Political Cycles and the Mix of Industrial and Residential Land Leasing

Abstract

This paper studies how political cycles change the mix of industrial and residential land in urban land leasing. The mixture of different types of land leasing in cities affects urban landscape, resident welfare, and economic sustainability. Using prefecture-level panel data from China and statistical regressions, this paper finds that cities lease out 3% more industrial land, as a percentage of total annual urban land leasing, when their party committee secretaries have been in office for no more than two years. In the same period, they lease out 2% less residential land. This is explained by the strategic behaviors of party committee secretaries to increase their chances of political promotion. Urban land leasing fuels local economic performance and increases the chance of city leaders’ promotion. While the economic benefits of residential land are immediate, those of industrial land cannot be reaped until two years later. This divided timeline results in more aggressive leasing of industrial land early on in party committee secretaries’ service terms, and that of residential land later on. Mayors’ service terms do not have the same effect. This political cycle distorts the temporal and spatial distributions of industrial and residential land in cities, and results in inefficient land use and unstable real estate markets.

1. Introduction

Urban land use is closely related to economic development [1], environmental sustainability [2,3], public health [4,5] and housing affordability [6]. Recognizing the importance of urban land use in the wellbeing of citizens, researchers have devoted much attention to search for the determinants of urban land use [7,8]. Several factors have emerged from the search: population and income growth, transportation modes and costs [9], spatial topology of the landscape [10], land use regulations [11] and climatic conditions [12]. Building on these determinants, researchers have also searched for ways to optimize urban land use. They concluded that urban land use should be fully responsive to the demand, with externalities taken into consideration [7].

In practice, however, urban land use is frequently interfered with by factors other than the demand; politics is found to be one such interfering factor [13,14]. Local governments across the world are in need of financial resources to support their departments and projects; thus, they prefer certain types of land use that contribute more to local tax. This has been found across democratic and non-democratic, developed and developing nations. For instance, U.S. city managers have given favorable zoning ordinances to retail land use and deliberately encouraged their expansion due to their abilities to encourage employment and generate local tax [15,16]. A similar phenomenon was identified in China. City governments encourage industrial land expansion to reap future taxes [17]. Land leased out by negotiation (mostly industrial land) barely contributes to local tax in the same year and the following year of provision, but the contribution almost doubles in the second year after provision and triples in the third [17].

In particular, political cycles are often translated into land use cycles. Politicians strive to win elections and get promoted [18]. In democratic countries, local leaders tend to deliver right before the end of their service terms to win the hearts of voters for the next election; this causes political business and fiscal cycles [19,20]. Similarly, they also court voters with favorable zoning ordinances and popular land use plans. Empirical studies have confirmed that in democratic countries with decentralized elections and party turnovers, urban land use plans are much more subjective to significant shifts right before or after an election. For instance, in the United States, the fate of land use plans and land preservation programs is frequently altered by political elections [21,22]. Similarly, in Canada, politicians charm their voters with popular land use policies right before the elections [23]. In non-democratic nations, while there is no party turnover, there may still be political business cycles [24]. China, for example, is a nation that has the potential to exhibit such a cycle. Local politicians in China do not get promoted for charming their voters, but do so by courting their superior politicians [25]. These local politicians are evaluated by superiors for their performances during their service terms [26]. After 1978, in the post-Mao era, the growth of local economy and fiscal revenue became the central concern of Chinese central government, and thereby important criteria for local politician evaluation [27]. Local politicians striving for promotion have adopted urban land leasing as an important instrument to fuel economic and fiscal growth [28]. As a result, urban land leasing may exhibit political cycles along with politician turnover, though not party turnover. Indeed, a recent empirical study confirmed that the amount of urban land leasing exhibits a political cycle along with the cyclical occurrence of the Provincial Party Congress, a congress that changes provincial leaders [29]. However, the amount of urban land leasing is a relatively crude measurement of local land leasing strategy aligned with political cycles. In fact, industrial and residential land leasing contribute differently to local fiscal and economic growth with bifurcated timelines. While industrial land leasing contributes to local tax and GDP, residential land leasing generates a considerable amount of timely revenue [17]. Moreover, the economic benefits of residential land are immediate, but those of industrial land cannot be reaped until two years later. Therefore, this paper asks the following questions: Do political cycles also prevail in the mix of industrial and residential land leasing in China? And if yes, how large is the effect and what incentivizes local politicians to do so? This paper is the first to ask these questions, and it provides another case of how the political cycle is factored into urban land use patterns in the setting of a developing and authoritarian country—China.

Aside from the politician evaluation system which makes a political cycle possible, China is a particular good setting for this study for two additional reasons. First, as a spacious country experiencing rapid urbanization, China has seen an unprecedented scale of urban land use change [30]. From 1990 to 2000, a third of Chinese territory has experienced land use changes [31]. Second, China’s urban land is not fully competitive, but monopolized by local governments on the supply side. This offers city politicians some degree of autonomy to steer the market power, which may strengthen the relationship between politics and urban land use.

This paper proceeds as follows. Section 2 introduces the institutional background on urban land markets and promotion of city politicians in China. Section 3 describes the methodology. Section 4 introduces the panel dataset of prefecture-level cities in China from 2009 to 2013, and presents descriptive statistics. Section 5 reports the results and carries out robustness tests. Section 6 discusses the results and concludes this paper.

2. Background

2.1. Urban Land Leasing in China

China retains the rights of land use conversion to the state, and exercises this right through city or county governments. As a result, city and county governments become monopolistic suppliers in urban land markets: they and only they can lease out urban land for industrial, commercial, and residential uses.

City governments became enthusiastic about urban land leasing as it generates a remarkable amount of local revenue and fuels economic growth. Since the first adoption of urban land leasing in the late 1980s in Shenzhen, city governments across the nation quickly picked up on this and started to convert agricultural land at an unprecedented rate. Consequentially, agricultural land in China took a big hit: 0.3 million hectares of agricultural land were converted to construction use every year from 1979 to 1999, and this rate quintupled to 1.5 million from 1999 to 2003 [32].

Local governments, nevertheless, are not unrestricted in their power of land conversion. They are constrained by conversion quotas and urban construction land quotas as well as land use plans and urban plans. These quotas were initiated in 1996 by the central government of China in order to protect agricultural land and curtail urban sprawl. During a certain period (usually 5–15 years), the amount of agricultural land conversion and the amount of urban land in each jurisdiction (i.e., province, city, and county) are required to be kept below the assigned quotas. These quotas, if strictly enforced, would significantly slow down the conversion of agricultural land, especially in the coastal regions where urban expansion is the fastest. In addition to quotas, city governments are also constrained by land use plans and urban plans. These plans designate plots to different uses in each jurisdiction within a certain period (usually 15 years); the autonomy of city governments can only be exhibited within the framework of these plans.

In practice, however, as the strict implementation of quotas and plans is not a key criterion in the promotion of city politicians, quotas and plans fall short in enforcement [33,34]. For example, Pudong district in Shanghai had drained its 2020 construction land quotas by 2011 but kept expanding nonetheless. Similar situations happen for one-third of Chinese cities [33]. City governments also tweak land and urban plans in their favor before they publish these plans, and use amendment and exemptions to circumvent these plans after their publication [35].

To summarize, city leaders have much autonomy in the supply of urban land under the current system, albeit constraints such as quotas and plans. As a result, to understand urban land use change in China, it is essential to look into the motivation of city leaders under China’s politician system.

2.2. Promotion of China’s City Politicians

Politicians around the world strive for promotions [18], and city leaders in China are no exception. City politicians in China are led by the secretary of the Committee of the Communist Party (CCP) and the mayor. The CCP secretary is the “boss” of the city and makes major decisions. The mayor is the “manager” of the city and runs day-to-day businesses. The political future of the CCP secretary and the mayor hinges on their evaluations against a set of criteria [25]. These criteria assess politicians from various aspects including morality, capability, diligence, achievements [36] and honesty. Among these criteria, “achievements” links city leaders’ political career to local economic development.

The complete list of the evaluation criteria is not public information, but the link between political career and local economic performance has been empirically established [25,37]. The growth of GDP and fiscal revenue are two key indicators identified by prior studies that positively affect local politicians’ career outcomes. Since 1978, China’s politicians have shifted their attention towards economic development and modernization. The economic development of local jurisdictions, measured by the growth of GDP, became tightly attached to the promotion of local politicians [26,37,38]. Fiscal revenue became another important factor after the fiscal decentralization in the 1990s. The fiscal decentralization made the fiscal revenue of China’s central government more reliant on local economic activities; for certain taxes, the central government extracts a portion out of local government revenues [36,39,40]. Thus, local government revenues became closely related to their revenue contribution to the central government and a key factor in politician promotion. Since then, career-oriented local politicians have endeavored to accelerate the growth of local GDP and fiscal revenue [41,42]. They did so by supporting local enterprises in the 1980s, developing large-scale economic zones after the 1990s, and reaping revenue from residential land leasing after the 2000s. The latter two strategies involve the leasing of industrial and residential land, to which this paper devotes its attention.

After the 1980s, local politicians started to take advantage of their key asset—land—to advance their political careers [17,43]. They have adopted a sophisticated strategy for land leasing: on the one hand, they tend to lease out abundant industrial land at low prices to generate stable GDP and tax flows; on the other, they limit the supply of residential land to maximize fiscal revenues [43]. City politicians also fine-tune the strategy along their tenures. City politicians, by law, serve a five-year term and can be promoted or demoted at or before the end of the term. The turnover of city CCP secretaries was frequent, and most CCP secretaries and mayors only serve four years in a given term. For example, researchers showed that in 2001–2009, the average turnover cycle for city CCP secretaries is 3.7 years [44]. Our sample (463 CCP secretaries who served in 242 prefecture cities during 2009 to 2013) echoes their finding; the average turnover cycle is 3.8 years. Figure 1 further shows the distribution of the sample CCP secretaries’ tenure before they ended their position. Nearly half of the sample CCP secretaries ended their service terms by the fourth year; more than 25% finished their whole five-year terms. Similarly, our sample of 550 mayors has an average turnover cycle of 3.3 years. With approximately three to four years of service term in mind, CCP secretaries and mayors can sensibly optimize economic strategies with a relatively predictable time horizon [45]. They are found to align public spending along the five-year tenure to score better economic outcomes [24]. Similarly, they can strategically align land leasing with their tenure to strive for a brighter career outcome. In fact, some media coverage reveals that industrial development is prioritized at the start of the service term of local politicians.

2.3. Land Leasing Strategies of Chinese Local Governments

Prior studies have found that Chinese local governments adopted specific strategies for leasing industrial, commercial and residential land. They tend to negotiate one-on-one with potential land users to lease out industrial land at relatively low prices. In contrast, they introduce more competition using methods such as bidding and auction in the leasing of commercial and residential land to yield high prices. Empirical studies have found that in 2002, 68% of urban land leased out by negotiation was industrial land, while 83% leased out through auction was commercial and residential land [17,46]. This mixed strategy is explained by the dual goals of local governments to (1) reap revenue within a short time frame with commercial and residential land leasing; and (2) attract external investment, encourage local employment and secure long-term local tax income through industrial land leasing [17]. Some studies further explained these dual goals at a deeper level [28,47]. They argue that it is the economic and political incentives that drive local politicians to deliberately chase these goals. On the one hand, local government revenue has been largely sustained by the revenue from commercial and residential land leasing since the late 1990s. On the other, the fiscal revenue and the external investments attracted by urban land leasing can promote local economy and help with the advancement of city leaders’ political careers.

The general strategy described above slightly differs across regions and city leaders. Cities with lower productivity tend to adopt less market-oriented methods to lease out land, because they have little leverage to attract external investments other than offering land at lower prices; western cities in China fall into this category [48]. Younger city leaders are found to lease out a larger amount of land because they are promising stars in the political job market [49], and aggressive urban land leasing can help them get promoted. None of these studies, however, have pointed out the alignment of urban industrial and residential land leasing strategy with the tenure of city leaders. This paper fills the gap.

3. Methodology

3.1. The Relationship between Land Leasing Patterns and CCP Secretaries’ Tenures

This paper examines three sets of relationships: the relationship between the mix of industrial–residential land leasing and CCP secretaries’ tenures, the relationship between industrial and residential land leasing and local economic performances, and the relationship between local economic performances and CCP secretaries’ political promotion. Note that while we focus the description on CCP secretaries, we have in fact done the same analyses for the tenures of mayors. The first relationship is studied with the following equation.

$$S_{itk}={\alpha }_{1}T{N}_{it}+{\alpha }_2{X}_{it-1}+{\alpha }_3{Z}_{it}+a_{ik}+{\epsilon }_{itk}$$

(1)

where $S_{itk}$ denotes the share of land use type k that is leased out, as a percentage of total urban land leasing in city i, year t. $T{N}_{it}$ is a dummy variable indicating whether the CCP secretary’s tenure up to year t in city i is no longer than two years. We use two years as the cut-off because the average length of a turnover cycle of Chinese city CCP secretaries, as well as that of our sample, is approximately four years. The two-year cut-off thus divides this cycle exactly in half and enables a comparison between the earlier and later periods. $X_{it-1}$ and $Z_{it}$denote city characteristics and individual characteristics of the CCP secretary who serves in city i year t, we lag one year for $X_{it-1}$ to avoid reverse causality. $a_{ik}$ denotes city fixed effects and ${\epsilon }_{itk}$ denotes the random error. Year fixed effects are excluded due to their high correlation with $T{N}_{it}$ (F-statistics over 25 and significant at the 1% level), but time-related shocks are accounted for with time trend polynomials in the robustness tests.

A conceptual framework that combines the standard urban economic theory [9,50] and the model of Lichtenberg and Ding [51] is adopted to guide the choice of control variables. The former states that urban land use is affected by the relative value of agriculture versus that of urban land, population, income and transportation cost, while the latter adds the fiscal pressure of local governments to the model. Following this basic framework, for $X_{it-1}$, we include variables affecting both the supply and demand of industrial and residential land. For demand-side variables, we include determinants of urban area in standard urban economic theory [9,50]: total population (POP), measured by the number of registered population in a prefecture city at the end of the year, and annual wage (WAGE) measuring income. They boost the demand for dwellings and residential land. We also control for various types of investments, including foreign direct investment (FDI), fixed assets investment (FAI) and infrastructure measured by road coverage (ROAD). These investments either complement or substitute industrial land in urban production function, and therefore affect urban land value and demand. Road coverage also measures transportation cost in the standard urban economic theory. Finally, we control for the economic structure of a city, measured by the share of GDP generated from the agricultural sector (AGDP) and the manufacturing sector (MGDP), which also affects the relative value and demand for urban land, as well as the demand specifically for industrial land [46]. Note that by controlling both AGDP and MGDP, we also implicitly control for the share of GDP from the service sector, which is associated with the demand for residential land.

For the supply side, we control for two economic variables: fiscal condition (FIS), measured by the fiscal income net of expenditure, and GDP per capita (PGDP). We have also adopted an alternative measurement of fiscal gap—the ratio of fiscal expenditure to income—and eventually found the results to be robust with that of the former measurement. Fiscal hardship triggers residential land leasing as it generates a considerable amount of fiscal revenue and helps local authorities sail through the financial difficulty. At the same time, a depressing economy leads to an increased supply of industrial land; this illustrates the local government’s efforts to attract firms and boost the economy [17]. We also control for the institutional constraint on land leasing, measured by construction land quotas (QUOTA). Construction land quotas are by law the maximum amount of construction land that can be provided during a certain period (usually 5–15 years) within a certain adminsitrative unit. The constraint of quotas, when binding, may change the behaviors of land leasing. A tight quota may break the possibility for a CCP secretary to achieve high growths of GDP and fiscal revenue. As a result, they may focus more on one goal and give up the other, and change land leasing patterns. As mentioned above, in reality, land quotas are not a key criterion in CCP secretaries’ promotion and are therefore loosely enforced. Their effectiveness on changing land leasing patterns remains an empirical question. The variable QUOTA is measured by the average annual construction land quotas (in hectares) imposed on prefecture cities by their provinces during the study period. The study period (2009–2013) spans across two planning periods, 2006–2010 and 2011–2020. We calculate the average annual quotas for 2009 and 2010 by dividing the quotas during the first planing period by five. Similarly, the average annual quotas for 2011–2013 are obtained by dividing the quotas during 2011–2020 by ten. After obtaining the average quotas for each of the five studied years, it is easy to calculate their average.

$Z_{it}$ is beyond the basic conceptual framework of urban economic theory; it is what we add to the model. We believe that aside from the fiscal pressure, the decision of local governments is also heavily affected by their leaders; as a result, the characteristics of these leaders may be associated with urban land leasing strategy. We include demographic variables such as age and age square, sex and ethnic group. Other than that, past education experiences and work trajectories are also accounted for. We control for education experiences with dummies of education levels, academic disciplines and studying experiences overseas. Work trajectories may suffer from selection bias, as the central government may arrange promising officials to specific positions to gain valuable experience for promotion, while futureless officials can be assigned to sinecures. To address this issue, we control for a rich set of variables related to work experiences, including service term (in months) and its square, years since first started work, years since first joined CCP, a city native (i.e., holding a position in the city where one was born), and work experience in specific institutions. Following prior studies [49,52,53,54,55,56], we include dummies indicating whether the official has worked in the central government, provincial governments, universities or research institutions, state-owned enterprises, China Communist Youth League, the organizational department, overseas institutions, the economic sector, and secretaries in previous terms. In fact, this list of control variables exhausts all variables used in prior studies. Thus, it should sufficiently account for all observable differences in CCP secretaries. Nonetheless, CCP secretaries may also have unobservable differences in their abilities. To deal with that, we add CCP secretary fixed effects in an alternative specification.

If CCP secretaries align land leasing with their tenures, we expect ${\alpha }_1<0$ for industrial land and ${\alpha }_1>0$ for residential land. We test the robustness across five different specifications. First, as mentioned above, we include flexible time trend polynomials up to the order of three, following prior studies [29,57]. These polynomials mitigate the concern that the influx of government stimulus in 2009 drives the greater share of industrial land in total urban land leasing. Second, as an alternative approach to deal with the complication of the government stimulus in 2009, we focus the analysis on cities with CCP secretaries serving the later part of their tenures in 2009 and 2010. For these CCP secretaries, the political cycle of their tenure does not coincide with the economic cycle induced by the 2009 stimulus. Third, we introduce CCP secretary fixed effects to control for time-invariance unobservable differences across CCP secretaries. Fourth, we probe into an alternative explanation of the political cycles. Political cycles in urban land leasing may be induced by other political events rather than the rollout of tenures. A prior study found that the cyclical occurrence of the Provincial Party Congress (PPC), instead of the tenures, explains the fluctuations of how much land is leased out across years [29]. To account for this possibility, we include a set of dummy variables indicating the year before, the year of, and the year after PPC. If PPC rather than tenure explains the mix of industrial and residential land leasing, we expect coefficients associated with these dummies to be significant and ${\alpha }_1$ insignificant. Last, we test the robustness against a more flexible specification as follows.

$$SubLan{d}_{itk}={\alpha }_{1}T{N}_{it}+{\alpha }_2{X}_{it-1}+{\alpha }_3{Z}_{it}+{\alpha }_{4}Lan{d}_{it}+a_{ik}+{\epsilon }_{itk}$$

(2)

where $SubLan{d}_{itk}$ stands for the amount of leasing of land type k in city i year t, and we control for the total amount of land leasing, $Lan{d}_{it}$, on the right-hand side. This is more flexible than Equation (1) as ${\alpha }_4$ is free to take any value.

3.2. The Divided Timelines of the Rollout of Benefits

This section tests whether the benefits of industrial and residential land leasing roll out with divided timelines. To be specific, we expect residential land leasing brings in immediate revenue growth and immediate or shortly delayed GDP growth. If this is not the case, city officials would have no reason to favor residential land when their service terms are ending. In contrast, industrial land leasing spurs the growth of GDP and fiscal revenue with a lag of approximately two to four years. With shorter lags, CCP secretaries would keep favoring industrial land in the latter half of their service terms, while with more extended lags, they would have given up industrial land completely and preferred residential land early on. These predictions are tested with Equation (3).

$$\begin{gathered} G_{it}={\beta }_1{S}_{it,k=m}+{\beta }_2{S}_{it-1,k=m}+{\beta }_3{S}_{it-2,k=m}+{\beta }_4{S}_{it,k=r} \\ +{\beta }_5{S}_{it-1,k=r}+{\beta }_6{S}_{it-2,k=r}+{\beta }_7{W}_{it}+a_i+{\epsilon }_{it} \end{gathered}$$

(3)

where $G_{it}$ denotes the growth rate of per capita GDP or fiscal revenue for city i year t. $S_{it,k=m}$ denotes the share of land type k leased out in city i year t, as a percentage of total land leasing. k = m indicates industrial land and k = r denotes residential land. Due to the limited length of the data, we can only include time lags up to two years. $W_{it}$ denotes city characteristics, $a_i$ denotes city fixed effects and ${\epsilon }_{it}$ denotes the random error.

For $W_{it}$, we only include land quotas due to their exogeneity, while other variables in $X_{it}$ are likely by-products of past land leasing. To maintain a causal interpretation of $\beta$, we exclude them from Equation (3). We also include lags of $W_{it}$ in an extended specification, and test for robustness against time trend polynomials and CCP secretary fixed effects. As mentioned above, we expect ${\beta }_1=0$, ${\beta }_2=0$, ${\beta }_3\ge$0, and ${\beta }_6=0$. For fiscal revenue, we expect ${\beta }_4>0$ and ${\beta }_5=0$, while for GDP, we expect ${\beta }_4\ge 0$ and ${\beta }_5\ge 0$.

3.3. Economic Performance and Political Promotion

Finally, we test the relationship between local economic performance and the career outcomes of city CCP secretaries with the following equation.

$$P_{it}={\gamma }_{0}Avg{G}_{itj=g}+{\theta }_{0}Avg{G}_{itj=f}+{\pi }_1{X}_{it}+{\pi }_2{Z}_{it}+{\pi }_{3}T{N}_{it}+a_i+{\epsilon }_{in}$$

(4)

where $P_{it}$ denotes whether a CCP secretary in city i year t is promoted. A CCP secretary is considered promoted if he or she: (a) kept the same level of position but was relocated to a higher-rank city (e.g., capital city); (b) was elected into the provincial party committee; (c) was assigned a provincial-level or sub-provincial-level position; or (d) was elected into the central government. $Avg{G}_{itj=g}$ and $Avg{G}_{itj=f}$ denote the average growth rates of per capita GDP and fiscal revenue, respectively, from the beginning of a CCP secretary’s service term until year t. ${\epsilon }_{in}$denotes the random error. We expect ${\gamma }_0>0$ and ${\theta }_0>0$.

4. Data and Descriptive Statistics

4.1. Data

This paper uses panel data of 242 prefecture cities, 465 CCP secretaries and 550 mayors in China from 2009 to 2013. While a city has many politicians, we focus primarily on CCP secretaries as they are the “boss” of cities and make major decisions. From interviews with government officials responsible for land leasing in multiple cities, news and anecdotes, we have learned that CCP secretaries play a significant role in determining land leasing in their jurisdictions. Some directly serve as the director the Land Banking Center, the governmental unit that prepares the plans for and implement land leasing. Others have the authority to approve or disapprove land leasing plans with their discretion. Many even negotiate with the higher-level government to develop more industrial parks, especially during their earlier years in office. A CCP secretary of Kaiping City has asked the Kaiping Land Resource Bureau to illegally convert farmland into industrial land in his first year in office. During his first three years in office (2003–2005), Kaiping expanded its industrial land by 1986 ha and developed several industrial parks [58,59]. Admittedly, mayors may also play an important role in land leasing, as they are the ones that run daily businesses of cities. It would be hard to say for sure whether it is the CCP secretary or the mayor that is more important in determining urban land leasing strategy. On the one hand, the CCP secretary has the ultimate decision right; whether he or she exercises this right in land leasing is at his or her own discretion. On the other hand, the mayor runs the city and is more likely to be directly involved with both the contemplation of the land leasing strategy and its implementation. Thus, it is an empirical question whether the tenure of the CCP secretary or that of the mayor would more significantly affect the mix of urban land leasing. We therefore test both, and the results show that only the tenures of CCP secretaries significantly affect the mix of industrial and residential land leasing. Therefore, empirically, in the 242 sample prefecture cities and the studied time frame (2009–2013), we find that only CCP secretaries’ promotion incentives stimulate a cycle in urban land leasing, and we thus focus our attention mainly to describe our findings for the CCP secretaries in the parts below.

The data come from five sources. Land leasing data, including the amount of annual urban land leasing, and industrial and residential land leasing, are obtained from the China Land and Resources Almanac. The China Land and Resources Almanac is the official statistical yearbook that documents land- and resource-related data. It is published annually and is by far the most comprehensive source for land-related data in China. Construction land quotas at the prefecture city level are compiled from provincial land use master plans from 2006 to 2020, obtained from official websites of the Ministry of National Resources of the People’s Republic of China, provincial governments and provincial Departments of Land and Resources. City characteristics are obtained from the China City Statistical Yearbook. The China City Statistical Yearbook is one of the most comprehensive statistical yearbooks that report economic data at the prefecture city level. CCP secretaries and mayors’ characteristics are collected from multiple online sources, including, websites of local mainstream media, and the official websites of local governments. We have collected all variables used in prior studies concerning city politicians and obtained complete records for 463 (out of 465) CCP secretaries and 550 mayors. We exclude prefecture cities whose CCP secretaries hold sub-provincial-level positions and whose CCP secretaries serve more than one service term. These CCP secretaries are likely on a different career trajectory than the rest. Aside from collecting quantitative data from official sources, we have also conducted interviews, gathered news and anecdotes to familiarize ourselves with the actual procedures of land leasing, the roles of mayors and CCP secretaries in these processes and the operation of politician evaluations. We have conducted face-to-face interviews with government officials in charge of city politician evaluations in Zhejiang Province in October 2015, and leaders of the Land Resource and Planning Bureau in Wuhan City in March 2016. These officials are chosen simply because we have connections with them and therefore have their consent to talk. We do not claim that these interviews can represent the situations across the nation, but simply use them as some background contextual knowledge to help develop the ideas and interpret the results.

Table 1. Descriptive statistics.

Variable Name	Variable Description and Units	Mean	Standard Deviation	Minimum	Maximum	Number of Observations
The share of industrial land	Industrial land leasing/total land leasing (%)	52.594	16.017	5.276	99.852	1197
The share of residential land	Residential land leasing/total land leasing (%)	34.231	13.579	0.052	89.274	1197
TN	Whether the CCP secretary is in the first two years of his/her service term (yes = 1)	0.455	0.498	0	1	1308
X:	City characteristics
POP	Total population (person)	391.521	225.891	18.59	1238.5	1187
WAGE	Annual wage (10,000 yuan = 1450 dollars in March 2017)	1.487	1.814	0.0004	32.063	1186
FDI	Foreign direct investment (million dollars)	315.579	477.654	0.25	3369.380	1133
FAI	Fixed assets investment (billion yuan)	63.072	47.684	2.111	304.392	1198
ROAD	Coverage of road (10,000 square meters)	879.023	685.493	14	3890	1193
AGPD	The share of GDP generated from the agricultural sector (%)	15.257	8.152	0.33	49.89	1198
MGDP	The share of GDP generated from the manufacturing sector (%)	44.569	18.773	0.17	90.97	1198
FIS	Piscal income-expenditure (million yuan)	−74.122	69.484	−1871.428	24.418	1198
PGDP	Per capita GDP (ten thousand yuan)	2.913	2.037	0.360	18.268	1194
Quota	Construction land quotas (hectare)	873.152	403.180	126	2802.544	1218
Z:	Politician characteristics
Age	Age of politician	52.506	3.606	41	61	1413
Age2	Age square	2769.878	374.493	1681	3721	1413
Male	Male = 1, female = 0	0.960	0.197	0	1	1417
Minority	Minority ethnics = 1, Han = 0	0.067	0.249	0	1	1413
Overseas study experience	Yes = 1	0.140	0.348	0	1	1417
Native	Politician born in the same city where he/she holds the position = 1	0.025	0.157	0	1	1416
Education:
Bachelor	Yes = 1	0.168	0.374	0	1	1417
Master	Yes = 1	0.607	0.489	0	1	1417
Doctor	Yes = 1	0.190	0.392	0	1	1417
Academic discipline:
Science, technology, agriculture, and medicine	Yes = 1	0.390	0.488	0	1	1417
Economics and business	Yes = 1	0.695	0.461	0	1	1417
Literature		0.324	0.468	0	1	1417
Work experience:
Central government	Worked in the central government (Yes = 1)	0.079	0.270	0	1	1417
Provincial governments	Worked in provincial governments (Yes = 1)	0.626	0.484	0	1	1417
University or research institute	Worked in a university or research institute (Yes = 1)	0.163	0.370	0	1	1417
State-owned enterprises	Worked as a leader of state-owned enterprises (Yes = 1)	0.344	0.475	0	1	1417
China Communist Youth League	Worked in the China Communist Youth League (Yes = 1)	0.333	0.471	0	1	1417
Organizational department	Worked in the organizational department (Yes = 1)	0.238	0.490	0	1	1417
Secretary	Worked as a secretary (Yes = 1)	0.601	0.489	0	1	1417

presents descriptive statistics of key variables.

The short study period is a concern if this period is atypical. To be specific, two things can be atypical about this particular period. First, after the 2008 recession, China’s central government has invested 4000 billion RMB (about 600 billion USD at that time) to fuel the real economy. This likely leads to massive industrial land leasing in 2009 and 2010, followed by an ease-off. If most CCP secretaries in the sample serve their first two years in 2009 and 2010, then mechanically we would find cities lease out more industrial land during early years of CCP secretaries’ tenure. This concern is relevant as in the sample, 71% CCP secretaries are in their first two years of service term in 2010, compared to 54% in 2011, 46% in 2013, 40% in 2009 and 35% in 2012. We adopt two approaches to deal with this concern. The first is to control for flexible time trend polynomials. If the investment and the resulted industrial land leasing fade in and out smoothly over time, their effects should be captured by the time trend polynomials rather than the politician’s tenure. The second approach is to focus the analysis on cities with CCP secretaries that started their service term on or before 2007. For these cities, their CCP secretaries were in the later three years of their service terms in 2009 and 2010, and their (new) secretaries were in the early years of their tenures during the ease-off period, 2012 to 2013. Thus, if politicians’ tenure rather than the 4000 RMB government stimulus affects the mix of industrial and residential land leasing, we expect the same results in the full sample versus in this subsample. We do the same for mayors.

Second, politician promotion during the study period may be different from other years. While we do not have land leasing data beyond the study period, we do have politician data in earlier years. We find that the probabilities of promotion and demotion for CCP secretaries during the study period are not significantly different from those during the previous five years (2004–2008). The same can be said for the impact of economic performance on CCP secretaries’ career outcomes. Similar conclusions hold for mayors’ promotion records. Thus, focusing on this relatively short period does not jeopardize this paper’s ability to draw more generalized conclusions.

4.2. Industrial and Residential Land Leasing Across Cities

The mix of industrial and residential land leasing varies across cities, as shown in Figure 2. While 75% of the 242 cities leased out more industrial land, the rest did the opposite. Among those that supplied more industrial (residential) land, variations are also impressive: the industrial-to-residential land leasing ratio in cities ranges from less than a half to more than six. We conclude that the patterns of land leasing are more heterogeneous than universal, and the determinants of these patterns are worthy of exploration.

4.3. Industrial and Residential Land Leasing Across CCP Secretaries’ Tenures

The explanation for the heterogeneous patterns we put forth is that CCP secretaries align land leasing with their tenures. Figure 3 plots the share of industrial land in total urban land leasing against the tenures of CCP secretaries. The share of industrial land gradually falls with the rollout of CCP secretaries’ tenures: over the 60 month (full) service term, the share of industrial land leasing decreases from 53% to 50%. This contrasts with the gradual increase of the share of residential land leasing in Figure 4. These patterns are less apparent for the service term of mayors.

5. Results

5.1. CCP Secretaries’ Tenure and the Mix of Industrial and Residential Land Leasing

Table 2. The share of industrial land leasing (in total land leasing) and the tenure of CCP secretaries.

	The Share of Industrial Land
	(1)	(2)	(3)
TN	−2.028 ** (0.837)	−1.948 ** (0.913)	−3.040 *** (1.056)
Xt-1:
FIS		−0.002 (0.003)	−0.0005 (0.003)
Quota		−0.006 ** (0.003)	−0.007 ** (0.003)
POP		0.021 (0.014)	0.022 (0.014)
AGPD		−0.268 (0.296)	−0.134 (0.306)
MGDP		0.048 * (0.028)	0.060 ** (0.030)
PGDP		0.150 (0.932)	0.457 (1.089)
WAGE		−0.109 (0.229)	−0.029 (0.241)
FDI		−0.001 (0.003)	0.0009 (0.003)
FAI		−0.072 ** (0.035)	−0.049 (0.039)
ROAD		0.001 (0.002)	−0.0003 (0.002)
Zt	No	No	Yes
F	5.87 **	3.41 ***	1.72 ***
R2	0.007	0.042	0.097
Number of observations	1081	870	807

City fixed effects are included. Standard errors clustered by cities in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively. Note:$X_{it-1}$ and $Z_{it}$ denote city characteristics (include the variables from FIS to ROAD) and individual characteristics of the CCP secretary who serves in city i year t. Corresponding to Equation (1).

and

Table 3. The share of residential land leasing (in total land leasing) and the tenure of CCP secretaries.

	The Share of Residential Land
	(1)	(2)	(3)
TN	1.518 ** (0.729)	1.519 * (0.790)	2.223 ** (0.932)
Xt-1:
FIS		0.004 (0.004)	0005 (0.003)
Quota		0.005 ** (0.003)	0.006 ** (0.003)
POP		0.0008 (0.014)	−0.00006 (0.015)
AGPD		0.310 (0.319)	0.236 (0.325)
MGDP		−0.045 * (0.027)	−0.053 * (0.029)
PGDP		−1.043 (0.857)	−1.308 (0.934)
WAGE		−0.021 (0.249)	−0.079 (0.274)
FDI		0.0008 (0.003)	−0.0002 (0.003)
FAI		0.048 (0.031)	0.030 (0.034)
ROAD		−0.004 (0.002)	−0.003 (0.002)
Zt	No	No	Yes
F	4.33 **	1.82 **	1.26
R2	0.005	0.030	0.072
Number of observations	1081	870	807

City fixed effects are included. Standard errors clustered by cities in parentheses. *, ** indicate significance at the 10%, 5% levels, respectively.

show the results of Equation (1) for industrial and residential land, respectively. Column (1) includes only the tenure dummy and city fixed effects. The share of industrial land in total annual urban land leasing is 2% greater when a CCP secretary is in office for no more than two years, compared to it during later years; this effect is statistically significant at the 5% level. In contrast, with the extension of the tenure, the share of residential land in total land leasing decreases by 1.5%. Columns (2) and (3) add city characteristics and CCP secretary characteristics, and the coefficients either remain the same or enlarge. With the full specification (in Column (3)), the share of industrial land in total urban land leasing is 3% greater and that of residential land is 2% smaller when a CCP secretary is in office for no more than two years. The same analysis for mayors’ tenure does not show a signifciant relationship between tenure and the mix of industrial and residential land leasing.

Aside from the tenure, land quota is the only supply-side variable that significantly affects the mix of industrial and residential land leasing. A tighter quota leads to a greater share of industrial land and a smaller share of residential land in total land leasing. This implies that land quotas, though insufficiently enforced, can still affect land leasing patterns. However loosely enforced, land quotas are still a constraint on local land leasing behavior, and may have some effect, though not as strong as intended. Here, the effect exists but the maginudes are very small. A hundred hectares larger quota (1/8 of the mean, and 1/4 of the standard deviation) only changes the share of industrial and residential land leasing by a magnitude of 0.5–0.7%. The finding that quotas change urban land leasing strategy is consistent with the findings of a prior study [60]. Moreover, the result indicates that when a tight quota breaks the possibility of achieving both GDP and revenue growth, longer term GDP growth is the top priority of CCP secretaries. This echoes prior studies that found GDP to be the prime determinant of political career in China [37,41,49]. The signs of coefficients on per capita GDP and fiscal condition are both as expected, but insignificant at the 10% level.

Two demand-side variables are statistically significant. The share of GDP generated by the manufacturing sector is significantly positively associated with the percentage of industrial land and negatively with that of residential land. This is sensible, as an economy relying on manufacturing likely requires more industrial land and less of other land types. Another significant variable is fixed assets investment, which discourages industrial land leasing, implying a substitution effect. The signs of the share of GDP generated by the agricultural sector are as expected, but the coefficients are insignificant. We also find the impacts of FDI, road coverage and wages to be both small and insignificant.

We test the robustness in

Table 4. The shares of industrial and residential land leasing (in total land leasing) and the tenure with flexible time trend.

	The Share of Industrial Land					The Share of Residential Land
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
TN	−3.005 *** (1.041)	−2.892 ** (1.039)	−2.101 * (1.106)	−3.119 * (1.618)	−2.429 * (1.359)	2.174 ** (0.918)	1.914 ** (0.915)	1.384 (0.940)	3.177 ** (1.379)	2.416 ** (1.218)
t	Yes	Yes	Yes	No	No	Yes	Yes	Yes	No	No
t2	No	Yes	Yes	No	No	No	Yes	Yes	No	No
t3	No	No	Yes	No	No	No	No	Yes	No	No
CCP secretary fixed effects	No	No	No	No	Yes	No	No	No	No	Yes
Xt-1	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Zt	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
F	1.76 ***	1.74 ***	1.98 ***		1.41 ***	1.43 **	1.57 **	1.69 ***		1.23 *
R2	0.101	0.103	0.118	0.124	0.267	0.084	0.094	0.103	0.127	0.240
Number of observations	807	807	807	488	807	807	807	807	488	807

Note: t represent time trend in year, t² means t square, t³ means t power of 3. City fixed effects are included. Standard errors clustered by cities in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

,

Table 5. The shares of industrial and residential land leasing (in total land leasing) and the tenure with PPC effects.

	The Share of Industrial Land		The Share of Residential Land
	(1)		(2)
TN	−2.869 **	(1.058)	2.120 **	(0.933)
The year before the PPC	−0.275	(1.280)	0.583	(1.211)
The PPC year	1.796	(1.452)	−0.567	(1.279)
The year after the PPC	0.943	(1.296)	−0.511	(1.094)
Xt-1	Yes		Yes
Zt	Yes		Yes
F	1.65 ***		1.18
R2	0.101		0.074
Number of observations	807		807

City fixed effects are included. Standard errors clustered by cities in parentheses ** and *** indicate significance at the 5% and 1% levels, respectively.

and

Table 6. The amounts of industrial and residential land leasing (in total land leasing) and the tenure.

	Industrial Land		Residential Land
	(1)		(2)
TN	−50.200 ***	(16.219)	25.109 **	(11.668)
Land	0.169 **	(0.062)	0.101 ***	(0.027)
Xt-1	Yes		Yes
Zt	Yes		Yes
F	8.66 ***		9.03 ***
R2	0.356		0.365
Number of observations	808		809

City fixed effects are included. Standard errors clustered by cities in parentheses ** and *** indicate significance at the 5% and 1% levels, respectively.

. In Table 4, we include flexible time trend polynomials up to the order of three in Columns (1)–(3) and (5)–(7). The results are robust except in Column (7), where the significance wanes. In Columns (4) and (8), we focus on cities with CCP secretaries serving the later part of their service terms in 2009 and 2010. We find the main results robust with this subsample, and conclude that the complication of the government stimulus during 2009 and 2010 is not a major concern. We further introduce CCP secretary fixed effects in Columns (5) and (10) to account for time-invariant unobservable CCP secretary characteristics and the results remain similar. Table 5 tests the results against the political cycle induced by PPC. We find that none of the PPC-related dummies shows a significant effect, while the coefficients of the tenure remain essentially the same. Table 6 estimates Equation (2). Fifty hectares more industrial land and 25 ha less residential land are leased out during the first two years of CCP secretaries’ tenures compared to later years. These magnitudes are close to what we obtain in Table 2 and Table 3. To put these numbers into perspective, 25 ha less residential land per year means a reduction of approximately 16 residential complexes in the first two years of a CCP secretary’s tenure. Fifty hectares of industrial land per year translates to approximately 10% of a national- or provincial-level industrial park and 53 industrial enterprises, each with annual sales more than 500 million RMB. Note that while the direction and significance of the results are largely robust, the numerical coefficients 3% and 2% are not completely stable across models. Therefore, we caution against the over-interpretation of these magnitudes; they are only suggestive.

5.2. The Effects of Industrial and Residential Land Leasing on Economic Performance

Table 7. Land leasing and economic performance.

	Growth Rate of per Capita GDP			Growth Rate of per Capita Revenue
	(1)	(2)	(3)	(4)	(5)	(6)
$S_{it,k=m}$	0.002 * (0.001)	0.003 ** (0.001)	0.003 * (0.001)	0.004 (0.003)	0.006 (0.005)	0.005 (0.005)
$S_{it-1,k=m}$	0.0009 (0.001)	0.001 (0.001)	0.001 (0.001)	0.002 (0.003)	0.003 (0.003)	0.004 (0.003)
$S_{it-2,k=m}$	0.002 * (0.001)	0.002 * (0.001)	0.002 * (0.001)	0.002 (0.002)	0.002 (0.002)	0.002 (0.002)
$S_{it,k=r}$	0.004 *** (0.002)	0.005 *** (0.001)	0.004 *** (0.002)	0.007 * (0.004)	0.010 ** (0.004)	0.008 * (0.004)
$S_{it-1,k=r}$	0.003 ** (0.001)	0.003 ** (0.001)	0.003 ** (0.001)	0.001 (0.003)	0.002 (0.004)	0.002 (0.004)
$S_{it-2,k=r}$	0.002 (0.001)	0.001 (0.001)	0.001 (0.001)	0.003 (0.002)	0.003 (0.003)	0.003 (0.003)
$W_{it}$	No	Yes	Yes	No	Yes	Yes
$W_{it-1}$	No	No	Yes	No	No	Yes
$W_{it-2}$	No	No	Yes	No	No	Yes
F	5.52 ***	4.39 ***	4.00 ***	0.62	0.72	0.93
R2	0.1236	0.129	0.1422	0.032	0.045	0.074
Number of observations	477	401	401	465	392	392

Note:$W_{it}$ denotes city characteristics, $W_{it-1}$ means one year lag of $W_{it}$, $W_{it-2}$ means two years lag of $W_{it}$. Corresponding to Equation (3). City fixed effects are included. Standard errors clustered by cities in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

estimates Equation (3). Columns (1) and (4) include only the shares of industrial and residential land in total urban leasing, their lags and city fixed effects. Columns (2) and (5) add land quotas, while Columns (3) and (6) further add lags of quotas. Despite the change of control variables, the results remain identical across columns.

Table 8. The expected and estimated signs of the coefficients in Equation (3).

Coefficient	Growth Rate of per Capita GDP		Growth Rate of per Capita Revenue
	Expected Sign	Estimated Sign	Expected Sign	Estimated Sign
	(1)	(2)	(3)	(4)
${\beta }_1$	0	+	0	0
${\beta }_2$	0	0	0	0
${\beta }_3$	0/+	+	0/+	0
${\beta }_4$	0/+	+	+	+
${\beta }_5$	0/+	+	0	0
${\beta }_6$	0	0	0	0

Note:${\beta }_1-{\beta }_6 \mathrm{are}$ the coefficients in Equation (3).

contrasts the estimated signs with expected signs. They are all consistent except ${\beta }_1$ for GDP: surprisingly, the share of industrial land in total land leasing immediately boosts GDP growth. This is not likely to be a direct growth effect from industrial development, as an enterprise obtaining a plot of industrial land would need some time to build up the infrastructure before it starts to produce. This instant effect on GDP growth is more likely caused by the fixed assets investments accompanied with land leasing, such as the buildup of infrastructure. The induced demand for infrastructure boosts GDP indirectly, through the increased production in other sectors. Indeed, after we add fixed assets investment as an additional variable, ${\beta }_1$ shrinks by half and loses statistical significance.

Table 7 and Table 8 also show that industrial land only significantly increases the growth of per capita GDP two years after provision and exhibits no effect on the growth of fiscal revenue even then. Thus, it makes perfect sense for a CCP secretary to be reluctant towards industrial land leasing after the third year of his or her service term because the benefits will only fall on the successor. Residential land leasing, on the contrary, immediately boosts the growth of per capita fiscal revenue and GDP, and continue to propel GDP one year after leasing. A rational CCP secretary thereby shifts towards the leasing of residential land during the latter years of his or her service term to reap these timely benefits. These results remain qualitatively robust with the inclusion of time trend polynomials and CCP secretary fixed effects, the results of which are available upon request.

5.3. The Effects of Economic Performance on the Promotion of City CCP Secretaries

Table 9. GDP and revenue growth and the probability of promotion.

	Probability of Promotion
	(1)	(2)	(3)	(4)
$Avg{G}_{inj=g}$	0.089 (0.056)	0.107 ** (0.048)	9.265 (12.546)	0.026 ** (0.012)
$Avg{G}_{inj=f}$	0.045 (0.029)	0.048 * (0.027)	7.789 (6.138)	0.012 (0.011)
TN	0.036 *** (0.10)	0.032 *** (0.009)	7.031 ** (2.752)	0.019 ** (0.009)
Xit	No	Yes	Yes	Yes
Zit	No	No	No	Yes
City fixed effects	No	No	Yes	No
Wald	10.68 **	23.07 **	47.64 ***	121.82
Log Likelihood	−127.368	−96.374	−8.930	−69.364
Number of observations	872	704	93	648

Note:$X_{it}$ and $Z_{it}$denote city characteristics (include the variables from FIS to ROAD) and individual characteristics of the CCP secretary who serves in city i year t. Corresponding to Equation (4). Margin effects instead of the coefficients are reported, except for Column (3) where marginal effects cannot be calculated due to the inclusion of city fixed effects. Robust standard errors in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

estimates Equation (4). Column (1) shows the baseline model, including only the average growth rates of per capita GDP, fiscal revenue and the tenure dummy. Both GDP and fiscal revenue raise the probability of promotion, though insignificantly so. Column (2) adds additional city characteristics and Column (4) further adds individual characteristics of CCP secretaries. While conditional on city characteristics, the effects of economic performance become both larger and statistically significant, the control of individual characteristics diminishes such effects. Column (3) adds city fixed effects to control for different promotion probabilities across cities, and the results remain qualitatively the same with those of Column (1).

In all specifications, the tenure dummy is significantly positive, indicating a greater probability for promotion in the later stage of the service term; this is consistent with the data in Section 2 that a typical turnover cycle for city CCP secretaries lasts approximately four years. These results are defendable with the inclusion of time trend polynomials and dummy variables indicating political cycles induced by PPC.

6. Discussion and Conclusions

This paper, using China as an example, shows how and why the mix of industrial and residential land leasing exhibits political cycles. Circle back to the questions we raise in the introduction: do political cycles prevail in the mix of industrial and residential land leasing in China? If yes, what are the magnitudes and what incentivizes local politicians to do so? A brief answer would be yes, there are political cycles, but the magnitudes are modest, and they prevail because of local politicians’ pursuit of political promotion. More specially, using panel data at the level of prefecture city, we find that during the first two years of CCP sectaries’ service terms, cities lease out a 3% greater share of industrial land and a 2% smaller share of residential land in total annual urban land leasing. Mayors’ service terms do not have the same effect. We also probe into the cause of this phenomenon: this strategy maximizes CCP sectaries’ chances of political promotion. We find that the benefits of residential land leasing are more immediate, while those of industrial land take time to reveal. Therefore, it is sensible for CCP sectaries to lease out industrial land earlier on and shift towards residential land later. These effects are defensible with time trend polynomials, a long list of control variables, and political cycles induced by the cyclical occurrence of the Provincial Party Congress.

This paper makes two contributions. First, it provides an additional piece of evidence that urban land use is affected by politics. It shows that political cycles affect real economies, not only in western countries via the turnover of parties [19,61,62], but even in China where party turnovers do not exist. As mentioned in the introduction, previous studies have found that land use programs and plans are subjective to change right before or after political elections in western democratic countries [20,21,22]. This paper shows that the mix of urban land leasing in a non-democratic nation is also subjective to political cycles, ones that are not induced by elections, but by the limited service term and the politician evaluation system. The magnitudes of land use change induced by political cycles in different nations are not directly comparable, as the size of the nations and the stage of urbanization are both different. Nevertheless, the mere fact that they face similar challenges in urban land use is interesting and important to know. Second, this paper also contributes to the studies on land leasing strategies of Chinese governments. While most studies focused on the amount of total land leasing [51,63,64], a few focused on land leasing prices and the mix of industrial and residential land use [17,43]. Consistent with previous studies [51,63,64], this paper finds that generating fiscal revenue to finance local economic growth is among the major rationale of Chinese local governments leasing out urban land. Previous studies have also argued that Chinese local governments strategically lease out industrial land by negotiation at low prices to attract manufacturing investments and reap local tax [43]. Specifically, researchers found that land leased out by negotiation (mostly industrial land) barely contributes to local tax in the same year and the following year of provision, but the contribution almost doubles in the second year after provision and triples in the third for Chinese prefectural cities during 1999 to 2003 [17]. These findings are partly replicated by this paper. We find in Table 7 that the share of industrial land leasing has not significantly increased the growth rate of per capita revenue (which includes local tax) up until the second year after provision. However, it has elevated the growth rate of per capita GDP, both in the year of provision and in the second year after provision. The general idea that industrial land leasing contributes to revenue and local economic growth with a time lag is consistent with the findings of this paper. We have also successfully replicated the result that industrial land leasing initiates no effect on revenue in the same year and the following year of provision. However, we also find null effect in the second year of provision, in contrast to the literature [17]. This difference is likely caused by the specific time frame; the economic downturn in 2008 likely prolonged the period before the return from industrial land leasing accumulates. Notably, none of the above studies have found the fact that local governments do not lease out industrial and residential land by a static mixture, but fine-tune the mixture to align with city leaders’ service terms. This paper fills the gap, quantifies such an effect, and explains the rationale behind it. By doing so, this paper enriches our understanding of Chinese local governments’ land leasing strategy and adds the timing of industrial and residential land leasing as an additional element to consider.

Besides academic contributions, this paper also speaks to critical social issues. First, it identifies a source of land use inefficiency. Since urban land use, instead of tracking the demand, exhibits a political cycle, it is unlikely to be socially optimal. Studies have found that in the former Soviet Union and China in the era before the open gate policy, the lack of land market has led to a skewed ratio of industrial and residential land use [65,66]. Industrial land was preferred while residential land provision was suppressed. Later on, in China, after the land market was introduced, the situation has improved [67], but government distortion remains significant and inefficiency in industrial land use continues to be a serious issue [68]. Consistent with these prior studies, this paper finds evidence for local governments still manipulating urban land leasing in recent years to serve their own needs rather than responding to market demand. This of course, like previous studies suggested, would lead to inefficient land use and notable social costs. For instance, cities may limit industrial land leasing during the later years of their leaders’ service terms, despite the interest of entrepreneurs to open up new plants. This may discourage entrepreneurship and hurdle employment growth.

Second, this paper also speaks to the spatial mismatch of urban land use in China. Studies have found that workers, especially low-income ones, usually reside far away from employment centers and industrial parks where they work [69,70]. Prior studies have demonstrated that the oversupply of industrial land versus the undersupply of land for public facilities and residential amenities in industrial parks contribute significantly to the issue of spatial mismatch in Chinese cities [69]. Such mismatch has detrimental effects on households, especially low-income ones [70]. What this article adds to the above line of literature is that it points out that the mismatch can be induced by the timing of industrial and residential land supply. To give an illustrative example, a city may have designed a new industrial park accompanied with nearby residential locations. The city leader, new in office, would lease out the industrial land first while withholding the residential land. Thus, when the enterprises in the industrial park start to recruit workers, unless there were already plenty of residential locations around (which is unlikely as most industrial parks in China are located in less populated areas), these workers will need to come from faraway locations. This is certainly consistent with the storyline of Zhuo et al. [69] that industrial parks suffer from insufficient land supply for residential purposes. Even if such land supply amplifies in the following years, the spatial mismatch has already formed and will be difficult to turn around.

Third, this paper also speaks to the unstable real estate markets and housing bubbles in China. With local governments monopolizing urban land markets, residential land leasing in China, as proposed by prior studies [71], has the potential to be used as a tool to stabilize real estate markets. This is a luxury that many nations don’t have. However, this luxury is instead used to cater to the needs of city leaders. Rather than stabilizing the real estate cycles, residential land leasing in China exacerbates the fluctuations of housing prices [72]. When housing prices are soaring, local governments tend to further restrict land leasing to reap high revenues. How much local government relies on fees collected from residential land leasing for revenue significantly affects the rise of land prices, and consequentially housing prices [72]. This conclusion is quite consistent with our findings in Table 7 that a greater share of residential land leasing immediately boosts the growth rates of both per capita revenue and GDP. As a result, waves of policies implemented by China’s central government to curb unaffordable housing prices have been plainly ineffective. Some suggest that affordable housing construction led by the governments can counter-act the problem of soaring housing prices, but empirical evidence has found that such an approach only works in cities where housing affordability issues are less serious [73]. The more a local government relies on land leasing fees, the less willing and likely it is to provide land for affordable housing construction.

The skewed ratio and timing of industrial and residential land leasing are economically, environmentally and socially unsustainable. The departure of land use patterns from market demand hurts local economic development, employment and income growth; economically, this is unsustainable, especially in a developing country like China where growth is important. The scale of urban land expansion, especially the development of large-scale industrial parks, encroaches into farmland, wetland and unused land, threatening environmental sustainability. Finally, as mentioned above, the skewed mix of industrial and residential land use and the messed-up timing of land provisions exacerbates the problem of unaffordable housing prices and spatial mismatch of working and residential locations. These problems disproportionately hurt low-income households and are thus socially unsustainable.

It is important to mention that while CCP secretaries’ service terms affect the mix of industrial and residential land leasing, and thus contribute to the above social issues, the magnitudes of their effects are only modest. In other words, they are clearly not the sole cause of these social problems, but do to some extent exacerbate these issues. This paper shows that as long as the criteria by which city politicians are evaluated remains heavily skewed toward economic indicators and city governments continue to monopolize urban land markets from the supply side, it would be difficult to align supply with demand, and these distortions are likely to persist. Policymakers can attack this problem by liberating urban land markets or updating the evaluation criteria for city politicians. If city governments are deprived of the market power, or social indicators, such as housing affordability, receive greater weight in politician promotion, we expect the behaviors of politicians to shift towards social optimality.

Admittedly, several limitations remain in this paper. First, while the absolute growth rates of fiscal revenue and GDP matter for the promotion of a city politician, the relative performance compared to his or her peers in neighbor cities matters too. Thus, the strategy of a city’s industrial land leasing is likely responsive to the strategies of its neighbor and peer cities, as they are essentially competing with each other to attract external investments. These issues are assumed away in this paper, but can be explicitly examined in future studies with a spatial econometric framework. How to properly specify the spatial weight matrix would be a challenge, but as we gain deeper understanding of the nature of competitions among cities and city leaders, future studies should be able to do a better job at this. Second, this paper assumes that politician idiosyncratic characteristics and their service terms affect the mix of industrial and residential land leasing independently. However, certain idiosyncratic characteristics may magnify the political cycle in land leasing induced by service terms. Identifying these characteristics can help target a subgroup of city leaders whose behaviors need to be more closely watched to avoid inefficient urban land use. This can be dealt with by simply introducing interaction terms of politician characteristics and service terms. However, due to the large number of politician characteristic variables, doing so would cost a significant amount of degrees of freedom, and therefore is only feasible in future studies with a much longer time span.