A longitudinal study of changes in intra-metropolitan employment concentration in Beijing: Decentralisation,reconcentration and polycentrification

Abstract

Chinese

This article explores the spatial trend of intra-metropolitan employment concentration in the Beijing metropolitan area from 2004 to 2013. We use multiple-year economic census data and construct a unique longitudinal dataset of employment at the subdistrict level, applying the locally weighted regression method to identify employment centres in Beijing. The results show that jobs continued to decentralise from the urban core to the suburbs in Beijing over the studied period; however, different from the common trajectory of Western metropolitan spatial evolution, decentralised jobs tended to be more concentrated in subcentres and the polycentric urban structure prevailed in Beijing. Polycentricity strengthened in Beijing from 2004 to 2008, but job dispersion increased at the expense of polycentricity during 2008–2013. However, job dispersion did not follow an unstructured model; instead, subcentres played an increasingly important role in structuring the dispersed jobs. Job dispersion in Beijing has been more associated with the scatteration of service jobs, while manufacturing jobs tended to be more concentrated in subcentres, which contributed to the polycentric development of the capital. We also observe the persistence in the location of employment centres in Beijing over the studied period, as well as substantial spatial changes in the boundaries of employment centres, which suggests that the employment concentrations within the metropolitan area are persistent but not static.

Keywords

Beijing employment centres land-use change polycentricity urban spatial structure

Introduction

Employment concentrations within metropolitan areas manifest agglomeration economies at the sub-metropolitan level that determine urban spatial structure. Numerous studies on Western metropolitan areas have shown that multiple employment centres emerged outside the CBD with a continuing decentralisation of economic activities, which has profoundly changed urban spatial structure and led to a more dispersed and polycentric urban form (e.g. Agarwal et al., 2012; Anas et al., 1998; Hajrasouliha and Hamidi, 2017). However, although it is generally agreed that contemporary metropolitan areas are characterised by decentralised employment and decline of the CBD, there is less agreement on the trajectory of metropolitan spatial trends. Two competing views of emerging urban spatial structure have been developed (for a thorough discussion, see Garcia-Lopez and Muniz, 2010; Giuliano et al., 2007; Lee, 2007). Many studies point out that contemporary metropolitan areas are transforming into polycentric spatial configurations, with decentralised employment re-concentrating into a number of suburban subcentres due to rising forces of agglomeration in both production and consumption (e.g. Garreau, 1991; Stanback, 1991). Meanwhile, other scholars question the general trend towards the polycentric pattern, instead predicting a rather generally dispersed land use pattern as the spatial outcome of the decreasing transportation costs and the rise of ICTs (Information and Communication Technologies) (e.g. Gordon and Richardson, 1996; Lang, 2003). Empirical evidence in the debate so far has mostly been based on research conducted in North America and Europe. Few studies have been carried out on metropolitan areas in the developing world, and much less is known about how cities in developing countries have changed over the decades, and whether the trend of Western metropolitan evolution is similar there.

China has undergone rapid urbanisation over the past three decades. Responding to global forces of economic restructuring and technology advancement, and driven by unprecedented urbanisation and massive urban land development and transport investment, Chinese cities are becoming more decentralised and facing significant spatial transformation and restructuring (Gaubatz, 1999; Ma, 2004; Wei, 2012). To tackle the challenges of decentralised urban development and the problems that arise with urban sprawl, polycentric development has been introduced as a major spatial planning tool in China (Cheng and Shaw, 2018). Meanwhile, large-scale suburban developments, such as industrial parks, economic zones, new towns and new districts, are widespread on Chinese urban fringes, meaning the concentration of economic activities in suburban planned areas, and facilitating the formation of employment subcentres outside the old urban cores. This raises the interesting question as to whether Chinese cities are becoming polycentric.

Examining the changes in employment concentration in Chinese cities is of interest in two ways. First, it adds further evidence to the debate on urban spatial restructuring from a developing and transitional urban context. Second, the existing empirical evidence on the subject comes mainly from North America, which tends to overstate the role of the market (Shearmur and Alvergne, 2003). As some have argued, planning and local policies, such as land use regulations and local development incentives, could potentially influence the location decisions of firms, and have a marked influence on the spatial trend of metropolitan development (Agarwal, 2015; Shearmur and Alvergne, 2003). In China, the state and local governments exert strong influences on urban development strategies and policies, and spatial planning has been used as a strategic tool to direct growth towards designated areas and to shape spatial, political and socio-economic urban transformation. Therefore, the spatial evolution of employment concentration in Chinese cities may show a trend different from that of their Western counterparts. The examination of Chinese cities can shed light on whether planning policy can influence the general trend of employment decentralisation and reconcentration in metropolitan areas, and provides a policy-based explanation of the trajectory of metropolitan spatial evolution.

This study presents an examination of employment centres in metropolitan Beijing, one of the mega-cities in China, over a decade from 2004 to 2013, and tests whether Beijing became more polycentric or more dispersed during the period. We use multiple-year economic census data of Beijing and construct a unique longitudinal dataset of employment at the subdistrict level in 2004, 2008 and 2013, which covers 245 subdistricts in Beijing. The locally weighted regression (LWR) method presented in McMillen (2001) is applied to identify employment centres in Beijing, as it provides empirical results comparable to those of US and European studies. After identifying employment centres, we analyse the changing pattern of employment centres over the years, and examine how the employment centres influence the overall employment distribution within the metropolitan area.

Literature review

The polycentrification of Chinese cities has been attracting wide research attention recently. Wu (1998) was among the first to empirically examine the polycentric urban form in China with a case study of Guangzhou. His study demonstrated that polycentric urban development has appeared in China since the introduction of urban land reform. In recent years, the polycentric urban structure has been examined in depth, with case studies mainly focusing on a few leading Chinese cities (Feng et al., 2009; Fragkias and Seto, 2009; Huang et al., 2015, 2017; Qin and Han, 2013; Sun et al., 2012; Wen and Tao, 2015; Yue et al., 2010; Zou et al., 2015). Among all cases, the most widely examined is Beijing. Although some studies emphasised that the spatial extent of decentralisation in Beijing was still quite limited as compared with that of cities in the US, and the main city centre still held great importance in determining the spatial pattern of Beijing, suburban subcentres have been identified and proved to be the key factors in explaining the spatial distributions of population, housing prices and land development in Beijing.

In spite of many studies on the polycentric spatial structure of Chinese cities lately, there is still a lack of longitudinal studies that empirically examine the changes in urban spatial structure in decentralised Chinese cities. In studies of US metropolitan areas, there is a debate as to whether these areas have become more polycentric or more dispersed. Although the fact of the emergence of suburban employment subcentres has been well established in US metropolitan areas, it is still disputed whether the emerging new urban pattern is characterised by a polycentric structure or if employment centres are becoming less significant in determining urban spatial structure and the polycentric structure has given way to a generally dispersed urban form. When examining the employment distribution of subcentres in Los Angeles from 1970 to 1990, Gordon and Richardson (1996) found that the share of metropolitan jobs in subcentres was small and had fallen over the time period, and argued that Los Angeles might be described as a dispersed rather than a polycentric metropolitan area. Similarly, Lee (2007) analysed the spatial trends in selected US metropolitan areas during the 1980s and 1990s, and discovered that generalised job dispersion was more common than subcentring in those areas. In their study investigating the spatial structure of employment in 356 US metropolitan areas, Hajrasouliha and Hamidi (2017) found that job dispersion was a dominant characteristic in almost 70% of all metropolitan areas, and polycentricity was mostly found in large metropolitan areas. Meanwhile, other studies have emphasised that intra-metropolitan employment concentrations have remained remarkably stable and agglomeration economies at the sub-metropolitan level have continued to be important in the urban space economy. Through analysing the employment pattern inside and outside centres in Los Angeles from 1980 to 2000, Giuliano et al. (2007) found that the share of employment in centres was quite stable although the main centre kept losing jobs, and that only in the outer suburban counties was the dominant trend dispersion. Redfearn (2009) also documented the medium- and longer-term stability of the location of employment centres in Los Angeles. In a more recent study, Kane et al. (2018) analysed the changes in employment centres in Los Angeles from 1997 to 2014 using more sophisticated data and methods, and found that despite a small proportion of total employment in centres, growth in centres outpaced overall employment growth in the region.

The quantitative analysis dealing with urban spatial restructuring has been significantly influenced by the research conducted in North America, which is based on the assumption of efficient markets and understates the role of government policy. Shearmur and Alvergne (2003) addressed this concern by examining the growth of employment centres across the Paris region, where a clear and consistent regional plan was implemented. They found that the five centres designated in the regional plan displayed the fastest employment growth, and the planning policy had a strong influence on the location of employment in Paris. Another study that examined the role of subcentres in the growth pattern of the Tokyo metropolitan area also found a considerable tendency towards polycentric development, with strong polycentric structure planning (Sorensen, 2001). Due to the less formal planning of metropolitan areas, US studies tended to find no evidence that local policies directly influenced employment centre growth (Agarwal, 2015). However, it is still arguable that strong government interventions and planning policies can facilitate the growth of employment centres in metropolitan areas to combat urban sprawl. As some have argued, the nature of urban spatial restructuring, the underlying driving forces and the processes involved, are culturally and historically specific in different countries and embedded in local economic and political systems (Freestone and Murphy, 1998). More comparative analysis looking across different institutional contexts and beyond the North American experience is important for understanding more thoroughly the changing nature of the spatial structure of contemporary metropolitan areas.

Study area, data and methodology

Study area and data

The study area is the Beijing metropolitan area, which is the capital city and one of the six mega-cities of China. We define the Beijing metropolitan area to include the central city and the adjacent suburbs, which consist of 12 urban districts with a total area of 9115 km². According to the administrative divisions of Beijing, the Beijing metropolitan area is differentiated into three regions: the inner city, the inner suburbs and the outer suburbs. Figure 1 shows a map of the Beijing metropolitan area, with the boundaries of the inner city, the inner suburbs and the outer suburbs, and it also shows the employment density distribution in the area in 2013.

Figure 1.

Study area and employment density distribution, 2013.

To identify urban spatial structure, we compile employment data from the 2004, 2008 and 2013 economic censuses of Beijing. The data are at the subdistrict level, which is the finest geographical unit for which census data are available for Chinese cities.

Employment centre identification

In this study, we apply the LWR method proposed by McMillen (2001) with some modifications to identify employment centres. In the first stage, a locally weighted regression procedure is used to smooth the natural logarithm of employment density throughout the metropolitan area. Following McMillen (2001), we use a large window size (50%) and potential employment centre areas are those subdistricts whose residuals are significantly greater than zero at the 5% significance level. The LWR method has the advantage of detecting local rises in the employment density surface; this is especially crucial for employment centre identification in Beijing’s case, because of the large size of the metropolitan area and its great variation in employment density.

In the second stage, to avoid the complexity of the semiparametric procedure proposed by McMillen (2001), we follow Lee’s (2007) approach and define employment centres as those clusters of potential centre areas that satisfy the minimum employment threshold. As such, employment centres identified have both local significance in terms of higher employment density than surrounding areas and regional impact in terms of their employment size. If the total employment of adjacent candidate subdistricts identified in the first stage is greater than 0.5% of the metropolitan employment, we consider them to be employment centres. Subdistricts sharing common boundaries are defined as adjacent to each other. In some cases, if an employment centre crosses district boundaries, we separate the centre at the boundary. This is because in China administrative districts have distinct policies and administrations which can affect the functions of employment centres.

Exploratory analysis of employment centres

After identifying employment centres, we perform an exploratory analysis to understand the spatial dynamics of employment concentration. First, the changes in employment centres in both area and employment are analysed by treating employment centres as discrete entities. We apply the persistence score proposed by Kane et al. (2018) to measure the degree to which the areas and boundaries of employment centres have changed over time. The persistence score is given by:

\begin{matrix} Persistence score = \\ \frac{{centre}_{t} \cap {centre}_{t + 1} \cap \dots \cap {centre}_{t + n}}{{centre}_{t} \cup {centre}_{t + 1} \cup \dots \cup {centre}_{t + n}} \end{matrix}

where the numerator is the area of subdistricts that are persistent in centres across all periods, and the denominator is the area of subdistricts that are present in centres at any period. Then the aggregate employment trends within and outside of centres are examined in both the metropolitan area and the three regions. These results are helpful to reveal the general spatial trends of employment concentration within the Beijing metropolitan area.

Employment centre influence

To test whether the Beijing metropolitan area has become more polycentric or more dispersed, we follow Garcia-Lopez and Muniz (2010) and apply a polycentric density function to estimate the influence of employment centres, which is given by:

Ln D_{i} = β_{0} + β_{1} d_{mc, i} + β_{2} d_{sc, i} + ε_{i}

where LnD_i is the logarithmic density of employment at the subdistrict i, d_mc,i is the distance of the subdistrict i to the nearest main employment centre and d_sc,i is the distance of the subdistrict i to the nearest employment subcentre. Unlike most Western cities with the CBD as the urban centre, the heart of Beijing is occupied by the old imperial city, meaning the city has two separate central business areas, one each in the eastern and western parts of the inner city. Both of the two main centres are identified using the LWR method along with all subcentres. β₁ and β₂ are the density gradients associated with the distances to the main centre and the subcentre; ε_i is the error term with the usual properties.

The density gradients are expected to be negative, which means urban density decreases as we move further away from employment centres, because the longer the distance to employment centres, the worse the access to agglomeration economies. A positive and/or a non-significant gradient would show the lack of influence of employment centres. In terms of spatial dynamics, if the subcentre gradient and its significance have increased, the polycentric spatial structure would have been strengthened in the metropolitan area, while if the gradients of both the main centre and the subcentre have lost significance and value, a more generally dispersed urban form would emerge.

Results

To reveal the spatial trend of employment concentration in Beijing, we first examine the decentralisation and reconcentration of employment by analysing the changes in employment shares of different groups of subdistricts. The spatial structure of Beijing and its changes are then analysed by examining the identified employment centres. We look into both the location and boundary changes of employment centres and summarise the changes in employment shares and densities within and outside of the centres over time. Lastly, we quantify the centre influence by using the density function to reveal the spatial restructuring trend of Beijing.

Trends of employment decentralisation and reconcentration

To reveal the overall spatial changes in employment distribution in the Beijing metropolitan area, all subdistricts are first divided into five groups by their distances from the city centre and the changes in each group’s share of metropolitan employment over the years are presented in Figure 2. It would seem that employment decentralisation has occurred since 2004. The employment share of subdistricts less than 5 km from the city centre has decreased gradually since then, while for subdistricts more than 5 km but less than 40 km away it has increased, especially in places 5–20 km from the city centre. Meanwhile, the figure shows that the spatial extent of employment decentralisation has expanded over years. From 2004 to 2008, the spatial extent was limited by 5 km from the city centre, while from 2008 to 2013, the spatial extent increased to 10 km from the city centre because the employment shares of subdistricts less than 5 km and 5–10 km from the city centre decreased. Figure 2 also shows the spatial distribution changes of manufacturing and service employment. The decentralisation trend is also evident for employment in both sectors, while the difference is that the spatial extent of employment decentralisation in manufacturing was larger, limited to 10 km from the city centre from 2004 to 2008 and further increasing to 20 km from the city centre during 2008 to 2013.

Figure 2.

Trend of employment decentralisation and concentration; changes in employment shares by distances from the city centre and by density quantiles.

The studies on US metropolitan areas generally found that employment decentralisation has occurred together with employment deconcentration (e.g. Lee, 2007). But this is not the case in Beijing. To examine the trends of employment concentration, all subdistricts are further grouped into four quantiles by their densities, and the changes in each quantile’s share of metropolitan employment are presented in Figure 2. From 2004 to 2008, the employment share increased in the highest density areas, and the most significant employment gain was in the first quantile. This indicates that jobs shifted towards the higher density areas in this period, meaning the metropolitan area became more concentrated. However, the deconcentration trend was observed from 2008 to 2013, with decreased job share in the highest density areas (the first quantile) and increased job share in the middle density areas (the second and third quantiles). This shows that the metropolitan area became slightly more dispersed in this period. Figure 2 also shows the different trends of employment concentration for manufacturing and services. The similar concentration trend to that of total employment is evident for services, but the spatial distribution change for manufacturing employment indicates a strong reconcentration trend, with the employment share in the highest density areas increasing continuously and significantly during both periods. This implies that the decentralised manufacturing jobs tended to be reconcentrated in subcentres, while the service jobs tended to become more dispersed in the latter period, which has led to the overall dispersion of employment in the Beijing metropolitan area.

Patterns and spatial dynamics of employment centres

The trend of employment decentralisation and reconcentration implies the emergence of a polycentric urban structure. The spatial structure in Beijing is further analysed by examining the identified employment centres. Both the main employment centres and the subcentres identified by the LWR method are presented in Figure 3. The results confirm the existence of two main employment centres and a group of subcentres in Beijing, with the number of subcentres being between 15 and 17 depending on the year of identification. The locations of employment centres are consistent with the local knowledge. The two main centres are in the inner city – the Finance Street area in the west (Centre2) and the CBD area in the east (Centre1). The subcentres in the inner suburbs are mostly located in the major sci-tech parks and industrial parks of Beijing. The subcentres in the outer suburbs are mainly suburban new towns with self-sustained economic activities and administrative functions, as well as a major employment subcentre around the capital’s international airport in the north-eastern suburbs (Centre13), a centre based on a giant state-owned enterprise, Yanshan petrochemical company, in the south-west suburbs (Centre18) and a centre based on the Beijing Economic and Technological Development Zone in the south-eastern suburbs (Centre15).

Figure 3.

Employment centres in the Beijing metropolitan area, 2004–2013 (the employment centres are numbered, and their ID numbers correspond to those in Table 1. See Table 1 for the growth and spatial changes of each centre).

Table 1.

Stability and growth of employment centres, sorted by distance from the city centre.

Centre ID	Distance from the city centre (km)	Regions	Status	Land area (km²)			Persistence score	Total employment in 2013 (1000)	Growth rate of centre employment density after emerged (%)
Centre ID	Distance from the city centre (km)	Regions	Status	2004	2008	2013	Persistence score	Total employment in 2013 (1000)	Growth rate of centre employment density after emerged (%)
1(main)	0.9	Inner city	Persistent	19.3	25.9	27.3	0.41	1149	46.4
2(main)	1.4	Inner city	Persistent	28.1	14.6	10.7	0.31	524	99.2
6	5.8	Inner suburbs	Persistent	7.6	5.4	2.9	0.38	66	33.3
4	6.7	Inner suburbs	Persistent	38.2	38.3	23.3	0.52	711	107.9
3	7.2	Inner suburbs	Emerged		6.6	6.6	0	178	−35.7
8	7.7	Inner suburbs	Emerged		7.4	7.4	0	110	−0.2
7	9.9	Inner suburbs	Persistent	5.0	14.7	5.0	0.34	84	20.1
9	11.3	Inner suburbs	Persistent	7.1	7.1	7.1	1	55	38.2
5	14.9	Inner suburbs	Persistent	9.8	9.8	9.8	1	259	307.8
10	16.8	Inner suburbs	Persistent	14.8	20.5	20.7	0	147	31.4
15	18.9	Outer suburbs	Persistent	74.6	74.6	74.6	1	305	193.9
16	20.7	Outer suburbs	Persistent	28.4	28.4	28.4	1	70	86.0
14	20.7	Outer suburbs	Persistent	12.9	23.1	39.7	0.25	106	−14.0
13	21.8	Outer suburbs	Persistent	162.5	116.1	114.6	0.42	353	174.3
19	25.3	Outer suburbs	Persistent	7.2	14.6	7.2	0.50	52	14.5
12	25.9	Outer suburbs	Persistent	55.1	55.1	55.1	1	68	24.5
17	29.0	Outer suburbs	Persistent	39.0	39.0	39.0	1	74	19.6
11	36.6	Outer suburbs	Persistent	33.3	33.3	33.3	1	141	61.7
18	40.7	Outer suburbs	Persistent	89.7	89.7	75.2	0.84	77	23.7
Total				632.5	624.1	587.8	0.62	4532	69.3

In terms of the locational shift of employment centres, it is shown that the locations of employment centres are quite stable over time. This is consistent with the argument of Redfearn (2009) about the persistence of urban form due to the important role of agglomeration economies. Over the period of a decade with pronounced urban growth, there were only two newly emerged employment subcentres in Beijing (Centre3 and Centre8). Although the locations of employment centres were persistent, their boundaries changed substantially over time. Table 1 shows the persistence score, the density growth rate and the total employment for each individual employment centre in Beijing. Only seven of all 19 centres did not experience any boundary change over the period from 2004 to 2013, and have the persistence score of 1. This is consistent with Kane et al. (2018), which argued that employment centres that are persistent in locations can be vibrant in space with complex spatial dynamics.

Unlike US metropolitan areas, the main centres in Beijing were not quite stable, with relatively low persistence scores, and they experienced dramatic spatial changes because of the rapid job decentralisation and urban renewal of the inner city. Overall, employment subcentres in the inner suburbs were more volatile, especially those with a 5–10 km distance from the city centre. Not only did new centres emerge in these areas, but the boundaries of persistent centres also changed substantially with spatial shifts, making their persistence scores relatively low. This is because the spatial extent of employment decentralisation in Beijing was mainly limited by 10 km from the city centre, as discussed above. Subcentres with a 10–20 km distance from the city centre were quite stable in their boundaries, with only one exception. Centre10 is the only persistent subcentre, with a persistence score of 0. The boundary of this centre has changed completely over the period 2004–2013, mainly because the centre was based on a giant state-owned enterprise, the Capital Steel company, which relocated out of Beijing in 2008, leading to a dramatic change in employment densities within the area. Subcentres in the outer suburbs were mostly more stable in their boundaries than those in the inner suburbs. The four subcentres that are largest in employment size, with a total workforce of more than 200,000, experienced the most rapid density growth, and these were developed as some of the most important economic bases in Beijing, including the Zhongguancun sci-tech park (Centre4), the Yizhuang economic development zone (Centre15), the Shangdi software industrial park (Centre5) and the economic zone around the airport (Centre13).

Spatial structure changes: Polycentrification vs dispersion

Regarding the overall spatial structure changes in the Beijing metropolitan area, Table 2 summarises the changes in employment shares and densities within and outside of employment centres over time. Overall, although the majority (over half) of metropolitan jobs were still dispersed outside employment centres, a fairly large proportion of total employment was concentrated in the employment centres, as compared with most of US metropolitan areas. The spatial structure of the Beijing metropolitan area is genuinely polycentric, since substantially more jobs were concentrated in subcentres than in the main centres, with the employment shares in the main centres between 15.9% and 18.8% and the employment shares in subcentres between 24.6% and 32.4%, depending on the year under discussion. The continuous job concentration with employment decentralisation has strengthened the polycentricity in Beijing from 2004 to 2008. Not only has the number of subcentres increased in the period, with two newly emerged subcentres, but the increase in employment in subcentres has been intense, and the employment share in subcentres grew from 24.6% in 2004 to 32.4% in 2008, while the employment share in the main centres decreased from 18.8% to 15.9%. However, in the latter period of 2008–2013, job dispersion increased at the expense of polycentricity, with the subcentres’ employment share decreasing from 32.4% to 27.1% and the employment share outside centres increasing from 51.7% to 57.1%.

Table 2.

Employment trends within and outside of the centres in Beijing, 2004–2013.

	2004	2008	2013
Number of centres	17	19	19
Main centre	2	2	2
Subcentres	15	17	17
Total employment (1000)	6631	7773	10,555
Percent in centres (%)	43.4	48.3	42.9
Main centre	18.8	15.9	15.9
Subcentres	24.6	32.4	27.1
Total area of centres (km²)	632.5	624.1	587.8
Main centre	47.4	40.4	38.0
Subcentres	585.2	583.6	549.9
Total area out of centres (km²)	8483.8	8492.2	8528.5
Average employment density in centres (1000/km²)	4.6	6.0	7.7
Main centre	26.3	30.6	44.1
Subcentres	2.8	4.3	5.2
Average employment density outside centres (1000/km²)	0.4	0.5	0.7
Manufacturing employment (1000)	1323	1201	1230
Percent in centres (%)	35.6	41.9	42.0
Main centre	3.0	2.0	1.4
Subcentres	32.6	39.9	40.6
Service employment (1000)	4565	5959	6157
Percent in centres (%)	44.5	49.3	45.1
Main centre	24.2	19.4	19.0
Subcentres	20.3	29.9	26.0

Although the employment share of the main centres decreased over time, their average employment density still increased due to the shrinkage of their total area. So the overall concentration of employment in the main centres has not decreased, but the scale of employment concentrated in the main centres has declined relatively. From 2004 to 2008, the total area of subcentres remained steady, and the employment share and average density of subcentres increased substantially, indicating the reconcentration and polycentrification of decentralised jobs. From 2008 to 2013, both the total area and the employment share of subcentres have decreased greatly, while the average density of subcentres still increased. Overall, the average densities of both the main centres and subcentres increased continuously during the two periods, and the average density growth rate of the subcentres throughout the whole 2004–2013 period was the highest compared with all other areas, reflecting the overall job concentration and polycentrification trend in Beijing. However, in the 2008–2013 period, the average density growth rate of non-centre areas was higher than that of centre areas, implying that jobs became more dispersed with the continuous employment decentralisation in this period.

Table 2 also shows the changes in employment shares within and outside of employment centres in both the manufacturing and service sectors over time. The employment share of subcentres related to total manufacturing jobs kept increasing over the whole period, while their share related to total service jobs increased in the early period and then decreased in the latter period. This suggests that the concentration of both manufacturing and service activities drove the growth of subcentres and the polycentrification of urban structures in Beijing from 2004 to 2008, but that in the latter period, service jobs became more dispersed while manufacturing still tended to be highly concentrated in subcentres. Although studies on US metropolitan areas have linked the emergence of suburban subcentres to the decentralised service functions, especially the knowledge-intensive services (e.g. Kane et al., 2018), the polycentrification of urban structure in Beijing has been more associated with the decentralised manufacturing activities. Manufacturing was usually targeted for its potential to promote suburban development in Beijing’s city plan, often in the form of industrial parks, sci-tech parks or economic development zones, driving the formation of suburban subcentres in the Beijing metropolitan area.

Employment centre influence and spatial restructuring

To test whether the Beijing metropolitan area has become more polycentric or more dispersed, or in other words, how the location pattern of employment within the metropolitan area has been affected by the structuring elements of the city, we measure the influence of employment centres on the employment density pattern in Beijing by estimating the density gradients of both the main centre and the subcentre. Table 3 shows that the estimated gradients are significant in all three years. The negative signs indicate that employment density decreased with distance to the main centre and the nearest subcentre, which indicates that employment centres do determine the intra-metropolitan employment distribution pattern.

Table 3.

Changes in the density gradients of employment centres in Beijing, 2004–2013.

Y	Ln employment density
	2004	2008	2013	2004	2008	2013
dmc	−0.103***	−0.102***	−0.0992***	−0.0888***	−0.101***	−0.101***
	(–18.42)	(–16.27)	(–16.00)	(–10.17)	(–11.82)	(–11.96)
dsc	−0.0559***	−0.0768***	−0.0835***
	(–4.37)	(–5.50)	(–6.07)
dsc*inner city				0.0758*	0.0426	0.0109
				(2.09)	(0.82)	(0.21)
dsc*inner suburbs				−0.0783**	−0.140***	−0.148***
				(–3.26)	(–5.35)	(–5.61)
dsc*outer suburbs				−0.0751***	−0.0766***	−0.0800***
				(–5.00)	(–4.98)	(–5.29)
constant	9.559***	9.659***	9.948***	9.295***	9.692***	10.04***
	(84.90)	(88.64)	(91.57)	(48.20)	(54.41)	(56.50)
adj. R-sq	0.779	0.800	0.800	0.795	0.814	0.812

Notes: t statistics in parentheses; * p<0.05 ** p<0.01 *** p<0.001.

From a longitudinal point of view, the density gradient of the main centre has decreased since 2004, showing a decline in its structuring role. On the other hand, the subcentre gradient has increased substantially over time, which shows that the subcentres had a growing influence on the locations of employment outside centres, and played a key role in structuring the employment distribution within the metropolitan area. Overall, the polycentric structure has been strengthened in Beijing over the years, not only with the emergence of new subcentres and the increase in subcentre employment share, but also with the greater influence of subcentres on the location of non-central jobs. In the meantime, with the continuous decentralisation process, the influence of the main centre has fallen.

To differentiate the influence of subcentres in different regions, we further create three dummy variables that indicate whether the subdistrict is in the inner city, the inner suburbs or the outer suburbs, and three interaction terms between the subcentre distance variable and the dummy variables. The coefficients of these interaction terms should reflect the different influence of subcentres in these different regions. Overall, the influence of subcentres on the employment distribution in the inner suburbs was stronger than that in the outer suburbs, and meanwhile the subcentre gradient increased substantially over time in the inner suburbs, while in the outer suburbs the subcentre gradient did not change much. This suggests that although the spatial changes of subcentres in the inner suburbs were most dramatic, their structuring role has been strengthened constantly, and the polycentric structure has been enhanced most significantly there.

Discussion

Changing patterns of employment concentration within contemporary metropolitan areas have often been linked to the declining cost of transportation, the increasing importance of proximity in the knowledge-based economy, as well as both production- and consumer-side agglomeration economies (Agarwal et al., 2012; Kane et al., 2018; Lee, 2007). The market forces behind the location choice of firms eventually impact the urban structure by weighing the benefits against the costs of agglomeration, and the change in intra-metropolitan employment concentration crucially depends on the change in the nature and geographical scope of agglomeration economies (Lee, 2007). However, the changing pattern of employment concentration in Beijing has not only been driven by market forces, but also by the government’s planning efforts.

In addition to market forces, the predominant role of planning and local government intervention is evident in the urban process of decentralised and polycentric development in Beijing. Since the 1990s, a decentralised clustered spatial scheme has been proposed in the city’s master plan, and several satellite towns, high-tech industrial parks and economic development zones have been designated since then in the suburbs to restrain the sprawl of the central city. The new city plan launched in 2004 continued the idea of decentralised and polycentric development, and new urban centres with different functions were planned to concentrate industries that were highly specialised, such as the industrial parks and development zones in Zhongguancun, Yizhuang, Shangdi and Shunyi. These strategic locations have been promoted in the urban development process by allocating more urban development land, facilitating the construction of transport facilities and encouraging new firms, especially high-tech firms, to locate in these areas through tax incentives and other subsidies (Liu and Liu, 2018). Over the decade, these places were developed as what are essentially large suburban employment subcentres, which facilitated the reconcentration and polycentrification of employment in Beijing with the employment decentralisation.

Initially, the decentralisation of the population and industries in Beijing was driven by the government-led industrial relocation projects and the state-funded housing programmes (Feng et al., 2008; Wu and Phelps, 2011). The relocation of industries and residents propelled the early-stage suburbanisation and polycentric development of Beijing. The establishment of land and housing markets and the inflow of migrant population since the late 1990s have further stimulated massive suburban growth, and the continuous decentralisation of the population with the further influx of migrants was an important driving force of the dispersion of service activities in the latter period. Meanwhile, attracted by tax benefits, the improved infrastructure and the advantage of industrial agglomeration, manufacturing industries were inclined to locate in suburban industrial parks and economic development zones, which played an important role in driving the polycentric development of Beijing (Zou et al., 2015).

The changing spatial trend of employment concentration after 2008 can be partly attributed to the 2008 Olympic Games held in Beijing. The Olympics acted as a catalyst for urban development and restructuring due to the massive investment in urban infrastructure and construction for the event, including the rail transit system, new highway and ring roads and some mega-projects such as the new airport terminal (Zou et al., 2015). These new constructions promoted urban expansion and the dispersion of the urban population and industries, which led to the overall employment dispersion in the latter period. Overall, the planning efforts and the market forces, with the establishment of the land and housing markets since the late 1990s, have together driven the changing pattern of employment concentration in Beijing, which should be understood within Beijing’s planning and development history.

Conclusion

This article has explored the spatial trends of intra-metropolitan employment concentration in the Beijing metropolitan area from 2004 to 2013, to provide further evidence to the debate on urban spatial restructuring from a developing and transitional urban context with strong planning interventions. The results show that jobs continued to decentralise from the urban core to the suburbs in Beijing over the period. Unlike in most of Western metropolitan areas, where employment decentralisation has occurred together with employment deconcentration, employment in Beijing became more concentrated with job decentralisation, as shown by the shift of jobs towards the highest density areas, especially in the early period of 2004–2008. An examination of employment trends within and outside of employment centres shows that a fairly large proportion of jobs were concentrated in centres in Beijing, with more jobs in subcentres than in the main centres, and the employment growth in subcentres has been intense and has outpaced the growth of non-central employment. Therefore, subcentring was a predominant spatial trend in Beijing, rather than the job dispersion that is more common in US metropolitan areas.

The results suggest that polycentricity was strengthened in Beijing during the 2004–2008 period. New subcentres emerged in the suburbs during the period, and employment within subcentres grew more intensively. Meanwhile, the location of non-central employment was more influenced by subcentres, as shown by the increasing density gradient of subcentres. Meanwhile, during the 2008–2013 period, subcentres’ employment share began to decline and employment outside subcentres grew more intensively. However, job dispersion did not follow an unstructured model; instead, subcentres played a more and more important role in structuring the dispersed jobs, and the growth of non-central employment tended to be more concentrated near the subcentres. Therefore, as argued by Garcia-Lopez and Muniz (2010), polycentricity and scatteration can be complementary to each other. Job dispersion in Beijing has been more associated with the scatteration of service jobs, while manufacturing jobs tended to be more concentrated in subcentres, which contributed to the polycentric development of the city. Our findings suggest that the spatial evolution of employment concentration within the metropolitan area is a continuously changing spatiotemporal process, which should be examined via longitudinal data.

Besides the overall spatial trends of employment concentration across the whole metropolitan area, this article also explored the location and boundary changes of individual employment centres by examining their stability and growth over time. The results show the persistence in the location of employment centres in Beijing over the decade. This stability not only suggests the important role of agglomeration economies in the spatial evolution of the metropolitan area, but is also attributed to the planning efforts of the municipal government. Those subcentres with the largest employment size and the rapidest density growth are all important economic bases designated and promoted as the growth poles in the city’s master plan. Meanwhile, substantial spatial changes in the boundaries of employment centres have also been observed, which suggests that the employment concentrations within the metropolitan area are persistent but not static. Our findings echo Shearmur and Alvergne’s (2003) argument that planning policy can have an effect on urban form, and that regions with a clear and consistent polycentric regional plan are more likely to see subcentring instead of general job dispersion, due to planning interventions to channel metropolitan growth to specific locations and facilitate the emergence and growth of employment centres.

Footnotes

Acknowledgements

The author is very grateful to the journal editor and reviewers for their insightful and constructive comments, which have considerably helped improve the article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is supported by grants from the National Natural Science Foundation of China (41371005). The usual disclaimer applies.

ORCID iD

Tieshan Sun

References

Agarwal

(2015) An examination of the determinants of employment center growth: Do local policies play a role? Journal of Urban Affairs 37: 192–206.

Agarwal

Giuliano

Redfearn

(2012) Strangers in our midst: The usefulness of exploring polycentricity. Annals of Regional Science 48: 433–450.

Anas

Arnott

Small

(1998) Urban spatial structure. Journal of Economic Literature 36: 1426–1464.

Cheng

Shaw

(2018) Polycentric development practice in master planning: The case of China. International Planning Studies 23: 163–179.

Feng

Wang

Zhou

(2009) The spatial restructuring of population in metropolitan Beijing: Toward polycentricity in the post-reform era. Urban Geography 30: 779–802.

Feng

Zhou

(2008) New trends of suburbanization in Beijing since 1990: From government-led to market-oriented. Regional Studies 42: 83–99.

Fragkias

Seto

(2009) Evolving rank-size distributions of intra-metropolitan urban clusters in South China. Computers, Environment and Urban Systems 33: 189–199.

Freestone

Murphy

(1998) Metropolitan restructuring and suburban employment centers: Cross-cultural perspectives on the Australian experience. Journal of the American Planning Association 64: 286–297.

Garcia-Lopez

Muniz

(2010) Employment decentralisation: Polycentricity or scatteration? The case of Barcelona. Urban Studies 47: 3035–3056.

10.

Garreau

(1991) Edge City: Life on the New Frontier. New York: Doubleday.

11.

Gaubatz

(1999) China’s urban transformation: Patterns and processes of morphological change in Beijing, Shanghai and Guangzhou. Urban Studies 36: 1495–1521.

12.

Giuliano

Redfearn

Agarwal

, et al. (2007) Employment concentrations in Los Angeles, 1980–2000. Environment and Planning A 39: 2935–2957.

13.

Gordon

Richardson

(1996) Beyond polycentricity: The dispersed metropolis, Los Angeles, 1970–1990. Journal of the American Planning Association 62: 289–295.

14.

Hajrasouliha

Hamidi

(2017) The typology of the American metropolis: Monocentricity, polycentricity, or generalized dispersion? Urban Geography 38: 420–444.

15.

Huang

Liu

Zhao

, et al. (2017) Emerging polycentric megacity in China: An examination of employment subcenters and their influence on population distribution in Beijing. Cities 69: 36–45.

16.

Huang

Liu

Zhao

(2015) Monocentric or polycentric? The urban spatial structure of employment in Beijing. Sustainability 7: 11,632–11,656.

17.

Kane

Hipp

Kim

(2018) Los Angeles employment concentration in the 21st century. Urban Studies 55(4): 844–869.

18.

Lang

(2003) Edgeless Cities: Exploring the Elusive Metropolis. Washington, DC: Brookings Institution Press.

19.

Lee

(2007) ‘Edge’ or ‘edgeless’ cities? Urban spatial structure in US metropolitan areas, 1980 to 2000. Journal of Regional Science 47: 479–515.

20.

Liu

(2018) Polycentric development and the role of urban polycentric planning in China’s mega cities: An examination of Beijing’s metropolitan area. Sustainability 10(5): 1–14.

21.

LJC

(2004) Economic reforms, urban spatial restructuring, and planning in China. Progress in Planning 61: 237–260.

22.

McMillen

(2001) Nonparametric employment subcenter identification. Journal of Urban Economics 50: 448–473.

23.

Qin

Han

(2013) Emerging polycentricity in Beijing: Evidence from housing price variations, 2001–05. Urban Studies 50: 2006–2023.

24.

Redfearn

(2009) Persistence in urban form: The long-run durability of employment centers in metropolitan areas. Regional Science and Urban Economics 39: 224–232.

25.

Shearmur

Alvergne

(2003) Regional planning policy and the location of employment in the Ile-de-France – Does policy matter? Urban Affairs Review 39: 3–31.

26.

Sorensen

(2001) Subcentres and satellite cities: Tokyo’s 20th century experience of planned polycentrism. International Planning Studies 6: 9–32.

27.

Stanback

(1991) The New Suburbanization: Challenge to the Central City. Boulder, CO: Westview Press.

28.

Sun

Han

Wang

, et al. (2012) Suburbanization and subcentering of population in Beijing Metropolitan Area: A nonparametric analysis. Chinese Geographical Science 22: 472–482.

29.

Wei

YHD

(2012) Restructuring for growth in urban China: Transitional institutions, urban development, and spatial transformation. Habitat International 36: 396–405.

30.

Wen

Tao

(2015) Polycentric urban structure and housing price in the transitional China: Evidence from Hangzhou. Habitat International 46: 138–146.

31.

(1998) Polycentric urban development and land-use change in a transitional economy: The case of Guangzhou. Environment and Planning A 30: 1077–1100.

32.

Phelps

(2011) (Post)suburban development and state entrepreneurialism in Beijing’s outer suburbs. Environment and Planning A 43: 410–430.

33.

Yue

Liu

Fan

(2010) Polycentric urban development: The case of Hangzhou. Environment and Planning A 42: 563–577.

34.

Zou

Mason

Zhong

(2015) Modeling the polycentric evolution of post-Olympic Beijing: An empirical analysis of land prices and development intensity. Urban Geography 36: 735–756.