Linking the UN SDGs and Sustainable Development Progress: The Case of China

Abstract

This study examines China’s progress in sustainable development (SD) from 1990 to 2019 toward the 2030 Agenda and 2030 carbon peak target. Through extended factor analysis with principal component analysis, the sustainable development index (SDI), including each economy, society, and environment subclass SDI, was derived, and the contribution and weight of each Sustainable Development Goal (SDG) to the SDI calculated. The main findings are: (a) A progressively improving SDI is derived, supporting the considerate achievement of China’s SD policies; (b) the SDI trend aligns with per capita GDP, implying that the state of economic development measured by per capita GDP well approximates the degree of SD; (c) most goals, especially those in the economy and society subclasses, show sustained improvement and contribute to the improvement of China’s SDI. However, it is still relatively fragile in the environment subclass and requires further strengthening. Therefore, China should continue strengthening ecological civilization construction to achieve the 2030 Agenda and 2030 carbon peak target.

JEL Classification C43, O10, O53

Keywords

UN Sustainable Development Goals China sustainable development factor analysis principal component analysis sustainable development index

Introduction

The international community has collaborated to enhance the well-being of the people worldwide by addressing three crucial dimensions: the economy, society, and environment. This endeavor has led to the establishment of Sustainable Development Goals (SDGs). Initially, in 2000, the United Nations (UN) introduced eight Millennium Development Goals (MDGs), targeted for completion in 2015. Building upon the progress made with the MDGs, in 2015, the UN embraced a more comprehensive approach, endorsing the 17 SDGs and setting a target completion year of 2030.

Since its reform and opening up in 1978, the Chinese government has ascribed great importance to sustainable development (SD) and formulated a series of policies and measures to ensure its execution. In September 2020, President Xi Jinping, during the 75th UN General Assembly, proposed the “dual-carbon” goals, stating that China aims to peak its carbon emissions by 2030 and achieve carbon neutrality by 2060. On September 22, 2021, the “Opinions on Thoroughly Implementing the New Development Philosophy and Doing Well in Carbon Peaking and Carbon Neutrality Work,” jointly issued by the Central Committee of the Communist Party (CCP) of China and the State Council, was introduced. It stipulates that by 2060, the proportion of non-fossil energy should exceed 80%.

Therefore, as SD transitions from theory to practice, it is imperative to assess the historical trends and progress of SD in China. As the largest developing country with a substantial population and limited resources per capita, China assumes a pivotal role in realizing the 2030 Agenda. The evaluation of the SD process gains added significance in the context of the “dual-carbon” goals. Assessing the SD progress becomes crucial not only for the overall SDGs but also for the specific achievement of the carbon peak target in 2030. However, only a few studies have measured the degree and weight to which the UN SDGs are reflected in SD.

To investigate SD progress on a national and regional level in China, several studies constructed SD index or utilized SD indicators based on their subjective evaluations (CCIEE, 2021; Chen, 1996; Kim et al., 2019; Zhang et al., 2019; Zhu et al., 2019). However, they did not use the UN 17 SDGs with 247 separate indicators as practical targets. Only a few studies utilized the UN SDGs to examine China’s SD performance (Xu et al., 2020; Sachs et al., 2022). Using 94 out of 247 indicators, Sachs et al. (2022) derive an aggregate sustainable development index (SDI) by taking the arithmetic mean of all standardized indicators of 163 countries. Albeit it enables a comparison of the relative achievements of SD among the countries in a particular year, the limitations include assigning equal weights to every SDG in SDI derivation, unavailability to conduct the objective time-series comparison for countries or separate indicators, and insufficient coverage of the indicators due to broad range of country selection.

Derivation of the SDI can be affected by various factors such as weights for relevant variables, range of selected indicators, data standardization, an imputation method for indicators in missing periods, etc. (see Singh et al., 2012 for an extensive review of various indexes). The most critical issue is the weights of the relevant indicators that reflect their contribution to the derived SDI. Most studies assume a constant weight for each variable with respect to the SDI (Chandrasekharan et al., 2013; Prakash et al., 2017; Sachs et al., 2022). However, this study introduces a novel approach to SDI measurement by assuming varying degrees of contribution from indicators. This is achieved by calculating weights for relevant indicators using an extended factor analysis (FA) with principal component analysis (PCA) (Barrera-Roldán Saldivar-Valdés, 2002; Fernando et al., 2012).

Furthermore, this study expands the existing literature on SD by thoroughly and objectively examining the status of SD in China from 1990 to 2019, utilizing 156 out of 247 UN SDG indicators. Compared to this study, the recent studies that derive the Chinese SDI do not use fruitful data or sophisticated methodology (Kim et al., 2019; Sachs et al., 2022; Xu et al., 2020; Zhang et al., 2019; Zhu et al., 2019). Most existing studies use indicators based on their subjective evaluations, while this study utilizes UN SDG indicators that are internationally recognized and reliable and further enable comparative analysis.

The main research results show an overall marked improvement in China’s SD. Substantial strides have been made across economic and social domains, especially evident in achieving Goals 1, 2, 3, 8, 9, and 17. In addition, some goals exhibited a steady trend in the early period but then demonstrated gradual improvement since the mid-late 2000s encompassing Goals 4, 6, 7, 10, 11, 15, and 16. However, the environmental aspect remains relatively fragile and needs to be further strengthened, especially in addressing Goals 13 and 14.

The remainder of this study is organized as follows. The next section reviews the existing literature, including China’s SD policies and frameworks. This is followed by the section describing the data used in the analysis and summarizes the SDG trend by subclass (economy, society, and environment). Next, we explain the methodology employed, that is, the extended FA with PCA method, followed by a summary of the estimation results, including overall SDI and subclasses SDI. The last section concludes with relevant policy implications.

Literature Review

Since the reform and opening up, China has been implementing national SD policies. China’s SD path can be divided into four stages: concept understanding (1980s), acceptance (1992–2002), advancement (2003–2011), and innovation (2012–present). In 1983, China established environmental protection as a basic national policy, promulgating a series of laws during this period that initially formed a legal framework for environmental protection.¹ Shortly after the United Nations Conference on Environment and Development (UNCED) held in Rio, in September 1992, the CCP of China and the State Council promulgated the “Ten Strategies for China’s Environment and Development,” which for the first time proposed the implementation of SD strategies in China. Later, China further incorporated SD in the Ninth Five-Year Plan (1996–2000)² and held various CCP symposiums related to SD during the second stage.

In July 2003, General Secretary Hu Jintao put forward the Outlook of Scientific Development, emphasizing “putting people first, fostering a comprehensive, coordinated, and sustainable development perspective, and promoting the comprehensive development of the economy, society, and individuals.” Since then, SD has become the basic requirement of the Outlook of Scientific Development. In 2005, a “Harmonious Society”³ was put forward as a strategic governance task. From this stage onward, China actively pursued an SD strategy and propelled research in the field of SD to new heights, maintaining a prolonged period of heightened prosperity.

Lastly, in the fourth stage, China embarked on a journey of profound implementation and innovation of the SD strategy. Under the leadership of Xi Jinping, the 18th National Congress of the Communist Party of China (CPC) in November 2012 emphasized the imperative of in-depth implementation of SD strategy. This marked the beginning of a concerted effort by government departments at all levels to execute SD strategies nationwide and rigorously evaluate their outcomes. For example, the “Proposal of the CPC Central Committee for the Formulation of the 13th Five-Year Plan (2016–2020) for National Economic and Social Development” was adopted in 2015, which combined “Green Development” with the development concepts of innovative development, coordinated development, open development, and inclusive development to form five concepts of development (Kuhn, 2016; Yang, 2015).

The aforementioned SD progress indicates China’s commitment and resolve to achieve its goal to execute the SD strategy. To measure the degree of SD policy performance in a country or region, a scientific, practical, and stable evaluation index system needs to be established. As SD moves from theory to practice, scholars in China, including the State Planning Commission, the National Bureau of Statistics, Tsinghua University, Peking University, and other research institutions, have researched China’s SDI system from different perspectives and scales.

Among them, the most representative recent research is the “Blue Book of Sustainable Development: Evaluation Report on the Sustainable Development of China (2021)” (CCIEE, 2021). Since 2018, this annual report constructs the China Sustainable Development Indicator System (CSDIS), in which the framework consists of five themes: economic development, social and people’s livelihood, resources and environment, consumption and emission, and governance and protection. CCIEE (2021) presents the continuous upward trend in China’s SDI between 2015 and 2019. While the CSDIS is divided across national, provincial, and metropolis levels, using an equal weighting method to derive the index along with subjective indicators makes SD progress incomparable with other countries.

Likewise, a number of studies utilized unique indicators based on their subjective evaluations to examine the SD progress of China (Chen, 1996; Kim et al., 2019; The Chinese Academy of Sciences, 1999; Zhang et al., 2019; Zhu et al., 2019). At the national level, focusing on the population, natural resources, ecological environment, and human settlements, Chen (1996) emphasizes the urgency and necessity of implementing SD in China. The Chinese Academy of Sciences (1999) utilizes five levels and 208 indicators for China’s SDI system and has been reporting accordingly and publishing annually since then. Furthermore, Zhu et al. (2019) assessed the overall changes in China’s SD using 53 evaluation indicators from 2012 to 2016. Their findings indicate that the overall SD continuously improved, and the growth trend was maintained throughout the years.

At the sub-national levels, Zhang et al. (2019) assessed the SD level of “Five Modernizations” (industrialization, informatization, urbanization, agricultural modernization, and greenization) and its determinants with respect to the scale and economic level by using the panel data of 283 prefecture-level cities and 17 indicators from 2006 to 2015. In addition, as a cross-country analysis, employing the cluster analysis, Kim et al. (2019) compared China’s SD performance with 41 other economic transition countries using economic, social, and environmental indicators between 1990 and 2014, demonstrating that China has achieved the most remarkable improvement in SD.

Moreover, some studies have attempted to utilize UN SDGs to examine China’s SD progress and construct an SD evaluation index system (Sachs et al., 2022; Xu et al., 2020). Xu et al. (2020) assessed 17 SDGs at national and provincial levels in China using 119 indicators from 2000 to 2015. They found that China’s SDI score increased at the national and provincial levels during this period. Furthermore, in the most representative SD annual report since 2015, Sachs et al. (2022) utilized 94 indicators of UN SDGs to compare the SD progress of 163 countries. In the SD global rank, China ranked 76 out of 149 countries in 2016, 54 out of 156 countries in 2018, 48 out of 166 countries in 2020, 57 out of 165 countries in 2021, and 56 out of 163 countries in 2022. Although these studies attempted to quantify the SD progress in China using the unified SDGs, they used standardized indicators to derive the SDI by taking the simple arithmetic mean of the standardized indicators, thus assigning the same weight to all indicators.

Various methodologies measure the degree of sustainability in terms of corporate achievement, environment, well-being, human development, and innovation (Singh et al., 2012). However, studies have yet to measure the degree to which the UN SDGs are reflected in SD, relying on traditional methods to derive the SDI instead. The traditional method calculates an SDI for each goal by assigning the same weight to all indicators (Barrera-Roldán Saldivar-Valdés, 2002; OECD, 2008; Sachs et al., 2022).

For instance, Sachs et al. (2022) standardized indicators by using the minimum and maximum values of each indicator for all countries. Thus, the standardized SDI score of the sample countries depends on the relative value with respect to that of other countries. Further, as this method utilizes a simple arithmetic average of the standardized indicators or goals, the contributions of all standardized indicators to the SDI are weighted equally. Therefore, the state of SD reflects the relative status among all sample countries and does not suggest time-series implications of the same indicator.

To address the issue of deriving the SDI with uniform weights in existing literature or the challenges in conducting time-series comparisons, this study makes the first attempt to utilize the extended FA with the PCA method to derive the SDI, including each economic, social, and environmental SDI, assigning a different weight to each SDG with respect to the SDI. Additionally, by forming and analyzing samples over the period 1990–2019, this study makes time-series comparisons possible.

Data Description

Table 1 lists the number of indicators reflecting SDGs used in this study. A total of 247 indicators reflect the 17 SDGs. Since 16 indicators overlap among the original 247 indicators, 231 separate indicators have been used (UN homepage).⁴

Table 1.

Data Description.

Category	Goals	Description	SDG Indicators	CHN
Society	Goal 1	No poverty	13	8(8)
Society	Goal 2	No hunger	14	13(13)
Society	Goal 3	Good health and well-being	28	26(25)
Society	Goal 4	Quality education	12	9(9)
Society	Goal 5	Gender equality	14	7(7)
Environment	Goal 6	Clean water and sanitation	11	9(7)
Economy	Goal 7	Affordable and clean energy	6	6(6)
Economy	Goal 8	Decent work and economic growth	16	13(13)
Economy	Goal 9	Industry, innovation, and infrastructure	12	10(9)
Economy	Goal 10	Reduced inequalities	14	12(10)
Economy	Goal 11	Sustainable cities and communities	14	3(3)
Environment	Goal 12	Responsible consumption and production	13	10(7)
Environment	Goal 13	Climate action	8	3(3)
Environment	Goal 14	Life below water	10	7(3)
Environment	Goal 15	Life on land	14	12(8)
Society	Goal 16	Peace, justice, and strong institution	24	11(9)
Society	Goal 17	Partnerships for the goals	24	18(16)
	Total		247	177(156)

Notes: (1) CHN is the number of indicators for China among 247 Sustainable Development Goal (SDG) indicators. (2) Out of the total of 177 indicators available, 156 indicators are used in the analysis by excluding the data with only 1 year or the same values over years. Some unavailable indicators are replaced by proxy variables.

To utilize accurate data with the longest period, data were collected from various sources, such as the UN, World Bank, ILO, WHO, OECD, IHME, and Ocean Health Index. To find the long-run trend of SD targets, data were collected from 1990 to 2019. CHN in the last column of Table 1 lists the number of indicators available for China.

In this study, 177 indicators with replacement by proxy data were collected, but only 156 indicators were used for the derivation of the SDI. 21 indicators (with four overlapping) out of the 247 indicators are on the number of countries or the number of developing countries. Therefore, since they are not targeted indicators of a specific country, those indicators are excluded from the analysis, so the number of indicators that are used for analysis in this study is 156 out of 177.

The collected data are modified as follows. First, as many values are missing for some indicators, the data with the lowest number of missing years are initially selected. Subsequently, the missing values are replaced with their imputed values. To be specific, by exploiting the interpolation method, the missing values between the collected values are imputed by its mean, and the missing values before and after the collected values are replaced by the first and the last collected data, respectively. When the initial year is not 1990, the missing values are replaced by the oldest data available to avoid the case where the imputed value can be negative even though it should have positive values.

Second, the arithmetic mean of the indicators within each SDG was calculated to avoid the problem of having different number of indicators in each SDG. Furthermore, the data where the smaller values reflect improvement are converted to their inverse values such that the larger values indicate improvement. Lastly, Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy is calculated to examine the adequacy of the utilized dataset on employing FA. The KMO scored 0.7243, in which the values above 0.5 are considered satisfactory.

The UN SDGs in Table 1 are categorized into three dimensions: society, economy, and environment. UN (2016) suggested the five pillars of SDGs (People, Planet, Prosperity, Peace, and Partnership); and Elkington (1994) classified the Triple Bottom Line (TBL; Society, Economy, Environment). The “people” dimension of the SDGs is referred to as the “Society” principle of the TBL, “Planet” as the “Environment,” and “Prosperity” as the “Economy”. In other words, Goals 1–5 are classified into “Society,” Goals 7–11 into “Economy,” and Goals 6, 12–15 into “Environment” categories. Furthermore, Goals 16 and 17 were originally classified into “Peace” and “Partnership” dimensions of the SDGs, respectively, are clustered into “Society” category in accordance with the existing literature (Cai Choi, 2020; D’Adamo et al., 2021; Vinuesa et al., 2020).

Figures 1 –3 present each SDG trend by category between 1990 and 2019. Figure 1 shows the SD trends of the Society subclass, that is, Goals 1, 2, 3, 4, 5, 16, and 17. Goals 1, 2, 3, 4, and 17 present a gradually improving trend, while Goal 5 shows a U-shaped trend over the years with a trough in 2006, and Goal 16 exhibits a quite stable pattern despite some fluctuation until 2013, showing a large fall in 2014, and then recovers at a fast pace.

Figure 1.

Trends of Sustainable Development Goal (SDG) in the Society Subclass.

Strictly positive attributes include poverty, hunger, and health-related indicators, which correspond to Goals 1, 2, and 3, respectively. These include a noticeable decrease in the poverty rate, prevalence of malnutrition, mortality rate, traffic death, and incidence of tuberculosis, malaria, and hepatitis B. Additionally, there is an increase in government health expenditure, social protection benefits, population with access to basic sanitation services, crop production index, physicians, and vaccination. Also, improvement in quality education indicators (for Goal 4) such as pre-primary and tertiary school enrollment and the pupil–teacher ratio are considered positive attributes.

Gender, peace, justice, and strong institution indicators such as the proportion of seats held by women in the national parliament, homicide, and the proportion of voting rights of China in international organizations have also shown improvement (for Goals 5 and 16). In addition, indicators that represent partnership for the goals have shown an increase in tax revenue, fixed internet broadband subscriptions, individuals using the internet, Public–Private Partnership (PPP) for infrastructure, statistical capacity score, and methodology assessment of statistical capacity (for Goals 17).

Negative attributes include the 2008 Great Sichuan earthquake⁵ (for Goal 1), an increase in the consumer food price index in 2019 (for Goal 2), an increase in daily smoking and alcohol consumption (for Goal 3), a decrease in primary school enrollment and primary completion rate in the early period (for Goal 4), a consistent decrease in the female labor participation rate (for Goal 5), an increase in age-standardized prevalence of women and men aged 18–29 years who experienced sexual violence by age 18 and the slavery score (for Goal 16). Interestingly, for Goal 17, the net official development assistance (ODA) has turned negative since 2011, indicating that China became a donor country. Foreign direct investment (FDI) inflows and total ODA for technical cooperation increased gradually at the same time.

Figure 2 presents the SD trends of the Economy subclass, which are Goals 7, 8, 9, 10, and 11. While Goals 7, 10, and 11 show quite a steady trend in the early period and continuously improving trend afterward (except a significant decrease in 2008 for Goal 11), Goals 8 and 9 gradually increased since 1990.

Figure 2.

Trends of Sustainable Development Goal (SDG) in the Economy Subclass.

Positive attributes include indicators related to affordable and clean energy such as access to electricity and non-solid fuels, international financial flows to developing countries in support of clean energy research and development (R D) and renewable energy production (for Goal 7), and economic growth and work-related indicators such as an increase in mean nominal monthly wage of employees, decrease in material footprint, adolescent labor force participation rate, and age-standardized all-cause disability-adjusted life year (DALY) rates attributable to occupational risks (for Goal 8).

Indicators related to industry, innovation, and infrastructure, such as an increase in R D researchers, mobile cellular subscriptions, official flows for infrastructure, air transport passenger volume, and a decrease in CO₂ emissions per unit of GDP, are also positive attributes (for Goal 9). As for inequality indicators, there was an increase in the number of refugees, financial soundness, and the proportion of voting rights in international organizations such as the International Monetary Fund (IMF), a decrease in the Gini index, deaths and disappearances recorded during migration (for Goal 10). Sustainable cities and communities also improved, represented by the decrease in particulate matter with a diameter of 2.5 µm or less (PM2.5) in urban areas and the population living in slums (for Goal 11).

Negative attributes include a decrease in the ratio of renewable energy consumption to total final energy consumption (for Goal 7), a slowing down of the economic growth rate (for Goals 8 and 10), and an increase in the death rate in 2008 due to exposure to natural forces (for Goal 11).

Figure 3 shows the SD trends of the Environment subclass, which includes Goals 6, 12, 13, 14, and 15. Every SDG trend exhibits different patterns throughout the period; Goal 6 remains stable until 2016 and improves quite rapidly. Goal 12 moderately increases until 2015 and shows a small-scale fluctuation later. Goal 13 shows a stable trend, with a large fall in 2008. Goal 14 shows a deteriorating trend over the period with a slight increase between 2008 and 2014. Goal 15 shows a steady or slowly improving trend until 2010, increases quite rapidly until 2015, then deteriorates at the end of the period.

Figure 3.

Trends of Sustainable Development Goal (SDG) in the Environment Subclass.

Positive attributes include improvement in water and sanitation-related indicators such as access to improved water and sanitation, water use efficiency, and water body extent (for Goal 6), and a decrease in domestic material consumption per capita, combustible renewables, and waste, and an increase in the installed renewable electricity generation capacity and average compliance rate, which are related to responsible consumption and production (for Goal 12). A decrease in CO₂ emissions per capita was a main driver of the improvement among the climate action indicators (for Goal 13). An increase in marine protection areas and an improvement in fisheries indicators were the positive attributes among the sea-environment indicators (for Goal 14). Furthermore, various land-environment indicators have improved, such as the forest areas, protected freshwater sites, mountain key biodiversity areas (KBAs) covered by protected areas, and grassland loss index (for Goal 15).

Negative attributes include an increase in annual freshwater withdrawal (for Goal 6), an increase in total municipal waste (for Goal 12), an increase in total greenhouse gas (GHG) emissions throughout the period, an increase in the death rate in 2008 (for Goal 13), an increase in aquaculture production (for Goal 14), a decrease of the Red List Index of species survival throughout the period, and a large decrease of total ODA for biodiversity in 2018 (for Goal 15).

Model Specification

The FA and the PCA are frequently used for index construction, particularly when dealing with variables that exhibit high correlation. These methods focus on the variance and interrelationships between the variables. Fabrigar et al. (1999) argue that the FA, acknowledging random errors (unique variance), offers a more realistic model of the structure or correlations than PCA, which assumes no unique variance (Abeyasekera, 2005; Barrera-Roldán Saldivar-Valdés, 2002; OECD, 2008; Singh et al., 2012).

Expanding upon this concept, Fernando et al. (2012) integrated FA with the PCA method. First, FA is applied on the original indicator variables to identify their structure and subsets. Then, PCA is utilized on these subsets to define variability and assign weights. Thus, this study employs this enhanced FA–PCA method, as proposed by Fernando et al. (2012), to formulate the aggregate SDI. This methodology is effective in extracting a concise set of variables, called factors, from a large set of indicators.

In FA, the following model in Equation (1) is assumed.

\begin{matrix} z_{1} = b_{11} f_{1} + b_{12} f_{2} + \dots + b_{1 k} f_{k} + μ_{1} \\ z_{2} = b_{21} f_{1} + b_{22} f_{2} + \dots + b_{2 k} f_{k} + μ_{2} \\ . \\ . \\ z_{q} = b_{q 1} f_{1} + b_{q 2} f_{2} + \dots + b_{q k} f_{k} + μ_{q} \end{matrix}

(1)

Equation (1) implies that the observed and standardized $z_{i}$ s are explained by a linear combination of latent variables $(f_{j})$ for j $= 1, 2, \dots k$ . Here f is called a latent factor. b_ij for i = 1, 2, …, q and j = 1, 2, …, k are called factor loadings which are unobserved constants. $μ_{i}$ for i = 1, 2, …, q are called specific factors. Because latent variables are not observed, factor loadings and variance–covariance of specific factors should be estimated. This study follows PCA for this estimation (Harman, 1976).

The following assumptions are made. First, specific factors are assumed to have a zero mean, $E (μ_{i}) = 0$ , and the diagonal variance–covariance matrix, $C o v (μ) = Ψ$ , where

\begin{matrix} C o v (μ_{i} μ_{j}) = σ_{μ_{i}} i f i = j \\ C o v (μ_{i} μ_{j}) = 0 i f i \neq j \end{matrix}

(2)

Second, the latent variables and specific factors are assumed to be independent. Hence, Equation (3) holds.

C o v (f_{i}, μ_{i}) = 0

(3)

Third, observed variables are standardized to have a zero mean and unit variance.

z_{i} = \frac{x_{i} - M e a n (x_{i})}{σ_{x_{i}}}

(4)

Here, $M e a n (x_{i})$ and ${σ_{x}}_{i}$ represent the arithmetic mean and standard deviation of the originally observed variable $x_{i}$ , respectively, and $z_{i}$ is the standardized value of indicator $x_{i}$ . The mean of $z_{i}$ is zero, that is, $E (z_{i}) = 0$ and the variance is 1, that is, $V a r (z_{i}) = 1$ . Thus, FA is invariant to the rescaling of the variables. The indicators of SDGs have different units; therefore, standardization by standard deviation enables comparison of variables on a uniform scale.

Fourth, latent factors are standardized to have a zero mean and identity variance–covariance matrix.

\begin{matrix} E (f_{1}) = 0 and C o v (f_{i}, f_{j}) = 1 i f i = j \\ C o v (f_{i} f_{j}) = 0 i f i \neq j \end{matrix}

(5)

In matrix form, Equation (1) can be rewritten as follows:

Z = B F + U

(6)

where Z is a vector of observed and standardized variables, Z $= {[z_{1}, z_{2}, \dots, z_{q}]}^{T}$ for i = 1, 2, …, q, where $z_{i}$ is known as a latent variable. B is an unobserved factor loadings matrix, composed of q × k matrix of coefficients, b_ij for i = 1, 2, …, q and j = 1, 2, …, k. F is a vector of unobserved factors, $F = {[f_{1}, f_{2}, \dots, f_{k}]}^{T}$ for j = 1, 2, …, k, where $f_{i}$ is known as a latent factor. U is a vector of error terms, $U = {[μ_{1}, μ_{2}, \dots, μ_{q}]}^{T}$ , where $μ_{i}$ is known as a specific factor. The correlation between the latent variables (Z) and latent factors (F) is called the factor loadings (B).

In Equation (6), all variables except for Z are not observed; therefore, they are estimated under specific constraints. Because $C o v (F) = E (F - E (F)) (F - E F^{T}) = E (F F^{T}) = I$ and $C o v (F U) = 0$ , the variance and covariance of Equation (6) can be found in Equation (7).

C o v (Z) = B B^{T} + C o v (U)

(7)

This study considers the following assumptions. First, observed standardized variables (Z) are standardized to have a zero mean and unit variance. Second, latent factors (F) are uncorrelated with each other. Third, specific factors (U) are assumed to have a zero mean, $E (μ_{i}) = 0$ , and are uncorrelated with each other. Finally, the latent factors (F) and specific factors (U) are assumed to be independent. Because latent factors, F, are not observed, this study derives them by extracting the common latent variables that are highly correlated, through the estimation of a linear combination of these variables. Each factor captures a certain amount of overall variance in the latent variables, that is, $q = k = 17$ in this study. The number of factors that are used in deriving SDI is selected by specific criteria such as the number of eigenvalues of the latent variables and/or the Kaiser (1974) criteria.

Then, orthogonal varimax rotation of axes of factors is conducted to maximize the variance of the factor to obtain a pattern of loadings on each factor that is as diverse as possible. This strategy amplifies the variance of the squared loadings within factors.

Finally, the SDI is derived by calculating the weight of each goal as Equation (8) and multiplying the derived weights and values of the 17 goals as Equation (9). The weight of each goal is calculated by the ratio of squared maximum factor loading for each goal to the sum of total squared maximum factor loadings.

W_{i} = weight of goal i = \frac{squared highest factor loading (B) for goal i}{\sum_{i =1}^{17} squared highest factor loading (B) for goal i}

(8)

SDI = W_{i} Z_{i}

(9)

Therefore, the total sum of all weights is 1, and the weight assigned to each goal indicates its contribution to the SDI. These weights reflect differences in importance or influence in the derivation of SDI.

Estimation Results and Derivation of the SDI

In applying the extended FA with the PCA to the indicators that reflect the SDGs, the first step involves deriving eigenvalues for all estimated latent factors and determining the accumulated proportions of variance explained by each factor. Table 2 presents the eigenvalues of 17 latent factors and their respective proportions. The eigenvalue measures the variance of the observed variables explained by each factor. The sum of all eigenvalues is the same as the number of variables because the variance of each variable is 1 by standardization. Consequently, the sum of all eigenvalues in the model is 17, aligning with the number of variables. The column for “Difference” in Table 2 denotes the disparity in eigenvalues between each factor and the factor that follows it.

Table 2.

Eigenvalues and Selection of Factors.

Factor	Eigenvalue	Difference	Proportion	Cumulative
Factor 1	11.7453	9.6070	0.6909	0.6909
Factor 2	2.1383	0.7269	0.1258	0.8167
Factor 3	1.4114	0.6191	0.0830	0.8997
Factor 4	0.7923	0.5040	0.0466	0.9463
Factor 5	0.2883	0.1181	0.0170	0.9633
Factor 6	0.1703	0.0441	0.0100	0.9733
Factor 7	0.1262	0.0215	0.0074	0.9807
Factor 8	0.1047	0.0107	0.0062	0.9869
Factor 9	0.0940	0.0435	0.0055	0.9924
Factor 10	0.0505	0.0250	0.0030	0.9954
Factor 11	0.0255	0.0052	0.0015	0.9969
Factor 12	0.0203	0.0076	0.0012	0.9981
Factor 13	0.0127	0.0025	0.0007	0.9988
Factor 14	0.0102	0.0029	0.0006	0.9994
Factor 15	0.0073	0.0049	0.0004	0.9998
Factor 16	0.0023	0.0018	0.0001	1.0000
Factor 17	0.0006	.	0.0000	1.0000
Total	17.0000		1.0000

Note: LR test: independent versus saturated: chi²(136) = 1,042.76 (Prob > chi² = 0.0000).

The proportion indicates the relative weight of variance explained by each factor in the total variance. Therefore, the sum of all proportions is 1. Specifically, Factor 1 explains 69.09% (=11.7453/17) of the total variance, implying a 69.09% contribution of Factor 1 in the variance of SDGs used in the study. The column for “Cumulative” in Table 2 shows the accumulated proportion of the total variance explained by each factor, along with all preceding factors. According to the “Cumulative” column, the first three factors together contribute 89.97% to the total variance.

As a second step, an optimal number of factors must be selected using appropriate criteria. This study uses the Kaiser (1974) criteria and the scree plot method. Using the Kaiser criteria, factors with eigenvalues higher than 1, which are the three factors listed in Table 2, are selected. Also, from the scree plot, the same three factors are selected for analysis.

In the third step, to find the position that encompasses the actual data better overall, orthogonal rotations are undertaken under the assumption that the factors are not correlated. As shown in Table 3, following the rotation, the three factors collectively account for 89.97% of the total variances. Here, the contribution of the first two factors is 81.66%, and the third factor contributes only 8.31 percentage points (represented as 0.0831).

Table 3.

Orthogonal Rotation and Variances.

Factor	Variance	Difference	Proportion	Cumulative
Factor 1	8.0087	2.1351	0.4711	0.4711
Factor 2	5.8736	4.4611	0.3455	0.8166
Factor 3	1.4125	–	0.0831	0.8997

Note: LR test: independent versus saturated: chi²(136) = 1,042.76 (Prob > chi² = 0.0000).

In the fourth step, factor loadings are derived. Loading values are the weights and correlation coefficients of each variable, with a factor ranging between –1 and 1. The higher the loading values, the more relevant they are in explaining the goals based on relevant factors. In other words, they reflect the degree of contribution of each factor to the 17 goals. Table 4 shows the list of loading values. The combination of goals and factors is determined by the maximum contribution of each factor to the goals. For example, Factors 1–3 contributed 0.8318, 0.4730, and 0.1554, respectively, to the first goal. Thus, Goal 1 is decided mainly by Factor 1. In the same way, Goal 2 is explained mainly by Factor 1. The combinations of all goals and factors are denoted in bold.

Table 4.

Rotated Factor Loading and Uniqueness.

Variable	Factor 1	Factor 2	Factor 3	Uniqueness
Goal 1	0.8318	0.4730	0.1554	0.0603
Goal 2	0.9310	0.3144	0.0407	0.0327
Goal 3	0.9321	0.3544	0.0147	0.0054
Goal 4	0.6808	0.6372	–0.1906	0.0941
Goal 5	0.3251	0.8143	–0.1959	0.1928
Goal 6	0.4566	0.8034	0.0804	0.1396
Goal 7	0.5679	0.7878	–0.0667	0.0524
Goal 8	0.8282	0.4735	0.0428	0.0880
Goal 9	0.8273	0.5373	–0.0485	0.0246
Goal 10	0.3320	0.9127	–0.0134	0.0565
Goal 11	0.2332	0.7963	0.5188	0.0424
Goal 12	0.9294	0.2574	–0.1028	0.0594
Goal 13	–0.0301	–0.0412	0.9883	0.0206
Goal 14	–0.8917	0.2594	–0.0469	0.1354
Goal 15	0.7383	0.5733	–0.1155	0.1129
Goal 16	0.0109	0.6649	–0.1565	0.5332
Goal 17	0.8608	0.4520	–0.0017	0.0547
Proportion	47.11%	34.55%	8.31%	–

In the fifth step, the combination of factors and goals is decided from factor loadings. Goals 1, 2, 3, 4, 8, 9, 12, 14, 15, and 17 are well explained by Factor 1. Goals 5, 6, 7, 10, 11, and 16 are explained by Factor 2, and Goal 13 is explained by Factor 3.

Using the three factors in Table 4, Figure 4 shows their trends. The first factor shows a continuously increasing trend with a small decrease in 2016, which is similar to the trends of Goals 1, 2, 3, 4, 8, 9, 12, 14, 15, and 17. In addition, similar to Goals 5, 6, 7, 10, 11, and 16, Factor 2 shows a U-shaped trend, with a turning point as of 2008. Factor 3 shows a trend similar to Goal 13, with a significant drop in 2008, followed by a steady trend.

Figure 4.

Trends of Selected Factors.

Then, as a sixth step, the weight to each goal is calculated by using the information of factor loadings and their squared values. Table 5 combines the maximum factor loading for each goal and derives the weight of each goal to the SDI (OECD, 2008). The weight of each goal is calculated by the ratio of squared maximum factor loading for each goal to the sum of total squared maximum factor loadings (see Equation (8)). This is because while the value of factor loading is a simple correlation between the goal and factor, the squared value of factor loading represents the proportion of variance of the goal explained by that factor loading. This is to normalize the weight to be between 0 and 1.

Table 5.

Contribution of Each Goal to Sustainable Development Index (SDI).

Goals	Factor Loading	Squared Factor Loading	Weight
Goal 1	0.8318	0.6919	0.0576
Goal 2	0.9310	0.8668	0.0721
Goal 3	0.9321	0.8688	0.0723
Goal 4	0.6808	0.4635	0.0386
Goal 5	0.8143	0.6630	0.0552
Goal 6	0.8034	0.6454	0.0537
Goal 7	0.7878	0.6206	0.0516
Goal 8	0.8282	0.6859	0.0571
Goal 9	0.8273	0.6844	0.0569
Goal 10	0.9127	0.8330	0.0693
Goal 11	0.7963	0.6341	0.0527
Goal 12	0.9294	0.8638	0.0719
Goal 13	0.9883	0.9767	0.0812
Goal 14	–0.8917	0.7951	0.0661
Goal 15	0.7383	0.5451	0.0453
Goal 16	0.6649	0.4421	0.0368
Goal 17	0.8608	0.7410	0.0616
Sum	12.4357	12.0214	1.0000

Note: The sum of squared factor loadings implies total variance explained by all 17 goals.

The denominator is the total variance explained by the three retained factors. Thus, the weight is calculated by dividing the values of the factor squared by the total variance explained by the three factors. Table 5 lists the factor loading selected in Table 4 and the calculated weights. For example, the weight of Goal 1 is 5.76% and that of Goal 2 is 7.21%. From this calculation, the maximum weight is 8.12% for Goal 13, which means that Goal 13 has the highest contribution to the SDI. In other words, this means that Goal 13 plays the most significant role and holds the greatest influence in the calculation of the SDI.

The SDI is calculated by multiplying the derived weights and values of the 17 goals. Figure 5 shows the annual trends of the calculated SDI. China has shown a progressive improvement in SD during 1990–2019 with a large-scale fluctuation in 2008, reflecting the period of global financial crisis and an increase in the death rate due to the 2008 Great Sichuan earthquake. As most of the goals exhibit improvement over the years, the index also displays an increasing trend.

Figure 5.

Sustainable Development Index.

Furthermore, as a robustness check, because Goal 13 is mainly decided by Factor 3, SDI using only Factors 1 and 2 is calculated and compared in Figure 6. Excluding Factor 3, SDI shows a smaller fluctuation in 2008 and faster growth afterward. In other words, overall, it closely resembles the trend in Figure 5. This indicates that robustness is established.

Figure 6.

Annual Trends of Sustainable Development Indexes (SDIs) and GDP Per Capita.

Figure 6 additionally shows an annual trend of GDP per capita for China since 1990. SDI and GDP per capita generally show similar trends, although the former shows a noticeable fall in 2008, and the latter shows a continuously improving trend over the period. Therefore, this implies that while the state of economic development measured by per capita GDP well approximates the degree of SD, there are some differences as it does not cover social and environmental aspects of SD of a country. However, it is essential to engage in discussions regarding whether these results are robust or mere coincidences. A cautious approach is needed when interpreting a simplistic correlation, such as the SDI naturally is increasing as per capita GDP rises. As introduced earlier, in reality, China has demonstrated a significant interest in SD since the early stages of its reform and opening up policy. Through various developmental strategies, China has consistently propelled SD through comprehensive initiatives and innovations. Thanks to these ongoing efforts, China proudly announced at the COP28 held in Dubai in December 2023, the attainability of achieving a carbon peak before 2030. Therefore, similar trends observed in China’s SDI and GDP per capita can be seen as more than a mere coincidence; it is a consequential outcome of the Chinese government’s concerted efforts not only in economic growth but also in simultaneous endeavors for SD.

Finally, SDI subclasses are derived in accordance with the three categories classified in Table 1: Economy, Society, and Environment dimensions of SD. In other words, Society subclass includes Goals 1–5, 16–17, Economy subclass includes Goals 7–11, and Environmental subclass includes Goals 6, 12–15. Each SDI subclasses are represented by the factors with eigenvalues greater than 1. As demonstrated in Figure 7, the Society and Economy subclasses SDI present a gradually improving trend throughout the period. However, the Environment subclass SDI exhibits a stagnation in the mid-2000s and has shown recovery since the 2010s, in which the magnitude of the growth is slower than the other two subclasses SDI.

Figure 7.

Sustainable Development (SD) Subclasses Indexes.

The impacts of SD policies could be inferred from Figure 6, such as the inclusion of gender equality as a basic national policy in 2012 (Society subclass SDI Goal 5), full implementation of “taking targeted measures for poverty-alleviation (jingzhunfupin)” policy in 2014 (Society subclass SDI Goal 1), commitment to new normal economic growth in recent years focusing on improving the imbalance of regional development and distribution (Economy subclass SDI Goals 8 and 10), and implementation of the “South-to-North Water Diversion Project” in 2002 (Environment subclass SDI Goal 6).

Furthermore, the 2008 Great Sichuan earthquake has been linked to a decrease or slowdown of the subclasses SDI, especially the Environmental subclass SDI. Natural hazards are predicted to increase due to climate change, and the “number of deaths, missing persons and directly affected persons attributed to disasters per 100,000 population” is one of the major overlapping indicators (1.5.1/11.5.1/13.1.1) in the SDG. Therefore, due to the increase in mortality from these disasters, including the COVID-19 crisis, China should cultivate climate change resilience for SD by proactively responding to the climate change crisis and implementing relevant policy measures. This is directly and closely associated with achieving the 2030 carbon peak goal.

Conclusion

This study investigates the trend of SD progress and establishes China’s SDI by utilizing 156 out of 247 indicators of UN SDGs. Unlike the traditional methodology that uses a constant weight across all goals, the SDI in this study employs the extended FA with the PCA, assigning distinct weights to each SDG with respect to the SDI. To the best of our knowledge, this method has been used for the first time in the computation of the SDI.

The main findings are as follows. First, a progressively improving SDI is derived. It demonstrated a low value at the beginning of the 1990s, reaching the lowest value in 2008, but then showed a gradual improvement. The result signifies the effective implementation of SD policies by China. Interestingly, the trend of SDI is quite similar to that of the per capita GDP, suggesting that the status of SD is well approximated to the level of per capita GDP. However, this result is a consequence of the Chinese government’s extensive and ongoing efforts to promote SD alongside economic growth. It is crucial not to interpret this as indicating a direct and positive correlation between GDP per capita and the SDI.

Second, most of the goals show a continuous improvement over the period, indicating that they directly contribute to the trend of the SDI. These are Goals 1, 2, 3, 8, 9, 12, and 17.

Third, some goals show a steady trend in the early period but then demonstrate a gradual improvement since the mid-late 2000s. These are Goals 4, 5, 6, 7, 10, 11, 15, and 16. Finally, some of the goals (Goals 13 and 14) related to the environment remain fragile and require substantial improvement.

Overall, at the national level, China’s SD situation improves steadily with its economic growth. Concrete progress is mainly reflected in the economic and social fields, including poverty reduction, hunger reduction, good health and education, economic growth, and international partnerships. This improvement cannot be separated from the country’s related laws, policies, strategies, and implementation of large projects. For example, the implementation of the “taking targeted measures for poverty-alleviation” and the “Law of the People’s Republic of China on Compulsory Education,” policies related to gender equality, new normal growth strategies, and “South-to-North Water Diversion Project” all had a positive effect on the improvement of related goals. This demonstrates that China has taken practical steps toward the adoption of SD-based paradigm that recognizes the balance between economic, social, and environmental dimensions.

However, the carrying capacity of the environmental field is still weak with a deteriorating trend of individual indicators (Goals 13 and 14), which harms economic and social activities. Thus, to effectively advance the UN’s 2030 SD Agenda and achieve the goal of a carbon peak by 2030, China must introduce systematic response plans to dynamically maintain a virtuous circle of the economy, society, and environment. Additionally, it should continue to innovate its SD strategy and steadfastly implement the construction of an ecological civilization. Furthermore, an integrated strategy for achieving the SDGs and appropriate financing mechanisms should be well implemented, with the provision of proper guidance and monitoring of progress from both the public and private sectors.

China is currently actively advancing its digital economy, and it is crucial to harness this momentum for SD. China has evolved from being a follower to a crucial participant, contributor, and leader in global SD. Unfortunately, the indicators in the UN SDGs do not consider the digital economy factors well. As the largest developing country, China’s experience and achievements in SD, including the goal of achieving a carbon peak by 2030, will hold exemplary value. To fully realize the 2030 SD Agenda, China should enhance its official SD data systems to enable precise evaluation of SD progress and ensure international comparability.

This study also establishes the foundation for future research that can assess SD progress through cross-country panel data analysis, utilizing the unified UN SDG indicators alongside a well-suited SDI construction method.

Footnotes

Data Availability

All the data was obtained through an online database system; the links are mentioned in the references section.

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

This research was supported by the Korea University Grant (K2313141) of Korea.

Notes

ORCID iD

Seon Ju Lee

Sung Jin Kang

Shijun Cao

References

Abeyasekera

(2005). Multivariate methods for index construction. In Household sample surveys in developing and transition countries: Design, implementation and analysis (pp. 367–388). UN Department of Economic and Social Affairs. https://unstats.un.org/unsd/hhsurveys/pdf/Chapter_18.pdf

Barrera-Roldán

, Saldivar-Valdés

(2002). Proposal and application of a Sustainable Development Index. Ecological Indicators , 2(3), 251–256. https://doi.org/10.1016/S1470-160X(02)00058-4

Cai

, Choi

(2020). A United Nations’ Sustainable Development Goals perspective for sustainable textile and apparel supply chain management. Transportation Research Part E: Logistics and Transportation Review , 141, 1–21. https://doi.org/10.1016/j.tre.2020.102010

Chandrasekharan

, Kumar

R. S.

, Raghunathan

, Chandrasekaran

(2013). Construction of environmental performance index and ranking of states. Current Science , 104(4), 435–439. https://www-jstor-org-443.web.bisu.edu.cn/stable/24089743

Chen

(1996). Sustainable development strategy reader [in Chinese]. China Planning Press.

China Center for International Economic Exchanges (CCIEE). (2021). Blue book of sustainable development: Evaluation report on the sustainable development of China (2021) . Social Sciences Academic Press (China).

D’Adamo

, Gastaldi

, Imbriani

, Morone

(2021). Assessing regional performance for the Sustainable Development Goals in Italy. Scientific Reports , 11, 24117. https://doi.org/10.1038/s41598-021-03635-8

Elkington

(1994). Towards the sustainable corporation: Win-win-win business strategies for sustainable development. California Management Review , 36(2), 90–100. https://doi.org/10.2307/41165746

Fabrigar

L. R.

, Wegener

D. T.

, MacCaluum

R. C.

, Strahan

E. J.

(1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods , 4(3), 272–299. https://doi.org/10.1037/1082-989X.4.3.272

10.

Fernando

, Samita

, Abeynayake

(2012). Modified factor analysis to construct composite indices: Illustration on urbanization index. Tropical Agricultural Research , 23(4), 327–337. https://doi.org/10.4038/tar.v23i4.4868

11.

Harman

H. H.

(1976). Modern factor analysis . University of Chicago Press.

12.

(2005, February 20). Talk delivered to provincial main leading cadres in seminar about enhancing capacity to build a socialist harmonious society . Renmin Ribao (People’s Daily).

13.

Kaiser

H. F.

(1974). An index of factorial simplicity. Psychometrika , 39, 31–36. https://doi.org/10.1007/BF02291575

14.

Kim

S. J.

, Kang

S. J.

, Jung

T. Y.

, Cao

(2019). China and sustainable development performance in economic transition. Global Journal of Emerging Market Economies , 11(3), 149–164. https://doi.org/10.1177/0974910119887037

15.

Kuhn

(2016). Sustainable development discourses in China. Journal of Sustainable Development , 9(6), 158–167. https://doi.org/10.5539/jsd.v9n6p158

16.

Organisation for Economic Co-operation and Development (OECD). (2008). Handbook on constructing composite indicators: Methodology and user guide . OECD. https://doi.org/10.1787/9789264043466-en

17.

Prakash

, Teksoz

, Espey

, Sachs

J. D.

, Shank

, Schmidt-Traub

(2017). The U.S. cities Sustainable Development Goals index 2017 . Sustainable Development Solutions Network.

18.

Sachs

J. D.

, Lafortune

, Kroll

, Fuller

, Woelm

(2022). Sustainable development report 2022: From crisis to sustainable development: The SDGs as roadmap to 2030 and beyond . Cambridge University Press.

19.

Singh

R. K.

, Murty

H. R.

, Gupta

S. K.

, Dikshit

A. K.

(2012). An overview of sustainability assessment methodologies. Ecological Indicators , 15(1), 281–299. https://doi.org/10.1016/j.ecolind.2011.01.007

20.

The Chinese Academy of Sciences (1999). 1999 China sustainable development strategy report [in Chinese]. Science Press.

21.

United Nations Centre for Regional Development (UNCRD). (2009). Report on the 2008 Great Sichuan earthquake . UNCRD. https://uncrd.un.org/sites/uncrd.un.org//files/report_2009_report_on_the_2008_great_sichuan_earthquake.pdf

22.

United Nations (UN). (2016). Transforming our world: The 2030 agenda for sustainable development . UN. https://sdgs.un.org/sites/default/files/publications/21252030%20Agenda%20for%20Sustainable%20Development%20web.pdf

23.

Vinuesa

, Azizpour

, Leite

, Balaam

, Dignum

, Domisch

, Fellander

, Langans

S. D.

, Tegmark

, Nerini

F. F.

(2020). The role of artificial intelligence in achieving the Sustainable Development Goals. Nature Communications , 11(233), 1–10. https://doi.org/10.1038/s41467-019-14108-y

24.

, Chau

S. N.

, Chen

, Zhang

, Li

, Dietz

, Wang

, Winkler

J. A.

, Fan

, Huang

, Li

, Wu

, Herzberger

, Tang

, Hong

, Li

, Liu

(2020). Assessing progress towards sustainable development over space and time. Nature , 577, 74–78. https://doi.org/10.1038/s41586-019-1846-3

25.

Yang

(2015, November 20–30). About the proposal for the 13th five-year plan [special session]. 6th International Electronic Symposium on Chinese Politics, China. http://politica-china.org/wp-content/uploads/6sei-yang-weimin-EN.pdf

26.

Zhang

, Yuan

, Tian

(2019). Sustainable development in China: Trends, patterns, and determinants of the “Five Modernizations” in Chinese cities. Journal of Cleaner Production , 214, 685–695. https://doi.org/10.1016/j.jclepro.2018.12.307

27.

Zhu

, Sun

, He

, Zhao

(2019). Are SDGs suitable for China’s sustainable development assessment? An application and amendment of the SDGs indicators in China. Chinese Journal of Population Resources and Environment , 17(1), 25–38. https://doi.org/10.1080/10042857.2018.1544753