Research Articles

Spatial pattern and its evolution of Chinese provincial population: Methods and empirical study

  • DENG Yu 1, 2 ,
  • LIU Shenghe 1, 2 ,
  • CAI Jianming 1, 2 ,
  • LU Xi , 3, * ,
  • Chris P NIELSEN 3
*Corresponding author: Lu Xi (1979-), Research Associate and Lecturer, specialized in renewable energy and sustainable development. E-mail:

Received date: 2014-06-06

  Accepted date: 2015-03-15

  Online published: 2015-12-31

Supported by

Key Program of the National Natural Science Foundation of China, No.71433008.Key Research Program of the Chinese Academy of Sciences, No.KZZD-EW-06-04.National Natural Science Foundation of China, No.41271174.National Key Technology R&D Program, No.2012BAI32B06.Beijing Planning of Philosophy and Social Science, No.13CSC011


Journal of Geographical Sciences, All Rights Reserved


China has been experiencing an unprecedented urbanization process. In 2011, China’s urban population reached 691 million with an urbanization rate of 51.27%. Urbanization level is expected to increase to 70% in China in 2030, reflecting the projection that nearly 300 million people would migrate from rural areas to urban areas over this period. At the same time, the total fertility rate of China’s population is declining due to the combined effect of economic growth, environmental carrying capacity, and modern social consciousness. The Chinese government has loosened its “one-child policy” gradually by allowing childbearing couples to have the second child as long as either of them is from a one-child family. In such rapidly developing country, the natural growth and spatial migration will consistently reshape spatial pattern of population. An accurate prediction of the future spatial pattern of population and its evolution trend are critical to key policy-making processes and spatial planning in China including urbanization, land use development, ecological conservation and environmental protection. In this paper, a top-down method is developed to project the spatial distribution of China’s future population with considerations of both natural population growth at provincial level and the provincial migration from 2010 to 2050. Building on this, the spatial pattern and evolution trend of Chinese provincial population are analyzed. The results suggested that the overall spatial pattern of Chinese population will be unlikely changed in next four decades, with the east area having the highest population density and followed by central area, northeast and west area. Four provinces in the east, Shanghai, Beijing, Tianjin and Jiangsu, will remain the top in terms of population density in China, and Xinjiang, Qinghai and Tibet will continue to have the lowest density of population. We introduced an index system to classify the Chinese provinces into three categories in terms of provincial population densities: Fast Changing Populated Region (FCPR), Low Changing Populated Region (LCPR) and Inactive Populated Region (IPR). In the FCPR, China’s population is projected to continue to concentrate in net immigration leading type (NILT) area where receives nearly 99% of new accumulated floating population. Population densities of Shanghai, Beijing, Zhejiang will peak in 2030, while the population density in Guangdong will keep increasing until 2035. Net emigration leading type (NELT) area will account for 75% of emigration population, including Henan, Anhui, Chongqing and Hubei. Natural growth will play a dominant role in natural growth leading type area, such as Liaoning and Shandong, because there will be few emigration population. Due to the large amount of moving-out labors and gradually declining fertility rates, population density of the LCPR region exhibits a downward trend, except for Fujian and Hainan. The majority of the western provinces will be likely to remain relatively low population density, with an average value of no more than 100 persons per km2.

Cite this article

DENG Yu , LIU Shenghe , CAI Jianming , LU Xi , Chris P NIELSEN . Spatial pattern and its evolution of Chinese provincial population: Methods and empirical study[J]. Journal of Geographical Sciences, 2015 , 25(12) : 1507 -1520 . DOI: 10.1007/s11442-015-1248-x

1 Introduction

In the past three decades, China has experienced a high speed of socioeconomic development, accompanied with dramatic changes in natural growth and migration of population (Zhang and Song, 2003). The annual natural growth rate between 2000 and 2010 was reported at 0.57% in China’s 6th national census, significantly less than that for the 1990s. Low natural growth rate might imply a decline in the “demographic bonus”, potentially weakening the economic growth in China (Ma and Zheng, 2011). Meanwhile, the amount of population migration reached 260 million, accounting for as high as 20% of total population in China in 2010 (NBS, 2001, 2011). Huge amount of migration raises concerns of social justice including equality in employment, education, housing as well as medical service, posing a great potential challenges to the society (Chen, 2012). A thorough investigation of the future spatial pattern of population and its evolutional trend is critical for the socioeconomic sustainable development in China (Liu et al., 2010).
There are three main methods in population forecasting: mathematical trend extrapolation, demographic-economic model, and cohort-component model (C-C model) (Andrew, 1984). Mathematical trend extrapolation relies on the pattern of historic population data to predict what will happen in the future (Keyfitz, 1968, 2005). This method assumes that the future growth of population follows the similar pattern with the past. The accuracy of population forecasting, therefore, is subject to the choice of baseline and underlying assumption (Nakicenovic, 2000). Demographic-economic model aims at forecasting population through linking population changes with economic conditions (Kenneth and Wendling, 1980; Thomas, 1981). The economic condition, involving the evaluation of interest rate, import-export demand and related policies, is equally complicated (Booth et al., 2006). This method is seldom applied for forecasting population on the national level. (Hyndman and Ullah, 2007). The C-C model divides the population by age groups, and takes effects of human fertility, mortality and natural growth into consideration, thus it provides a relatively accurate prediction of population (Stauffer, 2002). The key for this method is parameterizing the model to reflect the social and geographical characteristics of the population. The C-C model has been adopted by the United Nations (UN) to routinely forecast the future country-level population in the world. To be discussed later in details, the UN’s population forecasting for China demonstrated an impressive agreement with the data reported by China’s national censuses conducted in 2000 and 2010.
In addition, population forecasting involves micro-level and macro-level strategies (Long, 1985). The micro-level strategy is a bottom-up approach, with the higher-level population derived from individual geographical regions at the lower level. In contrast, macro-level strategy decomposes the national population through mathematical methodology into that at lower level, such as province, prefecture-level city or county city. Micro-level strategy exhibits a good performance on forecasting the changes in the future distribution of population for individual areas (Bach and Schraml, 1982; Wood, 1982) but fails in considering the impact of changes from the larger background. Besides, building up a nationwide population model based on micro-level strategy is difficult in practice for two reasons. First, data for the demographic components may not be available for all counties. Second, the future population at regional level might be underestimated or overestimated deliberately in China in order to secure more favorable policies from the central government for local development (Gans 1999). Both of them impair accuracy of population forecasting using micro-level strategy at national level, resulting in the governmental sub-optimal decisions as to where to develop the infrastructure, to make investment and to allocate subsidies. The macro-level strategy has advantages in calculating population with diverse types and patterns from differing historical periods and geographical areas (Harris and Moore, 1980). The application of macro-level strategy reported in the earlier literature often took the country as whole, or divided the total population into data grids. Few literatures emphasized on population forecasting for the jurisdiction level of province or county, which is critical for policy-making process at macro level.
The present study is intended to explore the dynamic spatial pattern for the future population in China between 2010 and 2050. Macro level strategy was adopted in this study to forecast the natural growth and migration at provincial level. Data relevant to China’s total population between 2010 and 2050 including total population, total urban population, and the urbanization rate on a national basis, are derived from the forecasting results using a C-C model developed by the Department of Economic and Social Affairs under the UN. These data are used as constraints to cap the natural growth and migration of population at the provincial level. Data from the 5th and 6th China’s national censuses conducted respectively in 2000 and 2010 provide a background of natural growth and migration of provincial population in China, serving an initial condition for this analysis.
Following this section, Section 2 describes data and methodology. The results are discussed in Section 3. Section 4 started with analyzing the spatial pattern and trend of Chinese provincial population. And an index system is introduced further to regionalize the changes in future spatial distribution of the population density, the ratio of accumulated net-migration to the peaking population, and accumulated natural population growth over the study period, with conclusion summarized in Section 5. The results reported in this analysis are expected to be helpful for a variety of key policy-making processes in China such as industrial development, ecological conservation, industrial upgrading and transfer, and environmental protection.

2 Data and methodology

Evolution of spatial distribution of the provincial population in China is attributed to a combined effects of natural growth and migration. Natural population growth for the past five years was tentatively extrapolated to the next five years, and then adjusted to make it in conformity with the UN’s forecasted population for the entire nation. The initial growth rates of provincial population between 2000 and 2010 were derived from the 5th and 6th national censuses. Migration population for each province was estimated separately according to the future national migration and migration matrixes. The latter were predicted using a Markov chain approach. We selected to neglect the contribution of natural growth from the migration population based on two considerations. First, the 6th national census data suggested an evident increasing trend for the whole family migration which has no contribution to the natural growth. Another reason is rooted in the fact that fertility rates were calculated on account of only registered population (Zhu, 2011).

2.1 Natural population growth

At the national level, we adopted the results of population forecasting for China by the UN as constraints to China’s future total population growth. Table 1 summarizes three key elements in the UN database forecasted from 2010 to 2050, including total population, urban population and the urbanization rate.
Table 1 Main indicators of population forecasting by the UN
2010 2015 2020 2025 2030 2035 2040 2045 2050
Total population 133972.49 136974.3 138779.2 139525.6 139307.6 138158.8 136090.6 133176.8 129560.4
Urban population 669192.56 761579 846363 911804 957649 984446 998581 1004090 1001612
Urbanization rate 0.49 0.55 0.61 0.65 0.69 0.71 0.73 0.75 0.77

Data source: World Population Prospects, the 2010 Revision

The process to predict population growth at provincial level involves three steps.
Step 1: Natural provincial growth rates between 2000 and 2010 were used as a first guess for the growth rates between 2010 and 2015. Population due to natural growth can be calculated using equation (1).
where Rn,2000s represents the natural population growth rate of a province on an average between 2000 and 2010; definitions of the other parameters and variables introduced here are summarized in Table 3.
Step 2: We adjust the forecasting results from Step 1 to ensure that the sum of provincial population matches the national population projected by the UN. The difference between them was allocated to each province based on its percentage of population to the national total. The process is described by equations (2)-(6).
where Pn,t is the population amount of province n in year t; Psum,t denotes the total provincial population in year t; PUN,t represents the national population forecasting from the UN; δsum, t is the difference between Psum,t and PUN,t; δn, t denotes the fraction of difference (δsum, t) allocated to each province; Pn,t represents the improved population of province n in year t.
Step 3: The natural growth rate for each province was then adjusted using results from Step 2. The updated growth rate is then used to extrapolate the population growth for the next five years. The process is expressed by equations (7) and (8):
where Rn,t represents the adjusted natural population growth rate; Pn,t+5 represents the first guess for the population of next five years, which will be adjusted by repeating Step 2.

2.2 Migration

2.2.1 Migration at national level
Urban population in China increased by 212.8 million from 2000 to 2010. The corresponding urbanization rate rose from 36% in 2000 to 50% in 2010. At national level, the migration consists of intra-urban migration and rural-to-urban migration, which accounted for 40 million and 74 million respectively in 2010. In the rural-to-urban transition, intra-province and inter-province migrations consisted of 40 million and 34 million population respectively. Rural-to-urban migration constitute the dominant component of inter-province migration (80%), with the rest compensation of intra-urban migration (8.5 million, 20%) (see Table 2). The inter-province migrations were estimated using equations using equations (9)-(12):
where ΔMR demonstrates the total population migrating from rural areas to urban areas; ΔMRinter is the inter-provincial migration component of the total rural-to-urban migration (ΔMR); ΔMinter represents the total inter-provincial migration. r1 is the ratio of ΔMR to changes in the urban population (ΔUP); r2 is the ratio of ΔMRinter to ΔMR; r3 is the ratio of ΔMRinter to ΔMinter.
Table 2 Net-migration population to the urban area at national level
Year Registered residence (million)
Total Urban area Rural area
Subtotal Intra-
Subtotal Intra-
Up to 2010 212.25 83.85 70.39 13.46 128.4 75.25 53.15
Up to 2000 98.12 43.39 37.93 5.46 54.74 35.44 19.30
2000-2010 114.13 40.47 32.46 8.01 73.66 39.81 33.85

Data source: The 5th National Population Census, 2000, and the 6th National Population Census, 2010

Table 3 Definitions of variables or parameters used in this analysis
Parameters or variables Definition
Rn,t Population natural growth rate in province n for year t
Pn,t Population amount for natural growth in province n for year t
P'n,t Improved population amount for natural growth in province n for year t
UP Total urban population
ΔMR Migration amounts from rural to urban area
ΔNM Net inter-provincial migration matrix
TPM Net transition probability matrix
M Migration amount
ΔM Total net migration amounts
δn, t Difference between provincial population projection and the data from the UNs in province n for year t
TPn,t Total population amount in province n for year t
Sn,t Area of province in province n for year t
DPn,t Population density of province in province n for year t
An Variance of population density of province n
Bn Ratio of new accumulated net-migration to maximum of the total population in province n
Cn Ratio of new accumulated net-migration to maximum of natural growth population of province n
2.2.2 Migration at provincial level
The approach of Markov chain was adopted to estimate the net-transition-probability migration matrix among provinces for the next five years. Constructing an initial probability matrix involves two steps: First, the net inter-provincial migration matrix between every two provinces between 2000 and 2010 was derived from the 5th and 6th national censuses; and second, this net inter-provincial migration matrix was divided by the net total migration amounts. This process is achieved through following equations (13)-(15):
where Minter,2000 represents the total inter-provincial migration between every two provinces up to 2000, and Minter,2010 for the total inter-provincial migration up to 2010. ΔNMinter denotes the net inter-provincial migration matrix between 2000 and 2010; TPM is the net transition probability matrix; Mn,t is the migration of province n in year t. Net inter-provincial migration for individual provinces can be calculated by multiplying net transition probability matrix with the national net inter-provincial migration population in the same years.

2.3 Spatial pattern of the future provincial population

As shown in equation (16), total provincial population equals the natural growth population plus inter-provincial migration. Population density for a specific province can be further derived from its total population dividing its area (equation 17). Here, we introduce three indicators constructed in equations (18)-(20), namely the population density (A), the ratio of accumulated net-migration to the peaking population (B), and accumulated natural population growth (C) over the study period to classify the changes of future pattern of provincial population. All provinces of China can be divided into three types of region: fast changing populated region (FCPR), slow changing populated region (SCPR) and inactive populated region (IPR). Sub-classes for the FCPR and the SCPR include net-immigration leading type (NILT), net-emigration leading type (NELT) and natural growth leading type (GLT).

3 Results

3.1 Natural population growth

China’s total population reached 1.265 billion in 2000 according to the 5th national census. The average natural growth rate for China was 0.565% from 2000 to 2010. Should the growth rate are maintained, China’s population would increase to 1.337 billion in 2010. This number is consistent with the population of 1.339 billion resulted from the 6th national census in 2010. The same method was applied to estimate the natural growth between 2000 and 2010 for each province without consideration of migration. The results were summarized in Table 4 (Column 3). Net migration of each province could be derived by subtracting the cumulative migration reported in the 5th national census from that in the 6th national census. As summarized in Table 4 (Column 4), the positive sign represents the net immigration population, and the negative sign means the net outmigration. The projected population in 2010 for each province would be obtained by the sum of projected natural growth population and net migration (Column 5). It is impressive for the agreement between projected population for 2010 and the reported population in the 6th national census. The projected population has errors within 5% as compared to the reported population for each province. The natural growth of future population between 2010 and 2050 were estimated using the same approach.
The growth rate of population projected for the period from 2010 to 2015 does not show significant changes as compared to the rate between 2000 and 2010. The population growth rate for each province tends to slow down after 2015, approaching to zero for most of the provinces during the period of 2020-2025. Moreover, the population of some provinces including Beijing, Tianjin, Shanghai, Heilongjiang, Jilin and Shenyang start decreasing. After 2025, more and more provinces have a shrinking-population. During 2045-2050, all provinces have a negative growth except Xinjiang. It is noteworthy that municipalities (except Chongqing) lead in reaching the peaking population, followed by other coastal provinces and some central provinces (Shanxi, Henan, Hunan, Hubei, Jiangxi, Anhui) in China. The final group comprises mostly of remote provinces of western China including Guangxi, Chongqing, Guizhou, Yunnan, Tibet (Xizang), Shaanxi, Gansu, Qinghai, and Ningxia. Xinjiang is a special case that its population becomes flat between 2045 and 2050.
Table 4 A comparison between reported population and projected population in 2010
Province Population 2000
Decade natural growth rate Projected natural growth population 2010 (million) Reported net migration 2000-2010 number
Projected population 2010 (million) a Reported population 2010
Percentage of projected population error b
Beijing 13.82 1.81 14.07 +4.40 18.47 19.61 5.81%
Tianjin 10.01 1.80 10.19 +2.07 12.26 12.94 5.26%
Hebei 67.44 5.99 71.48 -1.81 69.67 71.85 3.03%
Shanxi 32.97 5.90 34.91 -0.51 34.4 35.71 3.67%
Inner Mongolia 23.76 4.03 24.72 +0.33 25.05 24.71 1.38%
Liaoning 42.38 1.11 42.85 +0.09 42.94 43.75 1.85%
Jilin 27.28 2.33 27.91 -0.62 27.29 27.46 0.62%
Heilongjiang 36.89 2.35 37.76 -1.26 36.5 38.31 4.72%
Shanghai 16.74 1.01 16.91 +5.73 22.64 23.02 1.65%
Jiangsu 74.38 2.34 76.12 +3.50 79.62 78.66 1.22%
Zhejiang 46.77 4.44 48.85 +7.76 56.61 54.43 4.01%
Anhui 59.86 6.32 63.64 -4.81 58.83 59.50 1.13%
Fujian 34.71 6.05 36.81 +1.31 38.12 36.89 3.33%
Jiangxi 41.40 8.08 44.74 -1.76 42.98 44.57 3.57%
Shandong 90.79 5.27 95.57 -0.91 94.66 95.79 1.18%
Henan 92.56 5.42 97.58 -5.44 92.14 94.02 2.00%
Hubei 60.28 2.93 62.05 -2.68 59.37 57.24 3.72%
Hunan 64.40 5.35 67.84 -2.55 65.29 65.68 0.59%
Guangdong 86.42 7.65 93.03 +5.98 99.01 104.30 5.07%
Guangxi 44.89 7.98 48.47 -1.33 47.14 46.03 2.41%
Hainan 7.87 9.07 8.58 +0.05 8.63 8.67 0.46%
Chongqing 30.90 3.21 31.89 -1.96 29.93 28.85 3.74%
Sichuan 83.29 3.03 85.81 -1.38 84.43 80.42 4.99%
Guizhou 35.25 8.23 38.15 -2.10 36.05 34.75 3.74%
Yunnan 42.88 8.10 46.35 -1.07 45.28 45.97 1.50%
Tibet 2.62 11.17 2.91 +0.02 2.93 3.00 2.33%
Shaanxi 36.05 4.07 37.52 -0.61 36.91 37.33 1.13%
Gansu 25.62 6.38 27.26 -0.80 26.46 25.58 3.44%
Qinghai 5.18 9.76 5.69 +0.05 5.74 5.63 1.95%
Ningxia 5.62 10.52 6.21 +0.04 6.25 6.30 0.79%
Xinjiang 19.25 10.99 21.37 +0.24 21.61 21.81 0.92%

Note: a Projected population 2010=projected natural growth population 2010+reported migration 2000-2010 number

b Percentage of estimated population error = | reported population 2010-projected population 2010| / reported population 2010

Data source: The 5th National Population Census, 2000, the 6th National Population Census, 2010, and China Statistical Yearbook, 2011

3.2 Migration at national level

According to the UN’s population forecasting, the urbanization rate in China would grow at an annual rate of 1.2% from 2010 to 2020. In other words, China would experience a stable urbanization process in the 2010s but with a lower average annual growth rate than that for the 2000s. The growth rate for urbanization was projected to decrease to 0.8% during 2020-2030. Once the urbanization rate reaches 70%, the annual growth rate would decline to 0.4%. At the same time, the expansion of urban population will decrease as the growth of urbanization rate drops. Urban population will increase by 92.39 million between 2010 and 2015, as compared to the increase of 84.78 million between 2015 and 2020. Between 2045 and 2050, the growth in urban population decreases to negative value due to a sharply drop in growth rate projected for years after 2030 (Table 5).
Between 2000 and 2010, migration from rural to urban area accounts for 35% increase in urban population according to the data reported in the 6th national census. The experience of developed countries indicated that the population migrations increased first during the process of urbanization and then decreased after completion of the urbanization. China is facing a new situation where intra-provincial migration is the majority of population moving from rural to urban areas, and inter-provincial migration is minor. We assume that inter-provincial migration would reach an equilibrium when urbanization rate reaches 70%. The differences in terms of education, health, medicine, social insurance, and employment are assumed vanish between rural and urban areas and between different provinces. The contributions of rural-to-urban migration to the total growth of urban population are 30%, 30%, 20%, 10%,6%, 4%, 2% and 0% during 2010-2015, 2015-2020, 2020-2025, 2025-2030, 2030-2035, 2035-2040, 2040-2045 and 2045-2050 respectively. The result shows that the growth in inter-provincial migration sharply reduces from 17 million during the period of 2010-2015 to 2 million during 2025-2030. After the year of 2030, inter-provincial migration has reached regional equilibrium and the net inter-provincial migration was considered as zero.
Table 5 Provincial migration in China
Year Total population
Urban population
Urbanization rate Growth in
urban population
Migration from rural
to urban area
Growth in migration of inter-province
Inter- province Intra- province Total
2000 1259.95 456.35 0.36 - - - - -
2010 1339.72 669.19 0.50 212.84 34.23 40.26 74.49 46.07
2015 1369.74 761.58 0.56 92.39 12.74 14.98 27.72 17.34
2020 1387.79 846.36 0.61 84.78 11.69 13.75 25.42 15.91
2025 1395.26 911.80 0.65 65.44 6.01 7.07 13.08 8.19
2030 1393.08 957.65 0.69 45.85 2.11 2.48 4.59 2.87
2035 1381.59 984.45 0.71 26.80 0.00 1.61 1.61 0.00
2040 1360.91 998.58 0.73 14.14 0.00 0.57 0.57 0.00
2045 1331.77 1004.09 0.75 5.51 0.00 0.11 0.11 0.00
2050 1295.60 1001.61 0.77 -2.48 0.00 0.00 0.00 0.00

Data source: World Population Prospects, the 2010 and 2011 Revisions

3.3 Migration at provincial level

As illustrated in Figure 1, the net transition probabilities for all provinces in China were ordered from the highest positive values to the lowest negative values. Zhejiang, Guangdong, Shanghai, Beijing, Jiangsu, Tianjin and Fujian are the major net in-migration provinces in 2010. These migration-host provinces are all located in the east coast, with high socioeconomic development level. On the other hand, Henan, Anhui, Hubei and Hunan are the migration-source provinces in 2010, with the largest net transition probabilities respectively of -12.5%, -11.0%, -6.1% and -5.8%. Nearly all the migration-source provinces are located in central or northeast China with large number of surplus rural labor force. In contrast to these two categories of provinces, which have high absolute values of net transition probability, one third of the provinces maintains a net migration balance with probabilities fluctuating within the range of -1% and 1%. In addition, most of the total populations of these provinces were relatively small, and therefore their net-migration is negligible.
Figure 1 Net transition probability of each province
Net transition probability at provincial level for 2015, 2020, 2025, 2030, 2035 and 2040 were estimated using the Markov chain approach. The results indicate that net transition probability declines dramatically between 2010 and 2020, although maintaining the same migration direction. The migration probabilities for all provinces approach zero after 2030, reflecting the assumption that inter-provincial migration reaches a regional equilibrium when urbanization rate reaches 70% in 2030.

4 Spatial pattern of Chinese provincial population in 2010-2050

4.1 Evolution of provincial population density

Population density of each province for 2010, 2015, 2020, 2030, 2040 and 2050 was illustrated in Figure 2. The results suggested that the distribution of population in China maintains similar spatial pattern for next four decades. East China has the highest population density and followed by central China, northeast China and west China. Four provinces in the east, Shanghai, Beijing, Tianjin and Jiangsu, will remain the top in terms of population density, with an averaged population density of 3540, 1411, 1114 and 736 persons per km2, respectively; Shandong was No.5 in 2010 and will fall to No.7 in 2020, surpassed by Guangdong and Zhejiang in 2020. Hebei ranked No.10 in 2010, and No. 9 in 2020. The rank of Fujian increased from No.14 in 2010 to No.11 in 2050. As to southeast China, Hainan has the lowest population density, but it will surpass Liaoning in 2040 and Hubei in 2050, consequently reaching No.15 with an averaged population density of 269 persons per km2. In central China, Henan and Anhui have kept consistent in No.8 and No.10 respectively since 2015, with an averaged population density of 502 and 377 persons per km2, respectively. Hunan drops from No.12 to No.13 from 2010 to 2020, over the same period Hubei moves downwards from No.13 to No.16. Jiangxi rises to No. 14 after 2030. Shanxi keeps at No.18 all the time; The averaged population densities are ranging from 200 to 300 persons per km2 in Hunan, Hubei, Jiangxi and Shanxi. In northeast China, Liaoning drops 1 position separately in 2030 and 2035 from No.15 to No.17, with an averaged population density of 278 persons per km2, while Jilin would drop from No.23 to No.24 in 2035, with an averaged population density of 137 persons per km2. Heilongjiang keeps at No.26 over the entire study period, with an averaged population density of 80 persons per km2. In west China, almost all the provinces keep ranks below No.20 over the study period, except for Ningxia and Chongqing with a slight change between 2040 and 2050. The last three provinces are Xinjiang, Qinghai and Tibet, where the averaged population densities are 15, 9 and 3 persons per km2, respectively.
Figure 2 Provincial population density and its ranking change in 2010-2050

4.2 Regional characteristics

Based on evolution of population density change in each province, the ratio of new accumulated net-migration to the peaking population and new accumulated natural population growth, we divide China into three types of region: FCPR, LCPR and IPR. Furthermore, we classify FCPR into three types: NILT, NELT and NGLT. In addition, we categorize LCPR into net immigration type (NIT) and net emigration type (NET). The results are summarized in Table 6. China’s population is projected to increasingly concentrate in NILT area, which receives nearly 99% of newly accumulated immigration population. NELT area accounts for 75% of emigration population, including Henan, Anhui, Chongqing and Hubei. In LCPR, Fujian and Hainan belong to NIT, with growth in net-migration population as small as 1%. NET area will contribute 18% of emigration in future, while the percentage is significantly lower than that for NELT area in FCPR region. Population density for the majority of west provinces is likely to remain low and slightly declining.
Table 6 Regional type and its characters of spatial pattern
Regional type Criterion Ratio of new accumulated floating population (%) Ratio of new accumulated natural population growth (%) Province
Class Sub-classes A B C
Changing Populated
Net immigration leading type A>10 b>5% c>>1 99 18 Beijing, Tianjin,
Jiangsu, Shanghai, Zhejiang, Guangdong
Net emigration leading type B<-5% c<<-1 -75 14 Henan, Anhui, Chongqing, Hubei
Natural growth leading type b≈0% |c|≈0 -1 6 Liaoning, Shandong
Net immigration type 5<a<10 0<b<5% 0<c<1 1 5 Fujian, Hainan
Net emigration type -5%<b<0 -1<c<0 -18 31 Hebei, Shanxi, Jilin, Heilongjiang, Hunan, Jiangxi, Guangxi, Sichuan, Shaanxi
Inactive Populated Region a<5 0<b<5% |c|≈0 -6 26 Guizhou, Yunnan, Ningxia, Gansu, Inner Mongolia, Xinjiang, Qinghai, Tibet

Note: A represents variance of population density change; B denotes the ratio of new accumulated net-migration to maximum of total population amount; C demonstrates the ratio of new accumulated net-migration to maximum of natural growth population.

Figure 3 summarizes the evaluation of population density for the FCPR region in China. The ratio of new accumulated net-migration to the peaking population is above 10% in NELT area, with the highest number in Beijing (55%) followed by Shanghai (47%). As illustrated in Figure 3, the population densities in the six provinces in NELT first increase and then decrease reflecting the change of net immigration population amount. For Tianjin and Jiangsu, population densities reach the peak in 2020, with 1171 and 771 persons per km2, respectively. For Shanghai, Beijing and Zhejiang, population densities reach the peak in 2030, with 3825, 1499 and 697 persons per km2 respectively. The maximum population densities for these three provinces are 36%, 25%, and 35% higher than the corresponding values for 2010. For Guangdong, population density will reach the peak in 2035, with 684 persons per km2 as compared to the value, 570 persons per km2 for 2010. The ratio of accumulation emigration to the peaking population is below 15% in NELT area, with the highest ratio in Anhui (38%) and the second highest number in Henan (27%). The population densities in four provinces keep decreasing over the study period. For Henan, Anhui, Chongqing and Hubei, population densities in 2050 will decrease to 450, 348, 299 and 262 persons per km2 respectively. For Liaoning and Shandong, population densities drop to 243 and 592 persons per km2 in 2050, respectively.
Figure 3 Population densities forecasting for fast changing populated region in 2010-2050 Note: “—” denotes there’s net emigration population

5 Conclusions

We improved the macro-level strategy for population forecasting, and applied it to estimate the future natural growth and migration of population at provincial level in China. A comparison between the estimated natural growth and reported statistics from 2000 to 2010 suggested that the method proposed in this paper is applicable to make population forecasting. The method can be easily tailored for the county-level studies. It should be noted that the method based on the averaged natural growth to forecast the natural growth of population. Inter-provincial migration is assumed to reach an equilibrium after urbanization rate reaching 70% in 2030. The results suggested that inter-provincial migration probability is deduced to be around zero after 2030 through a Markov chain approach, which is consistent with our assumption.
The results from the analysis of the spatial pattern and evolution trend of Chinese provincial population suggest that the overall spatial pattern of Chinese population will be maintained in next four decades, with the east area having the highest population density and followed by central area, northeast and west area. Four provinces in the east, Shanghai, Beijing, Tianjin and Jiangsu, will remain the top in terms of population density in China, and Xinjiang, Qinghai and Tibet will continue to have the lowest density of population. Based on the analysis of the future trends in provincial population densities, we introduced an index system to classify the Chinese provinces into three categories: the FCPR, LCPR, and IPR regions. China’s population is projected to increasingly concentrate in NILT area where receives nearly 99% of new accumulated floating population. Policy-making and planning process in China should take into account the changes of population pattern in the FCPR region.

The authors have declared that no competing interests exist.

Andrew M I, 1984. Projection, forecast, and plan on the future of population forecasting.Journal of the American Planning Association, 50(2): 208-221.This article argues for basic changes in the way population forecasts are made and used in the planning process. The misuse and uncritical reliance on projections of current trends should end. More attention must be devoted to the systematic analysis of factors not considered by today's formal models, and a new generation of research methods and skills must be developed to study the future and planning's ability to shape it. Important initial steps in this reorientation from projections to forecasts are within the current resources and capabilities of planning agencies. Perhaps most important, however, the need to forecast is an invitation to planners to become again a source of ideas and inspiration for their communities about what might be and what ought to be.


Bach R L, Schraml L A, 1982. Migration, crisis, and theoretic conflict.International Migration Review, 16: 320-341.PIP: The nature of the distinction between the equilibrium and historical-structuralist positions on migration is examined. Theoretical and political differences in the two positions are considered both historically and in the context of the current global economic crisis. The proposal of Wood to focus on households as a strategy for integrating the two perspectives and for achieving a better understanding of migration and social change is discussed.


Booth H, Hyndman R J, Tickle Let al., 2006. Lee-Carter mortality forecasting: A multi-country comparison of variants and extensions.Demographic Research, 15(9): 289-310.We compare the short- to medium-term accuracy of five variants or extensions of the Lee-Carter method for mortality forecasting. These include the original Lee-Carter, the Lee-Miller and Booth-Maindonald-Smith variants, and the more flexible Hyndman-Ullah and De Jong-Tickle extensions. These methods are compared by applying them to sex-specific populations of 10 developed countries using data for 1986-2000 for evaluation. All variants and extensions are more accurate than the original Lee-Carter method for forecasting log death rates, by up to 61%. However, accuracy in log death rates does not necessarily translate into accuracy in life expectancy. There are no significant differences among the five methods in forecast accuracy for life expectancy.


Chen Daiyun, (in Chinese)

Gans H A,1999. Population Forecasting 1895-1945: The Transition to Modernity. Dordrecht, The Netherlands: Kluwer.

Harris R S, Moore E G, 1980. An historical approach to mobility research.Professional Geographer, 32: 22-29.Most existing work on residential mobility has assumed that the household relocation decision is an intrinsically significant object of inquiry. In contrast, we argue that mobility derives its significance primarily from the particular historical and locational contexts within which it occurs. We suggest, therefore, that future mobility research should be directed away from the development of a theory of mobility per se and toward a more explicit articulation of mobility studies to existing theories of urbanization and social change.


Hyndman R J, Ullah M S, 2007. Robust forecasting of mortality and fertility rates: A functional data approach.Computational Statistics & Data Analysis, 51: 4942-4956.ABSTRACT A new method is proposed for forecasting age-specific mortality and fertility rates observed over time. This approach allows for smooth functions of age, is robust for outlying years due to wars and epidemics, and provides a modelling framework that is easily adapted to allow for constraints and other information. Ideas from functional data analysis, nonparametric smoothing and robust statistics are combined to form a methodology that is widely applicable to any functional time series data observed discretely and possibly with error. The model is a generalization of the Lee&ndash;Carter (LC) model commonly used in mortality and fertility forecasting. The methodology is applied to French mortality data and Australian fertility data, and the forecasts obtained are shown to be superior to those from the LC method and several of its variants.


Kenneth B, Wendling R, 1980. The national-regional impact evaluation system: A spatial model of US economic and demographic activity.Journal of Regional Science, 20: 143-158.ABSTRACT Discusses a methodology developed at the Bureau of Economic Analysis that was designed to address multi-regional modeling problems. Called the National-Regional Impace System, it is comprised of 51 state econometric models that are integrated into a model of the US economy. -from Authors


Keyfitz N, 1968. Introduction to the Mathematics of Population. Reading, Mass: Addison-Wesley.This textbook is arranged in 6 parts: the life table (stationary population model); discrete analysis for 1 sex (recurrence equations, analysis of the projection matrix, projection matrix changing through time); continuous analysis (the renewal equation, parametrization, interrelationships of demographic variables in stable populations, reconciliation of matrix and integral equation, population time series); numerical techniques (interpolation and graduation, finite approximations); multiple populations in interaction (qualitative aspects, simultaneous differential equations, analysis by age); and probability models (sampling variance of demographic variables, birth and death processes, population distributions and individual behavior, the branching process). Additional chapters offer a bibliographic overview of mathematical demography advances since the book's publication in 1968, and explore some problems in the mathematics of population.

Keyfitz N, Caswell H, 2005. Applied Mathematical Demography. 3rd ed. New York: Springer.ButIacceptedtheinvitation.Key?tz ssuggestionthatweincorporatesome of my earlier work on matrix population models seemed like a good way to complement the methods presented in the book, and to expand the range of their applications. I am a demographer of plants and nonhuman animals. Some would call this an oxymoron, since the Greek rootdemos refers to people, and that s us. But there are precedents for taking more inclusive de?nitions of the Greek.Ecologyandeconomics,forexample,bothcomefromtherootoikos, referring to the household. Interpretingdemos as referring to individuals, whether they are persons or not, lets demography apply across species. There is a long tradition of such crossover. Alfred J. Lotka is acknowledged as a founding father of both demography and ecology. Raymond Pearl used demographic methods to analyze the e?ects of toxic substances and cro- ing on fruit ?ies. Lee (1987) compared the density-dependence of the vital rates of human and non-human animals. Today, in studies of senescence, reproduction, and individual heterogeneity, the boundaries between animal and human studies are becoming increasingly blurred (Wachter et al. 1997, Carey 2003, Wachter and Bulatao 2003, Carey and Tuljapurkar 2003).


Liu Shenghe, Deng Yu, Hu Zhang, 2010. Research on classification methods and spatial patterns of the regional types of China's floating population.Acta Geographica Sinica, 65(10): 1187-1197. (in Chinese)<p>With the rapid increase of the number and influence of floating population in China, it is urgent to understand the regional types of China's floating population and their spatial characteristics. After reviewing the current methods for identifying regional types of floating population, this paper puts forward a new composite-index method and its further modification method consisting of two indexes simultaneously: the net migration rate and gross migration rate. Further, those methods are empirically tested by using China's 2000 Census data at county level. The results show: (1) The composite-indexes method is much better than the traditional single-index method because it can measure the migration direction and scale of floating simultaneously and in particular it can identify the unique regional types of floating population with large-scale immigration and emigration. (2) The modified composite-indexes method, by using the share of a region's certain type of floating population to the total in China as weight, can effectively correct the over- or under-estimated error due to the rather large or small total population of a region. (3) The spatial patterns of different regional types of China's floating population are closely related to regional differentiation of their natural environment, population density and socio-economic development level. The three active regional types of floating population are mainly located in the eastern China with lower elevation, more than 800 mm precipitation, rather than in the region with higher population density and economic development level.</p>

Long J F, 1985. Migration and the phase of population redistribution.Journal of Development Economics, 17: 29-42.

Ma Hailing, Zheng Hong, (in Chinese)

Nakicenovic, 2000. Special Report on Emissions Scenarios, IPCC, Geneva and Cambridge University Press.

National Bureau of Statistics (NBS), 2001. China Statistical Yearbook. Beijing: China Statistics Press. (in Chinese)

National Bureau of Statistics (NBS), 2011. China Statistical Yearbook. Beijing: China Statistics Press. (in Chinese)

Stauffer D, 2002. Simple tools for forecasts of population ageing in developed countries based on extrapolations of human mortality, fertility and migration.Experimental Gerontology, 37: 1131-1136.Suitable assumptions for the Gompertz mortality law take into account the break in the time development observed recently by Wilmoth et al. They show how a drastic reduction in the birth rate and improved living conditions lead to a drastic increase in the fraction of old people in the population, and how immigration of half a percent of the population per year can mostly stop this increase.


Thomas P, 1981. An econometric model for forecasting regional population growth.International Regional Science, 6: 53-70.An econometric model for forecasting net migration and natural increase is proposed and then estimated using time-series data for Texas. The model is simulated five years out-of-sample and found to be quite accurate in forecasting future population growth. It outperforms simpler prediction methods, thus indicating that explicit modeling of net migration and natural increase is superior to modeling only total population." excerpt


Wood C H, 1982. Equilibrium and historical-structural perspectives on migration.International Migration Review, 6: 298-319.The equilibrium model of labor mobility and the historical-structural perspective on population movement are summarized and critiqued. A comparative analysis identifies the sources of the growing discontinuity in the contemporary literature on migration by exploring the theoretical and methodological implications of the contrasting paradigms of socioeconomic development in which each perspective is embedded. The last section outlines an alternative approach to the study of migration by shifting the unit of analysis to the household."


Zhang Kevin Honglin, Song Shunfeng, 2003. Rural-urban migration and urbanization in China: Evidence from time-series and cross-section analyses.China Economic Review, 14:;h2 class="secHeading" id="section_abstract">Abstract</h2><p id="">Since 1978, China has experienced a rapid and unprecedented process of urbanization, created by the history's largest flow of rural&ndash;urban migration in the world. This article attempts (a) to assess the role of the cityward migration in China's urbanization in 1978&ndash;1999 and (b) to empirically investigate factors behind the migration boom with time-series and cross-section data. We find that (a) rural&ndash;urban migration made dominant contributions to Chinese urban population growth; (b) while moving together with the Chinese economy, the causal link runs from economic growth to migration, not vice versa; (c) interprovince migrants were encouraged by the rural&ndash;urban income gap and discouraged by their geographic distances to destinations; and (d) the amount of intraprovince migrants is positively related to rural&ndash;urban income gap and urban population in that province.</p>


Zhu Baoshu, 2011. The changes in fertility of the migrants in the urban from the rural.South China Population, 26(6): 42-50. (in Chinese)As the main part of urbanization process,the migration from the rural to the urban generally brings about the drop of fertility or the stabilization of low fertility.As the migrants from the rural are different both from the urban residents and the rural residents,the migration of course can cause differential effects in the fertility in the rural and the urban.Usually the fertility of the migrants in the urban from the rural are lower than the fertility of the rural and higher than that of the urban.However,the resent studies show that the fertility of these migrants in urban even lower than the fertility of the urban residents,the significance of this fertility reverse need more and more further studies.This paper analyzes the fertility changes in the rural and the urban and the regional difference and estimates the total fertility rate of the migrants in the urban from the rural and discusses the child-bearing behaviors of the so called floating population.