**2. Literature Review**

Urban carbon-emission trends, peak forecasting, and the related influencing factors have attracted considerable research attention. Regarding carbon-emission trends, the research clearly indicates that emissions are on the rise in China. Zhou et al. [8], for

**Citation:** Tong, X.; Guo, S.; Duan, H.; Duan, Z.; Gao, C.; Chen, W. Carbon-Emission Characteristics and Influencing Factors in Growing and

Shrinking Cities: Evidence from 280 Chinese Cities. *Int. J. Environ. Res. Public Health* **2022**, *19*, 2120. https:// doi.org/10.3390/ijerph19042120

Academic Editors: Roberto Alonso González Lezcano, Francesco Nocera and Rosa Giuseppina Caponetto

Received: 18 January 2022 Accepted: 8 February 2022 Published: 14 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

633

example, found that carbon emissions increased continuously from 1992 to 2013 in all Chinese cities, growing faster in eastern China than in central and western China. Zhu et al. [9] found that carbon emissions doubled from 1997 to 2012 in Tianjin. However, while total carbon emissions have shown an overall growth trend, emissions are decreasing in certain sectors. For instance, using multiscale input–output analysis, Hung et al. [10] found that carbon emissions in Hong Kong's construction industry decreased from 2004 to 2011. Meanwhile, with China's stated aim to reach carbon peak by 2030, many studies have assessed its feasibility, arriving at two opposing views: "can achieve" and "cannot achieve." For example, using data for 50 Chinese cities, Wang et al. [11] predicted that emissions in those cities would peak between 2021 and 2025. Huang et al. [12] similarly found that under existing policies, Guangzhou would reach peak carbon in 2023. However, Lin et al. [13] found that the limited use of clean energy and ongoing rapid economic growth in Xiamen would cause it to reach carbon peak later than the 2030 target. Likewise, Zhang et al. [14] conducted three simulations on the timing of peak carbon in Baoding, and two of the scenarios indicated that peak carbon would not be achieved until 2040.

The factors affecting carbon emissions are another important area investigated in the research. Many studies have shown that socioeconomic factors, such as economic development, population growth, technology, industrial structure, and energy structure, have important effects on urban carbon emissions [8,15–17]. Generally speaking, economic development and population growth increase the demand for energy, resulting in an increase in carbon emissions. Ou et al. [18], for example, studied the socioeconomic factors affecting carbon emissions in cities with different developmental levels in China and found that economic and population growth increased emissions in cities at all levels. Taking 128 countries as samples, Dong et al. [19] found that countries with larger populations consumed more energy and thus generated more carbon emissions. Economic growth, however, does not always increase emissions. When economic development reaches a certain level, carbon emissions will decline; that is, the economy has an inverted U-shaped effect on carbon emissions. Studying 276 large cities around the world, Fujii et al. [20], for example, found that emissions first rose and then decreased with economic growth. Technological progress and industrial structure optimization typically improve energy efficiency and reduce emissions. Wang et al. [21], for instance, found that technology was negatively correlated with carbon emissions. Meanwhile, Li et al. [22] suggested that reducing the proportion of secondary industry and prioritizing the development of tertiary industry could be beneficial for reducing emissions in Chinese cities. Optimizing the energy mix means increasing the use of clean energy, which directly leads to a decrease in carbon emissions. Boluk et al. [23], for example, found that electricity generation using renewable energy helped to improve environmental conditions in Turkey. Similarly, Xu et al. [24] suggested that if China slows down its energy consumption and shifts toward low-carbon fuels, its emission targets could be feasible. Although the abovementioned socioeconomic factors affect carbon emissions, the effects are different. Economic development and population growth tend to increase emissions, while technological progress and industrial and energy structure optimization can decrease emissions.

The existing research tends to study all cities together without distinguishing between them. In fact, with ongoing economic development and population mobility, cities currently face two different development states: growth and shrinkage [25,26]. Different cities have different characteristics in terms of economies and population, and their effects on the environment are also different [27–29]. Studies have confirmed that growing and shrinking cities exhibit different energy-efficiency and carbon-emission characteristics. For example, after classifying growing and shrinking cities based on a population index, Xiao et al. [30] found that emissions in rapidly shrinking cities presented a continuously increasing trend, while growing cities reached their emission peaks during 2011–2013. Liu et al. [31] also used a population index to study emissions in growing and shrinking cities and found that urban shrinkage increased emissions and that the energy efficiency of shrinking cities was lower than that of growing ones.

Our review of the literature reveals limitations in the existing research. First, although many studies have focused on the characteristics of and factors affecting urban carbon emissions, they tend to ignore the potential differences between different types of cities. Second, although some studies have comparatively investigated the emissions of shrinking and growing cities, the classifications of those cities are mostly based on a single population index, lacking comprehensive classification. In truth, in addition to population changes, urban growth and shrinkage also involve economic, social, and land-use factors [32]. In light of the above, this study constructs an index, called "urban development degree," using economic, demographic, social, and land-use indicators. Then, cities are divided into growing and shrinking cities, and their carbon-emission characteristics and related influencing factors are investigated. The findings can provide a reference for emissionmitigation policies.

There are three contributions of this paper. First of all, we focus our research on the city level rather than the national or provincial level, which is a further complement to the micro-level research on carbon emissions. Secondly, we divide different types of urban development patterns by using a comprehensive indicator calculated from socioeconomic indicators, rather than researching all cities under a unified framework. Thirdly, we conduct an in-depth analysis of the influencing factors of carbon emissions in different types of cities, and analyze the reasons for the differences in carbon emissions, so as to provide a targeted reference for the formulation of carbon emission reduction policies and low-carbon development paths for cities with similar characteristics.

#### **3. Method**

#### *3.1. Research Framework*

Figure 1 presents a schematic framework of this study's methodological approach. The framework consists of three parts. The first divides the growing and shrinking city groups and calculates their carbon emissions. The second part compares the social development and carbon-emission characteristics of four groups of cities and uses extended STIRPAT (stochastic impacts by regression on population, affluence, and technology) to test the factors affecting carbon emission. Finally, the third part involves discussing the results and presenting the conclusions.

#### *3.2. Categorization of Shrinking and Growing Cities*

There are two ways to classify growing and shrinking cities. One is to consider the change in population over a certain period, where an increase in population is identified as urban growth, and a decrease is identified as urban shrinkage [6,28,33]. The other way is to consider the change in nighttime light over a certain period; when nighttime light brightens, the city is growing, and when it dims, the city is shrinking [34–36]. However, population division can only reflect changes in the urban population, and while nighttime light can reflect the economy and population, the data are not continuous in time and are characterized by uncertainty, which will affect the results. Therefore, some scholars try to use a comprehensive index to identify growing and shrinking cities. For example, Lin et al. [37] believe that urban growth and shrinkage are characterized by the changes in economy, population, land use and finances. Then, they construct a comprehensive index called urban development degree by using total population, economic growth, employment and unemployment and built-up area data to evaluate growing and shrinking cities in China. Zhang et al. [38] determine that the growth and shrinkage of cities are characterized by the changes in population, economy and social consumption, and shrinking cities often face economic downturns, shrinking population and declining spending power. Referring to Lin et al. [37] and Zhang et al. [38], this study, therefore, constructs an index of urban development degree (UDD) to divide growing and shrinking cities.

**Figure 1.** Research framework.

Calculating UDD requires the following demographic, economic, and social indicators: (1) Population includes three indicators, natural population growth rate, total population, and population density. Natural population growth rate refers to the difference between birth rate and death rate, which represents the change in natural population growth. Total population refers to the registered urban population, representing the change in the total population. Population density refers to the number of people per unit of land, representing the change in population density. (2) The economy includes per capita GDP (gross domestic product), per capita fiscal revenue, and GDP growth rate. Per capita GDP is the output level of unit population, and per capita fiscal income is the fiscal income of unit population, representing the level of urban economic development. GDP growth rate is the percentage increase in urban output, representing the speed of economic development. (3) The social and land-use dimensions include three indicators: total retail sales of consumer goods, per capita fiscal expenditure, and built-up area. Total retail sales of consumer goods represent a city's consumption capacity. Per capita fiscal expenditure is the level of fiscal expenditure per unit of population, representing the government's service capacity. Built-up area refers to the actual developed area in a city, representing the spatial change in urban land use. See Table 1 for details.


**Table 1.** Urban development degree indicators selection.

Equations (1)–(8) show the calculation process for UDD. When Xij is positive,

$$\chi'\_{\rm ij} = \frac{\chi\_{\rm ij} - \min \chi\_{\rm j}}{\max \chi\_{\rm j} - \min \chi\_{\rm j}},\tag{1}$$

When Xij is negative,

$$\chi'\_{\rm ij} = \frac{\max \chi\_{\rm j} - \chi\_{\rm ij}}{\max \chi\_{\rm j} - \min \chi\_{\rm j}},\tag{2}$$

$$\mathcal{Y}\_{\text{ij}} = \mathcal{X}\_{\text{ij}}^{\prime} / \sum\_{i=1}^{\text{m}} \mathcal{X}\_{\text{ij}}^{\prime} \tag{3}$$

$$\mathbf{e}\_{\mathbf{j}} = -\mathbf{k} \sum\_{i=1}^{\mathbf{m}} (\mathbf{Y}\_{\mathbf{i}\mathbf{j}} \times \ln \mathbf{Y}\_{\mathbf{i}\mathbf{j}})\_{\prime} \text{ \(}\mathbf{k} = \frac{1}{\ln(\mathbf{m})}\text{\)}\tag{4}$$

dj = 1 − ej, (5)

$$\mathcal{W}\_{\mathbf{j}} = \mathbf{d}\_{\mathbf{j}} / \sum\_{\mathbf{j}=1}^{n} \mathbf{d}\_{\mathbf{j}} \tag{6}$$

$$\text{index}\_{\text{it}} = \sum\_{\mathbf{j}=1}^{n} (\mathbf{W}\_{\mathbf{j}} \times \mathbf{X}\_{\mathbf{ij}}')\_{\text{'}} \tag{7}$$

$$\text{UDD}\_{\text{(it\_0it\_1)}} = \text{index}\_{\text{it\_1}} - \text{index}\_{\text{it\_0}} \tag{8}$$

where i represents the city, j represents the indicator, t represents the year, n represents the number of indicators, and m represents the number of cities. Equations (1) and (2) are the process for standardizing the original data. Xij and X ij represent the original data and standardized data of the j index of city I, respectively. Equations (3)–(6) show the process for calculating the index weight; Wj is the index weight. Equations (7) and (8) show the process for calculating UDD. indexit represents the urban development index, and UDD(it0,it1) represents UDD from t0 to t1.

Following the literature on growing and shrinking cities [6,28], we identify a city whose UDD(it0,it1) < 0 as a shrinking city and a city whose UDD(it0,it1) ≥ 0 as a growing city. In addition, based on research on group divisions and the development characteristics of urban populations and economies [29], we divide the cities into the following four groups: (1) rapidly growing cities (RGCs), UDD ≥ 0.05, which have rapid population, economic, and consumption development; (2) slightly growing cities (SGCs), 0 ≤ UDD < 0.05, in which the population increases and the economy and consumption develop steadily; (3) rapidly shrinking cities (RSCs), UDD < −0.02, which show sharp declines in population, economy, and consumption; and (4) slightly shrinking cities (SSCs), −0.02 ≤ UDD < 0, which are characterized by population decreases and slow growth in consumption and the economy.
