Abstract

This paper develops a theoretical framework to study the critical role that politics play in shaping the spatial dimension of China's urbanization and the related welfare implications. Utilizing a large data set of residential land transactions matched with city leaders in 200 Chinese cities from 2000 through 2011, the empirical analysis finds that a 1 standard deviation increase in the career-incentive measure leads to 9 additional kilometers of outward expansion, a 23% increase relative to the sample average. It also finds some suggestive evidence pointing to the distortionary impacts of overly strong incentives of city leaders on spatial expansion, consistent with the theory.

I. Introduction

RAPID urbanization has occurred in many developing countries in recent decades (Glaeser, 2014). China has experienced striking urban growth: from 1990 to 2015, its urban population grew from 302 million to 771 million, and real urban GDP increased at an annual rate of 10%.1 Along with this massive urban growth, the cities have undergone substantial spatial expansion. According to data based on U.S. Landsat TM/ETM images covering 200 major Chinese cities, the total built-up urban land area of these cities expanded from 18,652 square kilometers in 1990 to 51,891 square kilometers in 2010, increasing at an annual rate of 5.3%, far outstripping urban population growth.2 This rapid expansion in urban area produced profound social and economic impacts that may last for decades. While increasing city size can bring about higher productivity (Duranton & Puga, 2014), it also leads to longer commuting times and traffic jams (Newman & Kenworthy, 1999; Kahn, 2000; Harari, 2018). Furthermore, urban spatial expansion causes pollution and potential damage to the ecosystem (Glaeser & Kahn, 2004; Zheng & Kahn, 2013). Rural-to-urban land conversion in China has been notoriously characterized by undercompensation of peasants who lost their agricultural land to urban expansion (Feng, Lichtenberg, & Ding, 2015). Such undercompensation incentivizes cities to sprawl rather than undertake brownfield redevelopment.

This paper examines a key driver of urban expansion. In addition to the standard economic causes of urban spatial expansion that have been widely studied in the literature, such as population, income, transportation costs, and agricultural rent (Brueckner & Fansler, 1983; Glaeser & Kahn, 2004; McGrath, 2005; Burchfield et al., 2006), we introduce a new dimension: the critical role of city leaders' career-advancement incentives in shaping the spatial pattern of China's urbanization. Our analysis is motivated by two important institutional features of China's urban planning and development. First, China's city leaders, who hold critical control over local economic activities, play a central role in planning urban land development. They set the key parameters that determine how much new land is to be developed and where (Lichtenberg & Ding, 2009). Second, these leaders are placed in tournaments in which their promotions are evaluated by upper-level governments and closely linked with local economic outcomes such as total output and fiscal revenues (Li & Zhou, 2005). The root cause of these two institutional features lies in China's intergovernmental governance structure, which combines a regionally decentralized economic system with a politically centralized hierarchy (Xu, 2011).

City leaders care about the welfare of their citizens (just as typical social planners would), but they also care about their own political careers. Given that the advancement of a city leader depends on the city's economic performance, the city leader puts certain positive weight on that performance (e.g., total output) in the leader's objective function. This weight increases and overshadows their concern for welfare as their career-incentive intensity increases, which may cause distortion. One strategy that is frequently used to enhance economic performance is to raise fiscal revenues to finance public infrastructure. Taking land at the city edge is a key financing mechanism since it generates land sale revenues for the city treasury and the compensation fee for farmland is very low. while the demolition fee for brownfield redevelopment is very high (World Bank and Development Research Center of the State Council, 2014).

There is a trade-off for city leaders. On one hand, land development can enhance a city's economic performance. On the other hand, outward expansion generates social costs (e.g., pollution, lengthy commuting) and hurts social welfare. Social welfare also has a positive weight in the city leader's objective function. Land development incurs administration costs on the leader. Moreover, the upper-level governments impose quota restrictions on urban land expansion in order to prohibit the loss of arable land and maintain food security. City leaders have to expend costly efforts, such as devising creative rationales for extra quotas and networking with upper-level government officials, to lobby for permission to exceed typical expansion allowances (Xie, 2015).

This paper first develops a model that captures the driving forces of urban outward expansion. Our model predicts that the higher the career-incentive intensity of city leaders is, the more urban outward expansion occurs, which leads to higher output and larger population. Furthermore, city leaders with sufficiently strong career incentives tend to expand their cities outward beyond the socially optimal level, and this excessive expansion is detrimental to social welfare. We then test the key predictions of our model using a large database of over 30,000 completed residential land transactions in 200 Chinese cities from 2000 through 2011. The land transaction data contain detailed information on the use type, transaction date, sale price, and geographic location of each transacted land parcel. We focus on residential land because its sales revenues accounted for nearly three-quarters of the total land revenues produced through public auctions held by city governments, which are at the heart of our political economy story of urban land development. We match the land transaction data with data for 200 cities and 974 city leaders who took office during the same time period.

Our paper contributes to the literature by providing an original political economy story about China's urban spatial expansion. To the best of our knowledge, this paper is the first to show how local politicians under an authoritarian regime play a central role in shaping urban spatial expansion. Furthermore, we investigate the welfare implications of such career-incentive-driven outward expansion. Our paper thus enriches the urban sprawl literature (Brueckner & Fansler, 1983; Glaeser & Kahn, 2004; McGrath, 2005; Burchfield et al., 2006), as well as the literature on land use policies and regulations, for which most current research is set in the context of democratic countries (Gyourko & Molloy, 2015).3

Moreover, our study contributes to a large strand of literature on the incentives of Chinese local leaders and their impacts on economic development. While some studies emphasize the role of fiscal incentives (Qian & Weignast, 1997), others focus on promotion incentives (Li & Zhou, 2005; Xu, 2011). Our study extends this line of the literature by linking local leaders' career incentives to the spatial expansion of urban land development, which is arguably one of the major growth engines of Chinese cities (Henderson, 2005).

There is a small but growing body of recent literature on land development and regulations in urban China (Deng et al., 2008; Lichtenberg & Ding, 2009; Han & Kung, 2015; Brueckner et al., 2017; Cai, Wang, & Zhang, 2017). Unlike these studies, we explicitly model and empirically test the role of city leaders' career incentives in driving the outward expansion of city boundaries and its welfare implications. In addition, the urban land development of Chinese cities has involved pervasive rent seeking and corruption, as is well documented in the literature (Cai, Henderson, & Zhang, 2013; Cai et al., 2017). However, we argue that the empirical findings in this paper are unlikely to be driven by the rent-seeking and corruption motives of local leaders. We elaborate on this in section V.

The rest of the paper is organized as follows. Section II describes the institutional background of land development, as well as of local leaders' career incentives in China. In section III, we present a simple theory that demonstrates city leaders' decision-making processes with regard to urban land development. Section IV discusses the data and the main variables. We present our primary empirical results regarding the outward expansion of urban land in section V. In section VI, we discuss welfare. Section VII concludes.

II. Institutional Background

In this section, we present detailed institutional background on urban land development, local leaders' career incentives, and land-fiscal policies in China.

A. Urban Land Development Regulations in China

Land quota and urban growth control.

In 1978, at the beginning of the economic reforms, the arable land per capita in China was only 0.1 hectare, far lower than that in the United States (0.85 hectare).4 In the 1980s and early 1990s, there was an unprecedented loss of arable land in China due to large-scale industrialization and urbanization. This sizable decline fueled concerns about China's food security (Brown, 1995). At the same time, conflicts between local governments and farmers due to insufficient compensation for rural-to-urban land conversion intensified. In order to keep urban land expansion under control, the central government amended the Land Administrative Law in 1998 and enacted a new set of arable land protection provisions in order to prohibit additional losses. In the same year, a new ministry, the Ministry of National Land and Resources of the People's Republic of China, was established to strengthen the central government's control over land development.

Since 1998, an urban land quota system has been implemented via a hierarchical, top-down planning process. The central government makes the nation's long-term land development plan. The first plan covered the period from 1997 through 2010. The plan dictates the maximum amount of newly developed urban land for each province in the long term, as well as the minimum amount of arable rural land to be held in reserve. Under these two important constraints, provincial governments make their own long-term development plans. In practice, provincial governments also make short-term (five-year or annual) land development plans. According to these plans, they allocate land use quotas to cities under their administrative control. The quota constraints are primarily imposed on rural-to-urban land conversion. Within a city's boundary, filling in vacant and waste land spots and redeveloping brownfields do not require extra land quotas.

When a provincial government allocates land quotas to cities in its jurisdiction, it must follow certain general guidelines set by the central government. For example, a city's land quota should be proportional to the city's income and predicted future population growth. The city government makes decisions regarding the size and location of land to be developed in a city. When land development involves conversion of rural land to urban land, it must be approved by the upper-level government. If the conversion is within the original quota constraints imposed by the upper-level government, obtaining approval is relatively easy. Otherwise the city government has to exert extra effort to lobby the upper-level government for an increase in the land quota.5 The lobbying process involves extensive bargaining and negotiations with the upper-level government and even the central government. City leaders typically need to design and propose rationales for extra quotas, network with upper-level or central government officials, and take the risk of being scrutinized by the upper-level governments in cases where the constraints are violated. Various creative rationales for extra quotas have been developed by city governments, such as creating college towns, cultural and historic sites, or high-tech industrial parks at the city's edge. All of these involve low-density land use and justify more land allocation (Xie, 2015). In addition, sometimes city governments even tolerate some small-scale land development at a city's edge without any approval from the upper governments.

There is intense competition for quotas among cities within the same province. Some cities may receive more land quotas from the provincial government than others. When short-term provincial quotas are used up, provincial governments may yield to pressure from city governments by allocating future land quotas in advance in order to support urban economic growth in their provinces. As a result of bargaining and concessions, the long-term land quota of rural-to-urban land conversion was exhausted several years before the completion of the first long-term plan in numerous provinces. For example, by 2004, Shandong and Zhejiang provinces had already used up 80% and 99% (respectively) of the total land quotas allotted to them by the 1997–2010 long-term plan. The zeal of local governments for urban land development placed tremendous pressure on the central government, which led to the early ending of the 1997–2010 long-term plan for land development in 2006. Not surprisingly, the drafting of the next long-term plan for 2006–2020 quickly encountered lengthy intergovernmental bargaining.

Conversion of rural to urban land versus brownfield redevelopment.

When converting rural land to urban land, the city government has to compensate the farmers who were using the land before the conversion. By China's Land Administration Law, the total compensation fees should not exceed thirty times the average annual value of products generated from agricultural land in the three years just before the conversion. There was a consensus that the compensation was generally far below the market value of the land at the city edge. In some cases, the market value was 500 times greater than the compensation (Guo, 2001; Ding, 2007). This is mainly because the calculation of the compensation fees is based only on the average returns of agricultural products grown on the farmland. At city edges, farmers often plant high-return agricultural products (e.g., organic vegetables, fruits) or run sightseeing farms that cater to urban residents, or they build shabby lodging to rent out to migrant workers who work in the cities. Although undercompensation of farmers caused discontent and protest, in most cases the farmers had to accept the local governments' offers due to their lack of political power. After rural land is converted to urban land, it is at the disposal of the city government since, by law, the state owns all urban land, and all land sale revenues are turned in to the city treasury. This is why developing rural land at the city edge is lucrative to city governments: they can auction off the land at a much higher price after conversion.

By contrast, brownfield redevelopments near city centers have generally been very costly to local governments. According to World Bank and DRCSC (2014), low-density edge growth and leapfrog growth accounted for about 95% of China's urban growth, while infill and urban redevelopment were quite uncommon. Historically, the central city area was occupied by state-owned enterprises (SOEs) and public housing for their employees during the planning era (1949–1978). These public housing units were sold at very low prices to their residents in the 1998 housing reform. Although the state owns all urban land, the Chinese Constitution and the Land Administration Law are ambiguous about how local governments should execute their property rights and about which use rights apply to the current or former SOE employees living there. This gives great bargaining power to the current use right holders and sitting residents and significantly raises the compensation fees for brownfield redevelopment (Ding & Knaap, 2005).

Land use planning and regulations.

The authority in charge of general land use planning and guideline setting is the city's land reserve and allocation committee, whose members include the city's key leaders and bureau directors from relevant government agencies. Apparently the city's top leader has a decisive voice in key urban planning and land development strategies such as setting the urban spatial boundary. After creating the urban development strategies and guidelines, the committee typically delegates the routine decisions, such as use type and detailed restrictions on floor-to-area ratio (FAR),6 building height, and green area rate to the city's urban planning bureau for each parcel of land to be developed. After the urban planning bureau has set the detailed use regulations for a parcel, the land is turned over to the city's land bureau for auction. Note that what is actually for sale is the leasehold of the land. The typical leasehold length for residential-use land is seventy years, and for purely commercial use land it is forty years. Since 1988, the use rights of vacant urban land parcels have been allocated through leaseholds by city land bureaus.

B. Local Leaders' Career Incentives and Land Finance Policies

China's political system and local leaders' career incentives.

In China, there are four hierarchical city levels: provincial, deputy provincial, prefectural, and county. Our sample excludes all county-level cities. The hierarchical ranking of city leaders largely depends on the hierarchical level of the cities in which they take office, but there are exceptions. Specifically, the hierarchical ranking of city leaders consists of four possible levels. In descending order, these are politburo, province, deputy province, and prefecture. The top city officials are the party secretary and the city mayor. The party secretary is more powerful than the city mayor due to the ruling position of the Chinese Communist Party. We therefore refer to the city's party secretary as the city leader in this paper.

China adopted a one-level-down appointment system in 1984. Prefectural city leaders are under the supervision and control of provincial governments, and deputy-provincial or higher-ranking officials are subject to the evaluation and appointment of the central government. The appointment of a city leader (a party secretary in our case) can be a deliberative process, and many factors, such as the candidate's personality, expertise, and age, come into play. The evaluation criteria of local leaders include political loyalty, educational qualifications, age, and the economic performance of their regions. Since the early 1980s, regional economic performance measures (such as total output and fiscal revenues) have been key performance indicators for the career advancement of local leaders, as documented in the literature (Li & Zhou, 2005; Xu, 2011; Yao & Zhang, 2015; Tsai, 2016).

Local fiscal system and land finance policies.

Beginning in the late 1970s, China went through several waves of fiscal reforms aimed at decentralizing its fiscal system and fiscal management. Fiscal decentralization led to a decline in the central government's share of fiscal income, and it acted to turn this tide by implementing a tax-sharing reform in 1994.

The fiscal revenue of local governments consists of two categories: budgetary and extrabudgetary. Since the 1994 reform, budgetary revenue has been shared between local governments and their upper-level governments. Consequently, 75% of the value-added tax, the largest source of tax revenue, now goes to the central government. Corporate income tax, originally designated as a local tax in 1994, was reclassified as a shared tax between the central and local governments after 2000. Due to repeated reallocations of tax revenues in favor of the central government, local governments have felt increasing fiscal pressure. Land sale revenue is classified as extrabudgetary revenue and local governments are not required to share it. Before 2011, China had no property tax system. Since 2011, just two cities (Shanghai and Chengdu) have begun to levy property taxes on second houses and luxury villas. Property taxes are still in the experimental stage in China.

Against this backdrop, land sale revenues rose to prominence and became the largest source of extrabudgetary income for local governments. According to Chen and Kung (2016) while land revenue accounted for less than 10% of extrabudgetary revenue before 1998, it grew phenomenally over time. By 2008, land revenue accounted for 79% of extrabudgetary revenue and about 38% of total fiscal revenue. Fang et al. (2015) show that the average share of land revenues in city fiscal budgets increased further to 60% in 2009 and to more than 70% in 2010. Over the past two decades, city governments have increasingly relied on land-leasing revenues to finance provision of urban public goods such as infrastructure investment, which can help boost local economic performance. It is worth mentioning that about three-quarters of the land sale revenues created through public auctions come from the sale of residential land. City governments tend to deliberately lower the sale prices of industrial land parcels in order to attract new firms, which again may enhance economic performance.

III. Theoretical Framework

In this section, we present a theoretical framework to demonstrate how a city leader makes decisions regarding the expansion of a city's urban spatial boundary $S$. Our model assumes a monocentric city with linear form as in Duranton and Puga (2015). All workers commute to the central business district (CBD) to work. The city leader sets the optimal urban boundary. We assume no migration across cities for simplicity; however, rural workers in the areas surrounding a city can move into the city under the regulation of the city government in order to attain the population that corresponds to the planned urban spatial size.7 Let us consider a representative city.

A. Individual

An individual worker's utility comes from housing services as well as from consumption of a composite good that includes all other commodities. An individual worker's objective function is
$maxx(r),h(r)u(x(r),h(r))=x(r)θh(r)1-θs.t.w=x(r)+h(r)p(r)+2tr-κT(S),$
(1)
where $r$ indicates the location where the worker lives, $h(r)$ is housing consumption, $x(r)$ is consumption of the composite good, $w$ is wage income, $t$ is commuting cost per unit distance, and $p(r)$ is house price per unit floor area at location $r$. And $p(S)$ is the housing price at the city edge and is equal to the rural rents $p̲$. $T(S)$ is the social cost associated with outward urban expansion due to a variety of factors, such as pollution, congestion, damage to the ecosystem, and undercompensation of farmers, and we assume $TS>0,TSS>0$. $κT(S)$ is the part of the social cost that must be borne by urban residents, $0<κ<1$. Let $V$ denote the individual worker's indirect utility corresponding to the solution of the individual's objective function, equation (1).

B. Firm

Assume that the city's production is conducted by a representative firm. The firm's production function is
$Y=AGN,$
(2)
where $A$ is the city's production amenity, $G$ is the public infrastructure provided by the city government, and $N$ is labor, which is equal to the city's population size in equilibrium. Here, public infrastructure may consist not only of road infrastructure, but also infrastructure for telecommunications and industrial or technology parks. In general, $G$ improves the total factor productivity. We assume that the firm sells products in the national market and that there is free trade among cities. The product price is normalized to 1. Prefect competition among firms gives 0 profit, implying the following wage equation:
$w=AG.$
(3)

C. Land Sale Revenues and Government Expenditures

The city government expands the city to boundary $S$. We assume that the local government conducts land development. The net revenues from land development are equal to the total housing value minus the land compensation fee and construction cost.8 Informed by the institutional background discussed in section II, we assume that the compensation fee $C(S)$ is lower than the social cost of urban outward expansion: $C(S)=(1-κ)T(S),0<κ<1$. The remaining part of the social cost will be borne by the urban residents. All of the net revenue from land development is used to finance public goods $G$ and is equal to
$G=∫0Sp(r)F(r)dr-∫0Sq(F)F(r)dr-(1-κ)T(S),$
(4)

where $F(r)$ is the total floor area per land unit at each location $r$ (the floor-to-area ratio) and $q(F)$ is the construction cost per unit floor area.

Given the institutional background in urban China discussed in section II, we assume that the city government delegates an agent to set the floor-to-area ratio $F(r)$ for land development at each location $r$. In determining the ratio, the agent aims to maximize the value per unit of land:
$F(r)=argmaxFp(r)F-q(F)F.$
The construction cost $q(F)$ increases monotonically with $F$. Note that the assignment of FAR by the agent is similar to what a private developer would do (DiPasquale & Wheaton, 1996), except that the agent considers the possible externality costs included in $q(F)$. Let the construction cost per unit floor area be $q(F)=0.5×F$. Then the optimal FAR level is $F(r)=p(r)$.

D. Build Outward: City Leaders Set Urban Boundary to Maximize Their Own Payoff

In making a decision regarding the urban boundary of their city, city leaders consider their own payoffs, which depend not only on the urban residents' welfare but also on the city leader's political career. The city leader's ex post promotion likelihood increases with $λY$, where $λ$ captures the intensity of the leader's career-advancement incentives. The higher $λ$ is, the further away the local leader is from the career glass ceiling or the brighter their career prospects are. Thus, improving the city's economic performance (captured by $Y$) can more effectively increase the likelihood of promotion for a city leader with a higher $λ$ than of a city leader with a lower $λ$.9 We henceforth assume that the payoffs of a city leader consist of a weighted average of the city's output performance and the welfare of urban residents, where the weight on output performance, denoted as $μ,0<μ<1$, is positively related to the city leader's career-incentive intensity $λ$.

The city leader faces a trade-off. On one hand, expanding the city outward can help the leader's political career. On the other hand, expanding the urban boundary may create social costs that will hurt welfare. Moreover, pushing the urban boundary outward usually requires the city leader to seek extra urban land quotas from the upper-level government, which is costly. Suppose that the upper-level government initially sets the city boundary constraint at the socially optimal boundary that maximizes each urban resident's welfare, denoted as $S0$ (see appendix A for the derivation of $S0$). This reflects that when upper-level governments allocate land quotas to cities, they consider the welfare and the fundamentals of the cities' economies, which is consistent with the institutional background discussed in section II. A city leader who wants to expand the spatial boundary of their city beyond $S0$ needs to lobby the upper-level governments.

The lobbying efforts are reflected in the lobby cost function, denoted as $D(S)$. We assume that both the lobby cost and the marginal lobby cost are 0 for $S≤S0$. When the city expansion goes beyond the boundary constraint, the marginal lobby cost increases with $S$ for $S>S0$. In addition to the lobby cost, administration cost is also associated with land development that increases linearly with $S$. The total effort cost of the city leader is expressed as $E(S)=a×S+D(S)$, where $a>0$ is the city leader's marginal cost of administration effort in urban land expansion.

The objective function of a city leader can be described as follows:
$maxSR≡μY+(1-μ)V-E(S)≡μ×ANG+(1-μ)×V-E(S)s.t.(i)G=∫0S0.5×p2(r)dr-(1-κ)T(S)(ii)∫0SF(r)h(r)dr=∫0Sp(r)h(r)dr=N.$
(5)

In equation (5), (i) is the city government's budget constraint incorporating the optimal FAR, and (ii) means that the total housing supply should accommodate the total housing demand, which effectively determines the city population corresponding to the optimal urban boundary.

E. Solution and Predictions of the Model

We can derive the first-order condition of equation (5) for $S$:
$RS=μA(GSN+NSG)+(1-μ)VS-ES=0.$
(6)

Suppose that there exists an interior solution with $V>0$. And suppose that the second-order derivative of the objective function with respect to $S$, denoted as $RSS$, is negative. Then with the first-order condition, we can solve for the optimal urban boundary for equation (5), denoted as $S*$, and the corresponding $G$, $N$, and $Y$. These fully characterize the equilibrium. Next, we present two of our model's predictions. Note that $S*$ depends on the city's production amenity $A$, transportation technology associated with $t$, agricultural rents associated with $p̲$, and institutional factors such as $κ$ and $a$. Additionally and more important, it depends on the weight $μ$ that the city leader puts on total output in the objective function, which is closely associated with the city leader's career-incentive intensity.

Proposition 1.

Assume that $RSS<0$ and $VSS<0$. And assume the parameter space guarantees that $Ω≡θθ(1-θ)1-θ/p̲1-θ<1$.10 Then at $S*,dS*/dλ>0,dY*/dλ>0$, and $dN*/dλ>0$.

Proof.

See appendix B.

Intuitively, city leaders with greater career-incentive intensity put more weight on the total output, which may enhance their promotion chances, so they are willing to exert more effort to expand the city outward, generating more total output and population.

Proposition 2.

Assume that $RSS<0$ and $VSS<0$. As long as $YS(S0)>a$,11 there exists a threshold career incentive level $λ¯(S0)$ that corresponds to the threshold weight $μ¯=a/YS(S0)$, such that city leaders choose to expand the urban boundary beyond the socially optimal level; that is, $S*>S0$ if and only if $λ>λ¯(S0)$ or $μ>μ¯$. Regarding welfare, for $λ>λ¯(S0),dV*/dλ<0$; for $λ≤λ¯(S0),dV*/dλ≥0$.

Proof.

See appendix B.

Intuitively, at the socially optimal size $S0$, the marginal welfare gain from urban expansion is 0. However, developing more land can still create additional total output. If city leaders care enough about their own political career, it is wise for them to expand the city beyond $S0$ because the benefits generated by the additional output will dominate their additional effort cost. Regarding welfare, if the city leader's career incentives are great enough that the current equilibrium urban boundary is already beyond the socially optimal level, then an increase in the city leader's incentive intensity drives more excessive expansion, which is detrimental to social welfare.

IV. Data, Variable Construction, and Summary Statistics

A. Data

The sample for our analysis of urban land expansion contains over 30,000 residential land transactions completed through public auctions in 200 Chinese cities from 2000 through 2011.12 This database is provided by the China Index Academy, China's largest independent think tank focused on the real estate market. Figure A1 in the online appendix maps the 200 cities, among which 196 are prefecture-level and four are provincial-level cities (Beijing, Shanghai, Tianjin, and Chongqing). They represent 70% of all prefectural cities in China and are mainly located in the eastern and central regions. For each land parcel, we have information on the use type, total area, transaction date, sale price, auction type, address, and more. By taking advantage of the microlevel land data, we are able to obtain detailed geographic information on residential land developments. We first geocode each land parcel based on its address and infer the names of the streets that serve as its boundaries. Using this information, we then find the smallest polygon that contains the land parcel on Baidu Map (www.map.baidu.com). The geographic coordinates of the polygon's centroid are used as the coordinates of the land parcel, and we use them to calculate the distance to the city center for each land parcel.

Table 1.
Summary Statistics
MeanSDObservations
Dummy: Prefecture 0.819 0.385 974
Dummy: Deputy-province 0.161 0.368 974
Dummy: Province or above 0.020 0.138 974
Tenure length, year 3.729 1.889 974
Start age 49.898 3.946 974
Dummy: Promotion of city leader 0.351 0.478 735
Dummy: Central experience 0.033 0.178 974
Dummy: Graduate degree 0.616 0.487 974
90th percentile distance to city center, km 40.388 24.356 362
80th percentile distance to city center, km 31.721 21.159 362
50th percentile distance to city center, km 17.773 13.065 362
Mean distance to city center, km 20.133 12.372 362
City leader's career-incentive intensity 0.306 0.236 362
MeanSDObservations
Dummy: Prefecture 0.819 0.385 974
Dummy: Deputy-province 0.161 0.368 974
Dummy: Province or above 0.020 0.138 974
Tenure length, year 3.729 1.889 974
Start age 49.898 3.946 974
Dummy: Promotion of city leader 0.351 0.478 735
Dummy: Central experience 0.033 0.178 974
Dummy: Graduate degree 0.616 0.487 974
90th percentile distance to city center, km 40.388 24.356 362
80th percentile distance to city center, km 31.721 21.159 362
50th percentile distance to city center, km 17.773 13.065 362
Mean distance to city center, km 20.133 12.372 362
City leader's career-incentive intensity 0.306 0.236 362

In panel A, the hierarchical-level dummies indicate the city leaders' hierarchical ranks at the time they took office. In panel B, the means are calculated by weighting each observation by the number of residential land parcels sold during the office time span of each city leader.

We match the residential land transactions in each city with the data on city leaders during our sample period. We are then able to calculate the extent of outward development during each leader's term of office. Specifically, we collapse the land sample at the city-leader level and measure the urban boundary by the top percentiles in the distribution of the land distance to the city center among all the land parcels sold during each leader's office span. Panel B of table 1 shows that among the 362 city leaders in our matched sample, the average of the 90th percentile distance to the city center is around 40 kilometers.

The outward expansion of urban land development is a typical form of urban spatial expansion. Previous studies have generally measured the expansion by the change in total urban land area (Brueckner & Fansler, 1983; McGrath, 2005; Deng et al., 2008; Lichtenberg & Ding, 2009). A clear limitation of such aggregate measures is that the growth of total urban land area may be caused by either new developments filling in open space between existing developments or developments pushing the existing urban boundary farther away from the city center. The latter type of development is more consistent with the outward expansion of urban boundaries, and our outward measure is better able to capture it. Nonetheless, it is worth noting that our outward measure is positively correlated with the growth of urban land area based on several widely used data sources, such as the land classification data constructed every five years by the Institute of Geographic Science and Natural Resources at the Chinese Academy of Sciences based on the U.S. Landsat TM/ETM images13 and the City Statistical Yearbooks, as shown in table A1 of the appendix.

The data appendix provides data construction details.

B. Constructing Career-Incentive Measurement of City Leaders

As noted in section III, mandatory retirement age varies with the hierarchical ranking of a city leader, so both the age and hierarchical level of city leaders at the start of their office term largely determine their likelihood of promotion. We therefore estimate the effects of initial age and hierarchical rank at the start of office (start age and start level, respectively) on promotion likelihood using our sample of 974 city leaders.

We define a promotion dummy variable that equals 1 if the city leader was promoted to a higher-level position by the end of their term. Note that the higher-level positions exclude those in the local CPPCC or the promotion outcomes of 735 city leaders. Among them, 258 were promoted. Figure A2 shows the percentage of leaders promoted by start age for each start level. For prefecture-level city leaders, it is clear that the likelihood of promotion strictly decreases with start age. Overall, 40% of prefecture-level city leaders successfully climbed to a higher hierarchical level, only 6% of deputy-province-level city leaders were promoted, and about 70% of the province- or politburo-level city leaders were promoted.

We then regress the promotion dummy on start age, the dummies of start levels, and the interactions of start age and dummies of start levels. Column 1 of table 2 shows the LPM results, and column 3 reports the probit results. The relationships between promotion, start age, and start level are consistent with the patterns shown in figure A2. Variations in start age and hierarchical level explain about 20% of the variation in promotion. We use the estimated coefficients on start age, start level, and their interactions in column 3 to predict the prior promotion likelihood for each of the 974 city leaders. Note that we intentionally avoid including variables that are endogenous to the actual effort level of city leaders, such as tenure length and city economic performance. We assume that a leader's career incentives should increase with the predicted prior promotion likelihood. As such, we measure the intensity of city leaders' career incentives by their predicted ex ante promotion likelihood, which is based on their start age and start level.

Table 2.
Start Age, Start Level, and Promotion
LPM (1)LPM (2)Probit (3)Probit (4)
Start age −0.048*** −0.045*** −0.136*** −0.128***
(0.005) (0.005) (0.015) (0.016)
Dummy: Deputy-province −2.690*** −2.825*** −7.935*** −8.524***
(0.263) (0.278) (1.398) (1.537)
Dummy: Province or above −0.570 −0.773 −1.547 −1.279
(0.758) (0.817) (2.386) (3.121)
Start age $×$ Dummy: Deputy-province 0.048*** 0.050*** 0.136*** 0.146***
(0.005) (0.005) (0.027) (0.030)
Start age $×$ Dummy: Province or above 0.022 0.024* 0.060 0.051
(0.013) (0.014) (0.042) (0.053)
Dummy: Central experience  0.177**  0.672***
(0.076)  (0.260)
(0.034)  (0.112)
Constant 2.749*** 2.587*** 6.368*** 5.858***
(0.229) (0.249) (0.752) (0.802)
Observations 735 735 735 735
$R2$ 0.193 0.200
Adjusted $R2$ 0.188 0.192
Pseudo-$R2$   0.162 0.168
LPM (1)LPM (2)Probit (3)Probit (4)
Start age −0.048*** −0.045*** −0.136*** −0.128***
(0.005) (0.005) (0.015) (0.016)
Dummy: Deputy-province −2.690*** −2.825*** −7.935*** −8.524***
(0.263) (0.278) (1.398) (1.537)
Dummy: Province or above −0.570 −0.773 −1.547 −1.279
(0.758) (0.817) (2.386) (3.121)
Start age $×$ Dummy: Deputy-province 0.048*** 0.050*** 0.136*** 0.146***
(0.005) (0.005) (0.027) (0.030)
Start age $×$ Dummy: Province or above 0.022 0.024* 0.060 0.051
(0.013) (0.014) (0.042) (0.053)
Dummy: Central experience  0.177**  0.672***
(0.076)  (0.260)
(0.034)  (0.112)
Constant 2.749*** 2.587*** 6.368*** 5.858***
(0.229) (0.249) (0.752) (0.802)
Observations 735 735 735 735
$R2$ 0.193 0.200
Adjusted $R2$ 0.188 0.192
Pseudo-$R2$   0.162 0.168

***$p$$<$ 0.01, **$p$$<$ 0.05, *$p$$<$ 0.1. Heteroskedasticity-robust standard errors in parentheses are clustered at the city level.

We also know whether the city leader worked previously in the central government and had a graduate degree, both of which are considered relevant to promotion in the Chinese political economy literature (Li & Zhou, 2005; Xu, 2011). Comparing column 2 with column 1 of table 2 (or comparing column 4 with column 3), we can see that these two variables contribute little to the variation in promotion since the adjusted $R2$ remains almost unchanged when we include these two variables. Also, the estimates of the coefficients on the start age, start level, and their interactions are robust to including these two variables.

C. Comparison of Outward Expansion between Earlier and Later Leaders in a Turnover

Before doing regression analysis, we first compare the spatial patterns of urban land development between the earlier leader and the later leader when a city undergoes a turnover of leaders. To differentiate the incentive intensities of city leaders, we define city leaders to be high incentive if their career-incentive intensity is above the sample median (0.25), and low incentive otherwise. Then we divide all the turnovers into four types by the incentive intensities of the earlier and later leaders: high-to-low, low-to-high, high-to-high, and low-to-low. Further restricting the sample to city leaders who sold at least one residential land parcel during their terms, we identify 30 high-to-low-type turnovers, 24 low-to-high-type turnovers, 45 high-to-high-type turnovers, and 63 low-to-low-type turnovers.

For each type of turnover, table 3 reports the mean of the 90th percentile distance to the city center of all land parcels sold by the earlier leader during their term (column 1) and the mean of the 90th percentile distance of the later leader (column 2). Column 3 shows the mean difference in the 90th percentile distance between the earlier and later leaders for each turnover type. One can see that when a city experiences a turnover from a high-incentive to a low-incentive leader, the 90th percentile of distance decreases from 43 kilometers to kilometers km on average (the $t$-statistic of the mean difference is −2.92 and the $p$-value is 0.005). By contrast, when a city experiences a low-to-high type turnover, the city expands farther outward by about 11 kilometers on average (the $t$-statistic of the mean change is 2.83 and the $p$-value is 0.007). Notice that the mean difference in the outward expansion is small and insignificant for the other two turnover types where the earlier and later leaders do not exhibit large differences in incentive intensities. This exercise indicates that the outward expansion of urban development is closely related to city leaders' career incentives.

Table 3.
Comparison of Outward Expansion between Earlier and Later Leaders in a Turnover
90th Percentile Distance to City Center of Land Sold by the Earlier Leader (km)(1)90th Percentile Distance to City Center of Land Sold by the Later Leader (km) (2)Mean Difference in 90th Percentile Distance to City Center (km) (3) $=$ (2) − (1)
High-to-low 42.820 34.889 −7.931***
(2.224) (1.554) (2.713)
Low-to-high 35.206 45.736 10.530***
(3.186) (1.929) (3.724)
High-to-high 47.211 49.803 2.593
(1.840) (1.566) (2.416)
Low-to-low 31.609 31.989 0.380
(1.187) (1.070) (1.598)
90th Percentile Distance to City Center of Land Sold by the Earlier Leader (km)(1)90th Percentile Distance to City Center of Land Sold by the Later Leader (km) (2)Mean Difference in 90th Percentile Distance to City Center (km) (3) $=$ (2) − (1)
High-to-low 42.820 34.889 −7.931***
(2.224) (1.554) (2.713)
Low-to-high 35.206 45.736 10.530***
(3.186) (1.929) (3.724)
High-to-high 47.211 49.803 2.593
(1.840) (1.566) (2.416)
Low-to-low 31.609 31.989 0.380
(1.187) (1.070) (1.598)

In column 3, *** indicates that $p$$<$ 0.01. Standard errors are reported in parentheses. There are 30 high-to-low-type turnovers, 24 low-to-high-type turnovers, 45 high-to-high-type turnovers, and 63 low-to-low-type turnovers.

V. The Effect of Career Incentives on Urban Outward Development

We examine the relationship between a city leader's career-incentive intensity and the extent of the city's outward development during the leader's office term. We have 362 city-leader observations for final estimation. The main regression specification is given by
$zc,sp=βpλc,s+ηc+Xc,sϕ+uc,s,$
(7)
where $zc,sp$ represents the $p$th percentile of the distribution of the distance to the city center of land parcels sold during the office term of city leader $s$ in city $c$; $λc,s$ is the career-incentive intensity of city leader $s$ when he took office in city $c$; $ηc$ are city fixed effects, capturing the time-invariant characteristics of city $c$ that may affect its spatial expansion, such as geographic features, climate, and natural endowments (Burchfield et al., 2006); and $uc,s$ is an error term. The parameter of interest is $βp$. We expect $βp$ to be positive for uppermost percentiles as our model predicts that city leaders with higher career incentives are more likely to expand their cities outward.

In equation (7), $Xc,s$ represents a set of city-leader-level characteristics. The inclusion of $Xc,s$ addresses the concern that the placement of a high-incentive leader may be correlated with some omitted city-leader-level characteristics associated with the city's outward expansion. First, the timing of a city leader's arrival and departure may overlap with the enactment of some national or regional policies. In addition, when the turnover of a city leader coincides with that of a superior, the city leader may develop the city in a way that caters to the new boss. To deal with these issues, we include the fixed effect of the years when each city leader takes and leaves office (the office-start and office-end years), as well as the province-specific time trends of the office-start and office-end years, to capture the national and regional trends of urban outward expansion upon the turnover of city leaders. Second, the appointment of high-incentive city leaders may initially be systematically correlated with certain city characteristics such as higher population, greater production scale, larger built-up urban land area, or greater growth potential in these dimensions. These cities may automatically expand faster after the new leaders take office. Because of this, we control for the city's population size, GDP per capita, and built-up urban land area (all in logs) in the years just before the office-start years in all our regressions. In order to control for the pretrends of the city's economic development, we also include in the regression the growth rates of population, GDP per capita, and built-up urban land area over the two years before a city leader takes office.

All the regressions are weighted by the number of land parcels sold during the office span of each city leader. Because the career-incentive-intensity measure is a generated regressor, the standard errors are calculated on the basis of 1,000 bootstrap replications. Also, we cluster the bootstrap replications on provinces because the outward expansion of cities in the same province may be correlated.

Columns 1 to 4 of table 4 report the regression results for the 90th percentile, 80th percentile, median, and mean of distance to the city center. The greater the top percentile of the distance to the city center is, the farther outward the city expands. Consistent with the model's prediction, the coefficient $βp$ is positive and significant for both the 90th and 80th percentiles, suggesting that a city leader with greater career incentives tends to promote outward-oriented urban development. The effect of career incentives is large: a 1 standard deviation increase in the career-incentive intensity (0.236) leads to an increase in the 90th percentile distance to the city center of about 9.46 kilometers, representing about a 23% increase over the sample mean (which is about 40 kilometers as shown in panel B of table 1). Columns 3 and 4 show that the effects of career incentives on the median and mean distances to the city center are positive but relatively small in magnitude and statistically insignificant. Table A2 in the appendix presents the regression results for different percentiles of distance to the city center. The effects of career incentives on the lower percentiles turn gradually from being positive and significant to being negative and insignificant as we move from the 90th percentile to the 10th percentile of the distance distribution. This constitutes strong evidence that the positive effects of career incentives show up mainly in the expansion of urban boundaries as opposed to inner-city land development. The standard economic factors such as initial growth rates of GDP per capita and population are positively associated with urban outward expansion, which is consistent with the classic urban literature.

Table 4.
Outward Development and Career Incentives of City Leaders Dependent Variable: City-Leader-Level Outward Expansion Measures (km)
90th Percentile Distance to City Center80th Percentile Distance to City Center50th Percentile Distance to City CenterMean Distance to City Center90th Percentile Distance to City Center80th Percentile Distance to City Center50th Percentile Distance to City CenterMean Distance to City Center
Full SampleDrop 4 Provincial-Level Cities
(1)(2)(3)(4)(5)(6)(7)(8)
City leader's career-incentive intensity 40.087*** 25.532** 1.473 9.731 40.131** 25.550** 1.534 9.769
(15.221) (10.945) (8.967) (7.295) (16.334) (11.319) (13.645) (7.379)
City fixed effects
Observations 362 362 362 362 353 353 353 353
$R2$ 0.975 0.971 0.959 0.980 0.971 0.966 0.946 0.975
90th Percentile Distance to City Center80th Percentile Distance to City Center50th Percentile Distance to City CenterMean Distance to City Center90th Percentile Distance to City Center80th Percentile Distance to City Center50th Percentile Distance to City CenterMean Distance to City Center
Full SampleDrop 4 Provincial-Level Cities
(1)(2)(3)(4)(5)(6)(7)(8)
City leader's career-incentive intensity 40.087*** 25.532** 1.473 9.731 40.131** 25.550** 1.534 9.769
(15.221) (10.945) (8.967) (7.295) (16.334) (11.319) (13.645) (7.379)
City fixed effects
Observations 362 362 362 362 353 353 353 353
$R2$ 0.975 0.971 0.959 0.980 0.971 0.966 0.946 0.975

***$p$$<$ 0.01, **$p$$<$ 0.05, *$p$$<$ 0.1. Standard errors (in parentheses) are calculated on the basis of 1,000 bootstrap replications clustered at the province level. Each regression is weighted by the number of residential land parcels sold during the office term of each city leader. The baseline city-leader-level characteristics include the dummies of the years when the city leader takes and leaves office; the province-specific time trends of the office-start and office-end years; the city's population, GDP per capita, and built-up urban land area (all in logs) in the year just prior to the leader's office-start year; and the growth rates of population, GDP per capita, and built-up urban land area over the two years just before the city leader takes office.

Urban development in the four provincial-level cities of Beijing, Shanghai, Tianjin, and Chongqing may be heavily influenced by the will of the leaders in the central government. Also, the leaders of these four cities may have quite different career horizons as most of those in our sample eventually became central leaders of the country. Columns 5 to 8 of table 4 show that the results are fairly robust to the exclusion of these four large cities.

The urban land development of Chinese cities has involved pervasive rent seeking and corruption in land auctions (Cai et al., 2013). As such, outward expansion may create opportunities for corruption (e.g., through under-the-table deals with private developers in land transactions and development or deals with private firms during infrastructure construction). Corruption, then, might be a motive in city leaders' decisions regarding urban outward expansion. But we believe that the rent-seeking and corruption motives do not drive our key results, which show that city leaders with greater career-advancement incentives tend to expand the city farther outward. Generally the opportunity cost of corruption is higher for younger city leaders with stronger career incentives (if caught and convicted). In fact, there is much talk in China about the so-called 59-year-old phenomenon, which refers to cases where city leaders become very corrupt just one year before the retirement age of 60.14

B. Identification Challenges and Robustness Checks

Table 5.
Outward Development and Career Incentives of City Leaders: Robustness Checks Dependent Variable: City-Leader-Level 90th Percentile Distance to City Center (km)
(1)(2)(3)(4)(5)
City leader's career-incentive intensity 43.143*** 35.929* 41.794** 47.473 41.707**
(16.398) (20.384) (16.948) (39.583) (17.181)
(12.856)
Log (initial road stock) $×$ Office-start year trend −0.011
(0.764)
Log (initial population) $×$ Office-start year trend  −3.420
(2.457)
Log (initial GDP per capita) $×$ Office-start year trend  −3.383
(3.014)
Log (initial built-up land area) $×$ Office-start year trend  2.683
(2.199)
Immediate predecessor's career-incentive intensity   −4.466
(12.601)
Bartik variable    169.788
(395.636)
Dummy: Central experience     4.806
(10.675)
(5.052)
City fixed effects
Observations 360 362 359 299 362
$R2$ 0.976 0.978 0.979 0.992 0.975
(1)(2)(3)(4)(5)
City leader's career-incentive intensity 43.143*** 35.929* 41.794** 47.473 41.707**
(16.398) (20.384) (16.948) (39.583) (17.181)
(12.856)
Log (initial road stock) $×$ Office-start year trend −0.011
(0.764)
Log (initial population) $×$ Office-start year trend  −3.420
(2.457)
Log (initial GDP per capita) $×$ Office-start year trend  −3.383
(3.014)
Log (initial built-up land area) $×$ Office-start year trend  2.683
(2.199)
Immediate predecessor's career-incentive intensity   −4.466
(12.601)
Bartik variable    169.788
(395.636)
Dummy: Central experience     4.806
(10.675)
(5.052)
City fixed effects
Observations 360 362 359 299 362
$R2$ 0.976 0.978 0.979 0.992 0.975

See table 4 for notes.

Various pretrends of urban development.

There is a concern that city leaders with differing career-incentive intensities may be systematically appointed to cities that have undergone different pretrends of economic development. To avoid the confounding effect of such pretrends, we control for the growth rates of various city economic characteristics in our baseline regressions. As a robustness check, we add into our baseline regression (column 1 of table 4) the interactions of the city characteristics in the year just before the leader takes office (population size, GDP per capita, built-up urban land area) with the linear trends of the leader's office-start year. Column 2 of table 5 shows that the main results are very similar after controlling for the effects of pretrends of urban development.

In addition, a given city leader may simply carry on the spatial development policies of their immediate predecessor. If career incentives are correlated among city leaders appointed to the same city, then the estimates shown in table 4 would be biased. Column 3 of table 5 shows that the main results are robust to including the career-incentive intensity of the immediate predecessor.

We also conduct a falsification test by regressing the extent of outward expansion during the office span of the city leader's immediate predecessor on the career-incentive intensity of the city leader. Table A3 in the appendix reports the results. The estimated coefficients on the career-incentive intensity turn out to be negative and insignificant for all four percentile outcomes, contrary to what we found in table 4.

Moreover, changes in national industrial structure may cause the extent of urban outward expansion to differ across cities with different initial industrial compositions. Meanwhile, the career-incentive intensity of a city leader may be correlated with the city's initial industrial structure. To address this concern, we construct a Bartik-type variable that measures the effect of industrial structural change during each leader's office term (Bartik, 1991). For this exercise, we use the subsample of 299 city-leader observations covering those who took office in or after 2004, the year when the variable definitions of the city-level employment data agree with those of the national-level data. The regression results reported in column 4 of table 5 show that the effect of career incentives on outward expansion remains, although the significance level drops due to the small sample size.

Promotion incentives versus competence.

Local leaders promoted to the prefecture level at a younger age may be more competent than other leaders. This raises the concern that the ability or competence of the leaders may be a confounding factor in the estimation. Our model helps us differentiate the effects of career incentives and the effects of ability and competence on urban outward expansion. When a city leader's career incentives drive excessive expansion, the effect on welfare is negative. In the context of our model, however, it is hard to imagine that the ability or competence of city leaders is distortionary in the sense of lowering welfare. Actually, if we assume that local leaders with higher ability levels can manage their cities more efficiently, then a higher ability level will enhance social welfare. People may still argue that a more competent city leader may also have better connections with the central government, which lowers the leader's lobbying costs, all other things being equal. If this is true, given the career-incentive intensity, higher competence would indeed drive excessive urban expansion. Even in this nontypical scenario, however, our empirical results remain robust. As shown in column 5 of table 5, when we control for the central work experience of city leaders as a proxy for central connections as well as educational attainment as a proxy for ability, the effects of both central work experience and education are insignificant, and the urban expansion effect of career incentives is very similar to our baseline effect as reported in column 1 of table 4.15

There is also concern that the upper-level or central government may assign younger and more competent leaders to cities that are targeted for faster growth. To address this concern, we conduct a careful check to see whether there is any systematic correlation between the cities' preexisting economic conditions and the appointed city leaders' personal characteristics (such as age, level, and education). We run separate regressions of the city population, GDP per capita, and built-up urban land area just prior to the leader's office-start year (all in logs), as well as the growth rates of these variables over the two years before the office-start year, on the city leader's start age, start level, and education. As shown in table A4 in the appendix, we find no significant relationship between the pretrends of city characteristics and a city leader's start age, start level or education.

More robustness checks.

Our main results in table 4 are robust to including the age of the provincial leader (and its squared term) as of the city leader's office start year, which addresses the concern that city leaders display more loyalty toward provincial leaders who are older than they are and, in turn, are more favored by those older upper-level superiors in the allocation of resources (in terms of granting land quotas). Adding the economic performance of peer cities in the same province as controls in the regression does not change the main results.

A city leader who has greater career-incentive intensity may stay in office for a longer time. Meanwhile, the leader's tenure length may affect the number of land parcels sold during the leader's office term and their distance from the city center, which in turn influences outward expansion. Our main results remain robust to inclusion of the city leader's total tenure length and its squared term. Also, it is possible for a city leader to draft a new plan upon being reappointed for an additional term after the first five years. We conduct several robustness checks regarding this reappointment issue. The results remain robust.

We also rerun the regressions in table 4, either replacing the predicted career-incentive intensity with the city leader's start age and start level or using the logs of the distance measures as outcome variables. Our main results still stand. Appendix D discusses the details and reports the results of all the above checks.

VI. Welfare Discussions

Is the career-incentive-driven urban expansion in Chinese cities distortionary? Our theory demonstrates that career incentives motivate city leaders to exert effort to expand the city outward and build infrastructure, which boosts output production and attracts more residents. However, incentives that are too strong may lead to expansion beyond the socially optimal level and cause welfare loss. While we have found empirical evidence that local leaders' career incentives are positively related to cities' total industrial output and population (see appendix E for the impacts of career incentives on various economic outcomes), empirically testing whether career incentives lead to excessive expansion is challenging. This is because ideally, “excessive” outward expansion should be measured as total outward expansion minus the benchmark outward expansion that is socially optimal or is permitted by the original land quota. Unfortunately, we can observe neither the socially optimal level nor the land quota data for each city. However, informed by the institutional background, we know that a city's original quota depends on its future population growth as predicted by the economic conditions at the start of each leader's office term. Hence we construct an alternative measure of excessive expansion, that is, the urban land expansion beyond what would be justified by a city's population growth.16

For each city leader, we calculate the built-up urban area necessary to host the population increase in the coming office term of five years, as predicted by the economic conditions that existed at the beginning of the leader's term, under four scenarios.17 In the first scenario, we assume that each city's population will grow at the same rate in the next five years as it did over the two years just before the start of the office term. In the second scenario, we predict the city's population growth rate using the city's total road stock, fixed asset investment, and population size in the year just prior to the office term, as well as the city fixed effects. In both scenarios, we assume that the city's population density remains unchanged at the initial level, and we calculate the city's projected built-up urban area as the projected population divided by the initial population density. In the third and fourth scenarios, we predict the city's population growth in the same way we did in the first and second scenarios, respectively, but we let the population density change. Specifically, we assume that the population density changes at the same rate that it did over the two years just before the office term.

We define excessive expansion to be the gap between the actual built-up urban area at the end of a leader's term and the projected area if the gap is positive; otherwise, the excessive expansion is zero. We find that more than 60% of the city leader cases in our main regression samples have a positive gap. To shed light on the effects of career incentives on the likelihood and extent of excessive expansion, we run a Tobit regression using specification (7) with the outcome variable being the excessive expansion constructed above. The results reported in table 6 show that career incentives are positively correlated with the excessive expansion thus measured.

Table 6.
Excessive Expansion and Career Incentives of City Leaders: Tobit Regressions
Excessive Expansion (sq. km)
Scenario 1 (1)Scenario 2 (2)Scenario 3 (3)Scenario 4 (4)
City leader's career-incentive intensity 57.271*** 43.563*** 77.335*** 55.171***
(0.134) (0.067) (0.106) (0.060)
City fixed effects
Observations 361 356 361 356
Pseudo $R2$ 0.23 0.27 0.34 0.34
Number of positive excessive expansion 324 309 222 213
Excessive Expansion (sq. km)
Scenario 1 (1)Scenario 2 (2)Scenario 3 (3)Scenario 4 (4)
City leader's career-incentive intensity 57.271*** 43.563*** 77.335*** 55.171***
(0.134) (0.067) (0.106) (0.060)
City fixed effects
Observations 361 356 361 356
Pseudo $R2$ 0.23 0.27 0.34 0.34
Number of positive excessive expansion 324 309 222 213

***$p$$<$ 0.01, **$p$$<$ 0.05, *$p$$<$ 0.1. See table 4 for notes on controls.

Building density is another important dimension of spatial development that is relevant to welfare. In China, building density is usually measured by floor-to-area ratio (FAR). Utilizing microlevel data with information on the regulatory FAR for each land parcel, we run land-parcel-level regressions of FAR on city leaders' career incentives while controlling for the effects of transaction time and the land's distance to the city center (see appendix F for details). We find that on average, the career incentives of local leaders are negatively associated with the regulatory residential building density in Chinese cities (see column 1 of table A10 in the appendix). Furthermore, column 2 of table A10 shows that the greater the career-incentive intensity of the city leader is, the smaller the magnitude of the gradient of building density against the distance to the city center becomes. The distance gradient itself is negative, which is consistent with most literature. Hence, one can see that while career incentives drive outward expansion, building density flattens out at the same time, which indicates that the spatial layout of cities governed by high-incentive city leaders is less compact than that of cities under low-incentive ones. The flattening-out of a city may cause lengthier commuting times and a reduction in agglomeration forces. As Harari (2018) showed, less compact urban layouts are associated with lower welfare.18 In this light, we interpret the result on FAR as another piece of suggestive evidence on the welfare impacts of city leaders' incentives.

In summary, we find some suggestive evidence pointing to the distortionary impacts of overly strong incentives of city leaders on spatial expansion. But due to the limitations of the data, a rigorous welfare analysis requires further research.

VII. Conclusion

The spatial aspect of urbanization is essential to consider because it not only shapes the internal structure and the spatial distribution of economic activities within a city, but also affects biodiversity and environmental quality for decades to come. This paper studies the effect of city leaders' career-incentive intensities on urban spatial expansion under China's unique institutional background.

An intriguing question arising from our study concerns the welfare implications of the spatial pattern of urban land development led by city leaders. On one hand, total factor productivity increases when outward expansion creates more land sale revenues and thus channels more money into public infrastructure investment. Furthermore, the higher productivity associated with outward expansion can attract more people from rural areas into industrial and service sectors. On the other hand, commuting costs increase as the geographic footprint of a city expands. Other social costs associated with urban spatial expansion are traffic jams and environmental costs, in addition to inefficiency and negative redistribution effects arising from the undercompensation of peasants for the agricultural land at the city edge. Because of the increased focus of promotion evaluation criteria on jurisdictional economic performance, Chinese city leaders generally downplay those social costs, which causes inefficiency.

Our theoretical model predicts that city leaders with higher career incentives tend to expand urban spatial size farther outward, generating higher output and population. When the incentives are strong enough, there will be excessive outward expansion at the expense of social welfare. Exploiting a large microlevel land data set that features rich variations across cities and over time, we put the model to empirical tests and find results fairly consistent with the theory. Note that due to data limitations, we are unable to perform a comprehensive welfare assessment. We thus regard our empirical results concerning welfare as suggestive evidence.

This paper also creates possibilities for future research in several directions. For instance, it is worth further studying how the pattern of urban land development influences the spatial distribution of economic activities within cities. Baum-Snow et al. (2017) find that road infrastructure facilitates the decentralization of population and manufacturing production in Chinese cities. It would be interesting to examine how urban land development directed by city governments affects the provision of public infrastructure and how this in turn shapes the internal urban structure in terms of the locations of firms, employment, and population. It would also be worth investigating how residential land developments complement industrial land developments in boosting local output and how land developments enhance the formation of edge cities in suburban areas. A recent study shows that as large numbers of firms cluster in industrial parks at the city edge, they can enjoy localization economy benefits and workers can also live in nearby residential developments (Zheng et al., 2017). This implies that the downsides of the outward expansion that we have described may be mitigated by the creation of industrial parks at the edges of many cities.

Notes

1

The calculations are based on China Statistical Yearbooks of various years.

2

From 1990 through 2010, the total urban population of these 200 cities grew at an annual rate of 2.5%. The urban population data is obtained from the 1990 and 2010 Chinese censuses.

3

Regarding the political economy of urban land development in democratic countries, one strand of literature emphasizes that homeowners who care about their own home values are likely to support stricter land use regulations (Brueckner, 1995; Glaeser, Gyourko, & Saks, 2005; Hilber & Robert-Nicoud, 2013; Ortalo-Magné & Prat, 2014; Duranton & Puga, 2015). In addition, some studies argue that local developers may also influence local land use regulations by lobbying for pro-growth policies (Molotch, 1976; Fischel, 2008; Glaeser et al., 2005; Hilber & Robert-Nicoud, 2013), with Solé-Ollé and Viladecans-Marsal (2012) providing some empirical evidence.

4

The historical arable land and population data of China and the United States are from the World Bank's website (data.worldbank.org).

5

China's Land Administration Law (1998) requires that any conversion of farmland to nonagricultural use must be approved by higher-level authorities and must be offset by conversion or reclamation of other land to agricultural use so that the amount of land used for agriculture, adjusted for quality, remains constant (Feng et al., 2015).

6

Floor-to-area ratio (FAR) is the total floor space built on a land parcel divided by the total land area. In order to regulate building density, the city government typically imposes an upper limit on FAR, which specifies the maximum floor space that can be built per unit of land (Brueckner et al., 2017; Cai et al., 2017).

7

This assumption is based on the fact that in China, while cities face a highly elastic supply of workers from surrounding rural areas due to the large urban-rural income gap, city governments can attain their desired population by controlling the immigration of rural workers to a large extent, such as by rationing the temporary residence permits issued to them. At the same time, migration costs across Chinese cities (and especially across provinces) remain high (Tombe & Zhu, 2019; Henderson et al. 2018).

8

In urban China, the city government is the monopoly supplier of city land. It holds auctions to sell long-term leaseholds of land to private developers. Through competitive bidding, these land sales can grab the developers' profits from land development.

9

In China's political system, the likelihood of a politician's promotion hinges on their career incentive when they start office and on the politician's performance during their office term. The two factors complement each other. This is different from what is documented in the literature on Western political regimes (Besley, Persson, & Sturm, 2010; Solé-Ollé & Viladecans-Marsal, 2012).

10

This condition can easily be satisfied under reasonable parameter space.

11

This condition can easily be satisfied under reasonable parameter space.

12

For residential land sold through negotiations, the sale prices were far below the market value. This suggests that such negotiated sales are not intended to create fiscal revenues for local governments. Since local government revenues are at the heart of this paper's story, our residential land transaction data from public auctions are suitable for our research purposes.

13

Goldblatt et al. (2016) develop a methodology to quantify the urban land area in India using Google Earth Engine based on imagery from the U.S. Landsat database.

14

For example, see the article titled “The 59 Years Old Phenomenon” on Epoch Times (https://www.theepochtimes.com/the-59-years-old-phenomenon_1732969.html), and “Why 59-Year-Old Officials Are Being Charged with Corruption in China” on Vision Times (http://www.visiontimes.com/2014/03/01/why-59-year-old-officials-are-being-charged-with-corruption-in-china.html).

15

Xi, Yao, and Zhang (2018) find that career-concerned opportunism is mitigated by capabilities among city leaders in China.

16

We thank one anonymous referee for this suggestion.

17

In the calculation, we assume that at the beginning of each office term, the city leader expects that their term will last for five years, because an official term in China's administrative system is stipulated as five years, although this is not strictly binding.

18

Harari (2008) constructs a disconnection index to measure the degree of city compactness using nighttime light imagery data that delineate the spatially contiguous footprints of cities. Based on this measure, that paper studies the welfare impacts of city shape using a Rosen-Roback framework with free migration.

REFERENCES

Bartik
,
Timothy J.
,
Who Benefits from State and Local Economic Development Policies?
(
Kalamazoo, MI
:
W. E. Upjohn Institute for Employment Research
,
1991
).
Baum-Snow
,
Nathaniel
, “
Did Highways Cause Suburbanization?
Quarterly Journal of Economics
122
(
2007
),
775
805
.
Baum-Snow
,
Nathaniel
,
Loren
Brandt
,
J. Vernon
Henderson
,
Matthew A.
Turner
, and
Qinghua
Zhang
, “
this review
99
(
2017
),
435
448
.
Besley
,
Timothy
,
Torsten
, and
Daniel M.
Sturm
, “
Political Competition, Policy and Growth: Theory and Evidence from the US,
Review of Economic Studies
77
(
2010
),
1329
1352
.
Brown
,
Lester R.
,
Who Will Feed China? Wake-Up Call for a Small Planet.
(
New York
:
Norton
,
1995
).
Brueckner
,
Jan K.
, and
David A.
Fansler
, “
The Economies of Urban Sprawl: Theory and Evidence on the Spatial Sizes of Cities,
this review
65
(
1983
),
479
482
.
Brueckner
,
Jan K.
,
Shihe
Fu
,
Yizhen
Gu
, and
Junfu
Zhang
, “
Measuring the Stringency of Land-Use Regulation and Its Determinants: The Case of China's Building-Height Limits,
this review
99
(
2017
),
663
677
.
Burchfield
,
Marcy
,
Henry G.
Overman
,
Diego
Puga
, and
Matthew A.
Turner
, “
Causes of Sprawl: A Portrait from Space,
Quarterly Journal of Economics
121
(
2006
),
587
633
.
Cai
,
Hongbin
,
J. Vernon
Henderson
, and
Qinghua
Zhang
, “
China's Land Market Auctions: Evidence of Corruption?
RAND Journal of Economics
44
(
2013
),
488
521
.
Cai
,
Hongbin
,
Zhi
Wang
, and
Qinghua
Zhang
, “
To Build above the Limit? Implementation of Land Use Regulations in Urban China,
Journal of Urban Economics
98
(
2017
),
223
233
.
Chen
,
Ting
, and
J. K.-S.
Kung
, “
Do Land Revenue Windfalls Create a Political Resource Curse? Evidence from China,
Journal of Development Economics
123
(
2016
),
86
106
.
Deng
,
Xiangzheng
,
Jikun
Huang
,
Scott
Rozelle
, and
Emi
Uchida
, “
Growth, Population and Industrialization, and Urban Land Expansion of China,
Journal of Urban Economics
63
(
2008
),
96
115
.
DiPasquale
,
D.
, and
W. C.
Wheaton
,
Urban Economics and Real Estate Markets
(
:
Prentice Hall
,
1996
).
Ding
,
Chengri
, “
Policy and Praxis of Land Acquisition in China,
Land Use Policy
24
(
2007
)
1
13
.
Ding
,
Chengri
, and
Gerrit
Knaap
, “Urban Land Policy Reform in China's Transitional Economy” (pp.
9
37
), in
Chengri
Ding
and
Yan
Song
, eds.,
Emerging Land and Housing Markets in China
(
Cambridge, MA
:
Lincoln Institute of Land Policy
,
2005
).
Duranton
,
Gilles
, and
Diego
Puga
, “The Growth of Cities” (pp.
781
853
), in
Philippe
Aghion
and
Steven N.
Durlauf
, eds.,
Handbook of Economic Growth
, vol.
2
(
Amsterdam
:
Elsevier
,
2014
).
Duranton
,
Gilles
, and
Diego
Puga
“Urban Land Use” (pp.
467
560
), in
Gilles
Duranton
,
J. Vernon
Henderson
, and
William S.
Strange
, eds.,
Handbook of Regional Science and Urban Economics
, vol. 5 (
Amsterdam
:
North-Holland
,
2015
).
Fang
,
Hanming
,
Quanlin
Gu
,
Wei
Xiong
, and
Li-An
Zhou
, “
Demystifying the Chinese Housing Boom
,”
NBER working paper
21112
(
2015
).
Feng
,
Juan
,
Erik
Lichtenberg
, and
Chengri
Ding
, “
Balancing Act: Economic Incentives, Administrative Restrictions, and Urban Land Expansion in China,
China Economic Review
36
(
2015
),
184
197
.
Fischel
,
William A.
, “Political Structure and Exclusionary Zoning: Are Small Suburbs the Big Problem?” (pp.
111
136
), in
Gregory K.
Ingram
and
Yu-Hung
Hong
, eds.,
Fiscal Decentralization and Land Policies
(
Cambridge, MA
:
Lincoln Institute of Land Policy
,
2008
).
Glaeser
,
Edward L.
, “
A World of Cities: The Causes and Consequences of Urbanization in Poorer Countries,
Journal of the European Economic Association
12
(
2014
),
1154
1199
.
Glaeser
,
Edward L.
,
Joseph
Gyourko
, and
Raven E.
Saks
, “
Why Have Housing Prices Gone Up?
American Economic Review
95
(
2005
),
329
333
.
Glaeser
,
Edward L.
, and
Matthew E.
Kahn
, “Sprawl and Urban Growth” (pp.
2481
2527
), in
J. V.
Henderson
and
J. F.
Thisse
, eds.,
Handbook of Regional Science and Urban Economics
, vol.
4
(
Amsterdam
:
Elsevier
,
2004
).
Goldblatt
,
Ran
,
Wei
You
,
Gordon
Hanson
, and
Amit K.
Khandelwal
, “
Detecting the Boundaries of Urban Areas in India: A Dataset for Pixel-Based Image Classification in Google Earth Engine,
Remote Sensing
8
(
2016
),
634
.
Guo
,
Xiaolin
, “
Land Expropriation and Rural Conflicts in China,
China Quarterly
166
(
2001
),
422
439
.
Gyourko
,
Joseph
, and
Raven
Molloy
, “Regulation and Housing Supply” (pp.
1289
1337
), in
Gilles
Duranton
,
J. Vernon
Henderson
, and
William
Strange
, eds.,
Handbook of Regional Science and Urban Economics
, vol.
5
(
Amsterdam
:
Elsevier
,
2015
).
Han
,
Li
, and
James Kai-Sing
Kung
, “
Fiscal Incentives and Policy Choices of Local Governments: Evidence from China,
Journal of Development Economics
116
(
2015
),
89
104
.
Harari
,
Mariaflavia
, “
Cities in Bad Shape: Urban Geometry in India,
unpublished manuscript
(
2018
).
Henderson
,
J. Vernon
, “Urbanization and Growth” (pp.
1543
1591
), in
Philippe
Aghion
and
Steven N.
Durlauf
, eds.,
Handbook of Economic Growth
, vol.
1
(
Amsterdam
:
North-Holland
,
2005
).
Henderson
,
J. Vernon
,
Donglin
Su
,
Qinghua
Zhang
, and
Siqi
Zheng
, “
Local Factor Market Distortions in China,
unpublished manuscript
(
2018
).
Hilber
,
Christian A. L.
, and
Frédéric
Robert-Nicoud
, “
On the Origins of Land Use Regulations: Theory and Evidence from US Metro Areas,
Journal of Urban Economics
75
(
2013
),
29
43
.
Kahn
,
Matthew E.
, “
The Environmental Impact of Suburbanization,
Journal of Policy Analysis and Management
19
(
2000
),
569
586
.
Li
,
Hongbin
, and
Li-An
Zhou
, “
Political Turnover and Economic Performance: The Incentive Role of Personnel Control in China,
Journal of Public Economics
89
(
2005
),
1743
1762
.
Lichtenberg
,
Erik
, and
Chengri
Ding
, “
Local Officials as Land Developers: Urban Spatial Expansion in China,
Journal of Urban Economics
66
(
2009
),
57
64
.
McGrath
,
Daniel T.
, “
More Evidence on the Spatial Scale of Cities,
Journal of Urban Economics
58
(
2005
),
1
10
.
Molotch
,
Harvey
, “
The City as a Growth Machine: Toward a Political Economy of Place,
American Journal of Sociology
82
(
1976
),
309
332
.
Newman
,
Peter
, and
Jeffrey
Kenworthy
,
Sustainability and Cities: Overcoming Automobile Dependence
(
Washington, DC
:
Island Press
,
1999
).
Ortalo-Magné
,
François
, and
Andrea
Prat
, “
On the Political Economy of Urban Growth: Homeownership versus Affordability,
American Economic Journal: Microeconomics
6
(
2014
),
154
181
.
Qian
,
Yingyi
, and
Barry R.
Weingast
, “
Federalism as a Commitment to Market Incentives,
Journal of Economic Perspectives
11
(
1997
),
83
92
.
Solé-Ollé
,
Albert
, and
Elisabet
, “
Lobbying, Political Competition, and Local Land Supply: Recent Evidence from Spain,
Journal of Public Economics
96
(
2012
),
10
19
.
Tombe
,
Trevor
, and
Xiaodong
Zhu
, “
Trade, Migration and Productivity: A Quantitative Analysis of China,
American Economic Review
109
(
2019
),
1843
1872
.
Tsai
,
Pi-Han
, “
Fiscal Incentives and Political Budget Cycles in China,
International Tax and Public Finance
23
(
2016
),
1030
1073
.
World Bank and Development Research Center of the State Council
,
Urban China: Toward Efficient, Inclusive, and Sustainable Urbanization
(
Washington DC
:
World Bank
,
2014
).
Xi
,
Tianyang
,
Yang
Yao
, and
Muyang
Zhang
, “
Capability and Opportunism: Evidence from City Officials in China,
Journal of Comparative Economics
46
(
2018
),
1046
1061
.
Xie
,
Zhikui
,
(
Beijing
:
Commercial Press
,
2015
).
(In Chinese)
.
Xu
,
Chenggang
, “
The Fundamental Institutions of China's Reforms and Development,
Journal of Economic Literature
49
(
2011
),
1076
1151
.
Yao
,
Yang
, and
Muyang
Zhang
, “
Subnational Leaders and Economic Growth: Evidence from Chinese Cities,
Journal of Economic Growth
20
(
2015
),
405
436
.
Zheng
,
Siqi
, and
Matthew E.
Kahn
, “
Understanding China's Urban Pollution Dynamics,
Journal of Economic Literature
51
(
2013
),
731
772
.
Zheng
,
Siqi
,
Weizeng
Sun
,
Jianfeng
Wu
, and
Matthew E.
Kahn
, “
The Birth of Edge Cities in China: Measuring the Effects of Industrial Parks Policy,
Journal of Urban Economics
100
(
2017
),
80
103
.

Author notes

A supplemental appendix is available online at http://www.mitpressjournals.org/doi/suppl/10.1162/rest_a_00862.