m)$$

(10)

4.2. Model Builds

In this paper, five different VAR models are constructed to study the relationship between each variable, and the relationship between each variable is cross-verified by establishing several different models, as follows:

(1) To study the relationship between economic development, agroforestry development, and energy consumption, model 1 is built as shown in Equation (11).

$$Y_{GA{E}_{-}t}={\mu }_0+{\mu }_1{Y}_{GA{E}_{-}t-1}+{\mu }_2{Y}_{GA{E}_{-}t-2}+\dots +{\mu }_m{Y}_{GA{E}_{-}t-m}+{\epsilon }_t$$

(11)

where Y represents the variable studied, $Y_{GA{E}_{-}t}$ represents the variable of the current period, and $Y_{GA{E}_{-}t-m}$ represents the variable of the lag m period (m≥ 1), as shown in Equation (12).

$$Y_{GA{E}_{-}t}=\left[\begin{array}{c}GD{P}_t\\ AF{D}_t\\ E{C}_t\end{array}\right],{\epsilon }_t=\left[\begin{array}{c}{\epsilon }_{1t}\hfill \\ {\epsilon }_{2t}\hfill \\ {\epsilon }_{3t}\hfill \end{array}\right],{\mu }_0=\left[\begin{array}{c}{\mu }_{10}\hfill \\ {\mu }_{20}\hfill \\ {\mu }_{30}\hfill \end{array}\right],{\mu }_i=\left[\begin{array}{ccc}{\mu }_{11}^i& {\mu }_{12}^i& {\mu }_{13}^i\\ {\mu }_{21}^i& {\mu }_{22}^i& {\mu }_{23}^i\\ {\mu }_{31}^i& {\mu }_{32}^i& {\mu }_{33}^i\end{array}\right],(i=1,2,\dots ,m)$$

(12)

(2) To study the relationship between agroforestry development, energy consumption, and population, model 2 is constructed as shown in Equation (13).

$$Y_{AE{P}_{-}t}={\mu }_0+{\mu }_1{Y}_{AE{P}_{-}t-1}+{\mu }_2{Y}_{AE{P}_{-}t-2}+\dots +{\mu }_m{Y}_{AE{P}_{-}t-m}+{\epsilon }_t$$

(13)

where Y represents the variable studied, $Y_{AE{P}_{-}t}$ represents the variable of the current period, and $Y_{AE{P}_{-}t-m}$ represents the variable of the lag m period (m≥ 1), as shown in Equation (14).

$$Y_{AE{P}_{-}t}=\left[\begin{array}{c}AF{D}_t\\ E{C}_t\\ P{S}_t\end{array}\right],{\epsilon }_t=\left[\begin{array}{c}{\epsilon }_{1t}\hfill \\ {\epsilon }_{2t}\hfill \\ {\epsilon }_{3t}\hfill \end{array}\right],{\mu }_0=\left[\begin{array}{c}{\mu }_{10}\hfill \\ {\mu }_{20}\hfill \\ {\mu }_{30}\hfill \end{array}\right],{\mu }_i=\left[\begin{array}{ccc}{\mu }_{11}^i& {\mu }_{12}^i& {\mu }_{13}^i\\ {\mu }_{21}^i& {\mu }_{22}^i& {\mu }_{23}^i\\ {\mu }_{31}^i& {\mu }_{32}^i& {\mu }_{33}^i\end{array}\right],(i=1,2,\dots ,m)$$

(14)

(3) To study the relationship between agroforestry development, energy consumption, and CO${}_2$ emissions, model 3 is constructed as shown in Equation (15).

$$Y_{AE{C}_{-}t}={\mu }_0+{\mu }_1{Y}_{AE{C}_{-}t-1}+{\mu }_2{Y}_{AE{C}_{-}t-2}+\dots +{\mu }_m{Y}_{AE{C}_{-}t-m}+{\epsilon }_t$$

(15)

where Y represents the variable studied, $Y_{AE{C}_{-}t}$ represents the variable of the current period, and $Y_{AE{C}_{-}t-m}$ represents the variable of the lag m period (m ≥ 1), as shown in Equation (16).

$$Y_{AE{C}_{-}t}=\left[\begin{array}{c}AF{D}_t\\ E{C}_t\\ C{O}_{2t}\end{array}\right],{\epsilon }_t=\left[\begin{array}{c}{\epsilon }_{1t}\\ {\epsilon }_{2t}\\ {\epsilon }_{3t}\end{array}\right],{\mu }_0=\left[\begin{array}{c}{\mu }_{10}\hfill \\ {\mu }_{20}\hfill \\ {\mu }_{30}\hfill \end{array}\right],{\mu }_i=\left[\begin{array}{ccc}{\mu }_{11}^i& {\mu }_{12}^i& {\mu }_{13}^i\\ {\mu }_{21}^i& {\mu }_{22}^i& {\mu }_{23}^i\\ {\mu }_{31}^i& {\mu }_{32}^i& {\mu }_{33}^i\end{array}\right],(i=1,2,\dots ,m)$$

(16)

(4) To study the relationship between energy consumption, population, and CO${}_2$ emissions, model 4 is built as shown in Equation (17).

$$Y_{EP{C}_{-}t}={\mu }_0+{\mu }_1{Y}_{EP{C}_{-}t-1}+{\mu }_2{Y}_{EP{C}_{-}t-2}+\dots +{\mu }_m{Y}_{EP{C}_{-}t-m}+{\epsilon }_t$$

(17)

where Y represents the variable studied, $Y_{EP{C}_{-}t}$ represents the variable of the current period, and $Y_{EP{C}_{-}t-m}$ represents the variable of the lag m period (m ≥ 1), as shown in Equation (18).

$$Y_{EP{C}_{-}t}=\left[\begin{array}{c}E{C}_t\\ P{S}_t\\ C{O}_{2t}\end{array}\right],{\epsilon }_t=\left[\begin{array}{c}{\epsilon }_{1t}\hfill \\ {\epsilon }_{2t}\hfill \\ {\epsilon }_{3t}\hfill \end{array}\right],{\mu }_0=\left[\begin{array}{c}{\mu }_{10}\hfill \\ {\mu }_{20}\hfill \\ {\mu }_{30}\hfill \end{array}\right],{\mu }_i=\left[\begin{array}{ccc}{\mu }_{11}^i& {\mu }_{12}^i& {\mu }_{13}^i\\ {\mu }_{21}^i& {\mu }_{22}^i& {\mu }_{23}^i\\ {\mu }_{31}^i& {\mu }_{32}^i& {\mu }_{33}^i\end{array}\right],(i=1,2,\dots ,m)$$

(18)

(5) To study the relationship between economic development, agroforestry development, energy consumption, and CO${}_2$ emissions, Model 5 is built as shown in Formula (19).

$$Y_{GAE{C}_{-}t}={\mu }_0+{\mu }_1{Y}_{GAE{C}_{-}t-1}+{\mu }_2{Y}_{GAE{C}_{-}t-2}+\dots +{\mu }_m{Y}_{GAE{C}_{-}t-m}+{\epsilon }_t$$

(19)

where Y represents the variable studied, $Y_{GAE{C}_{-}t}$ represents the variable of the current period, and $Y_{GAE{C}_{-}t-m}$ represents the variable of the lag m period (m ≥ 1), as shown in Equation (20).

$$Y_{GAE{C}_{-}t}=\left[\begin{array}{c}GD{P}_t\\ AF{D}_t\\ E{C}_t\\ CO{2}_t\end{array}\right],{\epsilon }_t=\left[\begin{array}{c}{\epsilon }_{1t}\\ {\epsilon }_{2t}\\ {\epsilon }_{3t}\\ {\epsilon }_{4t}\end{array}\right],{\mu }_0=\left[\begin{array}{c}{\mu }_{10}\\ {\mu }_{20}\\ {\mu }_{30}\\ {\mu }_{40}\end{array}\right],{\mu }_i=\left[\begin{array}{cccc}{\mu }_{11}^i& {\mu }_{12}^i& {\mu }_{13}^i& {\mu }_{14}^i\\ {\mu }_{21}^i& {\mu }_{22}^i& {\mu }_{23}^i& {\mu }_{24}^i\\ {\mu }_{31}^i& {\mu }_{32}^i& {\mu }_{33}^i& {\mu }_{34}^i\\ {\mu }_{41}^i& {\mu }_{42}^i& {\mu }_{43}^i& {\mu }_{44}^i\end{array}\right]$$

(20)

where i = $1,\dots ,m.$

4.3. Model Testing

4.3.1. ADF Unit Root Test

This paper adopts time series data: first, the second-order difference of each variable; then, the ADF unit root test. The unit root test results are significant at a 1% level, as shown in

Table 4. ADF unit root test.

Variables	Z (t)	p-Value	Differential Orders	Total
GDP	−5.5470	0.0000	2	0.9900
AFD	−6.2300	0.0000	2	0.9630
EC	−9.3190	0.0000	2	−0.9440
PS	−4.5030	0.0002	2	0.9720
CO${}_2$	−5.8980	0.0000	2	0.9730
6	AFD → PS	0.0000	−0.9540	−0.9540
7	AFD → CO${}_2$	0.7740	0.5280	1.3010
8	EC → PS	−0.9800	0.0000	−0.9800
9	EC → CO${}_2$	0.3790	0.1630	0.5420
10	PS → CO${}_2$	−0.1670	0.0000	−0.1670

.

4.3.2. Stability Test

As shown in Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6, the variables of each model fall within the unit circle, indicating that the five models are more stable. Specifically, as shown in Figure 2, the inverse of the root modes of all variables in Model 1 are less than one, that is, the second-order differences of the variables economic development (GDP), agroforestry development (AFD), and energy consumption (EC) are all within the unit circle, which shows that the stability conditions of this VAR model are satisfied and its estimation results are robust.

As shown in Figure 3, the inverse of the root modes of all variables in Model 2 are less than one, that is, the second-order differences of agroforestry development (AFD), energy consumption (EC), and population size (PS) are all within the unit circle, which shows that the stability conditions of the VAR model are satisfied and its estimation results are robust.

As shown in Figure 4, the inverse of the root modes of all variables in Model 3 are less than one, that is, the second-order differences of the variables agroforestry development (AFD), energy consumption (EC), and carbon dioxide emissions (CO${}_2$) are all within the unit circle, which shows that the stability conditions of this VAR model are satisfied and its estimation results are robust.

As shown in Figure 5, the inverse of the root modes of all variables in Model 4 are less than one, that is, the second order differences of energy consumption (EC), population size (PS) and carbon dioxide emissions (CO${}_2$) are all within the unit circle, which shows that the stability conditions of the VAR model are satisfied and its estimation results are robust.

As shown in Figure 6, the inverse of the root modes of all variables of Model 5 are less than one, that is, the second-order differences of the variables economic development (GDP), agroforestry development (AFD), energy consumption (EC), and carbon dioxide emissions (CO${}_2$) are all within the unit circle, which shows that the stability conditions of this VAR model are satisfied and its estimation results are robust.

4.3.3. Johansen Cointeger Test

The time-series data between economic development, agricultural and forestry development, energy consumption, population, and CO${}_2$ emission are all stable after second-order difference processing and normalization processing. However, there may still be a long-term equilibrium relationship between them. Therefore, Johansen co-integration test is carried out in this paper. The test results are shown in

Table 5. Johansen co-integration test.

	Max Rank	Params	LL	Eigenvalue	Trace Statistic	Critical Value 5%
Model 1	0	12	51.10112		98.61640	29.68000
	1	17	85.40572	0.92122	30.00720	15.41000
	2	20	95.24057	0.51737	10.33750	3.76000
	3	21	100.40931	0.31810
Model 2	0	39	100.52055		52.10080	29.68000
	1	44	114.71232	0.69353	23.71730	15.41000
	2	47	123.93795	0.53643	5.26600	3.76000
	3	48	126.57095	0.19701
Model 3	0	30	68.16102		32.66860	29.68000
	1	35	76.24761	0.47635	16.49540	15.41000
	2	38	81.36629	0.33601	6.25810	3.76000
	3	39	84.49533	0.22145
Model 4	0	39	89.56117		41.90160	29.68000
	1	44	101.54564	0.63164	17.93270	15.41000
	2	47	106.69543	0.34894	7.63310	3.76000
	3	48	110.51198	0.27243
Model 5	0	20	94.52384		120.47980	47.21000
	1	27	128.77648	0.92091	51.97450	29.68000
	2	32	145.66508	0.71378	18.19730	15.41000
	3	35	151.58932	0.35521	6.34890	3.76000
	4	36	154.76375	0.20954

.

It can be seen from the results of Johansen co-integration test in Table 5 that none of the five models built in this paper exist. In each model, the trace statistic for each phase is greater than the critical value of 5%. This shows that each model has passed the test and requires further Granger causal testing.

4.3.4. Granger Causal Test

After determining that each model does not have a synergistic relationship, Granger causality is tested for each model. The results of Granger causation are shown in Appendix A 

Table A2. Results of Granger’s causality test.

Model	Excluded		Equation	chi2	df	Prob > chi2
Model 1	AFD	→	GDP	3.3493	2	0.1870
	EC	→	GDP	7.1586	2	0.0280 **
	ALL	→	GDP	20.8700	4	0.0000 ***
	GDP	→	AFD	7.0421	2	0.0300 **
	EC	→	AFD	8.5733	2	0.0140 **
	ALL	→	AFD	10.2840	4	0.0360 **
	GDP	→	EC	30.2920	2	0.0000 ***
	AFD	→	EC	3.1659	2	0.2050
	ALL	→	EC	39.5070	4	0.0000 ***
Model 2	EC	→	AFD	11.1580	5	0.0480 **
	PS	→	AFD	4.9533	5	0.4220
	ALL	→	AFD	15.4730	10	0.1160
	AFD	→	EC	45.8530	5	0.0000 ***
	PS	→	EC	9.6037	5	0.0870 *
	ALL	→	EC	50.6000	10	0.0000 ***
	AFD	→	PS	19.5970	5	0.0010 ***
	EC	→	PS	6.3804	5	0.2710
	ALL	→	PS	20.1500	10	0.0280 **
Model 3	EC	→	AFD	11.1580	5	0.0480 **
	PS	→	AFD	4.9533	5	0.4220
	ALL	→	AFD	15.4730	10	0.1160
	AFD	→	EC	45.8530	5	0.0000 ***
	PS	→	EC	9.6037	5	0.0870 *
	ALL	→	EC	50.6000	10	0.0000 ***
	AFD	→	PS	19.5970	5	0.0010 ***
	EC	→	PS	6.3804	5	0.2710
	ALL	→	PS	20.1500	10	0.0280 **
Model 4	PS	→	EC	14.1000	5	0.0150 **
	CO${}_2$	→	EC	36.0450	5	0.0000 ***
	ALL	→	EC	40.1260	10	0.0000 ***
	EC	→	PS	2.0216	5	0.8460
	CO${}_2$	→	PS	3.2389	5	0.6630
	ALL	→	PS	3.5842	10	0.9640
	EC	→	CO${}_2$	31.1310	5	0.0000 ***
	PS	→	CO${}_2$	6.5185	5	0.2590
	ALL	→	CO${}_2$	39.4350	10	0.0000 ***
Model 5	AFD	→	GDP	4.8862	2	0.0870 *
	EC	→	GDP	5.8958	2	0.0520 *
	CO${}_2$	→	GDP	2.4567	2	0.2930
	ALL	→	GDP	25.2260	6	0.0000 ***
	GDP	→	AFD	6.3174	2	0.0420 **
	EC	→	AFD	15.5890	2	0.0000 ***
	CO${}_2$	→	AFD	15.1700	2	0.0010 ***
	ALL	→	AFD	31.2320	6	0.0000 ***
	GDP	→	EC	20.7500	2	0.0000 ***
	AFD	→	EC	1.8891	2	0.3890
	CO${}_2$	→	EC	1.1055	2	0.5750
	ALL	→	EC	42.2300	6	0.0000 ***
	GDP	→	CO${}_2$	0.1936	2	0.9080
	AFD	→	CO${}_2$	1.9974	2	0.3680
	EC	→	CO${}_2$	4.4610	2	0.1070
	ALL	→	CO${}_2$	14.7870	6	0.0220 **

*** indicates significance at the 1% level. ** indicates significance at the 5% level. * indicates significance at the 10% level. Similarly hereinafter

. A table of the results of the Granger causality test is shown in Appendix A Table A2, where the direction of the arrow (e.g., →) indicates the direction from the cause to the effect.

Results from the Granger causal test in Appendix A Table A2 lead to the following conclusions:

(1)

In Model 1, energy consumption is the cause of economic development and is significant at 5%. Economic development is the cause of agroforestry development and is significant at 5%. Energy consumption is responsible for agroforestry development and is significant at 5%. Economic development is responsible for energy consumption and is significant at 1 percent.

(2)

In Model 2, energy consumption is responsible for agroforestry development and is significant at 5%. Agroforestry development is responsible for energy consumption and is significant at 1%. Population numbers are responsible for energy consumption and are significant at 10%. Agroforestry development is responsible for population numbers and is significant at 1 %.

(3)

In Model 3, energy consumption is responsible for agroforestry development and is significant at 5%. Agroforestry development is responsible for energy consumption and is significant at 1%. Population numbers are responsible for energy consumption and are significant at 10%. Population numbers are responsible for energy consumption and are significant at 10%. Agroforestry development is responsible for population numbers and is significant at 1 %.

(4)

In Model 4, the population is responsible for energy consumption and is significant at 5%. CO${}_2$ emissions are responsible for energy consumption and are significant at 1%. Energy consumption is responsible for CO${}_2$ emissions and is significant at 1%.

(5)

In Model 5, agroforestry development is the cause of economic development and is significant at 10%. Energy consumption is responsible for economic development and is significant at 10%. Economic development is the cause of agroforestry development and is significant at 1%. Energy consumption is responsible for agroforestry development and is significant at 1%. CO${}_2$ emissions are responsible for agroforestry development and are significant at 1%. Economic development is responsible for energy consumption and is significant at 1%.

In summary, economic development, energy consumption, and CO${}_2$ emissions are the factors affecting agroforestry development. Economic development, agroforestry development, population numbers, and CO${}_2$ emissions are the factors that affect energy consumption. Energy consumption is the factor affecting economic development and CO${}_2$ emissions. Agroforestry development is the factor that affects population size and energy consumption.

4.4. Impulse Response Analysis

The results of the impulse response of Model 1 are shown in Figure 7, where agroforestry development (D2_AFD) is affected by a shock of one unit standard deviation to the VAR system: Agroforestry development (D2_AFD) absorbs the shock by decreasing and then increasing by the corresponding percentage unit, and this effect does not subside until after period 20. Energy consumption (D2_EC) is affected by the shock, decreasing and then increasing, reaching a peak in period 4 and slowly decreasing after period 20. Economic development (D2_GDP) is affected by the shock and absorbs it instantly, decreasing and then increasing cyclically, before leveling off after period 25.

The impact on the VAR system of a one-unit standard deviation shock to energy consumption (D2_EC) is as follows: Energy consumption (D2_EC) absorbs the shock, decreases, and then increases by a percentage of the corresponding unit, with the impact leveling off after period 10. Agroforestry development (D2_AFD) is affected by this shock and absorbs it instantaneously, decreasing and then increasing, reaching a peak in period 3 and slowly decreasing after period 10. Economic development (D2_GDP) and agroforestry development (D2_AFD) show similar changes.

The impact on the VAR system of a one-unit standard deviation shock to economic development (D2_GDP) is as follows: Economic development (D2_GDP) absorbs the shock, decreases, and then increases by a percentage of the corresponding unit, and this effect does not subside until after period 25. Agroforestry development (D2_AFD) is affected by this shock and absorbs it instantaneously, rising and then falling, and then continuing to fluctuate upwards until period 25. Energy consumption (D2_EC) and agroforestry development (D2_AFD) show similar changes.

The results of the impulse response of Model 2 are shown in Figure 8, where agroforestry development (D2_AFD) is affected by a shock of one unit standard deviation to the VAR system: Agroforestry development (D2_AFD) absorbs the shock by first decreasing and then increasing by the corresponding percentage of units, and this effect fades after 20 periods. Energy consumption (D2_EC) is affected by the shock, and it decreases and then increases, reaching a peak in period 4 and then a minimum in period 5, before slowly decreasing in period 20. Economic development (D2_GDP) is affected by the shock and absorbs it instantaneously, rising and then falling, before leveling off after period 25.

The impact on the VAR system of a one-unit standard deviation shock to energy consumption (D2_EC) is as follows: Energy consumption (D2_EC) absorbs the shock, decreases, and then increases by a percentage of the corresponding unit, and the impact remains until period 25. Agroforestry development (D2_AFD) is affected by this shock and absorbs it instantly, rising and then falling, reaching a minimum peak in period 5 and slowly declining after period 10. Population size (D2_PS) is affected by the shock and decreases and then increases, leveling off after period 20.

The impact on the VAR system of a one-unit standard deviation shock to population (D2_PS) is as follows: The impact of this shock to population (D2_PS) is absorbed instantaneously, decreases gradually, and then continues to fluctuate upwards until it levels off after period 20. Agroforestry development (D2_AFD) absorbs the shock instantaneously, rises, and then falls until it reaches a minimum peak in period 9 and increases by the corresponding percentage of units, which gradually subsides until period 20. Energy consumption (D2_EC) and agroforestry development (D2_AFD) show similar changes.

The results of the impulse response of Model 3 are shown in Figure 9, where agroforestry development (D2_AFD) is affected by a one-unit standard deviation shock to the VAR system: Agroforestry development (D2_AFD) absorbs the shock, decreases, and then increases by the corresponding percentage unit, and this effect subsides until after period 15. The impact of the shock on CO${}_2$ emissions (D2_CO${}_2$) is absorbed instantaneously, decreases, and then increases, fluctuating cyclically until it plateaus after period 15. Energy consumption (D2_EC) is affected by this shock and absorbs it instantaneously, decreasing and then increasing, reaching a minimum peak in period 3 and then slowly decreasing and leveling off after period 15.

The impact on the VAR system of a shock to CO${}_2$ emissions (D2_CO${}_2$) of one unit standard deviation is as follows: Agroforestry development (D2_AFD) absorbs the shock, rising and then falling, reaching a minimum peak in period 8, and gradually leveling off after period 10. After period 15, the effect slowly declines. Energy consumption (D2_EC) is affected by this shock and absorbs it instantaneously, rising and then falling, before declining and leveling off after period 10.

The impact on the VAR system of a one-unit standard deviation shock to energy consumption (D2_EC) is as follows: Agroforestry development (D2_AFD) absorbs the shock instantaneously, increases, and then decreases, and this effect subsides until after period 10. The effect will continue to rise and fluctuate until it subsides after period 10. Energy consumption (D2_EC) and CO${}_2$ emissions (D2_CO${}_2$) show similar changes and increase by a corresponding percentage of units.

The results of the impulse response of Model 4 are shown in Figure 10, where CO${}_2$ emissions (D2_CO${}_2$) are affected by a shock of one unit standard deviation to the VAR system: CO${}_2$ emissions (D2_CO${}_2$) absorb the shock, first decreasing and then increasing by the corresponding percentage of units, and this effect subsides until after period 15. Energy consumption (D2_EC) is affected by this shock and absorbs it instantaneously, rising and then falling, peaking in period 3, before declining and leveling off in period 10. Population size (D2_PS) is affected by this shock and absorbs it instantaneously, rising and then falling, before leveling off after period 20.

The impact on the VAR system of a shock to energy consumption (D2_EC) of one unit standard deviation is as follows: CO${}_2$ emissions (D2_CO${}_2$) absorb the shock instantaneously, falling and then rising, peaking in period 2, and leveling off after period 10. Energy consumption (D2_EC) is affected by this shock and absorbs it instantaneously, decreasing and then increasing, before declining after period 15. Population size (D2_PS) is affected by this shock and absorbs it instantly, decreasing and then increasing, before decreasing and leveling off after period 20.

The impact on the VAR system of a unit standard deviation shock to population size (D2_PS) is as follows: CO${}_2$ emissions (D2_CO${}_2$) absorb the shock instantaneously, rising and then falling, gradually rising and peaking in period 9, and then leveling off after period 20. Energy consumption (D2_EC) is affected by this shock and absorbs it instantaneously, rising and then falling, and then continuing to fluctuate upwards until period 25. Population size (D2_PS) is affected by this shock and absorbs the shock until it falls, reaching a minimum peak in period 5, then rising and leveling off after period 10.

The results of the impulse response of Model 5 are shown in Figure 11, where agroforestry development (D2_AFD) is affected by a shock of one unit standard deviation to the VAR system: Agroforestry development (D2_AFD) absorbs the shock by first decreasing and then increasing by a corresponding percentage unit, and this effect remains until period 30. CO${}_2$ emissions (D2_CO${}_2$) will absorb the shock instantaneously, decreasing and then increasing, and will continue to fluctuate, and this effect will not subside until period 30. Energy consumption (D2_EC) absorbs the shock, decreases, and then increases by a percentage of the corresponding unit, and this effect does not subside until period 30. Economic development (D2_GDP) is affected by this shock and absorbs it instantaneously, decreasing and then increasing, and this effect does not subside until period 30.

CO${}_2$ emissions (D2_CO${}_2$) are affected by a shock of one unit standard deviation to the VAR system: Agroforestry development (D2_AFD) absorbs the shock instantaneously, rising and then falling, reaching a minimum peak in period 2, which subsides until period 10. The effect slowly declines after period 10. Energy consumption (D2_EC) is affected by this shock and absorbs it instantaneously, rising and then falling, before continuing to fluctuate upwards, with the effect slowly declining and leveling off after period 10. Economic development (D2_GDP) is affected by this shock and absorbs it instantaneously, decreasing and then increasing, and this effect remains in place until period 30.

The impact of a unit standard deviation shock to energy consumption (D2_EC) on the VAR system is as follows: Agroforestry development (D2_AFD) absorbs the shock instantaneously, decreasing and then increasing, and this effect subsides after period 10. The impact of the shock on CO${}_2$ emissions (D2_CO${}_2$) is absorbed instantaneously, decreasing and then increasing, and gradually subsiding after period 10. Energy consumption (D2_EC) is affected by this shock and absorbs it, decreasing and then increasing, with the effect declining and leveling off after period 10. Economic development (D2_GDP) is affected by this shock and absorbs it instantaneously, decreasing and then increasing, and this effect remains in place until period 30.

The impact on the VAR system of a one-unit standard deviation shock to economic development (D2_GDP) is as follows: Agroforestry development (D2_AFD) absorbs the shock instantaneously, rising and then falling, and the impact remains unabated until period 30. CO${}_2$ emissions (D2_CO${}_2$) absorb the shock instantaneously, falling and then rising, and the impact remains unabated until period 30. Energy consumption (D2_EC) is affected by this shock and absorbs the shock, rising and then falling, and this effect does not subside until period 30. Economic development (D2_GDP) is affected by this shock and absorbs it instantaneously, decreasing and then increasing.

4.5. Variance Decomposition

Analysis of the variance decomposition of the VAR model captures information on the relative importance of each stochastic disturbance term in the system that has an impact on the endogenous variables, the contribution of each structural shock to changes in the endogenous variables, and so on.

The results of the variance decomposition of Model 1 are shown in Figure 12. Agroforestry development (D2_AFD) has a positive shock effect on itself, reaching its strongest point in period 2, then gradually weakening, rising slightly in period 5, and then leveling off. Agroforestry development (D2_AFD) has a positive impact on energy consumption (D2_EC), which is strongest in period 2, rises slightly in period 5, and then plateaus. Agroforestry development (D2_AFD) exerts a positive shock on economic development (D2_GDP), which continues to be high, declines slightly in period 4, rises slightly in period 5, and then levels off.

Energy consumption (D2_EC) has a slight positive impact on agroforestry development (D2_AFD). Energy consumption (D2_EC) has a significant positive impact on itself, reaching its strongest point in period 2, then gradually decreasing and leveling off after period 5. Energy consumption (D2_EC) has a slight positive impact on economic development (D2_GDP), which is not significant.

Economic development (D2_GDP) has a significant positive impact on agroforestry development (D2_AFD) and shows a continuous trend of strength. Economic development (D2_GDP) has a significant positive shock effect on energy consumption (D2_EC), which reaches its strongest point in period 3, then decreases slightly, and finally plateaus. Economic development (D2_GDP) exerts a significant positive shock effect on itself, reaching its strongest point in period 2, then decreasing slightly, and leveling off after period 5.

The results of the variance decomposition of Model 2 are shown in Figure 13, where agroforestry development (D2_AFD) has a significant positive impact on itself, reaching its strongest point in period 2, and then slightly weakening and leveling off. Agroforestry development (D2_AFD) has a positive impact on energy consumption (D2_EC) and tends to increase gradually. Agroforestry development (D2_AFD) has a significant positive impact on population size (D2_PS), which tends to increase gradually.

Energy consumption (D2_EC) has a slight positive impact on agroforestry development (D2_AFD), but the impact is not significant. Energy consumption (D2_EC) has a significant positive impact on itself, reaching its strongest point in period 2, then decreasing and leveling off after period 5. Energy consumption (D2_EC) has a slight positive impact on population size (D2_PS), which is not significant. Population size (D2_PS) has a slight positive shock on agroforestry development (D2_AFD), which is insignificant. Population size (D2_PS) has a slight positive shock on energy consumption (D2_EC), which is insignificant. Population size (D2_PS) exerts a significant positive shock on itself, reaching its strongest point in period 2, then decreasing slightly and leveling off after period 5.

The results of the variance decomposition of Model 3 are shown in Figure 14, which shows that agroforestry development (D2_AFD) has a significant positive impact on itself, reaching its strongest point in period 2, and then slightly weakening and leveling off. Agroforestry development (D2_AFD) has a significant positive impact on CO${}_2$ emissions (D2_CO${}_2$), which tend to increase gradually. Agroforestry development (D2_AFD) has a positive impact on energy consumption (D2_EC), which tends to increase gradually.

CO${}_2$ emissions (D2_CO${}_2$) have a slight positive impact on agroforestry development (D2_AFD), but the impact is not significant. CO${}_2$ emissions (D2_CO${}_2$) have a significant positive impact on themselves, reaching their strongest point in period 2, then gradually weakening and leveling off after period 5. CO${}_2$ emissions (D2_CO${}_2$) have a slight positive impact on energy consumption (D2_EC), but the impact is not significant.

Energy consumption (D2_EC) has a slight positive impact on agroforestry development (D2_AFD), but the impact is not significant. Energy consumption (D2_EC) has a slight positive impact on CO${}_2$ emissions (D2_CO${}_2$), but the impact is not significant. Energy consumption (D2_EC) exerts a significant positive shock on itself, reaching its strongest point in period 2, then decreasing slightly and leveling off after period 5.

The results of the variance decomposition of Model 4 are shown in Figure 15. CO${}_2$ emissions (D2_CO${}_2$) have a significant positive shock effect on themselves, reaching their strongest point in period 2, then weakening slightly, and leveling off after period 5. CO${}_2$ emissions (D2_CO${}_2$) have a positive impact on energy consumption (D2_EC), but the impact is not significant. CO${}_2$ emissions (D2_CO${}_2$) have a positive impact on population size (D2_PS), which tends to increase gradually.

Energy consumption (D2_EC) has a significant positive impact on CO${}_2$ emissions (D2_CO${}_2$), with an increasing trend. Energy consumption (D2_EC) has a significant positive impact on itself, reaching its strongest point in period 2, then decreasing slightly and leveling off after period 5. Energy consumption (D2_EC) has a slight positive impact on population size (D2_PS), which is not significant.

Population size (D2_PS) has a slight positive impact on CO${}_2$ emissions (D2_CO${}_2$), but the impact is not significant. Population size (D2_PS) has a slight positive impact on energy consumption (D2_EC), but the impact is insignificant. Population size (D2_PS) has a significant positive impact on itself, reaching its strongest point in period 2, and then decreasing and leveling off.

The results of the variance decomposition of Model 5 are shown in Figure 16, which shows that agroforestry development (D2_AFD) has a significant positive impact on itself, reaching its strongest point in period 2, and then slightly weakening and leveling off. Agroforestry development (D2_AFD) has a significant positive impact on CO${}_2$ emissions (D2_CO${}_2$), which tend to increase gradually. Agroforestry development (D2_AFD) has a significant positive impact on energy consumption (D2_EC), which is strongest in period 2, then weakens slightly, and levels off after period 5. Agroforestry development (D2_AFD) has a positive impact on economic development (D2_GDP) and tends to increase gradually.

CO${}_2$ emissions (D2_CO${}_2$) have a slight positive impact on agroforestry development (D2_AFD), which tends to increase gradually. CO${}_2$ emissions (D2_CO${}_2$) have a significant positive impact on themselves, reaching their strongest point in period 2, and then weakening slightly and leveling off. CO${}_2$ emissions (D2_CO${}_2$) have a slight positive impact on energy consumption (D2_EC), which is insignificant. CO${}_2$ emissions (D2_CO${}_2$) have almost no significant impact on economic development (D2_GDP).

Energy consumption (D2_EC) has a slight positive impact on agroforestry development (D2_AFD), but the impact is not significant. Energy consumption (D2_EC) has a positive impact on CO${}_2$ emissions (D2_CO${}_2$), but the impact is not significant. Energy consumption (D2_EC) exerts a significant positive shock on itself, reaching its strongest point in period 2, then decreasing slightly and leveling off after period 5. Energy consumption (D2_EC) has almost no significant impact on economic development (D2_GDP).

Economic development (D2_GDP) has a significant positive impact on agroforestry development (D2_AFD), which continues to show a strong trend. Economic development (D2_GDP) has a significant positive impact on CO${}_2$ emissions (D2_CO${}_2$), which continues to show a strong trend. Economic development (D2_GDP) has a significant positive impact on energy consumption (D2_EC) and shows a continuous strong trend. Economic development (D2_GDP) exerts a significant positive shock on itself, reaching its strongest point in period 2, then decreasing slightly and leveling off after period 5.

4.6. Forecast

The test results of several VAR models are shown in

Table 6. Test Results of five VAR models.

Model	Equation	Parms	RMSE	R-sq	chi2	p > chi2
Model 1	D2_GDP	7	0.0295	0.6365	47.2757	0.0000 ***
	D2_AFD	7	0.0466	0.3736	16.1022	0.0132 **
	D2_EC	7	0.5078	0.8307	132.4766	0.0000 ***
Model 2	D2_AFD	16	0.0594	0.5832	33.5843	0.0039 ***
	D2_EC	16	0.6872	0.8760	169.5574	0.0000 ***
	D2_PS	16	0.0116	0.7241	62.9938	0.0000 ***
Model 3	D2_AFD	13	0.0473	0.6066	38.5426	0.0001 ***
	D2_EC	13	0.5621	0.8756	175.8880	0.0000 ***
	D2_CO${}_2$	13	0.0594	0.6530	47.0547	0.0000 ***
Model 4	D2_EC	16	0.7412	0.8558	142.3793	0.0000 ***
	D2_PS	16	0.0146	0.5584	30.3524	0.0107 **
	D2_CO${}_2$	16	0.0703	0.6566	45.8892	0.0001 ***
Model 5	D2_GDP	9	0.0298	0.6668	54.0341	0.0000 ***
	D2_AFD	9	0.0393	0.5989	40.3189	0.0000 ***
	D2_EC	9	0.5247	0.8374	139.0061	0.0000 ***
	D2_CO${}_2$	9	0.0664	0.3696	15.8298	0.0449 **

*** indicates significance at the 1% level. ** indicates significance at the 5% level.

. From the test results in Table 6, it can be seen that the five VAR models fit well overall, with most of the p-values significant at the 1% level and only a few p-values significant at the 5% level.

The results of several different VAR models for agroforestry development (AFD) are shown in Figure 17. The overall trend shows that China’s AFD declines in 2021, then continues to rise in 2022 and 2023, declines slightly in 2024, and continues to rise in 2025.

The results of several different VAR models for energy consumption (EC) projections are shown in Figure 18. The overall trend shows a decrease in 2021, an increase in 2022, a slight decrease in 2023, a significant decrease in 2024, and a slight increase in 2025.

The results of several different VAR model projections for CO${}_2$ emissions (CO${}_2$) are shown in Figure 19. The overall trend shows an increase in 2021, a slight decrease in 2022, a slight increase in 2023, a significant decrease in 2024, and an increase in 2025.

5. Discussion

5.1. Research Shortcomings

There are three main shortcomings in this paper: (1) There are many other factors that influence CO${}_2$ emissions, and in this paper we have only explored the impact of economic development, agricultural and forestry development, energy consumption, and population numbers on CO${}_2$ emissions. However, in reality, these factors may not be the main influencing factors, for example, there are also industrial production, power systems, and other production and operation activities that may emit more CO${}_2$. (2) We have only studied 31 years of data, from 1990 to 2020, in China, and although we have selected 103 secondary indicators, we believe that this amount of data is far from sufficient. (3) Our study is only based on a linear relationship, but the effects of economic development, agroforestry development, energy consumption, and population size on CO${}_2$ emissions may be influenced by a non-linear relationship.

5.2. Future Perspectives

In the future, we will add more variables to study the factors influencing CO${}_2$ emissions in China, not only at the macro level to explore the relationships between them, but also at the micro level to investigate the causes of CO${}_2$ emissions in detail. In addition, we will collect as much indicator data as possible and increase the time span in order to add value and meaning to our research.

6. Conclusions and Implications

6.1. Conclusions

(1) There is a direct and significant positive relationship between economic development (GDP) and agroforestry development (AFD). There is an indirect relationship between economic development (GDP) and energy consumption (EC), with a significant positive effect. There is an indirect relationship between economic development (GDP) and energy consumption (PS), with a significant negative effect. There is both a direct and indirect effect between economic development (GDP) and carbon dioxide emissions (CO${}_2$), with an overall significant positive effect.

Economic development provides the space and resource base for the development of agriculture and forestry. Energy is one of the essential factors of production, and although the intensity of energy consumption has been decreasing in recent years, the total consumption has been increasing. At the same time, the capital accumulation brought by economic development provides the basis for industrial development and production and manufacturing, leading to a new round of energy demand. Economic development reflects the level of social civilization and the demographic needs of social development, and therefore the demographic structure is constantly adjusting, with economic development determining the natural changes in population. Generally speaking, the higher the level of economic development, the lower the fertility level and the lower the birth and death rates of the population. Therefore, economic development can indirectly affect CO${}_2$ emissions through people’s basic carbon emissions; industrial, agricultural, and forestry production; and energy consumption. Economic development can also directly affect carbon dioxide emissions, which may be due to the accelerated circulation of capital, products, and factors due to good economic performance and more frequent economic activities, thus affecting the carbon emission performance of industries and enterprises.

Like economic development, agroforestry development is an important part of the economic system and exhibits the same impact relationships.

The relationship between agroforestry development (AFD) and energy consumption (EC) is direct and significantly positive. There is an indirect effect between AFD and population size (PS), which is significantly negative. There is both a direct and an indirect effect between AFD and CO${}_2$ emissions, with an overall significant positive effect.

The relationship between energy consumption (EC) and population size (PS) is direct and significantly negative. The relationship between energy consumption (EC) and carbon dioxide emissions (CO${}_2$) is both direct and indirect, with an overall positive relationship. The relationship between population size (PS) and carbon dioxide emissions (CO${}_2$) is direct and negative.

Fossil energy sources of all types are the main source of carbon; therefore, the use of energy to participate in combustion activities and industrial production and manufacturing increases CO${}_2$ emissions. The effect of energy consumption on population size may arise from the demographic restructuring due to economic growth. In contrast, the negative effect of population size on CO${}_2$ emissions may originate from the fact that the level of reduction of carbon emissions per capita is much higher than that caused by the growth of population size.

(2) The results of the Granger causality test show that economic development, energy consumption, and CO${}_2$ emissions are the causes of the constraints on agroforestry development. Economic development, agroforestry development, population size, and CO${}_2$ emissions are influences that constrain energy consumption. Energy consumption constrains economic development and CO${}_2$ emissions. Agroforestry development will affect the number of people and energy consumption.

(3) Over the next three years, China’s agroforestry development will be affected by the impulse response of economic development, energy consumption, and CO${}_2$ emissions, and will show a declining development trend. Over the next three years, China’s energy consumption will be influenced by the impulse response of economic development, agroforestry development, population size, and CO${}_2$ emissions, with a decreasing development trend followed by an increasing development trend. Over the next three years, China’s CO${}_2$ emissions will be influenced by the impulse response of energy consumption and agroforestry development factors, showing a downward and then upward development trend.

6.2. Suggestions

Here are three policy suggestions.

(1) The premise of improving the ecological protection system is to improve the ecological civilization system. We start with the following three suggestions: first, improve the assessment and evaluation system of economic and social development; second, delineate the ecological protection red line, and construct an accountability system; third, establish safety laws and regulations, to improve the ecological environmental protection management system. To establish a complete ecological civilization protection system, ecological civilization system construction needs to keep pace with the times, constantly improve, and be based on top-level design. In the continuous practice of exploration, one must keep pace with the times, constantly improving the ecological civilization protection system, which requires actively mobilizing the masses to participate, with everyone involved.

(2) To establish a unified national agricultural and forestry market, improve the construction of factors and resource markets; strengthen the quality of services and commodity markets; and unify rules, standards, and procedures for supervision. Local governments can combine local characteristics to introduce modern management, intensive production, and unified marketing and purchasing using a national unified market platform. Furthermore, the internal consumption of resources in small-scale production should be reduced, and more use should be made of technological innovation, industrial upgrading, and smooth national circulation of agricultural and forestry products.

(3) The Chinese government should increase financial subsidies for agriculture and forestry, and local governments should take into account the actual local situation; formulate guidelines and programs for the region; strengthen supervision, monitoring, and enforcement; and implement subsidies to encourage the agriculture and forestry industries to contribute to the reduction of CO${}_2$ emissions. At the same time, the government should vigorously develop new energy industries such as wind energy and solar energy to reduce the consumption of fossil energy; strengthen the integration of industry, academia, and research to promote the development of green low-carbon technology and clean energy technology; and strengthen publicity to cultivate civic awareness so that everyone can be a practitioner and advocate of low-carbon development.

6.3. Managerial Implications

(1) This paper constructs a more scientific and comprehensive indicator system from the perspective of asymmetric impact thinking, and empirically analyzes the relationship between economic development, agricultural and forestry development, population size, energy consumption, and the impact of CO${}_2$ emissions. This paper provides theoretical references and references for China’s economic development and the achievement of the carbon peak target under the epidemic.

(2) At a time when the global epidemic is raging, China’s economic development is also in the doldrums. At the same time, the Chinese government is faced with a dilemma between environmental protection and economic development. If the Chinese government continues to promote its “double carbon” strategy, economic development will inevitably suffer. Population decline is a foregone conclusion, consumption is weak, and fertility rates continue to weaken; population growth is the best option for boosting domestic demand. This shows that the level of quality economic development and green development in the country is still low. At the same time, as a large producer, high energy consumption promotes carbon emissions, which also indicates a low level of green production technologies. At present, China has undertaken carbon-neutral work in the areas of pilot construction and promotion of carbon sink markets, clean energy research and development, promotion of new energy vehicles, capacity reduction and differential tariffs, certification of “carbon-neutral enterprises”, technology, and financial subsidies. This paper argues that strengthening the development of agroforestry complex management, accelerating the modern management and upgrading of agroforestry complex management, and vigorously developing an agroforestry complex management economy is the best choice to balance economic development and environmental protection.

6.4. Social Implications

The international situation is volatile, and economic uncertainty has increased, affected by the COVID-19 epidemic and war. China’s aspirations to be carbon neutral by 2060 are even more daunting. To facilitate the acceleration of the carbon neutrality process, we have examined the impact of EC, AFD, EC, and PS on CO${}_2$ emissions. In order to better describe these elements in China, we have chosen as many variables as possible. There are two main points: (1) multiple variables make the trends of change more stable and help to avoid anomalous changes of variables; (2) the construction of EC, PS, and ADF will enrich the whole research framework. This is because it is in line with the current situation of the Chinese economy and the complexity of the economic system itself.