Dynamics of land use efficiency with ecological intercorrelation in regional development

Dynamics of land use efficiency with ecological intercorrelation in regional development

Landscape and Urban Planning 177 (2018) 303–316 Contents lists available at ScienceDirect Landscape and Urban Planning journal homepage: www.elsevie...

770KB Sizes 0 Downloads 4 Views

Landscape and Urban Planning 177 (2018) 303–316

Contents lists available at ScienceDirect

Landscape and Urban Planning journal homepage: www.elsevier.com/locate/landurbplan

Dynamics of land use efficiency with ecological intercorrelation in regional development ⁎

Zhan Wanga,b,c,d, Jiancheng Chena, , Wentang Zhenga,e, Xiangzheng Denga,c,d,f,

T

⁎⁎

a

School of Economics & Management, Beijing Forestry University, No. 35 Tsinghua East Road, Haidian District, Beijing 100083, China School of Environment, Education and Development, The University of Manchester, Oxford Road, Manchester M13 9PL, UK c Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, No. 11A Datun Road, Beijing 100101, China d Center for Chinese Agricultural Policy, Chinese Academy of Sciences, No. 11A Datun Road, Beijing 100101, China e North China University of Technology, Beijing 100144, China f Waikato Management School, University of Waikato, Gate 1 Knighton Road Private Bag 3105, Hamilton 3240, New Zealand b

A R T I C L E I N F O

A B S T R A C T

Keywords: Beijing-Tianjin-Hebei Income Inequality Landscape Land use efficiency

Arguments about side effects of economic growth in urbanization call for deeper research on land use efficiency (LUE) from the perspective of urban planning for the coordination of social production and environmental conservation. Rural-urban migration increases rural household earning from part-time jobs at urban area. This social transformation increases the transportation demands and the risk of regional environmental degradation through ecological intercorrelation among urban-rural ecosystems. In this research, we aim to study how urbanrural ecological intercorrelation can dynamically determine the edge effects between backward-wave effects and spillover effects to affect dynamics of land use efficiency on the pathway of regional development. We analyze the marginal percentage changes of population growth and rural/urban income growth influence the dynamics of land use efficiency of Beijing-Tianjin-Hebei region (BTH). Empirical analysis results show that the urban income rises have weak spillover effects, while rural income growth primarily influences land use efficiency changes when urban-rural ecological intercorrelation is weak. We also test with or without the innovation impacts, and find both methods reporting the violation of normal economic development that in fact backward-wave effects exceed spillover effects in BTH. It implies that urban income growth should drive more spillover effects when urban-rural ecological intercorrelation is strong, but in fact it fails in a highly urbanized region. Thus, it is debatable that the fast population growth is the root of environmental degradation, in fact, ecological intercorrelation determines the edge effects of regional economic scale. That affects the structural effects of urbanrural landscape changes being allocated by population and income rises dynamically. Policy implication for regional development is to identify landscape rights in advance to keep dynamics of land use efficiency in a relatively stable structure for coherently improving environmental quality and the standard of living.

1. Introduction

regional development (Boudeville, 1966; Krugman, 1993; Perroux, 1950; Romer, 1993). The backward-wave effect reflects the unbalanced gap rises when a core region’s economic growth is too fast than neighborhoods; while, the spillover effect presents a normal pathway when the core region’s economic growth drives wages growths to neighborhoods via regional trade and knowledge spillover. Smith (1977) further proposed the gradient development theory to explain these effects can occur at the same stage of regional development, so that their edge effects may become more significant to identify the complex issues in a polycentric structure. However, there are arguments in theoretical and empirical studies about the function of spatial correlation on edge effects changes. Thus, we study the dynamics of land use

Friedmann (1986) stated that the history of a city development presents some comprehensive outcomes from agriculturalization to industrialization such as agricultural land increases then decreases for the increasing demand of living standard. World development follows this pathway in social transformation, and pushing towards regional agglomeration in distinctive development pathways. However, there are many theoretical arguments about the best way of sustainable development. The development poles theory states the unbalanced economic growths may induce backward-wave effect or spillover effect in a region because geographical characters may have uncertain impacts on



Corresponding author at: School of Economics & Management, Beijing Forestry University, No.35 Tsinghua East Road, Haidian District, Beijing, 100083, China. Corresponding author at: School of Economics & Management, Beijing Forestry University, No.35 Tsinghua East Road, Haidian District, Beijing, 100083, China. E-mail addresses: [email protected] (Z. Wang), [email protected] (J. Chen), [email protected] (W. Zheng), [email protected] (X. Deng).

⁎⁎

http://dx.doi.org/10.1016/j.landurbplan.2017.09.022 Received 2 September 2016; Received in revised form 31 August 2017; Accepted 23 September 2017 Available online 09 October 2017 0169-2046/ © 2017 Elsevier B.V. All rights reserved.

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

externality with non-marketized characteristics (Wei, 1999, 2015). Because unbalanced economic growths consequently enlarge the gaps of environmental inequalities which cannot be entirely solved by liberal marketization. The human rights for holding resources ideally should, but, never reach the ideal equality. Because there are potential losses of resource use efficiency which can earn much high profits from rising price of these scarce resources allocated by that inequalities (Peck & Tickell, 2002). Consequently, these inequalities ultimately can induce inefficient spatial hierarchy such like increasing transportation demands that highly likely long last forever, so a key question in urban expansion is how structural landscape conversion plays a role in a process of economic development, and how we can approach to a better standard of living when an optimal efficiency can be improved in urban transformation. Arguments about side effects of economic growth in urbanization call for deeper research on land use efficiency (LUE) from the perspective of urban planning for the coordination of social production and environmental conservation. Zhuangzi’s philosophy advocates “unity of man with nature” developed from Laozi which delineates a system with some cohesive and coherent relationships between human beings and natural environment. (Chen & Wu, 2009; Morrow, 2016). Fu, Wang, Su, and Forsius (2013) pointed out that landscape conversion reflects ecosystem changes with complex linkages to human activities in the terrestrial system, and having significant impacts on improvement of human wellbeing in the process of urbanization. China are emerging many “hollow villages” and calling for rural revitalization (Liu & Li, 2017). This social transformation increases the transportation demands and the risks of regional environmental degradation through ecological intercorrelation among urban-rural ecosystems (Ewing et al., 2010; Wang, Deng, Wang, & Chen, 2017). Wang, Deng, and Wong (2016) defined this process of urbanization with ecological linkages as a new definition of ‘eco-urbanization’, which involves many spatial correlations such as “ecological flows, stocks, risks, utilization, conservation, functional changes, and economic cost-benefits for sustainable development across different scales and hierarchies through networks, nexus, and interdependence of both natural and social evolutionary processes.” How to define the land use efficiency (LUE) is priority when we try to answer whether ecological intercorrelation is good or bad for either estimation models or planning practices. Current urban planning schemes on mainland China stress the land price and its potential social production value. The LUE was used to calculate the Gross Domestic Product (GDP) per square meter, but in some cases, ecological infrastructure (EI) including dry/paddy (cultivated) land, woodland, grassland, water/wetland, and unused land have the function to enhancing environmental quality (Li et al., 2016), moreover, these neighbor EIs can influence the economic value of built-up area (Li et al., 2009). The hedonic housing price discusses the economic value of neighbor landscape at the urban fringes, such as housing price increases when beside woodland (Shonkwiler & Reynolds, 1986). That indicates people prefer to live the house surround by a good natural landscape, so that the additional economic value of natural landscape depends upon individual subjective satisfaction level. Individual willingness-to-pay then can represent potential consumption demand under the constrain of individual expenditure budget, so personal income is the predeterminate value of preferred landscape for a certain level of environmental quality. At this point, income and population in classical economic theory are endogenous variables to critically determine regional consumption demand with the opportunity cost of landscape conversion. To individuals, the less payment is the better to maximize their preferred utilities, and that means the higher economic efficiency can be reached. However, economic efficiency differs from resource-use efficiency. Economic efficiency is used to be understood in two parts: the technical efficiency and the allocative efficiency. The resource-use efficiency is used to calculate the ratio of outputs and resource-use with respect to the inputs in economic value, so that can be understood as the technical

efficiency (LUE) in Beijing-Tianjin-Hebei region of the People’s Republic of China (PRC) to understand how the function of ecological intercorrelation affects edge effects, and aiming to clarify her nature for future planning in a feasible sustainable manner. Why population and personal income are the core factors? Many other factors such as education, housing, recreation, and other consumption demands can contribute to urban transformation, but they are determinated by population and personal income (Deng, Huang, Rozelle, & Uchida, 2008; Harris & Shonkwiler, 1997; Heckman & Mosso, 2014; Zheng & Kahn, 2008). At different stages of regional development, the alternatives of population growth and income growth are arguable about which is the main diving force can have more backwardwave effects or more spillover effects to neighborhoods. Urban consumption demand drives more low-skilled job opportunities increase. This induced rural labors prefer to work at neighbor urban area for earning higher wages before 2000 s on mainland PRC (Zhang & Song, 2003). With urban accessibility improved, increasing rural-urban migration via long-distance transportation move to cities for relatively higher income. Urban economic growth may bring about more backward-wave effects to those large cities, and enlarging the inequality of urban-rural income. We have seen urban environmental degradation, even though urban residents yearn for environment quality improvement when their income and demand increasingly for higher quality of life (McConnell, 1997). Consequently, there are more people moving to the cities where have better environment, so that further polarizes the inequality of environmental quality in a polycentric region. The structural changes in between urban resident disposable income and rural resident net income represent the structural changes of consumption demands in a real business cycle. These demands influence on market products sales, and leading to producers on the supply side to adjust labors for minimizing production cost. This leads rural-urban migration changes structural labor supply with responses to urban demand changes in remuneration (Zhang, Shen, & Zhao, 2014). Moreover, Hukou classifies rural or urban residents, which is the unique identification of population on mainland PRC. National statistics of rural and urban population is the ‘basing points’ to record the base number of labor. Although some official statistics at city-level have errors, the structure of rural-urban population present the evidence of rural-urban labor structural changes in regional development (Gibson & Li, 2017). However, new urban residents do not only be labor force but also are consumers. For instance, when housing price rises, we have not seen annual housing demand decreased, while in fact housing price is continually increasing in many cities of China because new urban residents are increasing faster. It may not mean classic economic theory fails. When we consider income and population as driving factors to regional economic scale over time, if the dynamic changes of factors can bare relatively balanced systematic changes, it means the changes of factors are not systematically asymmetric. Then, we may find a large proportion of new urban residents can afford a new house in a better urban environment when their income rises. While, if the factors changes cannot synchronize, we may see the opposite outcomes, such as urban water shortage, low quality of public service, and lack of per capita public resources. An economic model can provide a dynamic approach to study these structural changes for specifying percentage changes in each driving factor. Land use changes can reflect conflicts and structural changes in land use planning, but lack of explanatory mechanism that can spatially reflect landscape hierarchy in urban planning (Wu, 2004). Studies on urban transformation need to absorb some methodologies that can reflect structural changes driven by more human-centered factors to urban planning adjustment. For instance, old cities have remnant trails and sites that limit new settlements having the equal rights for the same share of resource utilization as old residents have, so that makes the redevelopment of old cities become very hard to approach the target of ‘sustainable urbanization’. Because there is a paradox that neoliberalism space creates housing market but also generates environmental 304

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Fig. 1. Nominal GDP (a) and resident population (b) in Beijing-Tianjin-Hebei, 1949–2015.

value of land use via ecological intercorrelation may have some uncertain implicated effects on neighborhoods. In the rest of this paper, we firstly introduce study area by using statistics of key indicators to describe social development, and pointing out our research questions in Section 2; in Section 3, we discuss the methodology, and proposing empirical models to analyze land use efficiency changes with regards to the percentage changes in population and income rises; then, we present and test the results in Section 4; after a comprehensive discussion given in Section 5, we write a short conclusion in the last Section 6.

efficiency but with the innovation changes to be shaped. Differently, the allocative efficiency measures the marginal changes of outputs caused by the marginal changes of a specific input factor in percentage changes of units that can either be volumes or values. Such like the Jevons’ paradox describes the higher resource use efficiency can bring about a higher level of resource consumption (Alcott, 2005). It indicates that increasing LUE may endogenously bring about a higher probability of land use changes to gain more satisfactions or profits from housing development. Therefore, in this study, we hold the technical efficiency as a constant to test the allocative efficiency of LUE how can be influenced via ecological intercorrelation and dynamically having impacts on inequalities of the increasing income or population at county level in BTH over time. That means we will estimate LUE changes by controlling its resource-use efficiency as a part of endogenous changes to analyze the allocative efficiency changes of LUE in light of the percentage changes in income and population. Future cities may try to find some vertically spatial planning schemes to improve the living quality for citizen wellbeing. Land is an eonian kind of scarce resources. State powers tend to have fixed administrative boundaries, so the ground land tends to be fixed in a region. Most of us currently cannot move to water area or other planets. The marginal cost of ground land less likely comes out a downwards turning point when population and income are increasing. The plot ratio of housing somehow presents the relationship between ground land and housing space, but the rent of housing differs from the rent of land as the land price. Even if the housing market can be considered as a perfectly competitive market in which individuals follow the rule of diminishing marginal utility; land market cannot be considered as a free market because property rights in the space cannot be completely wellclarified according to ground land area. This means classified property rights in private ownerships have uncertain spatial boundaries to be seen as public goods. Therefore, land resource is in fact the resources shared by both private and other ‘public’ owners. In this case, we think that different types of property rights regard to land should be addressed by identified and redefined different types of LUE. Because spatial diversity induces landscape diversity, to increase every unit of a type of land use in economic value can have uncertain impacts on per unit of GDP in social production on other types of land use. Therefore, we analyze the structural changes of land use efficiency to show different types of land have various contributions to regional economic growth because regional demand preference allocates the economic

2. Study area and data description In the whole region of Beijing-Tianjin-Hebei (BTH), a fragmented structure of economic growths increasingly brings about tensions in environmental degradation among sub-centers. Urbanization in Beijing was faster than Tianjin and Hebei. The agglomeration in BTH region has experienced a consistent population growth since 1949 and a rapidly economic growth since 1980s. According to the China compendium of Statistics edited by the national economy comprehensive statistics division of the PRC National Bureau of Statistics (NBS), during 1949–2015, the Beijing resident population increased with an annual average growth rate at 2.4–2.5% from 4201 thousand to 21.7 million, the Tianjin resident population increased with an annual average growth rate around 2% from 4025 thousand to 15.5 million, and the Hebei resident population increased with an annual average growth rate at 1.3–1.34% from 30.9 to 74.3 million (Fig. 1b). The Beijing nominal GDP in 108 CNY increased with an annual average growth rate at 13.3–16.2%, until 2015 which was near to 8300 times in 1949, Tianjin had an average growth rate at 12.2–14.4%, and Hebei had an average growth rate at 9.7–11.6% (Fig. 1a). Urban-rural income inequality rises in BTH. After the economy reform in 1978, the annual average growth rate of urban resident disposable income in Beijing (gBurinc) was 13.1–14.7% higher than that in Tianjian (gTurinc) and Hebei (gHurinc) (Table 1). Until 2015, Beijing urban income on annual average was 52859 CNY (in corresponding year) over 144.7 times than that in 1978. While, the Beijing annual average growth rate of rural resident net income (gBruinc) was 11.9–13.2% lower than Tianjian (gTruinc) at 12.62–14.16% and Hebei (gHruinc) at 12.04–13.56% (Table 1). By comparing the increasing value of nominal rural income to nominal urban income, the urban 305

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

engineering projects for ‘ecological construction’ to improve the environmental quality. The first green barrier is ‘five rivers and ten roads’ which increased 64.3 km2 green land, and until 2004, the total amount of green land in Beijing reached 102.3 km2. The second green barrier is one km outside of the sixth ring road in ‘suburban mountains’, covered 1650 km2. The third green barrier links to ‘the Taihang Mountain greening project’ and ‘the abandoned mine ecological restoration in BTH region project’, and the greening cover ratio of the third green barrier had exceeded 70% before 2008. Moreover, in Beijing downtown area including Dongcheng, Xicheng, Chaoyang, Haidian, Fengtai, and Shijingshan, the green space landscape area had been improved from 29% to 47% during 2000–2015. Until the end of 2015, there were orchard 1329.02 km2, woodland 7184.93 km2, grassland 853.39 km2 in the ecological conservation area located at far suburban area in Beijing adjacent to the Heibei and Tianjin, including the districts of Fangshan, Shunyi, Tongzhou, Changping, Daxing, Mentougou, Huairou, Pinggu, Miyun, and Yanqing. Although green space has increased in Beijing during the past several decades, it is quite hard for increasing the greening rate in old downtown area in the future without governmentsupported ‘big actions’ of urban renewal. With urban population and income increasing continually in urban area, demands of green space are uncertain to rural neighborhoods. Furthermore, sub-regional planning schemes can further lead to uncertain outcomes due to unbalanced economic growths in BTH region. Tianjin planned during 2006–2020 to implement an intensive land use planning scheme, which targeted about orchard land should increase 13.5 ha and forest land should increase 223.5 ha during 2010–2020, and that GDP per square kilometer in built-up downtown area must be over 0.4 billion CNY (in corresponding year), and GDP per square kilometer in industrial and mining land at township level should not be lower than 1.1 billion CNY before 2020. Moreover, the average annual growth rate of built-up area should be lower than 0.05% if one per cent of GDP increases in Tianjin; and the average annual growth rate of built-up area should be lower than 0.55% if one per cent of population increases. The ecological conservation area including cultivated land, orchard, forest, grassland, wetland, and unused land in total should be over a half of the total amount of area in Tianjin until 2020. In this plan, a higher plot ratio is prospection. However, before the strategic plan of Xiongan New Special Zone released on April 1st in 2017, Hebei Province as a supporter is less supervised by central planning with comparing Beijing and Tianjin. Because remote sensing data can provide practical evidence of land use changes, we can calculate the annual average changes in each type of EI land use in square meter per capita of resident population in three regions of BTH (Table 4).

Table 1 Annual average growth rate of nominal urban and rural income (CNY in corresponding year) in Beijing-Tianjin-Hebei, 1978–2015. Region

Beijing

Tianjing

Hebei

Unit: % (code)

rural (Bruinc)

urban (Burinc)

rural (Truinc)

urban (Turinc)

rural (Hruinc)

urban (Hurinc)

exponential growth rate moving average growth rate

11.89

13.09

12.62

11.78

12.04

11.97

13.17

14.66

14.16

13.12

13.56

13.36

income gap (urinc) is larger than the rural income gap (ruinc) (Fig. 2). Thus, intuitively, urban income rises may be the driving force of economic inequality to further influence the dynamics of land use efficiency in BTH. These income inequalities induced land use per capita changes may bring about uncertain environmental quality changes via ecological intercorrelation in BTH region. The built-up land in BTH increases continually during 1988–2008 (Table 2), which excludes the constructed land for urban green space in land use data source which is reclassified and produced by (Liu, 1996; Liu, Liu, Zhuang, Zhang, & Deng, 2003). The surface land use changes for EI including: dry/paddy (cultivated) land, woodland (all kinds of forestry land), grassland, water/wetland, and unused land intermittently decreased in Beijing and Hebei. However, the green land belongs to the construction land in the planning index (Table 3). Land developers can increase plot ratio to develop a residence community but may let neighborhoods suffer from ecological land losses under a more complexity spatial design in a small region. Thus, it may fail to guarantee the articles and regulations of green land use would be implemented practically. Moreover, it is unclear that increasing ecological infrastructure in Beijing has impacts on neighborhoods in Tianjin and Hebei. From the 1980s to 1990s, the growth rate of urban expansion was over 40% to the west 16 km of Beijing, the average expansion growth rate was over 20% to the east 10 km, to the north 12 km, and to the south 12 km, especially to the northeast 18 km was over 50% (Zhou, Zhang, Wu, & Niu, 2006). Environmental degradation in neighborhoods such like sand storms in Beijing occurred frequently in springs at the beginning of 21st century. Air pollution smog engulfed BTH, even the indictor of annual blue-sky days has no longer been published since 2012. Beijing municipal government has taken actions on environmental conservation in recent decade. There were three key

Fig. 2. Nominal Rural vs Urban income (CNY in corresponding year) in Beijing-Tianjin-Hebei, 1978–2015.

306

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Table 2 Land use change (km2) in Beijing-Tianjin-Hebei, 1988–2008. Region

Year

dry/paddy

woodland

grassland

water/wetland

built-up

Unused

Beijing

1988–1995 1995–2000 2000–2005 2005–2008 1988–1995 1995–2000 2000–2005 2005–2008 1988–1995 1995–2000 2000–2005 2005–2008

−882.70 −57.76 −398.67 −113.28 −163.87 −48.88 −453.89 −143.93 −1641.05 −121.22 −452.74 −439.59

118.98 −0.90 −10.47 −13.68 −0.61 −0.50 −0.78 −0.18 25.66 −48.20 30.60 6.03

−68.80 −0.29 5.79 −4.69 −23.70 5.93 −6.43 −0.08 −310.72 −12.43 −119.67 −73.95

113.25 0.92 −25.30 −10.70 119.22 34.93 24.70 88.37 −114.82 16.17 −74.73 −2.76

719.57 57.73 428.66 142.35 57.07 28.72 440.12 61.93 2170.48 169.46 659.36 537.09

−0.29 0.29 0.00 0.00 11.89 −20.20 −3.71 −6.11 −129.54 −3.78 −42.83 −26.82

Tianjin

Hebei

population increased from 16330 to 21705 thousand. Until 2015, the ratio of urban resident population had reached 86.5%. Beijing Gardening and Greening Bureau (Capital Greening Office) made great efforts to increase EIs, and tried to decrease the proportion of built-up area (not covered by EIs) from 20.26% to 18.6%. During 2007–2015, the total amount of green land almost double increased from 45590.74 to 86808.64 ha; however, the green land per capita of total resident population still decreased from 48 to 39.84 m2. Until 2015, the green land per capita of urban resident population in Beijing was just 16 m2. The Capital Greening Office bulletin annual statistics that city center of Beijing including Xicheng District and Dongcheng District had woodland 984.14 ha, woody land 1539.53 ha, wetland 289 ha, and greening rate reached 19.13% in Xicheng District and 14.62% in Dongcheng District, which are much lower than other districts in Beijing (Fig. 3). Thus, although EI has been improved in Beijing, the increases of urban population have great burdens on EI per capita, so that somehow lowers residential satisfaction of environmental quality and subjective living standard in Beijing. Therefore, although these remote sensing data based on long-distance satellite pictures reflect physical ground land use changes, it is usually very hard to directly provide policy options for reflecting downscaling community complex landscape from the perspective of urban planning. It is necessary to research the past LUE changes and to provide evidence for the future coordinate strategies of land use structure and supporting regional economic development with population and income rises in a sustainable manner. Based on a reliable dataset of historical changes in land use, population, and income at 165 district/county level in BTH, the empirical analysis employs the data description of calculated LUE in logarithmetics at 165 district/county level (Table 5). The statistics of economic indicators at district/county level in BTH are published by China's economic and social development statistical database of CNKI.net. Again, land use data source is reclassified and produced by (Liu, 1996; Liu et al., 2003). we aim to deeply understand what is the main driving force to dynamic land use

Table 3 “GB137-90 Article 4.2.1 The preparation and revision of overall urban planning, residential, industrial, road and square, and green space four kinds of main land use per capita indicators should comply with the provisions of the Table 4.2.1, The Planning Construction Land per capita Index” in 1991. Type

land use standard (m2/per capita)

residential industrial roads and squares green land (public green land)

18.0–28.0 10.0–25.0 7.0–15.0 ≥9.0 ≥7.0

However, these data show that the actual land use changes in EI and built-up area disobeyed the sub-regional planning regulations in the past several decades. Because ecological intercorrelation may take uncertain environmental effects to neighborhoods when sub-centers are not cohesively and coherently developing in a region (Wang, Deng, Wang et al., 2017). To identify increasing demands for ecological infrastructure (EI) in Beijing is the hardcore for understanding the landscape conversion with ecological intercorrelation in the process of eco-urbanization in BTH. Beijing as the capital city of PRC is one of ‘eyes of the storm’ in the world where the standards of living and environmental quality absorb world attention. With income rising in Beijing in the past several decades, the EI per capita was not consistently increasing, and its growth rate was slower than the growth rate of urban population. This intensifies environmental inequality due to the increase of environmental risks and potential economic losses due to the increase of living cost, so that the subjective standard of a livable city deviates from citizen’s expectation. Beijing in 1990s, the woodland per capita of registered population was 0.058 ha; until 2015, the woodland per capita of resident population declined to 0.0501 ha. After Beijing won the Olympic bid in 2002, the population increased sharply. From 2007–2014, resident

Table 4 Land use changes in annual average (m2) per capita of resident population in Beijing-Tianjin-Hebei, 1988–2008. Region

Year

dry/paddy

woodland

grassland

water/wetland

built-up

Unused

Beijing

1988–1995 1995–2000 2000–2005 2005–2008 1988–1995 1995–2000 2000–2005 2005–2008 1988–1995 1995–2000 2000–2005 2005–2008

−120.10 −7.00 −70.88 −7.20 −38.10 −34.40 −523.82 −17.48 −38.37 −8.72 −38.21 −39.96

16.19 −0.11 −1.86 −0.87 −0.14 −0.35 −0.90 −0.02 0.60 −3.47 2.58 0.55

−9.36 −0.03 1.03 −0.30 −5.51 4.17 −7.42 −0.01 −7.26 −0.89 −10.10 −6.72

15.41 0.11 −4.50 −0.68 27.72 24.58 28.50 10.73 −2.68 1.16 −6.31 −0.25

97.90 6.99 76.21 9.05 13.27 20.21 507.93 7.52 50.75 12.19 55.64 48.83

−0.04 0.03 0.00 0.00 2.76 −14.21 −4.29 −0.74 −3.03 −0.27 −3.61 −2.44

Tianjin

Hebei

307

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Fig. 3. Beijing Green land per capita of resident population in 2015.

resource-use efficiency of LUE changes caused by innovation impacts on their own endogenous changes to analyze their allocative efficiency changes of LUE in light of the percentage changes in income and population increases. In this section, we will state why we choose this methodology. Evaluation of the efficiency of resource utilization requires a clear definition firstly (Auzins et al., 2013). LUE cannot be estimated by the elasticity of land price with respect to impact factor changes unless the land price is correctly published. For instance, Chen, Chen, Xu, and Tian (2016) analyzed intensive land use efficiency would decrease during 2006–2020 by using the Data Envelopment Analysis (DEA) method. They found more ecological land use have been claimed for improving regional environmental quality because lots of regional land resource planning were published by local government for remarking achievements in one's official career. However, their conclusion is questionable because of the limitation of DEA method which assumes highly homogenous characteristics of the input factors, but ignoring heterogeneities in different economic bases in sub-regions, such as Henan Province and Hunan Province are all in the central of China, but the former has 107 million population, and the latter has 67.8 million until the end of 2015; so, they have very distinct regional economic bases. Xie and Wang (2015) studied urban industrial land use efficiency in BTH including ten cities: Beijing, Tianjin, Shijiazhuang, Qinhuangdao, Tangshan, Langfang, Baoding, Cangzhou, Zhangjiakou and Chengde. While, it indeed can be separated two parts by technical efficiency and scaled economic efficiency in urban area; which means the allocation efficiency of every decision-making unit (DMU) can be constituted by the pure productivity of technical factors and the economic efficiency of scale production. Thereby, it is debatable that the DEA method is suit for transboundary study on resource allocation efficiency mainly because there are backward-wave effects or spillover effects in regional economic development which have strong impacts on the economic efficiency of scale production. Tu, Yu, and Ruan (2014) employed indicators that were published by the PRC Ministry of Land and Resources, and using empirical econometric model to estimate the marginal effects on industrial LUE changes. They found industrial LUE is highly influenced by the year of land lease and land size in Hangzhou. It indicates that to estimate LUE needs to focus on regional characters. Meng et al. (2008) developed the evaluation method of industrial LUE by involving five aspects of indicators including: intensive land-use, industrial scale, economic benefit, social influence, and ecological impact. They further provided evidence that geographical characters

Table 5 Logarithmetics data description of Land use efficiency (lue) in dry/paddy (cultivated) land (agr), woodland (for), grassland (gras), water/wetland (wate), built-up land (bdup), and unused land (us) of Beijing-Tianjin-Hebei, 1988–2008. Variable Obs = 165

Mean

Std. Dev.

Min

Max

lueagr_8895 luefor_8895 luegras_8895 luewate_8895 luebdup_8895 lueus_8895 lueagr_9500 luefor_9500 luegras_9500 luewate_9500 luebdup_9500 lueus_9500 lueagr_0005 luefor_0005 luegras_0005 luewate_0005 luebdup_0005 lueus_0005 lueagr_0508 luefor_0508 luegras_0508 luewate_0508 luebdup_0508 lueus_0508

4.6594 1.5483 1.4548 1.4052 4.4510 0.3343 1.6564 0.2723 0.2144 0.3614 1.4323 0.0741 2.3121 0.0060 0.0118 0.4530 2.3185 0.1506 1.7279 0.1818 0.0829 0.0904 1.7869 0.0614

1.5212 1.9784 2.2430 1.9077 1.5510 1.6348 1.8008 1.2115 1.1659 1.6765 1.5786 0.8105 1.8832 1.5648 1.6875 1.7342 1.6043 0.7141 1.6978 0.8226 1.1148 1.3569 1.5522 0.5595

0.0000 −2.5446 −5.0751 −2.6668 0.0000 −6.9925 −3.0009 −4.8397 −5.6044 −6.1824 −2.2653 −7.5133 −7.9241 −7.1902 −8.6548 −7.6209 −1.8005 −1.4674 −7.9835 −1.9465 −7.2970 −6.6498 −4.9839 −1.5649

7.9404 7.2046 7.7928 6.4766 8.6349 7.2035 6.7968 5.7531 4.6421 6.6735 6.1358 4.3293 10.1664 4.4752 6.8100 9.5032 10.5819 4.3098 5.7378 5.5728 6.5746 4.4600 6.1797 5.7913

efficiency (LUE) changes? how do backward-wave effects or spillover effects affect regional economy development? what spatial planning can contribute to future cities in this region? and why or why not it works at this developing stage? 3. Methodology and empirical model As the Jevons’ paradox claimed, to increase land use efficiency (LUE) may endogenously bring about a higher probability of land use changes to gain irrational satisfactions or profits from housing development. Thereby, in this research, we hold the technical efficiency as a constant to test the allocative efficiency of LUE how that can be influenced via ecological intercorrelation and dynamically having impacts on inequalities of increasing income or population at district/county level in BTH over time. That means we will try to control the part of 308

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

capital as natural resource, and monetary dividend as financial productivity can also be considered as a part of social wealth to increasingly accumulate capital endogenously. Thus, how effectively allocate these resources to different social classes becomes a critical issue in social production to approach to an ideal equality-society. However, regional characters are naturally unequal. Life is equal, but no doubt s/ he holds unequal natures and resources. Thereby, individual income presents the level of a part of earning from working as a labor, a part of earning from natural resources use, and a part of earning from individual financial assets. Even though Marxism and liberalism argue about the increase of income can present regional economy scale, the average personal income is used to present the demand-side growth power of regional productive growth in traditional development economics (Harris & Shonkwiler, 1997). As we mentioned in the beginning of introduction, those development economists proposed the industrial interdependence theory to test regional growth poles having backwardwave effects or spillover effects on neighborhoods via spatial correlations over time, and that implicit spatial correlations can be considered as the ecological intercorrelation. Therefore, we propose our models as follows. The demand of i type of land use change (LUC) in a district/county j of BTH at time t can be presented by a function of local population (pop) times urban income per capita (urinc) and rural income per capita (ruinc) respectively with their elasticity parameters α, β and γ, and other unobserved factors (technical efficiency) A in the following Eq. (1).

convey ecological impact factors with structural effects on LUE changes over time. Zhou, Zhang, Wu, and Niu (2013) found quite low LUE of peri-urban land near to many large cities during 2004–2008 by using a Panel Data model. However, because their conclusion was based on mixed land use structure to calculate the LUE, it may distort the results to reflect real situation in many cities of China. Because urbanization is not only the issue of increasing GDP on per unit of land use but also to increase the standard of living in a region, the ‘urbanized land’ of LUE rises may not label a livable city with a better standard of living space for wellbeing. Demand of living space for well-being is sorely needed to be identified at regional scale. Peters, Picardy, Darrouzet-Nardi, and Griffin (2014) studied LUE for livestock production in US which indicates more living space can increase livestock production because animals can loiter in a larger space, and within some higher quality of living amenities can let them become stronger and healthier. However, there are very few researches about the demand of living space for human wellbeing because regional characters are too distinct to be compared with others. The total amount of rural residential land is more than the total urban land on mainland of PRC. Wang, Wang, Su, and Tao (2012) analyzed LUE of rural residential land, and found that “average land area for rural residential properties is high (288 m2)” which is over the regulated 233 m2 (0.35 mu) per household, and over 10% of rural households own more than one residential property which in fact break the regulation by the PRC Ministry of Land and Resources. Furthermore, current China is facing severer situation that many rural residents are leaving their rural houses and moving into cities. This makes urban land price increased sharply. Du, Thill, and Peiser (2016) studied LUE in Beijing and concluded that current increasing and price is better for improving intensive land use. However, when more rural residents move into cities, urban vs rural inequality of living space and living standards would be significantly influenced by their income gaps, and having uncertain impacts on the land use changes at urban fringes where have relatively lower land price and highly likely would be covered from cultivated land, woodland, grassland, or other ecological infrastructure land use. Spatial correlation is useful for studying structural changes of land use efficiency. Ecological network is constituted by ecological intercorrelation, and conveying environmental effects to neighborhoods. Such as environmental effects can have strong spillover effects on neighborhoods which are composited by observed and unobserved parts (Beames et al., 2015; Bryan et al., 2015; Fetzel, Niedertscheider, Haberl, Krausmann, & Erb, 2016; Kytzia, Walz, & Wegmann, 2011; Wang, Deng, Wang et al., 2017). Landscape as the observed land use structure would be influenced by environmental effects of unobserved to observed changes. Theoretical spatial correlation usually is explained by physical distance between two sites (Zeng, Zhang, & Xu, 2016), but advanced transportation may shorter the travel time to increase the spatial correlation with less environmental effects. Suppose that spatial correlation can be increased also by population and income rises, we can test the marginal effects of impact factor changes by comparison of LUE with spatial correlation and without spatial correlation. If we firstly assume every unit of each type of land use change is independently occurred, every unit of LUE can be allocated by the increases of population and income per capita, and constantly influenced by other unobserved factors such like the technology for regional development. In this case, the allocative efficiency can be caught by the increases of population and income per capita, and the technical efficiency can be caught by a constant innovation factor to present social progress being smoothly allocated over time. In classical economic theory, labor and capital can be substituted by each other under the assumption of liberal market. Karl Marx criticized this capitalism and proposed social production is driven by the factors of labor (population), capital, technology, and land. He assumes the capital is accumulated by social wealth which is the sum of individual’s wealth from social production. It indicates other kinds of capital, such like natural

LUCijt = Aijt ∗ popjtα ∗ urincjtβ ∗ ruincjtγ

(1)

Suppose each type of land use efficiency (LUE) in planning definition is determinated by the total increases of GDP in a district/county j of BTH, and B is unobserved factors of the i type of land use change (LUC) by every unit of regional production increased in the district/ county j at time t in Eq. (2).

LUEijt = ln

LUCijt ΔGDPjt

,

and Bijt = ln

Aijt ΔGDPjt

(2)

Empirical model (I) is reached by modifying the Eq. (1) with adding a quadratic term of population (pop) in a district/county j of BTH, where a non-linear relationship between population and LUE is assumed (André & Platteau, 1998; Kögel & Prskawetz, 2001), and εijt is the disturbance term in Eq. (3).

LUEijt = Bijt + α ln popjt + α ln popjt2 -1 + β ln urincjt + γ ln ruincjt + εijt (3) We aim to specify the relationships between LUE changes relatives to percentage changes in population and income with and without ecological intercorrelation. Zitti et al. (2015) stated that study on systematic effects of LUE can significantly clarify urban-rural gradients because multi-scalar of land use reflects various socio-economic behaviors with mixed ecological functions throughout linkages. Because structural effects as theoretical correlation can be tested in the estimation of variance-covariance matrix in Eq. (4) (Zellner, 1962). Suppose that structural effects present the theoretical ecological intercorrelation. We will test whether the ecological intercorrelation exists among residuals of system equations in statistical significance, and an alternative hypothesis of that every unit of each type of land use changes is not independently occurred, but to some extant influenced by other types of land use changes. Under this circumstance, the structural effects should exist over time.

V (b*) = (XT Σ −1X)−1

309

11 T ⎡ σ X1 X1 ⎢ σ 21XT X 2 1 ⎢ = ⎢ σ 31XT3 X1 ⎢ ⋮ ⎢ 61 XT X ⎢ σ 12 1 ⎣

σ 12 XT1 X2 ⋯ ⋯ σ 16 X1T X 6 ⎤ σ 22 XT2 X2 ⋯ ⋯ σ 26 XT2 X 6 ⎥ ⎥ σ 32 XT3 X2 ⋯ ⋯ σ 36 XT3 X 6 ⎥ ⎥ ⋮ ⋱ ⋱ ⋮ ⎥ T T 62 66 σ X12X2 ⋯ ⋯ σ X 6 X 6 ⎥ ⎦

(4)

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

where X = [ln pop ; βlnurinc ; γlnruinc], and θi = [ Aij αij βij ], b* = [ θ1 * θ2 * ⋯ θ6 *]T = (XT Σ −1X)−1XT Σ −1 (LUE ) , and Σ−1 = Σσ ⊗ I. If ecological intercorrelation exists, it should be as strong as better to have close spatial correlations among six main types of ecological infrastructure on cultivated land, woodland, grassland, water/wetland, and unused land. Because LUE has the geological constraints from the supply side (Chen & Han, 2014), it is unsustainable that to increase ‘floor-land ratio’ (Ding, 2001) of urbanization within the fixed boundaries, so that opening land market may not work in China context. Theoretically, a type of land can be converted to all other types of land use, but there are uncertain what land use changes can improve environmental quality, and uncertainly what impacts have on neighborhoods via ecological intercorrelation. For instance, urban forest has evapotranspiration-cooling effects and carbon dioxide sequestration to generate oxygen (Jim & Chen, 2009), and ameliorating gaseous and particulate pollutants to provide recreation amenities (Yang, Dai, & Wang, 2015), so that it has positive effects to the surrounding environment; but also have unknown negative effects due to biodiversity and local culture, for instance, Dang and Luo (2013) concluded that rural residential land use in China was inefficient with inequality issue due to the squire culture. Therefore, residents in urban or rural may have different preferences on land use changes in different regions. Because it is debatable that intensive land use is appropriate to each region under the same regulations. This empirical model (I) straightforward reflects the dynamic changes from the demand-side to analyze the supply constraints of the structural changes in LUE for supporting police options in urban planning. For instance, according to the Construction Urban Land Classification and Planning Standards edited by the former PRC Ministry of Urban and Rural Construction of Environmental Protection and published by the current PRC Ministry of Construction in 1991, the maximum residential land use standard was 28 m2/per capita (recall Table 2). However, the planning standard for intensive land use was eased during the last decade. Until 2012, the subsidy standard of residential construction land in Beijing was 30 m2/ per capita of registered population who can be employees of governments, state-own institutes or companies with Beijing Hukou. Thus, we may ask how future planning regulations can be appropriate to meet the increasing demand in urbanization? and how that can be offered from the supply side constraints? Rather than to boom real estate sector per se by hyping new urban residents up (Lu & Wan, 2014). Thereby, an economic model is efficient to analyze the dynamic changes from the demand-side to provide the supply constraints of the structural changes (Arsel & Dasgupta, 2013; Wang, Deng, Bai, Chen, & Zheng, 2016). For example, Wang et al. (2015) employed an economic model to test the increasing water use may harm rural income of China because urban water use increased sharply in the last two decades. It implies that increasing resource use efficiency in urban area induces the inequality of resource allocation but can make rural income rise in the future. It implies that efficient policy options may consider from the demand side to minimize the side effects of the structural changes. Moreover, once we have the same qualification of land use dataset, such like we use in this research, how much percentage changes in human-centered factors have impacts on the structural changes in LUE is the first key question that we need to clarify for seeking appropriate schemes of urban-rural coordinative development planning in eco-urbanization. Thus, we designed empirical model (II) to better understand their relationships without ecological intercorrelations. Empirical model (II) is a modified nonlinear seemingly unrelated regression (NSUR) model by employing the quasi maximum likelihood estimation (MLE) to test the assumption that the nonlinear relationships between dependent variables and independent variables. We assume that the independent variables and estimators are an implicit function of dependent variable and disturbance term υijt in Eq. (5). LUEijt = hij(b, xjt) + υijt

This assumption can be tested by whether the dependent variables are Gaussian conditioned on independent variables or with the condition that generalized method of moments (GMM) estimators on the expectation value of the correlation of disturbance term with respective to any measurable transformation of the independent variables (Moon & Perron, 2006). It indicates that there are no structural effects in the transformation process to distort the estimated distribution of probability changes in consumption demand factors over time, which means the H0: no autocorrelation can be tested and cannot be rejected. Magalhães and Seifert (2015) concluded that NSUR and the conventional method can estimate unbiased biomass by using the parallel independent variables in all panels in component models of the trees from different sample sites. This means that all panels have the homogenous weights to “enforce additivity” among neighborhoods which are closely adjacent enough. Fu et al. (2016) estimated biomass equations with two or more predictors and found nonlinear error-in-variable models are more efficient. Thereby, we write the density function (6) of observations to describe the probability distribution of the implicit function (5). n



f (LUEijt θ)

i=1

= L (θ LUE ), and

T

ln L = − 2 [M ln(2π) + ln Σ + tr(Σ −1 W)], (6)

where the covariance matrix of the ordinary least squares (OLS) residuals 1 W= − T V T V , E[Wik] = σik, n = 6, M as the number of disturbance scalars, and E [υt υtT ] = Σ is observed disturbance matrix. For better controlling assumed ecological intercorrelations, we adopt the feasible generalized least squares (FGLS) estimators instead of the generalized method of moments (GMM) to estimate the coefficients −1 −1 βˆ = [X T Uˆ X ]−1X T Uˆ (LUE ) when the nonlinear error-in-variable models is held by the improved estimates of the variance-covariance 1 matrix σˆik = T (LUEijt − Xit bit )T (LUEkjt − Xkt bkt ) , and here Uˆ = W⊗ Inn ; sequentially, run s = 16000 itineration of MLE to reach the estimated parameters in the concentrated log-likelihood function T ln L (b, ωˆ (b)) = − 2 [M (1 + ln(2π )) + ln Uˆ ], where ωˆ (b) is the vector of W. We suppose C is “some arbitrarily chosen positive number” (Oberhofer & Kmenta, 1974), if the Eq. (7) is satisfied. M

M

C ≥ tr(Uˆ ) = [ ∏ σˆii]T [tr(Rˆ )]T ≥



i=1

i=1

σˆiiT λ1T ≥ λ1T λ 0MT

where Rˆ is a M*M matrix holding a typical element ρˆik = σˆik

(7)

σˆii σˆkk ,

M

then C λ1T ≥ [ ∏ σˆii]T is held for reaching the estimators Σˆ of MLE. i=1

Sequentially, the likelihood ratio test is presented by LR = T (ln diag W − ln Σˆ ) ; and the Wald statistics is presented by ˆ ]−1RT ]−1 (Rbˆ-q) with setting H0 : Ris the matrix of λW = (Rbˆ-q)T [R [X T UX parameters in b are equivalent and q is an identical matrix (in this case R is 5*5 matrix, and q is 5*1 scaler). 4. Analytical results Population growth in BTH region was consistent during 1949–2015, however, which is not a consistent impact factor to the land use efficiency (LUE) changes in recent two decades. The empirical model (I) reports the quadratic term of population growth (lnpop) in the previous time period (L1.lnpop2) as the control variable does not show statistically significant (Table 6). It indicates that the population growth is not a primary impact factor of land use efficiency changes during 1995–2005. In other words, population growth was not always the main cause of LUE changes before 2005. While, after Beijing won the Olympic bid, population growth becomes the determinate impact factor of the increase of the ecological infrastructure land per ha GDP changes by recalling Eq. (2). With population increases, LUE of cultivated land and built-up land decreased sharply before 2005, and increased after 2005 in BTH (Fig. 4). It indicates that fast population growth indirectly

(5)

310

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Table 6 Empirical Model (I) results of land use efficiency in Beijing-Tianjin-Hebei, 1988–2008. Empirical Model (I)

LUE_8895

LUE_9500

LUE_0005

LUE_0508

Land Type

variables

code

Coef.

P>z

Std.Err.

Coef.

P>z

Std.Err.

Coef.

P>z

Std.Err.

Coef.

P>z

Std.Err.

cultivated land (dry/paddy)

_cons lnpop lnruinc lnuninc L1.lnpop^2 _cons lnpop lnruinc lnuninc L1.lnpop^2 _cons lnpop lnruinc lnuninc L1.lnpop^2 _cons lnpop lnruinc lnuninc L1.lnpop^2 _cons lnpop lnruinc lnuninc L1.lnpop^2 _cons lnpop lnruinc lnuninc

b0_ag b1_ag b2_ag b3_ag b4_ag b0_fo b1_fo b2_fo b3_fo b4_fo b0_gr b1_gr b2_gr b3_gr b4_gr b0_wa b1_wa b2_wa b3_wa b4_wa b0_bd b1_bd b2_bd b3_bd b4_bd b0_un b1_un b2_un b3_un

−14.064 2.219 −0.841 1.816 −0.113 3.059 −0.556 −1.045 1.727 −0.005 38.923 −6.697 −2.363 3.589 0.225 11.231 −3.767 −0.708 2.936 0.121 −26.662 3.772 −0.227 1.480 −0.171 6.728 −0.305 −0.468 0.123

0.190 0.119 0.060 0.025 0.038 0.829 0.767 0.079 0.108 0.947 0.007 0.000 0.000 0.001 0.002 0.398 0.032 0.208 0.004 0.072 0.015 0.009 0.618 0.072 0.002 0.202 0.248 0.346 0.891

10.7345 1.4241 0.4469 0.8075 0.0546 14.1792 1.8784 0.5943 1.0737 0.0720 14.4593 1.9040 0.6229 1.1252 0.0729 13.2748 1.7545 0.5624 1.0161 0.0673 10.9363 1.4506 0.4557 0.8233 0.0556 5.2699 0.2640 0.4961 0.8936

3.106 −0.261 0.772 −0.531 0.004 14.499 −1.132 −0.906 0.478 0.020 −15.207 2.666 −0.861 0.640 −0.107 13.107 −0.165 −1.004 −0.353 0.003 −2.575 0.068 1.014 −0.284 −0.013 0.682 0.087 −0.276 0.051

0.803 0.871 0.080 0.451 0.946 0.086 0.298 0.002 0.316 0.620 0.053 0.008 0.002 0.148 0.004 0.256 0.912 0.014 0.587 0.961 0.815 0.962 0.009 0.646 0.797 0.793 0.548 0.200 0.882

12.4759 1.6063 0.4415 0.7040 0.0591 8.4504 1.0880 0.2991 0.4769 0.0400 7.8611 1.0129 0.2771 0.4419 0.0372 11.5433 1.4864 0.4084 0.6511 0.0546 11.0329 1.4227 0.3872 0.6174 0.0523 2.6037 0.1445 0.2159 0.3421

15.451 −1.399 0.494 −0.503 0.035 −5.568 0.077 0.690 0.230 −0.019 12.919 −0.584 −0.208 −0.755 0.020 8.682 −0.297 0.475 −0.792 −0.005 7.865 −1.182 0.544 0.092 0.027 6.219 −0.203 0.040 −0.401

0.097 0.052 0.287 0.572 0.210 0.488 0.900 0.085 0.764 0.416 0.179 0.433 0.663 0.412 0.480 0.331 0.657 0.290 0.359 0.836 0.280 0.036 0.134 0.895 0.203 0.085 0.117 0.841 0.290

9.3030 0.7202 0.4639 0.8916 0.0276 8.0221 0.6194 0.4003 0.7694 0.0237 9.6055 0.7461 0.4785 0.9198 0.0287 8.9339 0.6698 0.4490 0.8631 0.0255 7.2799 0.5635 0.3630 0.6978 0.0216 3.6055 0.1294 0.1973 0.3792

10.036 0.687 0.209 −1.576 −0.025 −2.360 0.310 −0.557 0.649 −0.017 10.588 −0.306 −0.640 −0.143 0.001 1.328 −0.189 −0.306 0.389 0.001 1.474 1.961 −0.573 −0.864 −0.072 1.514 −0.190 −0.222 0.311

0.199 0.269 0.624 0.055 0.297 0.546 0.320 0.009 0.114 0.161 0.042 0.461 0.024 0.794 0.936 0.841 0.718 0.397 0.577 0.956 0.827 0.000 0.117 0.222 0.000 0.482 0.050 0.126 0.258

7.8168 0.6219 0.4259 0.8222 0.0238 3.9071 0.3117 0.2125 0.4102 0.0119 5.2065 0.4150 0.2833 0.5469 0.0159 6.6127 0.5240 0.3615 0.6975 0.0200 6.7347 0.5374 0.3661 0.7070 0.0206 2.1539 0.0967 0.1451 0.2745

woodland (forest/…)

grassland

water/wetland

built-up

unused

Bold values are the statistical significance according to the values in the column “P > z”. Fig. 4. Every per cent of population increases induced land use efficiency percentage changes in Beijing-Tianjin-Hebei (1988–2008) by Empirical Model (I).

changes of land use efficiency (LUE) even though urban income rises in a highly-urbanized region improved built-up LUE. Empirical model (I) reports that during 1995–2008, the LUE of woodland, grassland, and water land are not consistently positively increased by every per cent of urban income increases (lnuninc) (Fig. 5). During 1988–1995, the urban income has statistically significant impacts on LUE of ecological infrastructure land except unused land; but after 1995, urban income growth is not the primary impact factors of LUE changes (Table 6). We firstly assume that the equation of each type of land use efficiency changes is independently from other types of land use, but a command of

impels intensive land use in urban area. Every per cent of GDP growth with structural effects induces more land converted to built-up area after 2005. Moreover, LUE of woodland and water/wetland increased during 1988–2008. It is explainable that afforestation, artificial wetland, reservoirs, and open ground water channels increased for the Olympic Games (Ma, Zhang, Zhang, Zhao, & Li, 2012). It infers that demand of ecological infrastructure may primarily increase due to some planning issues or some other impact factors other than population growth. Urban income growth is not the determinator of the structure 311

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Fig. 5. Every per cent of urban income increases induced land use efficiency percentage changes in Beijing-Tianjin-Hebei (1988–2008) by Empirical Model (I).

Table 7 Empirical Model (I) test results of land use efficiency in Beijing-Tianjin-Hebei, 1988–2008. Empirical Model (I)

LUE_8895

LUE_9500

LUE_0005

Equation RMSE R-sq DW Test RMSE R-sq DW Test RMSE R-sq 1 lueagr 1.4439 0.0935 1.4558 1.7114 0.0197 1.5075 1.7346 0.0367 2 luefor 1.9203 0.0521 1.4258 1.1593 0.1439 1.6642 1.4969 0.0330 3 luegras 2.0142 0.1887 1.7858 1.0738 0.1351 1.7287 1.7895 0.0146 4 luewate 1.8179 0.0864 1.6754 1.5829 0.0747 1.4362 1.6794 0.0210 5 luebdup 1.4723 0.0935 1.1445 1.4999 0.0477 1.4948 1.3575 0.0468 6 lueus 1.6089 0.0255 1.7807 0.8363 0.0125 1.2891 0.7378 0.0304 NL-SUR System Autocorrelation Tests Harvey LM Test Ho: No Autocorrelation in eq. #: bij = 0 P-Value > Chi2(1) LM Test Rho P-Value LM Test Rho P-Value LM Test Rho 1 lueagr 11.887 0.072 0.001 5.431 0.036 0.020 7.831 0.057 2 luefor 13.317 0.081 0.000 0.152 0.001 0.697 2.992 0.022 3 luegras 1.784 0.011 0.182 0.071 0.001 0.790 7.259 0.053 4 luewate 4.017 0.024 0.045 3.414 0.023 0.065 1.720 0.013 5 luebdup 29.673 0.180 0.000 6.892 0.045 0.009 9.906 0.072 6 lueus 1.944 0.012 0.163 0.787 0.005 0.375 0.169 0.001 Ho: No Overall System Autocorrelation: b11 = b22 = bMM = 0 P-Value > Chi2(6) Harvey LM Test 62.621 0.000 16.746 0.010 29.876 NL-SUR System Heteroscedasticity Tests Ho: Homoscedasticity − Ha: Heteroscedasticity P-Value > Chi2(1) luefor: Engle LM ARCH Test: E2 = E2_1 0.275 0.600 0.403 0.526 9.398 luefor: Hall-Pagan LM Test: E2 = Yh 5.647 0.018 0.257 0.612 4.311 luefor: Hall-Pagan LM Test: E2 = Yh2 5.097 0.024 14.271 0.000 0.137 luefor: Hall-Pagan LM Test: E2 = LYh2 2.928 0.087 3.516 0.061 2.912 Overall System NL-SUR Heteroscedasticity Tests: Ho: No Overall System Heteroscedasticity P-Value > Chi2(15) Breusch-Pagan LM Test 195.584 0.000 285.256 0.000 180.008 Likelihood Ratio LR Test 216.199 0.000 343.200 0.000 207.135 Wald Test 84.997 0.000 135.099 0.000 90.286 NL-SUR Breusch-Pagan Diagonal Covariance Matrix LM Test: Ho: Diagonal Disturbance Covariance Matrix (Independent Equations) Ho: Run NLS − Ha: Run NL-SUR P-Value > Chi2(15) Lagrange Multiplier Test 195.584 0.000 285.256 0.000 180.008

LUE_0508 DW Test 1.1776 1.3162 1.3961 1.2950 1.1992 1.8256

RMSE 1.6505 0.8233 1.0977 1.4007 1.4187 0.5629

R-sq 0.0384 0.0677 0.1011 0.0120 0.1261 0.0609

DW Test 1.1220 1.5135 1.8010 1.9172 1.1251 1.6821

P-Value 0.005 0.084 0.007 0.190 0.002 0.681

LM Test 13.017 0.171 0.257 2.425 20.511 3.595

Rho 0.086 0.001 0.002 0.016 0.135 0.024

P-Value 0.000 0.679 0.613 0.119 0.000 0.058

0.000

39.976

0.000

0.002 0.038 0.711 0.088

0.543 3.938 10.199 3.148

0.461 0.047 0.001 0.076

0.000 0.000 0.000

202.934 241.135 171.836

0.000 0.000 0.000

0.000

202.934

0.000

network is not strongly connecting urban and rural area in a region, so that core regions at county level in BTH should but may not have significantly influences on the welfare redistribution to their neighborhoods. Thus, we use a command of LMANLSUR of Overall System NLSUR Autocorrelation Tests to report Harvey (LM) Test and DurbinWatson (DW) Test, and a command of LMHNLSUR of Overall System NL-SUR Heteroskedasticity Tests for testing the efficiency of each equation in the model (Shehata, 2012), because we guess the results may be distorted by autocorrelation and heteroskedasticities. When we try to make feasible urban planning for a city, we want to skim these influences from neighborhoods. Thereby, we test without these ecological intercorrelations by using employing model (II) which is supposed to should not have disturbing autocorrelation and

LMCOVNLSUR for reporting the test results of Breusch-Pagan Lagrange Multiplier Diagonal Covariance Matrix Test (Shehata, 2014) show that the hypothesis of independent equations in empirical model (I) is rejected (Table 7). It infers that ecological intercorrelation may exist among different types of land use efficiency changes in regional economic development. For instance, if population transporting among adjacent regions conveniently, there may have high ecological intercorrelation to transboundary development. However, when the ecological intercorrelation is weak to allocate demands to support better economic performance, local urban income does not have strongly spillover effects to raise up neighborhoods development. It implies that the relatively higher urban income of less population in urban area is a weak driving force to regional development when an advanced road 312

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Table 8 Comparison of Empirical Model (I) & (II) results of land use efficiency in Beijing-Tianjin-Hebei, 1988–2008. Comparison of Empirical Model (I) & (II)

LUE_8895

Land Type

Coef. Model I

variables

code

LUE_9500 Coef. Model II

P>z Model II

Coef. Model I

LUE_0005 Coef. Model II

P>z Model II

Coef. Model I

_cons b0_ag −14.064 3.106 15.451 lnpop b1_ag 2.219 0.518 0.3940 −0.261 0.081 0.8980 −1.399 lnruinc b2_ag −0.841 −0.736 0.1050 0.772 0.772 0.1620 0.494 lnuninc b3_ag 1.816 1.375 0.0520 −0.531 −0.447 0.5410 −0.503 L1.lnpop^2 b4_ag −0.113 −0.048 0.0250 0.004 −0.008 0.7250 0.035 woodland _cons b0_fo 3.059 14.499 −5.568 (forest/…) lnpop b1_fo −0.556 −0.317 0.7490 −1.132 0.576 0.1620 0.077 lnruinc b2_fo −1.045 −1.086 0.1190 −0.906 −0.903 0.0440 0.690 lnuninc b3_fo 1.727 1.938 0.0620 0.478 0.795 0.0970 0.230 L1.lnpop^2 b4_fo −0.005 −0.014 0.7350 0.020 −0.042 0.0710 −0.019 grassland _cons b0_gr 38.923 −15.207 12.919 lnpop b1_gr −6.697 −2.355 0.0020 2.666 0.841 0.0360 −0.584 lnruinc b2_gr −2.363 −2.703 0.0000 −0.861 −0.864 0.0210 −0.208 lnuninc b3_gr 3.589 5.131 0.0000 0.640 0.330 0.3920 −0.755 L1.lnpop^2 b4_gr 0.225 0.059 0.0400 −0.107 −0.040 0.0340 0.020 water/wetland _cons b0_wa 11.231 13.107 8.682 lnpop b1_wa −3.767 −2.616 0.0020 −0.165 1.371 0.0490 −0.297 lnruinc b2_wa −0.708 −0.820 0.1130 −1.004 −1.002 0.0120 0.475 lnuninc b3_wa 2.936 3.470 0.0000 −0.353 −0.062 0.9200 −0.792 L1.lnpop^2 b4_wa 0.121 0.077 0.0180 0.003 −0.053 0.1190 −0.005 built-up _cons b0_bd −26.662 −2.575 7.865 lnpop b1_bd 3.772 0.655 0.2860 0.068 −0.229 0.6960 −1.182 lnruinc b2_bd −0.227 −0.014 0.9790 1.014 1.013 0.0180 0.544 lnuninc b3_bd 1.480 0.548 0.4660 −0.284 −0.344 0.5360 0.092 L1.lnpop^2 b4_bd −0.171 −0.051 0.0350 −0.013 −0.003 0.9010 0.027 unused _cons b0_un 6.728 0.682 6.219 lnpop b1_un −0.305 −0.255 0.2850 0.087 0.099 0.0570 −0.203 lnruinc b2_un −0.468 −0.623 0.2190 −0.276 −0.277 0.3090 0.040 lnuninc b3_un 0.123 1.001 0.1090 0.051 0.112 0.6560 −0.401 System NL-SUR Non Normality Tests: Jarque-Bera LM Test Ho: Normality − Ha: Non Normality P-Value > Chi2(2) 1 lueagr 37.109 0.000 10.656 0.005 2 luefor 15.738 0.000 437.833 0.000 3 luegras 14.041 0.001 387.117 0.000 4 luewate 16.305 0.000 92.848 0.000 5 luebdup 38.680 0.000 6.631 0.036 6 lueus 431.590 0.000 13800 0.000 Likelihood-ratio test: Assumption: P-Value > Chi2(6) (lue_m22 nested in lue_m1) 13.55 0.0351 16.26 0.0124 cultivated land (dry/paddy)

LUE_0508 Coef. Model II

P>z Model II

−0.878 0.484 0.728 0.018

0.4190 0.2060 0.3530 0.6850

−0.182 0.692 −0.171 −0.010

0.7670 0.0850 0.6910 0.6330

−0.139 −0.217 0.269 0.006

0.9240 0.7230 0.7820 0.9190

−0.102 0.468 −0.042 −0.010

0.9010 0.3760 0.9380 0.7640

−0.932 0.539 0.727 0.020

0.2230 0.0890 0.2180 0.5480

−0.140 0.033 0.181

0.4150 0.8560 0.3190

436.082 509.350 415.516 227.165 32.719 2603.8 8.30

Coef. Model I

Coef. Model II

P>z Model II

1.118 0.155 −0.742 −0.042

0.2240 0.7570 0.3490 0.2320

0.208 −0.545 0.453 −0.013

0.5930 0.0890 0.1800 0.4070

0.162 −0.696 0.729 −0.017

0.8230 0.0530 0.2180 0.5550

−0.156 −0.314 0.514 0.000

0.7430 0.4020 0.3210 0.9970

2.030 −0.581 −0.745 −0.075

0.0010 0.0600 0.1590 0.0010

−0.183 −0.231 0.471

0.0970 0.1820 0.0810

0.000 0.000 0.000 0.000 0.000 0.000

451.131 925.701 2707.004 234.888 56.257 22100

0.000 0.000 0.000 0.000 0.000 0.000

0.2172

10.24

0.1148

10.036 0.687 0.209 −1.576 −0.025 −2.360 0.310 −0.557 0.649 −0.017 10.588 −0.306 −0.640 −0.143 0.001 1.328 −0.189 −0.306 0.389 0.001 1.474 1.961 −0.573 −0.864 −0.072 1.514 −0.190 −0.222 0.311

Bold values are the statistical significance according to the values in the column “P > z”.

of land use efficiency changes following non-normal distribution (Table 8). It means that the changes of LUE without structural effects in model (II) are influenced by population and rural income that violates the normal rules of economic development. This implies that increasing rural income theoretically should not be but in fact be the key impact factor to urban development if urban-rural ecological intercorrelation is weak in a region. We also test with or without the impacts of a constant technique efficiency term, and find both methods report the same violation. With a constant technique efficiency term in the model (I), population and urban income intuitively and seemingly drive the LUE changes because urban population and income are both higher than that of rural residents in this region. However, without a constant technique efficiency term in the model (II), rural income is statistically significant to the woodland and grassland LUE changes, even to the built-up LUE changes after 1995 (Fig. 6 and Table 8). Thus, it illustrates that ecological intercorrelation determines the allocative efficiency of economy scale driven population and income growth when resource use efficiency is smoothly improved over time in a region. Population and income growth drive the allocative efficiency changes of LUE, but urban-rural ecological intercorrelation determines the edge effects between backward-wave effects and spillover effects. The rural income can critically increase the allocative efficiency when ecological intercorrelation is weak currently, but in the future, resource use efficiency may increase economic scale effects when ecological intercorrelation becomes stronger. After 2005 in BTH, every per cent

heteroskedasticities in the estimated probability distribution of LUE changes in the implicit function, so that urban income rises could have significant impacts on local ecological infrastructure land use efficiency changes when the endogenous changes of resource use efficiency are controlled. Recall Empirical Model (II) that assumes without ecological intercorrelations to analyze the modified NL-SUR with MLE method. The theoretical null hypothesis of model (II) is that the autocorrelation and heteroskedasticities are struck out by running s = 16,000 itineration of the MLE. It indicates that structural effects can be allocated smoothly by the probability density function of the implicit function, so that the independent variables in percentage changes can be allocated without distortion, and to reach an unbiased variance-covariance matrix of MLE. Without ecological intercorrelations, model (II) also reports that urban income rises in BTH still do not show the statistical significance as a driving force to the LUE changes (Table 8). However, the rural income rises do have statistical significance as a driving force to the woodland, grassland, and built-up LUE changes (Fig. 6). It indicates that the increase of rural income has critical influences on regional resource allocation efficiency changes if the ecological intercorrelation is weak among adjacent regions. Moreover, because NL-SUR estimated by MLE does not require the data to follow the assumption of normal distribution, we test each equation of model (II) by employing the command of LMNNLSUR of Overall System NL-SUR Non-Normality Tests (Shehata, 2012). It is striking that the test results report each type 313

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

Fig. 6. Every per cent of rural income increases induced land use efficiency percentage changes in Beijing-Tianjin-Hebei (1988–2008) by Empirical Model (II).

investments such like ODI and FDI are used to implement at frim level; and infrastructures are financed by government bonds such like PRC did in 1990s. Business men always try to be close to governors for seeking some ‘imaginary’ guarantees because governmental planning policy easily affect these transboundary investments and public funds. That may further enlarge the inequality issues. Therefore, if we think ecological intercorrelation conveying a philosophy in regional governance, the dynamics of land use efficiency reflect these social impacts and distorting the ideology of local development. In this way, why not let local government directly invest some large transboundary projects through public private partnership to plan a longer term regional development? Strong ecological intercorrelation increase regional homogeneity and decrease heterogeneity that can increase the spillover effects of urban income to drive urban-rural coordinate development, but rural income rises may be weak to drive regional development in a sustainable manner. Because most people are prone to stay relatively far rural area when they have high enough income and enjoy countryside landscape and easily transporting to urban core. Otherwise, they may prefer to seek higher income jobs in cities, and most of new urban residents in mega cities may move to urban fringes, such like living in between the third and the sixth ring road in Beijing. In that case when the far countryside has weak ecological intercorrelation with the urban core, local industrial transformation may determine local resource reallocation to alleviate rural population moving to the city centers. Alternatively, when considering multi-targets in a city planning, to keep each type of land use efficiency (LUE) in an appropriately consistent range may be a smart plan. By holding the accumulative percentage changes in all type of land use as one, it means the growth rate of GDP and the growth rate of each type of land use changes can approach to be a very small number with a limiting value to zero in a longtime transformation. On this condition, the growth rate changes in population, rural income, and urban income can be compensated among each other to present a certain small number in land use efficiency changes relatives to the growth of GDP in a certain development stage. For instance, after 2005 in BTH, the growth rate of pooled changes of built-up land use efficiency is 1.96–2% if one per cent of population increases during 2005–2008. It means every per cent of population growth are highly statistically significant to the built-up land use efficiency increases around two per cent, so that the relative GDP growth decreases two per cent with increasing one per cent of builtup land use. Recall in the introduction, we stated that “Tianjin

increase in population growth are highly statistically significant to the built-up land use efficiency increases around two per cent, which means the relative GDP growth decreases two per cent with increasing one percent of built-up land use. Furthermore, the urban income rises cannot be the determinator to have more spillover effects on neighborhoods when the ecological intercorrelation is weak. Rural income rises can have some positive edge effects, but cannot compensate the loss in GDP growth. 5. Discussion In this research, we aim to study how urban-rural ecological intercorrelation can dynamically determine the edge effects between backward-wave effects and spillover effects to affect land use efficiency in the pathway of regional development. We study theoretical literatures and designing two empirical economic models. The analytical results show our models can follow the theory that resource use efficiency allocates economic scale effects by population and income growth via ecological intercorrelation in a region over time. This indicates that urban income rises drive backward-wave effects exceed the spillover effects when urbanrural ecological intercorrelation is weak, vice versa may drive the spillover effects exceed the backward-wave effects when urban-rural ecological intercorrelation is strong. In Beijing-Tianjin-Hebei region (BTH), to increase rural income can critically increase the allocative efficiency in land use when ecological intercorrelation is weak currently, but in the future, urban income may increase economic scale effects to lower land use changes and keep GDP growth when ecological intercorrelation becomes stronger. Structural changes of land use efficiency interpret downscaling transboundary development conveying social impacts via ecological intercorrelation. It is debatable that intensive land use changes are good or bad in urbanization, but economic growth is as faster as better even if with some uncertain side effects such as environmental degradation and large migration. These rapid changes in social transformation are used to throw governors off their guard in national and regional planning. Local government always focus on own business due to political responsibilities, and chasing to increase built-up land use efficiency. These unbalanced development targets easily lead to inconsistent planning polices unfit for long-term regional development. This induces administrative tensions reflect demerits against coordinative development, and to some extent aggravating income inequality and environmental inequality. By following market economy, transboundary 314

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

6. Conclusion

municipal planning Scheme 2006–2020 set targets that the growth rate of built-up land should be lower than 0.05% if one per cent of GDP increases; and the growth rate of built-up land should be lower than 0.55% if one per cent of population increases.” Thus, these targets of planning may not be fulfilled because the urban income rises cannot be the determinator to have more spillover effects on neighborhoods, and even if rural income rises can have some positive edge effects, that cannot compensate the loss in GDP growth. Even if we consider that the spillover effects by all urban and rural income rises can counteract the backward-wave effects by population growth, the pooled built-up land use efficiency can be lowered to 0.45–0.6% when one per cent of population increases during 2005–2008. This implies that income growth generates over 60 percent of the spillover effects on dynamically keeping LUE as a small number relative to GDP growth. Therefore, population growth is not the root of all problems about environmental degradation in social transformation, instead to balance income growth in both rural and urban area is the kernel, and the ecological intercorrelation can be determinants. Whether the increase of land use efficiency is good or bad depends on how to define the term of LUE because intensive land use may humble the planning targets to be come true in some cases. When urban income rises significantly faster than rural income, the demand of luxury housing is also increasing. It is questionable that the strategy of intensive built-up land use can be well implemented because re-planning or regulation violation often occurs. Some smart housing developers can still build more villas in suburban area by modifying interior construction land use ratio to reach the regulated average plot ratio and then selling villas at quite higher price. Furthermore, the meaning of land use efficiency needs to be redefined based on more ecological indicators for stricter regulations to different functions of land in urban planning. In this research, we intent to prove that LUE conveys structural effects of ecological functions such like forest and grass land because rich people prefer to pay more to enjoy a livable environment. Our 2016 summer survey results show that residents in Beijing have strong preferences to pay for ecological infrastructure improvement which even are stronger than to have higher quality of public services. It infers that some ecological indicators such as emission mitigation function, soil nitrogen-fixing intensity, etc., can contribute to estimate land use efficiency from the demand side. However, in the past, these research findings are very less likely to be emphasized by local governments in their long-term planning for a city. landscape rights in protection and modification are much unclear in many regions, and PRC still does not have the relevant laws. Because carbon sequestration market cannot perfectly reflect spatial misallocation of land cover with these functions, individual consumption demands cannot be directly involved into this kind of market. In other words, there is a gap between spatial allocation and individual consumption demands for different functions of land use in China. Researchers from transportation economics want to bridge this gap, but there are so many endogeneities in their theoretical frameworks. The basic point is that there are no clear articles about the definition of landscape rights for environmental conservation, and of course which cannot be owned by private consumers. European countries like UK and France in 1970s sequentially published the Law for environmental protection in which landscape rights are defined and used to be the reference of evidence for suing mental compensations. Suppose these rights of landscape can be clearly defined and managed by individual willingness in China, why not the willingness of environmental protection has a market value? At this point, land use efficiency thus can be divided by two parts into market and non-market values which are based on private owned and public owned landscape rights to reflect the total market value of regional environmental conservation. This individual level market based on shared private and public landscape rights may be more efficient than current the firm-level carbon sequestration market.

In this study, we review the structural changes of land use changes in Beijing-Tianjin-Hebei region after 1949; and analyze the pooled growth rate changes of land use efficiency induced by every per cent changes in population and both urban and rural income over four-timeperiods during 1988–2008. We find that the changes of population and income per capita determine the changes of land use efficiency in statistical significance. The growth rate of built-up land use efficiency changes is 1.96–2% if one per cent of population increases during 2005–2008, so that the target of regional planning cannot be fulfilled. Moreover, the income rises can make the spillover effects counteract backward-wave effects of population growth, and let the pooled growth rate of land use efficiency in built-up land lower to 0.45–0.6% during 2005–2008. Unbiased results of empirical model (II) show that increasing rural income is statistically significant to the changes of land use efficiency which is violating the normal rules of regional development in economic theory, because it should not occur if the ecological intercorrelation is strong in a rapidly urbanized region. It indicates that the backward-wave effects of rural income rises humble the spillover effects of urban income rises. We also test with or without the impacts of a constant technique efficiency term, and find both methods report the same violation. It infers that backward-wave effects exceed spillover effects in Beijing-Tianjin-Hebei region. Thus, to the pathways of development in different regions, ecological intercorrelation is determinant to coherently keep economy scale of land use efficiency in an optimal range for improving the standard of living in sustainable development to become a livable city. Conflicts of interest The authors declare no conflict of interests. Acknowledgments This research was partly financially supported by the P.R. of China National natural science foundation of international (regional) cooperation and exchange programs (Grant No. 71561137002), and the P.R. of China National Natural Science Funds for Distinguished Young Scholar (Grant No. 71225005), and the Key Project of National Natural Science Foundation of P.R. China (Grant No. 7153000125); and the first author was funded by the Joint-PhD program funded by China Scholarship Council (Grant No. 201606510044) for six-month studying in the Department of Urban Planning at The University of Manchester; to editors and reviewers, we revised this paper in token of gratitude. References Alcott, B. (2005). Jevons' paradox. Ecological Economics, 54(1), 9–21. http://dx.doi.org/ 10.1016/j.ecolecon.2005.03.020. André, C., & Platteau, J. P. (1998). Land relations under unbearable stress: Rwanda caught in the Malthusian trap. Journal of Economic Behavior & Organization, 34(1), 1–47. http://dx.doi.org/10.1016/S0167-2681(97)00045-0. Arsel, M., & Dasgupta, A. (2013). Structural change, land use and the state in China: Making sense of three divergent processes. The European Journal of Development Research, 25(1), 92–111. http://dx.doi.org/10.1057/ejdr.2012.26. Auzins, A., & Geipele, I. (2013). Measuring land-use efficiency in land management. Advanced Materials Research, 804, 205–210. http://dx.doi.org/10.4028/www. scientific.net/amr.804.205 [Trans Tech Publications]. Beames, A., Broekx, S., Heijungs, R., Lookman, R., Boonen, K., Van Geert, Y., ... Seuntjens, P. (2015). Accounting for land-use efficiency and temporal variations between brownfield remediation alternatives in life-cycle assessment. Journal of Cleaner Production, 101, 109–117. http://dx.doi.org/10.1016/j.jclepro.2015.03.073. Boudeville, J. R. (1966). Problems of regional economic planning. Edinburgh UP: Edinburgh University Press. Bryan, B. A., Crossman, N. D., Nolan, M., Li, J., Navarro, J., & Connor, J. D. (2015). Land use efficiency: Anticipating future demand for land-sector greenhouse gas emissions abatement and managing trade-offs with agriculture, water, and biodiversity. Global Change Biology, 21(11), 4098–4114. http://dx.doi.org/10.1111/gcb.13020.

315

Landscape and Urban Planning 177 (2018) 303–316

Z. Wang et al.

efficiency and planning in Shunyi, Beijing. Landscape and Urban Planning, 85(1), 40–48. http://dx.doi.org/10.1016/j.landurbplan.2007.09.004. Moon, H. R., & Perron, B. (2006). Seemingly unrelated regressions. The New Palgrave Dictionary of Economics, 1–9http://mapageweb.umontreal.ca/perrob/palgrave.pdf. Morrow, C. M. (2016). Metaphorical language in the zhuangzi. Philosophy Compass, 11(4), 179–188. http://dx.doi.org/10.1111/phc3.12314. Oberhofer, W., & Kmenta, J. (1974). A general procedure for obtaining maximum likelihood estimates in generalized regression models. Econometrica, 42(3), 579–590. Peck, J., & Tickell, A. (2002). Neoliberalizing space. Antipode, 34(3), 380–404. http://dx. doi.org/10.1111/1467-8330.00247. Perroux, F. (1950). Economic space: theory and applications. The Quarterly Journal of Economics, 89–104. http://dx.doi.org/10.2307/1881960. Peters, C. J., Picardy, J. A., Darrouzet-Nardi, A., & Griffin, T. S. (2014). Feed conversions, ration compositions, and land use efficiencies of major livestock products in US agricultural systems. Agricultural Systems, 130, 35–43. http://dx.doi.org/10.1016/j. agsy.2014.06.005. Romer, P. (1993). Idea gaps and object gaps in economic development. Journal of Monetary Economics, 32(3), 543–573. http://dx.doi.org/10.1016/0304-3932(93) 90029-F. Shehata, E. A. E. (2012). LMANLSUR: Stata module to perform overall system NL-SUR autocorrelation tests. Statistical Software Components. Shehata, E. A. E. (2014). LMCOVNLSUR: Stata module to perform breusch-Pagan lagrange multiplier diagonal covariance matrix test after (NL-SUR) regressions. Statistical Software Components. Shonkwiler, J. S., & Reynolds, J. E. (1986). A note on the use of hedonic price models in the analysis of land prices at the urban fringe. Land Economics, 62(1), 58–63. http:// dx.doi.org/10.2307/3146563. Smith, V. L. (1977). Control theory applied to natural and environmental resources an exposition. Journal of Environmental Economics and Management, 4(1), 1–24. http:// dx.doi.org/10.1016/0095-0696(77)90012-2. Tu, F., Yu, X., & Ruan, J. (2014). Industrial land use efficiency under government intervention: Evidence from Hangzhou, China. Habitat International, 43, 1–10. http:// dx.doi.org/10.1016/j.habitatint.2015.10.014. Wang, H., Wang, L., Su, F., & Tao, R. (2012). Rural residential properties in China: Land use patterns, efficiency and prospects for reform. Habitat International, 36(2), 201–209. http://dx.doi.org/10.1016/j.habitatint.2011.06.004. Wang, Z., Deng, X., & Chen, J. (2015). Impacts of sparing use of water on farmer income of China. Physics and Chemistry of the Earth, Parts A/B/C, 89, 18–24. http://dx.doi. org/10.1016/j.pce.2015.09.001. Wang, Z., Deng, X., Bai, Y., Chen, J., & Zheng, W. (2016). Land use structure and emission intensity at regional scale: A case study at the middle reach of the Heihe River basin. Applied Energy, 183, 1581–1593. http://dx.doi.org/10.1016/j.apenergy.2016.09.109. Wang, Z., Deng, X., Wang, P., & Chen, J. (2017). Ecological intercorrelation in urban–rural development: an eco-city of China. Journal of Cleaner Production, 163, S28–S41. http://dx.doi.org/10.1016/j.jclepro.2016.02.120. Wang, Z., Deng, X., & Wong, C. (2016). Integrated land governance for eco-urbanization. Sustainability, 8(9), 903. http://dx.doi.org/10.3390/su8090903. Wei, Y. D. (1999). Regional inequality in China. Progress in Human Geography, 23(1), 49–59. http://dx.doi.org/10.1191/030913299676254572. Wei, Y. D. (2015). Spatiality of regional inequality. Applied Geography, 61, 1–10. http:// dx.doi.org/10.1016/j.apgeog.2015.03.013. Wu, J. (2004). Effects of changing scale on landscape pattern analysis: Scaling relations. Landscape Ecology, 19(2), 125–138. http://dx.doi.org/10.1023/B:LAND. 0000021711.40074.ae. Xie, H., & Wang, W. (2015). Spatiotemporal differences and convergence of urban industrial land use efficiency for China’s major economic zones. Journal of Geographical Sciences, 25(10), 1183–1198. http://dx.doi.org/10.1007/s11442-015-1227-2. Yang, J., Dai, G., & Wang, S. (2015). China’s national monitoring program on ecological functions of forests: An analysis of the protocol and initial results. Forests, 6(3), 809–826. http://dx.doi.org/10.3390/f6030809. Zellner, A. (1962). An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association, 57(298), 348–368. Zeng, C., Zhang, A., & Xu, S. (2016). Urbanization and administrative restructuring: A case study on the wuhan urban agglomeration. Habitat International, 55, 46–57. http://dx.doi.org/10.1016/j.habitatint.2016.02.006. Zhang, K. H., & Song, S. (2003). Rural–urban migration and urbanization in China: Evidence from time-series and cross-section analyses. China Economic Review, 14(4), 386–400. http://dx.doi.org/10.1016/j.chieco.2003.09.018. Zhang, J., Shen, M., & Zhao, C. (2014). Rural renaissance: Rural China transformation under productivism and post-productivism. Urban Planning International, 29(5), 1–7. Zheng, S., & Kahn, M. E. (2008). Land and residential property markets in a booming economy: New evidence from Beijing. Journal of Urban Economics, 63(2), 743–757. http://dx.doi.org/10.1016/j.jue.2007.04.010. Zhou, L., Zhang, Z., Wu, L., & Niu, S. (2006). Landscape pattern gradient analysis of green space along the extended directions of urbanization in beijing. Forest Resources Management, 5, 47–52 [in Chinese]. Zhou, Y., He, C., Huang, Z., & Wang, W. (2013). Geography and urban land use efficiency: Empirical study based on panel data of 277 chinese cities during 2004–2008. Urban Development, 7, 007 [in Chinese]. Zitti, M., Ferrara, C., Perini, L., Carlucci, M., & Salvati, L. (2015). Long-term urban growth and land use efficiency in southern Europe: Implications for sustainable land management. Sustainability, 7(3), 3359–3385. http://dx.doi.org/10.3390/su7033359.

Chen, J., & Han, X. (2014). the evolution of the housing market and its socioeconomic impacts in the post-reform People's Republic of China: A survey of the literature. Journal of Economic Surveys, 28(4), 652–670. http://dx.doi.org/10.1111/joes.12076. Chen, X., & Wu, J. (2009). Sustainable landscape architecture: Implications of the Chinese philosophy of unity of man with nature and beyond. Landscape Ecology, 24(8), 1015–1026. http://dx.doi.org/10.1007/s10980-009-9350-z. Chen, Y., Chen, Z., Xu, G., & Tian, Z. (2016). Built-up land efficiency in urban China: Insights from the general land use plan (2006–2020). Habitat International, 51, 31–38. http://dx.doi.org/10.1016/j.habitatint.2015.10.014. Dang, G., & Luo, W. (2013). Equity and efficiency of rural residential land use system in China: An analysis of survey data. Modern China Studies, 20(2), 103 [in Chinese]. Deng, X., Huang, J., Rozelle, S., & Uchida, E. (2008). Growth, population and industrialization, and urban land expansion of China. Journal of Urban Economics, 63(1), 96–115. http://dx.doi.org/10.1016/j.jue.2006.12.006. Ding, C. (2001). The benchmark land price system and urban land use efficiency in china. Chinese Geographical Science, 11(4), 306–314. http://dx.doi.org/10.1007/s11769001-0045-x. Du, J., Thill, J. C., & Peiser, R. B. (2016). Land pricing and its impact on land use efficiency in post-land-reform China: A case study of Beijing. Cities, 50, 68–74. http://dx. doi.org/10.1016/j.cities.2015.08.014. Ewing, R., Greenwald, M., Zhang, M., Walters, J., Feldman, M., Cervero, R., ... Thomas, J. (2010). Traffic generated by mixed-use developments—Six-region study using consistent built environmental measures. Journal of Urban Planning and Development, 137(3), 248–261. http://dx.doi.org/10.1061/(ASCE)UP.1943-5444.0000068. Fetzel, T., Niedertscheider, M., Haberl, H., Krausmann, F., & Erb, K. (2016). Patterns and changes of land use and land-use efficiency in Africa 1980–2005: An analysis based on the human appropriation of net primary production framework. Regional Environmental Change, 16(5), 1507–1520. http://dx.doi.org/10.1007/s10113-0150891-1. Friedmann, J. (1986). The world city hypothesis. Development and Change, 17(1), 69–83. http://dx.doi.org/10.1111/j.1467-7660.1986.tb00231.x. Fu, B., Wang, S., Su, C., & Forsius, M. (2013). Linking ecosystem processes and ecosystem services. Current Opinion in Environmental Sustainability, 5(1), 4–10. http://dx.doi. org/10.1016/j.cosust.2012.12.002. Fu, L., Lei, Y., Wang, G., Bi, H., Tang, S., & Song, X. (2016). Comparison of seemingly unrelated regressions with error-in-variable models for developing a system of nonlinear additive biomass equations. Trees, 30(3), 839–857. http://dx.doi.org/10.1007/ s00468-015-1325-x. Gibson, J., & Li, C. (2017). The erroneous use of China's population and per capita data: A structured review and critical test. Journal of Economic Surveys, 31(4), 905–922. http://dx.doi.org/10.1111/joes.12178. Harris, T. R., & Shonkwiler, J. S. (1997). Interdependence of retail businesses. Growth and Change, 28(4), 520–533. http://dx.doi.org/10.1111/1468-2257.00070. Heckman, J. J., & Mosso, S. (2014). The economics of human development and social mobility. Annu. Rev. Econ. 6(1), 689–733. Jim, C. Y., & Chen, W. Y. (2009). Ecosystem services and valuation of urban forests in china. Cities, 26(4), 187–194. http://dx.doi.org/10.1016/j.cities.2009.03.003. Kögel, T., & Prskawetz, A. (2001). Agricultural productivity growth and escape from the Malthusian trap. Journal of Economic Growth, 6(4), 337–357. http://dx.doi.org/10. 1023/A:1012742531003. Krugman, P. (1993). First nature, second nature, and metropolitan location. Journal of Regional Science, 33(2), 129–144. http://dx.doi.org/10.3386/w3740. Kytzia, S., Walz, A., & Wegmann, M. (2011). How can tourism use land more efficiently? A model-based approach to land-use efficiency for tourist destinations. Tourism Management, 32(3), 629–640. http://dx.doi.org/10.1016/j.tourman.2010.05.014. Li, F., Liu, X., Hu, D., Wang, R., Yang, W., Li, D., & Zhao, D. (2009). Measurement indicators and an evaluation approach for assessing urban sustainable development: A case study for China's Jining City. Landscape and Urban Planning, 90(3), 134–142. http://dx.doi.org/10.1016/j.landurbplan.2008.10.022. Li, F., Liu, X., Zhang, X., Zhao, D., Liu, H., Zhou, C., & Wang, R. (2016). Urban ecological infrastructure: an integrated network for ecosystem services and sustainable urban systems. Journal of Cleaner Production. http://dx.doi.org/10.1016/j.jclepro.2016.02. 079. Liu, Y., & Li, Y. (2017). Revitalize the world’s countryside. Nature, 548(7667), 275–277. http://dx.doi.org/10.1038/548275a. Liu, J., Liu, M., Zhuang, D., Zhang, Z., & Deng, X. (2003). Study on spatial pattern of landuse change in China during 1995–2000. Science in China Series D: Earth Sciences, 46(4), 373–384. http://dx.doi.org/10.1360/03yd9033. Liu, J. (1996). The macro investigation and dynamic research of the resource and environment in China. Beijing: Science and Technology Press158–188. Lu, M., & Wan, G. (2014). Urbanization and urban systems in the People's Republic of China: research findings and policy recommendations. Journal of Economic Surveys, 28(4), 671–685. http://dx.doi.org/10.1111/joes.12078. Ma, C., Zhang, G. Y., Zhang, X. C., Zhao, Y. J., & Li, H. Y. (2012). Application of Markov model in wetland change dynamics in Tianjin Coastal Area, China. Procedia Environmental Sciences, 13, 252–262. http://dx.doi.org/10.1016/j.proenv.2012.01. 024. Magalhães, T. M., & Seifert, T. (2015). Biomass modelling of Androstachys johnsonii Prain: A comparison of three methods to enforce additivity. International Journal of Forestry Research, 2015. http://dx.doi.org/10.1155/2015/878402. McConnell, K. E. (1997). Income and the demand for environmental quality. Environment and Development Economics, 2(4), 383–399. http://dx.doi.org/10.1017/ S1355770X9700020X. Meng, Y., Zhang, F., An, P., Dong, M., Wang, Z., & Zhao, T. (2008). Industrial land-use

316