Land Value and Parcel Size in China:
Regional Variation and the Impact of Supply Constraint
Helen X. H. Bao[(]Department of Management Sciences,
Faculty of Business,
City University of Hong Kong
Kowloon, Hong Kong
Tel: (852) 2788 7269
Fax: (852) 2788 8560
E-mail: / Yong Tu
Department of Real Estate
National University of Singapore
Singapore
Tel: 65 68744455
Fax: 65 67748684
E-mail:
Structured Abstract Prepared for the AREUEA 2006 Meeting
Abstract
The purpose of this study is to cast light on the relationship between land price and parcel size under restricted land supply in China. Theoretical and empirical evidences indicate that the pattern of value-size relationship is generally convex first (due to subdivision) and then concave (caused by land assembly). These studies were conducted without considering any supply constraints. However, there are regions where land supply is strictly controlled by the government (e.g., a leasehold system). Under this circumstance, the value-size pattern may be more straightforward as land assembly and subdivision are more difficult to achieve than that in a freehold system. The hypotheses constructed in this paper are as follows. First, the concave functional form will not be observed under restricted land supply. Second, a log-linear functional form is sufficient to model the price-size relationship in a leasehold system.
The hypotheses are empirically tested with data from nine major cities in China between January 2002 and February 2005. Thin-plate splines semiparametric models are constructed to reveal any non-linear patterns in the data. To address the regional heterogeneity, separated models are estimated for the three regions in China (Eastern, Central and Western regions). This will facilitate the understanding of the regional dynamic in the China land market.
This study adds value to the literature by investigating land value-size relationship under restricted land supply. The empirical findings also provide insights into determinants of land value in developing countries, such as in China. It is particular useful for companies and investors who are interested in venturing into the China market. More specifically, foreign companies need to determine the land value when their Chinese partners, usually state-owned enterprises, using state land as their stake in the joint venture. The findings in this study are of both theoretical and practical importance.
Keywords: Land value; Parcel size; Semiparametric; Splines; China.
J.E.L. Classification: C14, R14
Introduction
Urban theories usually emphasize the distance from Central Business Districts (DBD) when assessing land values (see, for example, O’Sulivan, 2003). The effect of parcel size received relatively less attentions in the literature and in practice. Assessors often overlook parcel size factor when predicting land values. Therefore studies on the relationship between land price and parcel size are of both theoretical interests and practical significance.
The theoretical work by Colwell and Sirmans (1993) argued that the value-size relationship is “convex (concave) whenever land assembly (subdivision) is possible”. More specifically, the value-size function will exhibit convexity for smaller land parcels and concavity for larger ones. Under this framework, a log-linear hedonic function with a varying parameter for the logarithm transformed parcel size is sufficient to capture the price variation (Thorsnes & McMillen 1998, page 235). However, there is hardly a unanimous voice from the abundant empirical evidences in the literature. For example, Brownstone and De Vany (1991) found a negative relationship between parcel size and land price, suggesting a strictly concave pattern. On the other hand, Colwell and Munneke (1999) identified a convex price structure within the urban center of Cook County, Illinois, US. As argued by Thorsnes & McMillen (1998), the discrepancy in the empirical evidences might be resulted from the small sample size and inflexible functional form. Table 1 in the Appendix gives a summary of four selected empirical studies on land value and parcel size. Apparently most studies are based on rather small data sets and simple functional form. Technically, the data set should cover a broad range of parcel size to facilitate a thorough investigation. A small data set that consists of land transaction records in a single city cannot fulfill this task. Inflexible functional form may also introduce misleading results. Thorsnes & McMillen (1998) is the only example that used a flexible functional form to capture the nonlinear relationship between parcel size and land value.
The Colwell and Sirmans (CS) framework is established with an assumption that land parcels can be traded freely in the market. Some researchers refined this model by considering other factors that may affect the price-size relationship. For example, Colwell and Munneke (1999) suggested that the distance to CBD has an effect on the degree of concavity of the value-size function. However, it is unclear whether the restrictions on land supply will have an effect on the value-size relationship. Existing empirical evidences are also primarily based on the US market, where all land are freehold properties. There are countries and regions where land supply is strictly controlled by the government (e.g., a leasehold system). Would the CS framework still be valid when land supply constrains allows very limited land assembly and subdivision? This paper intends to answer this question with empirical evidences from China.
The China Land Market
The setting of this paper is in the People’s Republic of China, one of the fastest growing countries in the world. After China opened its door in early 1980s, the country has been experience fast economic growth and drastic changes. Urban grown and the development of land market are accelerated by the economic reform (see, for example, Anderson & Ge 2004, and Ding 2004).
All land in PRC is owned by the government. Land transaction is illegal before 1987, when the first land use right was granted in Shenzhen. In 1980s, as PRC gradually opens its door, a leasehold land system has been experimented in coastal cities first, and then extended country-wide. Before 2000s, public land is distributed primarily through private treaties, where land price are determined by negotiations between the government and the land users. As the land market is still at its infant stage, the government found the challenges of determining land price are tremendous. Also, the private treaty transaction channel also gives rise to corruption and bribery. New regulations are formulated recently to fade out private treaty as a mean of public land distribution. Nowadays public auction and tender are the major channels to sell public land.
A land lease will be issued to the winners in public auctions or tenders. The lease term is 70, 50 and 40 years for residential, commercial and industrial development respectively. A premium will be charged upon the grant of the land lease. There is no annual rent attached to the lease. Leaseholders are required to pay miscellaneous fee annually, which vary among provinces. However, the annually fee is nominal and not associated with the land price. All land leases are non-renewable.
The land market in China is a two-tier system consists of a primary market where government grants public land to users, and a secondary market where land users exchange their land. By law all land sales must be publicly announced before and after the transaction. The land sales records are also open to public review.
The Effect of Restricted Land Supply
If all land are privately owned and can be traded freely, land assembly and subdivision would evolve as a mean to maximize the land values. Additional costs will be generated by land assembly and subdivision though. For example, to subdivide a land parcel, certain infrastructures such as street and electric should be installed. To recover these costs, the aggregated price of subdivided land parcels should be higher than the original land price, and the price of the assembled land mass will be greater than the aggregated land price before assembly. Using a simple arbitrage equilibrium, Colwell and Sirmans (1993) shown that land assembly causes convexity and subdivision contributes to concavity in the value-price function.
If land is solely owned by the government, as the case in China, land users are usually granted the right of use, income and transfer of the land, instead of the ownership. A land lease will be issued to the user with conditions on how the land should be used. No revision on the conditions is allowed without government approval, and additional charges will be imposed upon approval. Consequently, land users in a leasehold system have much less flexibility to utilize their land. Subdivision and assembly can be difficult or even infeasible, as these may not be allowed in the lease conditions. Even the government approves the changes of lease conditions, the additional costs may make the subdivision or assembly financially impractical.
In the absence of subdivision and assembly, one may suspect that the relationship between parcel size and land value would be close to linear or even disappear. However, the land supply under a leasehold system is usually very restricted. This causes the expectation of future shortage of land supply. Land is scarce, and therefore highly valuable in a leasehold system. Smaller land parcels are more affordable and sold at a premium. The value-size function should exhibit certain convexity.
Data
The data used in this paper are obtained from the State Land Administration in nine major cities from China, as shown in Table 2. The central government divided the country into three regions: Western, Central and Eastern. The Eastern region consists of 12 costal provinces that are the pillar of national economy. This region covers only less than 15% of the total land area, but contains more than 40% of the population and contributes to over 60% of annual national GDP. The 10 provinces in the Western region, however, are among the most under-developed cities in China. Although the land mass in this region is almost one half of the total land size in China, its contribution to the national economy is merely about 10% in terms of GDP. We suspect that the land-size pattern will be different in these regions in that the land supply is much restricted in the coastal area. Three major cities are selected from each region in order to have a thorough understanding of the overall land value determination mechanism in China.
The sampling period in this study is from 2002 to 2004. To account for regional heterogeneity, indicators of economic and social development in each city are also included in the model. These statistics are obtained from the Statistics Year Book published by each city in various years. Table 3 gives definition of variables and Table 4 contains descriptive statistics for the variables.
Methodology
To better review the non-linear value-price pattern in the China market, a semiparametric functional form is adopted using thin-plate spline technique. The model to be estimated is as follows.
(1)
where L_PRICE is the log-transformed sales price per squared meter, L_AREA is the log-transformed parcel size, X is the matrix of all regressors expect for parcel size, and are the unknown parameters and functional form to be estimated. Model (1) combines a known function of the parameters of interest with a nonparametric function.
Under the spline smoothing framework, and are found through minimizing
+ (2)
The first term in (2) measures goodness of fit while the second term penalizes the roughness of the resultant surface so that a trade-off between the fidelity to data and smoothness can be accomplished; is a measure of the overall roughness of . One common smoothing splines estimator is the thin-plate spline, for which,
(3)
l is the smoothing parameter that controls the smoothness of the resultant surface. The value of l will be determined through a cross-validation procedure introduced by Craven and Wahba (1979). Statistical properties of splines have been well developed and discussed in detail in Wahba (1990) and Green and Silverman (1994).
Future Work
The conceptual framework of this paper needs to be refined. A theoretical framework will be developed to reflect property rights development in China Urban land market. Also, empirical analysis of the China data set will be finished. Some statistical tests are necessary to investigate whether the semiparametric model introduces significant improvement over a traditional log-linear model. Finally, the two hypotheses in this paper will be tested with the empirical evidences.
Appendix:
Table 1: Summary of selected empirical works on the price-size relationship
Authors / Sampling period / Country / Region / Sample size / Functional formBrownstone and DeVany (1991) / 1977-1985 / Southern California, US / 85 / Log-linear
Kowalski and Colwell (1986) / 1975-1983 / Detroit, US / 24 / Box-Cox
Thorsnes Paul and Deniel P. McMillen (1998) / 1980-1987 / Portland, Oregon, US. / 158 / Kernel Semiparametric Model
Colwell and Munneke (1999) / 1986-1993 / Cook County, Illinois, US / 2678 / Log-linear
Table 2: Land transaction records in nine major cities from China
EASTERN / Beijing / Beijing / 1203
Shanghai / Shanghai / 273
Guangdong / Guangzhou / 276
CENTRAL / Jilin / Changchun / 162
Jiangxi / Nanchang / 76
Hunan / Wuhan / 136
WESTERN / Sichuan / Chongqing / 785
Guizhou / Guiyang / 61
Sichuan / Chengdu / 100
Total / 3072
Table 3 Variable Definition
AUCTION / =1 if the land is sold through auction and 0 otherwise
TENDER / =1 if the land is sold through tender and 0 otherwise
AREA / Lot size (square meter)
COM / =1 if the land is leased for commercial development
RES / =1 if the land is leased for residential development
PUB / =1 if the land is leased for public projects
MIX / =1 if the land is leased for mixed usage
PLOT_RATIO / Plot ratio of the land parcel
PRICE / Sales price in 10,000 RMB
Y2003 / =1 if the land is leased in year 2003
Y2004 / =1 if the land is leased in year 2004
Y2005 / =1 if the land is leased in year 2005
DISTAREA / Area of the district where the land is located (sq. km)
POPDEN / Population density (person per square km)
MIGRANTS / Number of temporary residents (10000 persons)
LIBRARY / Capacity of public library (number of volumes)
INDUSTRIAL / Gross output value of industry (10000 RMB)
CONSUMPTION / Total retail sales of consumer goods (10000 RMB)
FDI / Total amount of foreign capital actually used (10000 USD)
INCOME / Annual discretionary income per urban residents (RMB)
HOSPITAL / Number of hospital beds
GDP / Gross Domestic Product (10000 RMB)
SCHOOL / Number of kindergartens and primary schools
TEACHER / Number of full-time primary school teachers
Table 4 Data Description