-
Hegang is located in the northeast of Heilongjiang Province, China. It is one of the central cities in the northeast of Heilongjiang Province and an important national energy city. Fig. 1 shows the geographical location of Hegang. Hegang covers an area of 14 684 km2 and has six municipal districts, namely Dongshan District, Nanshan District, Xiangyang District, Gongnong District, Xingshan district, Xingan district. Additionally, there are two counties under the jurisdiction of Hegang, respectively, Luobei and Suibin. Within the 21 resource-based cities in Northeast China defined by The National Plan for Sustainable Development of Resource-based Cities (2013–2020), Hegang is classified as a recession group (He et al., 2017).
According to the Seventh National Census of China, the total population of Hegang is about 891 thousand in 2020. Compared with approximately 1058 thousand people in the Sixth National Census in 2010, the population of six districts and two counties in the city decreased about 167 thousand, a decrease of 15.81%, and the annual average decline rate is 1.7%. Among the population composition of Hegang, the elderly population aged 60 and above have increased 63 thousand comparing with the Sixth National Census, and the proportion increased by 9.75%. The proportion of the population aged over 60 is 1.1% higher than that of the whole province and 5.62% higher than that of the whole country. In recent years, the GDP of Hegang has shown a trend of slow growth, increasing from 25.1 billion yuan (RMB) in 2010 to 34.02 billion yuan (RMB) in 2020 (Hegang Municipal People’s Government, 2021). The proportion of industrial structure in Hegang has evolved from 24: 50: 26 in 2010 to 30.5: 29.3: 40.2 in 2020 (Hegang Bureau of Statistics, 2010–2020). Fig. 2 shows the change trend of population and industrial contribution rate in Hegang. Finally, it is worth mentioning that Hegang is faced with many problems, such as lagging behind of economic development, resource depletion, flabby extended and substitute industry, an unbalanced industrial structure, brain drain, unemployment, poverty and so on.
-
Shrinking cities emphasize that population loss is caused by structural crises, such as post industrialization, economic and social transformation, and other complex factors interact in different cities and regions, thus breaking the relatively balanced pattern of the urban system itself (Alves et al., 2016). Fig. 3 depicts the conceptual model of urban shrinkage. There is an association between city shrinkage and its external environment. The external impacts of cities may be globalization or the adjustment of production structure, or governance arrangements operating on different spatial levels (Haase et al., 2016). Moreover, the development and shrinkage of cities often interact with each other. Literatures suggest that population loss and economic recession are the main characteristics of urban shrinkage (Wolff and Wiechmann, 2018; Li et al., 2020). In industry dimension, the hypothesis of structural change holds that the continuous development of the economy will lead to the transformation of the economic center and the upgrading of the industrial structure (Liu et al., 2007). In terms of built environment, urban shrinkage is also accompanied by the abandonment of public service facilities (Wu and Sun, 2017). Therefore, variable selection in this paper mainly includes four dimensions: demographic profile, economic performance, industry structure and built environment.
In demographic profile dimension, population aging widely exists in developed and developing countries, and there is an obvious correlation with the shrinkage of cities (Feldhoff, 2013; Liu and Yang, 2017). Researches have confirmed that the aggravation of population aging trend is one of the important formation mechanisms of urban shrinkage in China (Zhang et al., 2018). The dependency ratio is defined as the ratio of dependents (under 18 and over 60) to the working age population (between 18 and 60). The larger the dependency ratio, the more dependent people the labor force bears per capita (Hartt, 2018a). As an important population factor, resident population can reflect the overall situation of the population in China. The natural population growth rate can reflect the tendency and speed of population growth. Therefore, the natural growth rate, aging rate, dependency ratio and resident population were selected as the demographic profile dimension factors affecting the urban shrinkage of Hegang. In economic performance dimension, GDP per capita is generally used to measure the degree of economic development of a country or region (Zhang et al., 2013). Local financial revenue is directly related to the industrial structure and economic operation quality of the city. At the macro level, investment is one of the ‘troikas’ to promote economic development (Li and Zou, 2018). Total fixed assets investment plays a key role in reflecting changes in economic structure and quality. Foreign direct investment can reflect the openness of a particular region. Total retail sales of consumer goods can not only reflect the domestic demand and consumption capacity of the city, but also reflect the operation of the urban economy. Affected by the economic recession caused by urban shrinkage, the overall unemployment rate of city is generally high (Nefs et al., 2013; Mallach, 2017). Therefore, GDP per capita, local financial revenue, total fixed assets investment, foreign direct investment, total retail sales of consumer goods and unemployment rate were selected as the economic performance factors. In industry structure dimension, changes in the industrial structure play an important role in economic development (Yu, 2015). This paper uses the proportion of secondary industry and the proportion of tertiary industry to reflect the industrial structure of Hegang. As a typical resource-based city, Hegang has more than 30 types of mineral resources, with coal and graphite being the most important resources to the country. Subjecting to data availability limitations, we choose production value of mining industry to reflect the mining industrial situation in Hegang. Resource dependence refers to the dependence of economy on resources. The ratio of mining industry employees to the total population at the end of the year is used to measure resource dependence (Li and Zou, 2018). In built environment dimension, good public facilities are more likely to attract people to settle down in a city (Guimaraes et al., 2016; Guo and Li, 2019). Per capita public green area and per capita road area were selected to reflect the infrastructure facilities of Hegang. The consequences of urban shrinkage impacts on urban development and the housing market are various, such as vacancies and the disposal of large numbers of buildings (Du et al., 2020b). Similarly, the economic recession that accompanies the shrinkage of cities will lead to a downturn in the real estate market (Haase et al., 2014; Lee et al., 2016; Wang and Ye, 2020), which in turn will affect sales of commercial housing and the start of new housing. Sales area of commercial housing and housing construction starts rate were chosen as the representative variables. Housing construction starts rate is defined as the ratio of new construction area to housing construction area for the current year. Urban shrinkage is generally reflected in the loss of population, which in turn causes the government to reduce the corresponding level of public services (Deng and Zhang, 2020). This is particularly evident in education services, where the cities in population outflow lack the incentive to upgrade the level of education services in the region. Medical service is directly related to the public’s health and is the most basic need of the public (Zhong et al., 2016). With the continuous advance in means of data collection, the amount of POI data in recent years has been showing an upward trend. In order to reduce the error that may be caused by use of absolute quantity. Here, we choose the ratio of POI of science and education cultural services to POI in this year, and the ratio of POI of medical care services to POI in this year represent the education and medical service facilities in Hegang. POI data is more granular than traditional statistics. Mixed and multifunctional land use have been identified as being able to promote urban vibrancy (Koster and Rouwendal, 2012; Yang et al., 2017). Referring to the research of Liu and Long (2016), using point of interest (POI) data, a mixed index can be computed to denote the degree of mixed land use. This study included 21 variables (Table 1). The data sources mainly include the Heilongjiang Statistical Yearbook (Heilongjiang Bureau of Statistics, 2006–2020), Hegang Statistical Yearbook (Hegang Bureau of Statistics, 2006–2020), China City Statistical Yearbook (Department of Urban Surveys of National Bureau of Statistics of China, 2006–2020), China Urban Construction Statistical Yearbook (Ministry of Housing and Urban-Rural Development of the People’s Republic of China, 2006–2020), Statistical Bulletin of National Economic and Social Development of Hegang City (Hegang Bureau of Statistics, 2005–2019). Moreover, POI data are obtained from the AutoNavi, a leading navigation and location-based service provider in China. All data of each variable was standardized, which makes the across time periods comparison more robust.
Variable type Variable Unit Source Demographic profile Natural growth rate ‰ China City Statistical Yearbook Aging rate % Statistical Bulletin of National Economic and Social Development of Hegang City Dependency ratio % Resident population Ten thousand China Urban Construction Statistical Yearbook Economic performance GDP per capita Yuan (RMB) per person Heilongjiang Statistical Yearbook Local financial revenue Ten thousand yuan Hegang Statistical Yearbook Total fixed assets investment Hundred million yuan Heilongjiang Statistical Yearbook Foreign direct investment
Total retail sales of consumer goodsTen thousand yuan
Ten thousand yuanChina City Statistical Yearbook
China City Statistical YearbookUnemployment rate % Hegang Statistical Yearbook Industry structure Proportion of secondary industry % China City Statistical Yearbook Proportion of tertiary industry % China City Statistical Yearbook Resource dependence % China City Statistical Yearbook Production value of mining industry Ten thousand yuan Hegang Statistical Yearbook Built environment Per capita public green area
Per capita road aream2
m2Heilongjiang Statistical Yearbook
Heilongjiang Statistical YearbookSales area of commercial housing
Housing construction starts rate
Ratio of EPOIs
Ratio of MPOIskm2
%
%
%Hegang Statistical Yearbook
Hegang Statistical Yearbook
The AutoNavi
The AutoNaviMixed land use % The AutoNavi Notes: POI refers to Point of Interest. Ratio of EPOIs represents the ratio of the number of science, education and cultural service POIs to all POI quantities in this year. Ratio of MPOIs represents the ratio of the number of medical care services POIs to all POI quantities in this year. The data of aging rate and dependency ratio are from Economic and social development statistical bulletin of Hegang. The data collected annually over a span of 15 years from 2005 to 2019 Table 1. List of include demographic profile, economic performance, industry structure and built environment variables to research population shrinkage of Hegang
-
The cross-correlation function can not only calculate the correlation of two variables at one point in time, but also calculate the correlation at multiple points in time (Hoekveld, 2012). Cross-correlation between two functions means that an infinite integral that multiplies two functions by making complex conjugate and reverse translation respectively. In other words, an infinite integral multiplied by the second function after conjugation and translation of the first function. Cross-correlation analysis can reveal the intensity and direction of correlation between two time series. The cross-correlation coefficient is calculated with Equation (1) or Equation (2).
$$ \mathop R\nolimits_{fh1} (x) = \int_{ - \infty }^\infty {f(x)h(x + x'){\text{d}}} x $$ (1) $$ \mathop R\nolimits_{fh2} (x) = \int_{ - \infty }^\infty {f(x - x')h(x){\text{d}}x} $$ (2) where
$x$ stands for the integral variable, x′ stands for the parameter variable.$ {R}_{fh}\left(x\right) $ is a cross-correlation function. The cross-correlation of two functions$ f\left(x\right) $ and$ h\left(x\right) $ is defined by the infinite integral with x. -
Cross-correlation can measure the time and direction, as well as the strength of the relationships between variables. Although cross-correlation analysis has the above advantages, there are still some limitations. That is, using cross-correlation alone can lead to ambiguous inferences when studying several time series simultaneously (Baccalá and Sameshima, 2001; Hartt, 2018a). In addition, because the seemingly open nonlinear system may be closed due to a wide range of feedback loops, there is a risk of false negative results (Billings, 2013). To overcome the limitations of cross-correlation analysis, we introduce partial-correlation analysis to eliminate the risk of false correlation as much as possible. Partial-correlation is a multi-factor system of geographic systems, also known as ‘net correlation’, ‘pure correlation’ or ‘conditional correlation’ (Vargha et al., 2013). The change of one element will affect the change of other elements, so there are different relations among them (Park et al., 2015). Partial-correlation is the net correlation or pure correlation of two random variables when the effects of other variables are excluded (Kim et al., 2011). In other words, it is the conditional correlation of two random variables when the rest variables in the same system take a given value.
$$ R = \left( {\begin{array}{*{20}{c}} {{r_{11}}}&{{r_{12}}}&{{r_{13}}} \\ {{r_{21}}}&{{r_{22}}}&{{r_{23}}} \\ {{r_{31}}}&{{r_{32}}}&{{r_{33}}} \end{array}} \right) = \left( {\begin{array}{*{20}{c}} 1&{{r_{12}}}&{{r_{13}}} \\ {{r_{21}}}&1&{{r_{23}}} \\ {{r_{31}}}&{{r_{32}}}&1 \end{array}} \right) $$ (3) $$ \mathop R\nolimits_p = \left( {\begin{array}{*{20}{c}} 1&{{r_{12.3}}}&{{r_{13.2}}} \\ {{r_{21.3}}}&1&{{r_{23.1}}} \\ {{r_{31.2}}}&{{r_{32.1}}}&1 \end{array}} \right) $$ (4) where R stands for correlation coefficient matrix, Rp stands for partial-correlation coefficient matrix. Suppose that there are three elements or variables X1, X2, X3 and the correlation coefficient matrix is shown in the Equation (3). r12 represents the correlation coefficient between variables X1 and X2 and so on. Because it is a symmetric correlation matrix, only r12, r13 and r23 need to be calculated. The partial-correlation coefficient matrix is shown in the Equation (4). In fact, the partial-correlation coefficient matrix is also a symmetric matrix. The number after the subscript dot represents the variable that remains unchanged. For example,
$ {r}_{12.3} $ means the X3 remains unchanged. -
The rise of complex network research further promotes the development of network visualization technology and puts forward higher requirements for it (Sun et al., 2010). Moreover, through the method of network visualization, the feedback of each factor is displayed intuitively, and the dynamic relationship between key variables in the system is further studied. The first network visualization method is used to obtain the feedback relationship between variables in the network. Each variable selected is represented by a point in the graph, and the colors of the points represent different variable types. In addition, each line in the figure connecting the two variables represents a significance test passing the cross-correlation and partial-correlation. Positive relationships are indicated by red lines, and negative relationships are indicated by black lines. As a result, lines with arrows indicate that there is a time delay between the two variables in the cross-correlation analysis. Because there are a large number of edges in the whole network, the time lag associated with each relationship is not visually represented. This paper is inspired by system dynamics for a better understanding of the multidimensional process of urban shrinkage and the feedback mechanism of the key factors of urban shrinkage. Therefore, the second network visualization method is used to check the cyclic feedback relationship of key factors. According to the feedback characteristics of mutual cause and effect of the internal components of the system, system dynamics explores the root cause of the problem from the internal structure of the system (Forrester, 2003; Saeed, 2014). The variables that precede a change in the key variable are indicated on the left side by arrows pointing in key variables, and the variables that follow a change in the key variable are indicated on the right side by arrows pointing away from key variables (Hartt, 2018a). In addition, the variables of the complete loop indicate that there is no time delay between the correlations of the two variables in the cross-correlation analysis. Similarly, positive relationships are indicated by red lines, and negative relationships are indicated by black lines.
Complex Pathways to Population Shrinkage: Case Study of Hegang City, China
doi: 10.1007/s11769-022-1276-2
- Received Date: 2021-09-10
- Accepted Date: 2022-01-08
- Available Online: 2022-05-26
- Publish Date: 2022-05-05
-
Key words:
- urban shrinkage /
- resource-based cities /
- Hegang /
- complex pathways /
- China
Abstract: With the transformation of China’s industry and demographic structure, urban shrinkage, especially the resource-based cities due to their single industrial structure, tend to emerge gradually. Using empirical and quantitative methods, this paper explores the evolution mechanism of urban shrinkage in Hegang, a resource-based city in China. Our findings suggest that there are many correlations or cyclic relationships among variables, which provide an empirical confirmation for the complexity of urban shrinkage process. The result shows there is a time delay of about four to five years between the mining industry and the changes of demographic profile, economic performance and built environment variables. The development of Hegang has formed a path dependence on resource-based industries. Furthermore, the time lags between demographic profile and economic performance variables are not obvious, and the real estate market has a certain sensitivity to perceive population loss and economic change. Besides, market led public service facilities are more sensitive to the changes of population outflow and economic recession than government led public service facilities. The study findings could offer insights for other resource-based cities in developing countries to employ on the economic development policies issues.
Citation: | WANG Tingting, WU Kang, YAO Cuiyou, LIU Xiaoxiao, 2022. Complex Pathways to Population Shrinkage: Case Study of Hegang City, China. Chinese Geographical Science, 32(3): 418−437 doi: 10.1007/s11769-022-1276-2 |