Analysis of the Government’s Decision on Leasing Different Lands under Public Ownership of Land

Abstract

Using the multinomial logit model, this paper investigates the factors influencing the government’s decision to lease different types of land in Shenzhen, China, including residential, industrial, commercial, and public service land. The aspects of the land attributes, economy and government at the district level, and land accessibility are considered as the influencing factors. Regarding the factors as the variables, the influencing factors supporting the district government decision to lease different types of land and the probability that a type of land will be consider to be leased by the government are investigated via the multinomial logit model. Using data of factors from 2005 to 2021 in Shenzhen, China, the results of the model can be obtained. After discussing and analyzing the results, it is shown that the land attribute, land accessibility, and economy and polity at the district level affect government decisions on leasing land; furthermore, industrial land is more likely to be leased by the district government than other types of land. Lastly, implications and suggestions for the district government are discussed.

1. Introduction

For different countries, the approach or behavior of leasing land is different because of different social systems and periods of economic development [1]. Studying the decisions of a district government on leasing different land types is important for supporting the policies of economic development, urban sustainability, and urban planning [2,3,4].

In China, land leasing is implemented by the local government, which delegates the right to lease land to the district governments. The district governments decide on land leasing according to the district’s conditions and land planning [1]. There are various land uses, such as residential, commercial, industrial land, etc., and their characteristics, prices, and surrounding environments are different. Thus, it is very important for the government to decide which type of land to lease. In recent years, the revenues from leasing land have made up a high proportion of total government revenues in China. According to the Statistical Yearbook in China, state-owned land leasing revenue increased from CNY 2386.1 billion in 2011 to CNY 8705.1 billion in 2021. The increase in local income through land leasing has led to problems such as land oversupply and housing affordability issues in the land and real estate markets [5]. In China, the demand for construction land is exhibiting a marked upward trend, particularly in economically prosperous provinces. It is imperative to enhance the efficiency of allocating construction land within the country [6]. Furthermore, a balance must be considered between preserving arable land and meeting the demands for construction land [7]. Thus, the government should consider the factors influencing the leasing of various types of land when they decide, balancing public revenues and sustainable development of a city.

Government behavior affects land leasing through administrative hierarchy, choices of lease method, and land supply decisions because the government can provide funds, boost basic construction, absorb citizens and employment, and improve economic development [8,9]. There is competition for land leasing between district governments to absorb investment and obtain profits [1]. In eastern China, government competition for land leasing began earlier, and that of western China lags behind the national level [10]. Political tenure influences land leasing and a city’s economy [11]. A newly appointed mayor is more motivated to stimulate long-term economic growth and provide more land for industrial use. Long-term mayors prefer to lease more land to the service industry to obtain immediate benefits [12,13]. Local governments prefer regulating the price of industrial land, and they also like raising the commercial and residential land prices, which is favorable to maximizing their interests [8]. Mayors supply commercial and residential land for higher financial benefits [14]. Furthermore, the local government’s interventions affect land supply [15], and residential land supply decreased after the Chinese government strengthened its land market intervention [16].

Land leasing is associated with urban economic development [2,3,15]. It can promote urban economic growth, and land newly made available may have a lagged effect on economic growth due to the requirement of construction time after land is made available [4]. Economic development is affected by land leasing via government financial relationships, which further affects the land lease price and scale [17]. In addition, increasing land transfer income accelerates economic growth directly [18]. As the primary concern of economic development, the gross domestic product (GDP) positively influences land supply and land for the service and industrial sectors [10,19]. Investment in fixed assets and workers’ average annual wages correlates with land supply, and government fiscal decentralization positively affects land supply [20]. The proportion of industrial land leases boosts GDP growth due to the investment in fixed assets, such as infrastructure construction, generated from land leasing [9]. Meanwhile, it is essential to attract foreign direct investment to enhance Chinese economic competition and development, and companies interested in foreign direct investment often need land to build factories and business districts, which promotes urban sprawl [21].

From the aforementioned research, most of the literature has focused on land use efficiency, factors influencing land lease prices, and government intervention in the land market. Most publications considered certain land types and did not involve different land use types. Also, most studies considered factors that influence land leasing, such as economic development or government intervention. Few studies focused on government decisions or strategies for leasing land and the probability of government decisions. This study considers four types of land uses: In addition to residential, industrial, and commercial land, public facilities land is included to analyze the government’s decision on leasing land. Moreover, more factors, which combine with the land attribute, district-level economy and polity, and land accessibility, from micro to macro level, are considered.

Shenzhen is a prefecture-level city under the jurisdiction of Guangdong Province, a state-planned city, a mega-city, a special economic zone approved by the State Council, a national economic center city, and a national innovative city. Many land policies, such as the compensated leasing of land use, reform of mixed land use for secondary and tertiary industries, and the ‘Industry Upstairs’ policy, have been piloted in Shenzhen. With the city’s rapid urbanization, the shortage of construction land is increasingly apparent. For example, the manufacturing industry is gradually overflowing to other cities in Guangdong Province. Thus, the government needs to improve the efficiency of construction land and promote urban sustainable development by balancing leasing different types of land.

Based on the data analysis method, this paper investigates the factors influencing the government’s decision to lease different types of land in Shenzhen, China, including residential, industrial, commercial, and public service land. The factors influencing land leasing include various aspects, such as land attributes, economy and government, and land accessibility. Considering these factors as the variables, the multinomial logit model is presented to analyze the factors influencing the district government’s decision on land leasing. The results of the model were obtained according to the data from 2005 to 2021 in Shenzhen, China, and the factors affecting the government’s decision on leasing land are presented herein. The land data were obtained from the official website of China Land Market; the data on the district level were collected from the Shenzhen Statistical Yearbook and the yearbooks of every district in Shenzhen; the data for the government tenure were obtained from the official website of each district government and past news reports in Shenzhen; and the accessibility data were computed using the software ArcGIS 10.2 according to the location of each piece of land obtained from the Baidu Map. Lastly, the implications and suggestions for the district government are derived.

This study gives recommendations for the governments on their decisions on leasing land, which can improve the efficient use of construction land, promote sustainable urban development, and balance government revenues and reasonable land allocation. The methodology and policy implications might be applied to Chinese metropolitans and countries with similar backgrounds and policies of public land ownership.

The innovative points of this study are as follows: (1) Four types of land, including residential, industrial, commercial, and public service land, are involved in investigating the factors influencing the government’s decision to lease; (2) the aspects of land attributes, economy and government at the district level, and land accessibility are considered when selecting the influencing factors; and (3) the multinomial logit model is proposed to analyze the influencing factors that the district government considers when leasing different types of land and the probability of which type of land is considered to be leased.

2. Background of Land Leasing in Shenzhen

In the past 40 years, Shenzhen has carried out three land reforms. The first time was in 1987, when land leasing was changed from free charge to compensation, pioneering the marketization of land resources [22]. Shenzhen publicly auctioned the first piece of state-owned land use right, realized the separation of state-owned land ownership and use rights, and formulated a new system for the land allocation by market-oriented approaches, which directly promoted the revision of the Constitution in 1988 and improved the formation of the national market for land use right.

The second was the implementation of unified expropriation and transfer of rural land to construction land in 1992 (Luohu, Yantian Futian, and Nanshan Districts) and 2004 (Bao’an and Longgang Districts), which took the lead in realizing the nationalization of the whole land, and nominally eliminating the collective ownership of land in Shenzhen. In 1992, the Interim Provisions on Rural Urbanization in Shenzhen Special Economic Zone was promulgated by the municipal government to implement the unified urbanization and expropriation work, comprehensively realizing the transformation of rural areas into cities and villagers into citizens [23]. Moreover, in 1993, the ‘five unified’ land management system was established in which the government implements unified requisition, reserve, unified development, unified allocation, and unified management.

The third time was in 2012, when the former Ministry of Land and Resources and the Guangdong Provincial Government authorized and implemented The Overall Plan for the Reform of Shenzhen’s Land Management System (2012–2020) [24]. Shenzhen has successively promulgated the Opinions on Optimizing the Allocation of Space Resources to Promote Industrial Transformation and Upgrading, the Opinions of Shenzhen Municipal People’s Government on Improving the Management of State-owned Land Supply, and the Measures of Shenzhen Municipality on the Management of Land Supply for Industrial and Other Industries. A land resource optimization allocation system was established, supported by differentiated land supply methods, ensuring industrial development space, accelerating residential land supply, and standardizing land price calculation and management.

In addition, in 2012, the supply of stock land in Shenzhen exceeded the supply of new land for the first time. The secondary development of stock land, which mainly focuses on urban renewal, land rehabilitation, and low-efficiency land redevelopment, has gradually become an important means of land resource marketization.

With the reform of land leasing in Shenzhen, the housing price is generally high, construction land is increasingly lacking, industrial upgrading is increasingly important, and urban renewal space is urgently needed. Therefore, the government must consider the balance of leasing various land use types and improve land use efficiency. Figure 1 is the process of the land reforms in Shenzhen.

3. Methodology

This paper investigates the influencing factors on the district government decisions to lease different types of land, including residence, industry, commerce, and public service land, and the probability of which type of land the government will lease is also analyzed. The methods of data analysis used in this paper are data collection, data cleaning, computation based on geographic information, and mathematical models.

In light of the contextual background of land leasing in Shenzhen and a thorough examination of the existing literature, this paper presents the influencing factors of the land attributes, economy, and government at the district level, as well as land accessibility. Considering the factors from the macro and micro levels, the district government’s decision on land leases is analyzed more clearly.

Considering the selected factors as the variables, the multinomial logit model is applied to analyze the behavior of the district government in deciding on leasing land. The advantage of the multinomial logit model is that it can handle multiple categories and obtain the probability of each category, which is very useful for problems that require determining probability. It can handle nonlinear relationships, thereby improving the predictive ability of the model [25]. It can also process large datasets and complex feature spaces because this model can be quickly trained using optimization algorithms such as gradient descent. In addition, this model has good interpretability and robustness [26]. This model has been recognized as an effective choice method, and has been applied to the studies of land use change [27], land development [25,28], land management [29], land classification [30], land policy [31], etc.

The data of the variables, including land data for residence, industry, commerce, and public service, district-level data of foreign investment in actual use, investment in fixed assets, average wage of employees, government fiscal gap, district government tenure, and accessibility data of land, were collected from 2005 to 2021. The land data were obtained from the official website of China Land Market. The data on the district level were collected from the Shenzhen Statistical Yearbook and the yearbooks of every district in Shenzhen. The data for the government tenure were obtained from the official website of each district government in Shenzhen and past news reports. The accessibility data were computed using the software ArcGIS according to the location of each piece of land. The longitude and latitude information are obtained from the Baidu Map.

The data for continuous variables are processed by standardization of the z-score method based on the software Stata 12.0. The multicollinearity of the variables was estimated using the variance inflation factor (VIF).

According to the data obtained, the results of the multinomial logit model were determined via the software Stata. Then, the influencing factors for the district government’s decision on land leases and the probability of the decision are discussed. Finally, some policy implications are presented from the discussions. Figure 2 shows the process of the methodology in this study.

4. Multinomial Logit Model of Influencing Factors for Land Leasing

4.1. Classification of Influencing Factors

According to the influencing factors discussed in previous publications, and by analyzing the background of land leasing in Shenzhen, the various aspects of factors, including the land attributes, district-level economy and government, and land accessibility, are considered, combined with macro and micro levels.

Table 1. The classification of influencing factors.

Classification	Variable
Land attribute	Area for a piece of land
	Transaction price for a piece of land
	Location of the district
	Supply mode of a piece of land
District-level economy and polity	Foreign investment in actual use of the district
	Investment in fixed assets of the district
	Average wage of employees of the district
	Fiscal gap of the district
	District chief tenure
Land accessibility	Distance between a piece of land and the location of the district government
	Distance between a piece of land and the location of Shenzhen Citizen Center
	Distance between a piece of land and Shenzhen Bao’an International Airport
	Distance between a piece of land and the nearest metro station
	Distance between a piece of land and the nearest municipal key high school
	Distance between a piece of land and the nearest public park
	Distance between a piece of land and the nearest industrial park

shows the summary of the classification of influencing factors.

4.1.1. Land Attribute

The land attribute includes information on the location, area, supply mode, and transaction price for a piece of land.

The land attribute reflects the basic land characteristics. The location of land is important to land price [32], and land leasing has led to a decrease in the intensity of development in urban centers, with more land being used for suburban areas [33]. The area of land, land use, and land transaction fee affect land prices positively [34]. The increase in land transfer income directly accelerates economic growth [18]. The location of the land affects regional economic growth [3].

The land supply modes affect the decisions of government on land leases. Listing is the main method of leasing to government competition [10]. The land leased by tender and auction is, on average, lower than that leased by listing [35]. The listing method of industrial land prices is considerably cheaper than other open-outcry supply methods, because the listing has subsidies for land prices [36].

4.1.2. District-Level Economy and Polity

The district-level economy and polity involve the district chief tenure, foreign investment in actual use, investment in fixed assets, average wage of employees, and government fiscal gap.

Land leasing plays a significant role in urban areas’ economic growth and development [2,3,4]. The allocation of industrial land contributes to an increase in GDP growth, as it stimulates investment in fixed assets like infrastructure development, which is facilitated by land leasing [12]. There is a positive relationship between the supply of land and the investment in fixed assets, as well as the average annual wages of workers [19]. It is essential to attract foreign direct investment to enhance Chinese economic competition and development [21]. The government fiscal gap has a positive impact on land prices [37]. If the central government allocates lower budget revenue to local governments, they are more likely to encourage more land transactions, thereby obtaining more off-balance-sheet benefits [38].

Government intervention has an impact on land supply [15]. Mayoral tenure is correlated with the number and proportion of land leased to the service industry due to economic growth as an essential indicator of political performance [13]. Urban land leasing promotes the development of the local economy and increases the opportunities for the promotion of urban leaders, with the aim of maximizing measured economic performance and fiscal rent [11]. The economic benefits of residential land are immediate, while those of industrial land will not be apparent until two years later [12]. Land leases and the economic development of a city are influenced by political tenure. Compared with land for the tertiary sector, a local official with a longer tenure prefers supplying industrial land in order to achieve long-term profits [13,39].

4.1.3. Land Accessibility

The accessibility on land is essential in land price and leasing [40]. The aspect of land accessibility involves the distance of a piece of land to the district center (i.e., the location of the district government), city center (i.e., the location of Shenzhen Citizen Center), Shenzhen Bao’an International Airport, and the nearest metro station, public park, municipal key high school, and industrial park.

Shenzhen Citizen Center is the center of Shenzhen. It is located in the Futian District. It is a central area of Shenzhen, covering an area of 910,000 square meters, which is near the central business district. It is a comprehensive building integrating the functions of the government, the People’s Congress, museums, halls, and other functions. In addition, it is an administrative center and an important landmark building in Shenzhen. The district center is the location of the government in each district. Generally, there are relatively comprehensive and complete functions around the location of the district government. There is an airport in Shenzhen, which is Shenzhen Bao’an International Airport. It is located in Bao’an District and was officially opened in October 1991. As an important transportation hub, the airport affects land leasing [41,42]. The land price is higher when it is near the city center and higher when it is near the subway [35]. City center and public infrastructures, including schools and subways, affect land prices [43]. Public and industrial parks influence land leasing prices [1], especially for residential, commercial, and industrial land.

4.2. Multinomial Logit Model of Influencing Factors for Land Leasing

The multinomial logit model was applied to investigate the factors influencing the district government’s decision-making process regarding land leases and the associated probability of determining the specific type of land to be leased. It is an analysis method for multiple classification problems and helps determine probability [26]. Its principle is to classify the samples into various non-overlapping categories, and the logit model can estimate the relative probability among different categories. The relative probability that the sample belongs to each category can be obtained by calculating the logit value corresponding to each category. Finally, the corresponding category can be classified with the highest probability. In addition, this model can improve its predictive ability and have good interpretability and robustness [26]. It has been recognized as a practical choice method for land studies [27,28,29,30,31]. In this paper, the model is proposed as:

$$P(y_i=t\left|\mathit{x})\right.=\frac{\exp (y_{it})}{1+{\displaystyle \sum _{k=2}^N\exp (y_{ik})}},(t=1,2,3,4),$$

(1)

where $y_{it}$ is the types of land uses; $i$ is the district; and $t=1$ means residential land, $t=2$ means industrial land, $t=3$ means commercial land, and $t=4$ is for public service land.

$$\begin{array}{ll}{y}_{it}& =\frac{P(y_i=t\left|\mathit{x})\right.}{P(y_i=1\left|\mathit{x})\right.}\\ & ={\beta }_{t1}A{L}_i+{\beta }_{t2}T{P}_i+{\beta }_{t3}LC{1}_i+{\beta }_{t4}LC{2}_i+{\beta }_{t5}MD{1}_i+{\beta }_{t6}MD{2}_i+{\beta }_{t7}F{V}_i\\ & +{\beta }_{t8}FX{2}_i+{\beta }_{t9}W{G}_i+{\beta }_{t10}F{G}_i+{\beta }_{t11}D{C}_i+{\beta }_{t12}D{G}_i+{\beta }_{t13}C{C}_i\\ & +{\beta }_{t14}A{P}_i+{\beta }_{t15}M{S}_i+{\beta }_{t16}K{H}_i+{\beta }_{t17}P{P}_i+{\beta }_{t18}I{T}_i+a_t\end{array}$$

(2)

where $N$ is the quantities of use types of land ($N=4$), $a_t$ is the constant, $L{A}_i$ is the area for a piece of land, $T{P}_i$ is the transaction price for a piece of land, $LC{1}_i$ is the location of the district (1 when it is center district; 0 otherwise), $LC{2}_i$ is the location of the district (1 when it is suburban district; 0 otherwise), $MD{1}_i$ is the supply mode of a piece of land (1 when it is leased by listing; 0 otherwise), $MD{2}_i$ is the supply mode of a piece of land (1 when it is leased by auction; 0 otherwise), $F{V}_i$ is the foreign investment in actual use of the district, $F{X}_i$ is the investment in fixed assets of the district, $W{G}_i$ is the average wage of employees, $F{G}_i$ is the fiscal gap, and $D{C}_i$ is the district chief tenure. The following parameters are the distances: $D{G}_i$ is the distance between a piece of land and the location of the district government, $C{C}_i$ is the distance between a piece of land and the location of Shenzhen Citizen Center, $A{P}_i$ is the distance between a piece of land and Shenzhen Bao’an International Airport, $M{S}_i$ is the distance between a piece of land and the nearest metro station, $K{H}_i$ is the distance between a piece of land and the nearest municipal key high school, $P{P}_i$ is the distance between a piece of land and the nearest public park, and $I{T}_i$ is the distance between a piece of land and the nearest industrial park.

The base outcome of the model was set as industrial land use, because its characteristics show differently from residence, commerce, and public facilities land.

5. Data Collection and Processing

There are nine administrative districts, including Bao’an District (BA), Futian District (FT), Guangming District (GM), Longgang District (LG), Longhua District (LHA), Luohu District (LH), Nanshan District (NS), Pingshan District (PS), and Yantian District (YT), and a new district, Dapeng New District (DP), in Shenzhen. The districts are classified into center, urban peripheral, and suburban districts according to the Comprehensive Plan of Shenzhen City (2007–2020). The center districts are NS, LH, FT, and YT, the urban peripheral districts are BA, LHA, and LG, and the suburban districts are GM, PS, and DA. The location of Shenzhen and the districts in Shenzhen are presented in Figure 3.

There were 936 land leases in Shenzhen during 2005 and 2021. The use types of land refer to land for residence, industry, commerce, and public facility. The land data were obtained from the official website of China Land Market (https://www.landchina.com/ accessed on 1 April 2023, which the Real Estate Registration Center established under the jurisdiction of the Ministry of Natural Resources of China. The data collected include the location, area, use type, supply mode, and transaction price of the pieces of land. Because the public auction can show the government’s decisions more clearly and the policy of abolishing the negotiation of industrial land leasing was implemented in 2004, the public auction, including tender, listing, and auction, is considered the supply mode of land. Also, the start year of this study is 2005, which is after the land policy reform. The land leasing distribution from 2005 to 2021 is shown in Figure 4.

The data on the district level were collected from Shenzhen Statistical Yearbook and the yearbooks of every district in Shenzhen. The data collected are GDP, foreign investment in actual use, investment in fixed assets, average wage of employees, and government fiscal gap. The data for the government tenure is the tenure of the district chief, and the relevant data source is from the government’s official website in districts of Shenzhen and past news reports. The statistical yearbooks can be obtained in local libraries, the government’s official website, or the Bureau of Statistics in Shenzhen.

The accessibility data, including the land distances to essential locations, mentioned in Section 4.1, were computed through the software ArcGIS according to the location of each piece of land. These important locations were obtained from the Baidu Map, which provides the longitude and latitude. The data can be obtained through the Baidu coordinate picking system or Baidu API. The land data, the data on the district level, the data for the government tenure, and accessibility data can be accessed by anyone and are free of charge. Figure 5 shows the locations of the district center, city center, airport, and municipal key high schools. The distribution of metro stations, public parks, and industrial parks is shown in Figure 6, Figure 7 and Figure 8, respectively.

In Figure 5, the city center is Shenzhen Citizen Center, the district center is the government of each district, and the airport is Shenzhen Bao’an International Airport. There are 23 municipal key high schools confirmed by Shenzhen Education Bureau, and most schools are located in the NS, LH, and FT districts.

Figure 6 shows the changes and process of metro development in Shenzhen. The first line of metro was opened at the end of 2004. During 2005 to 2007, the metro development was at early stage, which involved Line 1 and Phase 1 of Line 4. From 2008 to 2009, Line 1 was expanded. Lines 2 and 3 were established successively in the initial stage. In 2011, Lines 1, 2, 3, and 4 were extended, and construction began on Line 5. In 2016, Lines 7, 9, and 11 were opened. Lines 5 and 9 had extensions in 2019. After that, Lines 6, 8, and 10 were newly opened, and Lines 2, 3, and 4 were expanded. In 2021, Line 20 was built up, and there were a total of 12 metro lines. Compared with 2005, metro development was becoming more mature, covering an increasingly large area and districts in 2021 in Shenzhen.

As shown in Figure 7, there are 849 parks in Shenzhen, and in general, the distribution of the parks is relatively evenly on the district level. There are a large number of parks in the LG, BA, and NS districts. In addition, the parks in the center districts are more densely concentrated.

As shown in Figure 8, there are 3050 industrial parks in Shenzhen. More industrial parks are located in the BA, LG, and LHA districts, which are the urban peripheral districts. The number of industrial parks in the center districts is relatively tiny.

Table 2. The descriptive statistics and results of VIF for the variables.

Variable	Unit	Observation	Mean	Standard Deviation	VIF
LA	m2	936	55,768.55	214,525.6	1.04
TP	CNY	936	5.87 × 108	1.71 × 109	1.19
LC1	-	936	0.2628	0.4404	4.68
LC2	-	936	0.4594	0.4986	2.9
MD1	-	936	0.906	0.292	5.34
MD2	-	936	0.0759	0.2649	5.13
FV	CNY	936	4.04 × 109	3.71 × 109	2.9
FX	CNY	936	5.73 × 1010	3.81 × 1010	3.14
WG	CNY	936	53,193.11	48,811.59	2.02
FG	CNY	936	4.13 × 109	5.61 × 109	2.14
DC	Year	936	2.7019	1.402	1.86
DG	m	936	26,336.94	15,400.37	1.1
CC	m	936	22,816.33	10,868.27	1.15
AP	m	936	31,302.95	17,437.83	3.28
MS	m	936	9670.762	10,576.31	3.33
KH	m	936	7743.832	6564.297	2.71
PP	m	936	568.0807	477.1932	1.22
IT	m	936	413.4986	284.0499	1.11

presents the descriptive statistics for the variables. The indices of observation, mean, and standard deviation for the variables are calculated to show distribution of the data.

To eliminate the different units and dimensions of the data, the data for continuous variables were processed by standardization of z-score method based on the software Stata. The formula of standardization of z-score is:

$$x^{*}=\frac{x-M}{D}$$

(3)

where $x$ is a variable, $M$ is the mean value of the sum of $x$, and $D$ is the standard deviation of the sum of $x$.

Before the regression, the multicollinearity of the variables in Equation (1) was estimated by the VIF. The results in Table 2 show that the VIF of the variables is smaller than 6. Thus, there is no multicollinearity.

6. Regression Results and Discussion

6.1. Results of Multinomial Logit Model

Obtained using Stata, the regression results of Equation (1) are shown in

Table 3. The regression results of Equation (1).

Use Type of Land	Variable	Coefficient	Standard Error	z-Statistics
Residence	Land attribute
	LA	−0.48389 **	0.24323	−1.99
	TP	5.20527 ***	0.5223	9.97
	LC1	−1.14154 *	0.60294	−1.89
	LC2	−0.38853	0.3977	−0.98
	MD1	−3.7681 ***	1.09203	−3.45
	MD2	−1.76371	1.12347	−1.57
	District-level economy and polity
	FV	0.50322 ***	0.19466	2.59
	FX	−0.0425	0.21282	−0.2
	WG	−0.34157 **	0.18009	−1.9
	FG	−0.19716	0.15878	−1.24
	DC	−0.13244	0.11801	−1.12
	Land accessibility
	DG	0.01091	0.11308	0.1
	CC	0.07255	0.11663	0.62
	AP	0.34622 *	0.20778	1.67
	MS	0.151217	0.19103	0.79
	KH	−0.12554	0.17377	−0.72
	PP	−0.28383 **	0.12258	−2.32
	IT	0.53705 ***	0.12085	4.44
	Constant	3.71491 ***	1.13899	3.26
Commerce	Land attribute
	LA	0.02375	0.0763	0.31
	TP	5.09271 ***	0.52014	9.79
	LC1	1.511134 ***	0.56176	2.69
	LC2	−0.63308	0.43899	−1.44
	MD1	−2.52471 *	1.36919	−1.84
	MD2	0.1827	1.40468	0.13
	District-level economy and polity
	FV	0.16115	0.18481	0.87
	FX	0.1187	0.20661	0.57
	WG	−1.00793 ***	0.18718	−5.38
	FG	−0.56158 ***	0.16581	−3.39
	DC	−0.20856 **	0.10145	−2.06
	Land accessibility
	DG	0.145388	0.12794	1.14
	CC	0.1055	0.133	0.79
	AP	0.15993	0.21269	0.75
	MS	−0.46884 *	0.24879	−1.88
	KH	−0.34838 *	0.20117	−1.73
	PP	−0.03533	0.13721	−0.26
	IT	0.5278 ***	0.12078	4.37
	Constant	1.93225	1.41641	1.36
Public facility	Land attribute
	LA	−7.79811 **	3.35715	−2.32
	TP	2.67607 ***	0.89797	2.98
	LC1	2.1762 **	0.94555	2.3
	LC2	−0.33912	0.7075	−0.48
	MD1	8.49464	444.9765	0.02
	MD2	12.10961	444.9768	0.03
	District-level economy and polity
	FV	0.01449	0.30191	0.05
	FX	0.63509 **	0.33771	1.88
	WG	−1.13683 ***	0.28248	−4.02
	FG	0.02592	0.30297	0.09
	DC	−0.01292	0.17854	−0.07
	Land accessibility
	DG	0.81151 ***	0.27388	2.96
	CC	−0.41962 *	0.24774	−1.69
	AP	0.64499 **	0.35764	1.8
	MS	−0.57104	0.44592	−1.28
	KH	−0.036325	0.33692	−0.11
	PP	−0.58756 *	0.32495	−1.81
	IT	0.54483 ***	0.19991	2.73
	Constant	−13.22932	444.9776	−0.03
Observation	936
$R^2$	0.3452

Note: ***: Significant at the 1% level; **: significant at the 5% level; *: significant at the 10% level.

.

According to the land attributes in Table 3, LA is significant for commercial land and public facilities land, showing significance at the 0.05 level, with negative coefficients (−0.4839 and −7.7981, respectively). However, it is insignificant for residential land. TP is very significant for residential, commercial, and public facilities land, with all positive coefficients. It is very significant for LC1 on commercial land (1.5111), and it is significant for residential (−1.1415) and public facility (2.1762) land. MD1 shows a very significant aspect of residential land. Also, it is significant for commercial land with a negative coefficient.

From the aspect of district-level economy and polity, FV is very significant for residential land with a positive coefficient and insignificant for the other two types of land. FX shows significance for public facilities land with a positive coefficient. WG is very significant for commercial and public facilities land, and it is significant for residential land with a negative coefficient. DC is significant for commercial land with a negative coefficient.

About land accessibility, DG shows very significant for public facilities land with a positive coefficient. AP is significant for residential and public facilities land with a positive coefficient. KH shows significance for commercial land with a negative coefficient. PP shows significance for the residential and public facilities land with a negative coefficient. IT is very significant for residential, commercial, and public facilities land with a positive coefficient.

The multinomial logit model is employed to further estimate the possibility that the district government makes decisions regarding the type of land to be leased.

Table 4. Possibility that the district government decides on the type of land to be leased.

Use Type of Land	Observation	Mean	Standard Deviation	Min	Max
Residential	936	0.17522	0.19971	5.72 × 10−8	0.95529
Industrial	936	0.58867	0.32842	0	0.99302
Commercial	936	0.19551	0.22808	0.00069	0.96835
Public facility	936	0.0406	0.06428	0	0.48823

shows the corresponding results.

Table 4 shows a 17.52%, 58.87%, 19.55%, and 4.96% possibility for the district government deciding to supply land for residential, industrial, commercial, and public facility purposes, respectively.

6.2. Discussion on Multinomial Logit Model

Regarding land attributes, land with a higher transaction price is more likely to be supplied by the district government for the use of residential, commercial, and public facility purposes than industrial land. Compared with industrial land, land with a smaller area for residential and public facility purposes is more likely to be leased by the district government because the industrial land area is generally bigger than other land types. Industrial production often requires a large space to accommodate the storage of production lines, machinery and equipment, raw materials, and finished products [44]. Furthermore, industrial production may require additional space for logistics, transportation, and employee parking. The utilization efficiency of industrial land is relatively low, and the building density and plot ratio per unit area are usually lower than residential and public facilities land. This is because industrial production requires a larger space to meet the needs of production and operation. The development cost of industrial land is relatively low, and for investors, the development of industrial land may bring higher economic returns. The planned area of industrial land is relatively large, as industrial development is the main driving force of economic growth.

Compared with the other three land types, the industrial land price is lower due to government intervention. Investors in the industrial sector, such as the manufacturing sector, are highly mobile, so governments have to compete to attract them. Under strong competitive pressure, the governments will offer more favorable policies [45], including cheap land, subsidized infrastructure, tax breaks, relaxed environmental policies, and even labor deregulation. The follow-up production of industrial enterprises can bring long-term revenue, stimulate employment, and promote regional GDP growth.

Compared with industrial land, residential land is more likely to be leased via the district government in urban peripheral districts and suburban districts and land leased for commerce and public facility in center districts. With the development of cities, land leasing has led to a decrease in the intensity of development in urban centers, and more land is being used for suburban areas [33]. In center districts, the population and housing need to be balanced, and construction land is scarce, resulting in an inadequate housing supply. With the development of urban peripheral and suburban districts, residential land is leased more in these districts to relieve the housing pressure in center districts. In addition, residential areas usually require larger spaces to accommodate more residents [32]. At the same time, the land resources in the urban peripheral and suburban districts of cities are relatively abundant, making them more suitable for large-scale residential development. Urban fringe and suburban areas usually have less traffic pressure and are more suitable for building large residential areas. In addition, the government can also promote the development of urban fringe areas and suburbs by constructing new transportation routes and infrastructure. Commercial and public facilities, such as hospitals, schools, and government agencies, usually need to be located in areas with convenient transportation and high pedestrian flow to provide services to citizens [46]. Urban centers are usually more suitable for the aggregation of these facilities. The center districts are always the commercial hubs, formulating the center business district. The public facilities land involves economic, cultural, educational, health, sports, research, and design institutions and facilities, and the construction of these facilities is more perfectly in center districts, which promotes the comprehensive development of center districts. The agglomeration of commercial and public facilities can drive the economic development of surrounding areas, improve land use efficiency, and increase government tax revenue. Therefore, the government may prioritize the transfer of commercial and public facilities land in the central area.

Unlike industrial land, the government at the district level tends to have land leases for residential and commercial purposes by tender and auction. Industrial land is generally leased by listing to show more transparency and avoid negotiation, which is in agreement with the National Minimum Price Standard for Industrial Land Transaction, implemented by the Ministry of Natural Resources of China to prevent the low price and set the price standard of industrial land in 2007. Governments prefer the listing mode to attract investment from priority enterprises and provide more subsidies to state-owned companies and enterprises through the method of listing, thereby reducing their land purchase prices [36]. Residential and commercial land often face higher market demand, encouraging developers to provide better development plans and higher bids, which helps improve the quality and efficiency of land development. Through tender and auction, the government can propose specific requirements for the use of land, such as building density, greening rate, and public facilities, to ensure that land development meets urban planning and long-term development goals.

From the aspect of district-level economy and polity, residential land may be leased through the district government while its foreign investments are high. When there is a low situation of foreign investment in the district, the government attracts foreign investment through industrial land leases to obtain long-term incomes. The increase in foreign investment is usually accompanied by the influx of foreign employees and professionals, increasing the housing demand and improving infrastructure [21]. The government’s transfer of residential land can meet the housing needs of these new residents. The increase in foreign investment may drive the process of urbanization, lead to urban population growth, and thus increase the demand for residential land. Therefore, the population growth and urban development demand brought about by foreign investment are more inclined to support a transfer of residential land.

Compared with industrial land, it is more likely for the government to supply land for public facility with a high investment in fixed assets of the district. When the investment of the district government in fixed assets is high, the workload of fixed assets and their expenses will be high. Thus, the government is willing to gain revenues and funds from leasing land, such as public facility, with higher prices. The increase in fixed assets investment may mean that the government has more financial resources to construct and improve public facilities [12], such as schools, hospitals, parks, and transportation infrastructure, which are essential to improving residents’ quality of life and social welfare. Compared to industrial land, the development risk of public facilities land may be lower because it is usually not affected by market fluctuations. The government may wish to balance the impact of industrial development and promote economic diversification through investment in public facilities.

Compared with industrial land, the land for commercial or residential purposes is more likely to be leased through the government at the district level when the employees’ average wage is low, which can promote the district’s economic development by leasing commercial and residential land to achieve more economic benefits. Land leasing is beneficial for the development of industrial enterprises, but may have a crowding-out effect on the income of the household sector [44]. Commercial and residential land development can provide more employment opportunities for local residents, especially in the construction, service, and retail industries. The development of commercial and residential projects may increase the income level of local residents [20], thereby driving consumption and economic growth. In an economic downturn or a slow industrial growth, the government may wish to stimulate economic growth and alleviate economic pressure by transferring commercial and residential land.

Commercial land is more likely to be supplied by the district government than industrial land when a district chief has a shorter tenure. The district government can obtain more short-term revenues from leasing commercial land because the prices are generally higher than those of industry [8]. Commercial land development often brings more direct and obvious social and economic benefits, such as commercial prosperity, increased employment, and increased taxation, which generates more substantial and direct income [39], and can serve as achievements during the tenure of the district mayor. Commercial land usually brings higher land transfer fees and tax revenue, a vital consideration for district chiefs who must balance their budgets and increase fiscal revenue during their tenure. Developing industrial land requires a longer time for planning, construction, and production. In contrast, the development cycle of commercial land is relatively short, which is more in line with the short-term and long-term time frames. Industrial projects may involve higher investment risks and more extended investment payback periods, while the development risks of commercial land are relatively low.

Based on the results of land accessibility, compared with industrial land, public facilities land is more likely to be supplied by the government while far away from the district government. Transferring public facilities land in areas far from government centers can help equalize services [46]. Areas far from the central area may have better natural environment and air quality, making them more suitable for building public facilities such as schools, hospitals, and parks. Building public facilities in areas far from the central area can alleviate traffic pressure in the central area and promote balanced development of transportation infrastructure. With the acceleration of urbanization, areas far from the central area may experience rapid population growth, requiring more public facilities to meet the needs of the new population, guiding the transfer of population and resources to the outskirts of the city, and alleviating population and environmental pressure in the central area [46].

While the land is close to Shenzhen Bao’an International Airport, industrial land is more likely to be leased by the government than residential land. The environment near the airport can provide more convenient internal and external transport of the products [42], and many industrial parks or logistics parks, such as Zhongcheng Future Industrial Park and Fuhai Technology Industrial Park, are located near Bao’an International Airport. Industrial land can help attract and promote aviation-related industries, such as aviation manufacturing, maintenance, logistics, etc., thereby driving economic growth. Certain safety zone requirements around the airport, such as clearance requirements, restrict the construction of high-rise buildings, which limits residential development in these areas.

While the land is near municipal key high schools, commercial land is more likely to be supplied by the government than industrial land. The environment near the key high schools is more suitable for developing commerce. For example, two shopping malls near Guangming High School can attract students and parents, and some educational institutions are set up in these malls. There is a demand for commercial facilities such as bookstores, stationery stores, catering, and educational services around the school, and commercial land can provide more community service facilities, such as supermarkets, banks, post offices, etc. These facilities can provide convenience for students and surrounding residents. The traffic flow around the school may be relatively high, and the development of commercial land can better cope with this flow by providing necessary transportation and parking facilities. Commercial land can be used to construct supporting facilities related to education [43], such as educational and training institutions, student activity centers, etc., which can enrich students’ learning and lives. The government may consider the long-term development and expansion needs of schools to avoid restrictions on the future development of schools caused by the development of industrial land.

The government prefers leasing residential and public facilities land over industrial land near public parks. Parks are green spaces in cities that provide critical ecological services, such as purifying air, regulating climate, and providing leisure spaces. The government tends to protect the environmental quality of these areas and avoid potential pollution from industrial activities. The national ministry stipulates the percentage of green area coverage that the residential greening rate should be at least 30% in the Urban Residential Areas Planning and Design Code. Also, nearby green spaces and public parks are very important for the residents to have a better environment. The possibility of developing residential properties near the park is higher, which may be due to the attraction of the park [47]. The development of residential and public facilities land can improve the residents’ quality of life, and provide a comfortable living environment and convenient public services, which complement the leisure and health functions of parks. Therefore, residents may be more willing to choose residential areas close to parks because such living environments are more pleasant. The government’s transfer of residential land can meet market demand while increasing land transfer income. Public facilities such as libraries, community centers, sports venues, etc. can work together with parks to form the social and cultural center of the community.

Compared with the other three types of land, industrial land is more likely to be transacted through the government while the land is close to the industrial park. Industrial parks generate noise and pollution; thus, residential and other types of land are usually far away from industrial parks, and industrial land is mainly close to the industrial parks. Furthermore, industrial development is in opposition to mixed use, retail, commercial, office, and residential development [48]. From Figure 6, some industrial parks, such as Baoneng Science and Technology Park and Shenzhen Bay Science and Technology Ecological Park, form an agglomeration effect in Shenzhen. The concentration of industrial land can promote the agglomeration of related industries, form an industrial chain, and improve production efficiency and competitiveness. To reduce operating costs, industrial enterprises can share the infrastructure within the industrial park, such as water supply, power supply, wastewater treatment, etc. Providing industrial land near industrial parks can reduce the commuting distance and time of employees, and improve their quality of life, as well as improve land use efficiency [38]. The government may wish to better control and reduce the risks that industrial activities may bring by centralizing the management of industrial land.

From the discussion above, the attributes of the land, district-level economy and polity, and land accessibility affect government decisions on leasing land. Regarding the land attributes, the land lease price affects leasing positively; the land area and mode of leasing influence leasing land negatively; and the district’s location positively influences residential land while negatively impacting commercial and public facilities land. Regarding the district-level economy and polity, foreign investment and investment in fixed assets positively affect leasing land, and the government fiscal gap, average wages of employees, and tenure of district chief influence leasing land negatively. For land accessibility, the distances between land and district and city centers, and the closest metro station, municipal key high school, and public park affect leasing land negatively, and the distance between a piece of land and the airport (or the closest industrial park) positively affects leasing land.

From Table 4, the proportion of industrial land leasing is far more than land for residential, commercial, and public facility purposes. Therefore, industrial land tends to be leased through the district government in Shenzhen more often than other types of land uses. Industry, especially manufacturing, is booming in Shenzhen [49]. According to Shenzhen Statistical Yearbook, Shenzhen’s total industrial output value above the designated size jumped from CNY 2 trillion in 2011 to CNY 4.5 trillion in 2022, ranking first among China’s large and medium-sized cities. Shenzhen has always valued the development of industrial and high-tech industries. Therefore, the district government in Shenzhen supplies more industrial land to develop industries, such as factories, manufacturing, and other emerging industries, and to achieve long-term benefits. This helps attract and cultivate key industries and promote economic growth, which is beneficial for the development of industrial enterprises [44]. The development of emerging industries has caused the government to transact more industrial land and focus on the efficiency of existing industrial land. By developing concentrated and contiguous industrial land, Shenzhen can promote the clustering of upstream and downstream enterprises in the industrial chain, enhance the effect of industrial clusters, and enhance overall competitiveness. Developing industrial land can drive employment, increase taxes, promote sustained regional economic growth, and directly contribute to local government’s fiscal revenue. Public facilities land is supplementary to residential, commercial, and other land. It is usually needed to meet specific social service functions, and its planning and construction need to comprehensively consider various factors, including population distribution, service demand, etc. Compared to industrial and commercial land, public facilities land may involve a more prolonged investment and return cycle, and the government may need to plan and sell more carefully. The government’s transfer decisions may be influenced by market demand, and the market demand for industrial and commercial land may be more urgent and significant. Thus, the probability of it is relatively low among the four land types.

6.3. Validation of the Results

In order to validate the results in Table 3, we eliminated the variable, i.e., municipal key high school, from Equation (1).

Table 5. Validation results.

Use Type of Land	Variable	Coefficient	Standard Error	z-Statistics
Residence	Land attribute
	LA	−0.47209 *	0.24175	−1.95
	TP	5.21884 ***	0.52144	10.01
	LC1	−0.92726 *	0.50529	−1.84
	LC2	−0.2535	0.35737	−0.71
	MD1	−3.78762 ***	1.0927	−3.47
	MD2	−1.80104	1.12337	−1.60
	District-level economy and polity
	FV	0.50517 ***	0.19505	2.59
	FX	−0.04481	0.21298	−0.21
	WG	−0.35545 **	0.17499	−2.03
	FG	−0.20145	0.15781	−1.28
	DC	−0.13863	0.11799	−1.17
	Land accessibility
	DG	0.01601	0.11219	0.14
	CC	0.07177	0.11654	0.62
	AP	0.40482 **	0.18119	2.23
	MS	0.0967	0.16668	0.58
	PP	−0.29055 **	0.1222	−2.38
	IT	0.51426 ***	0.1189	4.33
	Constant	3.62806 ***	1.13144	3.21
Commerce	Land attribute
	LA	0.02321	0.07652	0.30
	TP	5.09976 ***	0.51943	9.82
	LC1	2.05596 ***	0.45962	4.47
	LC2	−0.29543	0.39009	−0.76
	MD1	−2.57864 *	1.37914	−1.87
	MD2	0.15296	1.4135	0.11
	District-level economy and polity
	FV	0.16215	0.18445	0.88
	FX	0.12431	0.20571	0.60
	WG	−1.06458 ***	0.18428	−5.78
	FG	−0.54493 ***	0.16205	−3.36
	DC	−0.19209 **	0.1013	−1.90
	Land accessibility
	DG	0.15538	0.12763	1.22
	CC	0.10237	0.13277	0.77
	AP	0.30805	0.19191	1.61
	MS	−0.66484 ***	0.22536	−2.95
	PP	−0.04623	0.13203	−0.35
	IT	0.49009 ***	0.11743	4.17
	Constant	1.65224	1.41819	1.17
Public facility	Land attribute
	LA	−7.7537 **	3.34713	−2.32
	TP	2.63477 ***	0.89546	2.94
	LC1	2.28759 ***	0.82636	2.77
	LC2	−0.31891	0.6475	−0.49
	MD1	7.84074	329.5867	0.02
	MD2	11.4504	329.5872	0.03
	District-level economy and polity
	FV	0.01451	0.30191	0.05
	FX	0.64599 *	0.33708	1.92
	WG	−1.15594 ***	0.28005	−4.13
	FG	0.02617	0.30231	0.09
	DC	−0.01949	0.17589	−0.11
	Land accessibility
	DG	0.80619 ***	0.27134	2.97
	CC	−0.41221 *	0.2457	−1.68
	AP	0.68288 **	0.3404	2.01
	MS	−0.582612	0.39872	−1.46
	PP	−0.60387 *	0.32865	−1.84
	IT	0.5385 ***	0.19724	2.73
	Constant	−12.60416	329.588	−0.04
Observation	936
$R^2$	0.3436

Note: ***: Significant at the 1% level; **: significant at the 5% level; *: significant at the 10% level.

illustrates the validation results.

Regarding the results in Table 5, compared with the results in Table 3, the significant variables and the sign of the coefficient of the variables are the same, except that the variable of metro station for commercial land shows very significant in Table 5, and it is significant according to Table 3. Thus, the results of the model in Equation (1) are proved to be valid.

Compared with other similar studies, the result that land with a larger area for industry is more likely to be leased by the district government is in accordance with Dai et al. [44]. The results of lower industrial land prices are identified by Jens [45]. Residential land is more likely to be leased in urban peripheral and suburban districts and land leased for commerce and public facility in center districts, which is the same as Cheng [1], Sharma et al. [32], and Wu et al. [46]. The government tends to have land leases for residence and commerce by tender and auction, which is in accordance with Zhang et al. [36]. The government attracts foreign investment through industrial land leases to obtain long-term incomes, which is identified by Jin et al. [21]. The result that the district government prefers leasing commercial land in short is the same as Yuan et al. [8] and Cai [39]. Industrial land is more likely to be leased around the airport, consistent with Cheng [42]. The result that the government prefers leasing residential and public facilities land near public parks is in agreement with Zheng et al. [43] and Brambill et al. [47]. Industrial land is more likely to be transacted through the government when the land is close to the industrial park, which is in accordance with Cheng [42].

7. Conclusions and Implication

This study employs a multinomial logit model to explore the influencing factors guiding the government’s decisions to lease various types of land in Shenzhen, China.

The data were collected from 2005 to 2021. The land attribute, district-level economy and government, and land accessibility are considered as the influencing factors in the model. The results of the multinomial logit model were obtained using the software Stata. After analyzing the results, the influencing factors that the government decides to lease different land types in Shenzhen of China were obtained.

It is shown that the land attribute, district-level economy and polity, and land accessibility affect government decisions on leasing land. Regarding the land attributes, the land lease price affects leasing positively; the land area and mode of leasing influence leasing land negatively; and the district’s location positively influences residential land while negatively impacting commercial and public facilities land. Regarding the district-level economy and polity, foreign investment and investment in fixed assets positively affect leasing land, and the government fiscal gap, average wages of employees, and tenure of district chief influence leasing land negatively. For land accessibility, the distances between land and district and city centers, and the closest metro station, municipal key high school, and public park affect leasing land negatively, and the distance between land and the airport (or the closest industrial park) positively affects leasing land. In addition, compared with other types, industrial land tends to be leased through the district government.

The following policy recommendations are derived from the results.

Firstly, the government should encourage foreign investment and provide preferential policies to absorb the investment from foreign companies to supply more industrial land in the long term, and land for commercial and residential purposes would be leased to provide immediate benefits in stimulating economic development. The government can increase fixed assets investment and improve infrastructure construction and public service level. At the same time, the government can finely manage its finances, arrange fiscal revenue and expenditure reasonably, and reduce the negative impact of fiscal gaps on land transfer. Furthermore, a land transfer responsibility system can be established during the tenure of the district chief to ensure the continuity and stability of land leasing decisions.

Secondly, more residential land should be leased through the government in urban peripheral districts and suburban districts to alleviate the oversupply of residential land in center districts and improve urban livability. The center districts can be built as an area with commerce as the core and other public facilities as auxiliary.

Thirdly, the transportation and public facilities in urban peripheral and suburban districts should be balanced with the residential land supply. The metro stations should be strengthened, the accessibility of land to important locations should be enhanced, and the facilities for schools, public parks, green areas, etc., should be increased with the residential land supply and corresponding population through government investment. Industrial parks can be planned around existing industrial parks to form industrial agglomeration, promoting industrial development and reducing the waste of existing resources. The government can formulate specialized industrial land planning and policies to ensure the effective utilization and industrial development of industrial land, and strengthen the environmental protection and green development of industrial land.

Fourthly, the government can establish a monitoring and evaluation system for land leasing decisions to regularly evaluate policy effectiveness and promptly adjust inappropriate policy measures. In addition, when formulating land leasing policies, regional development strategies and plans should be comprehensively considered to ensure that land leasing is consistent with regional development goals.

This study provides recommendations for the governments on their decisions on leasing land for improving the efficient use of construction land, promoting sustainable urban development, and balancing government revenues and reasonable land allocation.

The limitations of this study are as follows: (1) The subdivided land use types have yet to be considered in this paper to obtain more detailed government decisions on leasing different types of land; (2) the multinomial logit model used in this study is a supervised learning method, and an unsupervised learning method has not been considered. Future research should consider the subdivided land use types to analyze more detailed government decisions on leasing land, and machine learning methods with unsupervised learning can also be applied to predict the land use structure and pattern.

The methodology and policy implications might be applied to Chinese metropolitans and countries with similar backgrounds and policies of public land ownership.