Faculty of Economics and Business
Master's Thesis
for MSc International Financial Management
Double Degree with Uppsala University 2011-2013
Determinants of Housing Price Inflation in
China and 15 OECD Countries
Xueyang Liu
S2093960
23
th, January, 2013
Contends
Abstract ... - 1 -
1.Introduction ... - 1 -
2. Literature Review... - 6 -
1. Market Development of China from 1949 to 2002 ... - 6 -
2. Important Policies to Constrain Housing Price from 2003 to 2011 ... - 8 -
3. Key Determinants of Housing Price ... - 9 -
1) Demand Factors ... - 10 -
2) Supply Factors ... - 12 -
4. Country specific ... - 13 -
3. Empirical Study ... - 15 -
1. Country Level Model ... - 15 -
1) Regression Model ... - 15 -
2) Data Sources and Description of Country Level Model... - 16 -
2. City Level Model ... - 19 -
1) Regression Model ... - 21 -
2) Data Sources and Description of Country Level Model... - 22 -
4. Empirical Results and Analysis ... - 23 -
1. Country Level Model ... - 23 -
2. City Level Model ... - 29 -
5. Discussion ... - 33 -
6. Conclusion and Recommendations ... - 36 -
Abstract
The purpose of the study is to explore whether there any differences exist in the responsiveness of housing price deviations to the changes in housing price determinants at both a country level and a city level. Motivated by this, I take a comparative approach by using OLS regression models which apply panel data of different objects in this study. The estimation result of the country level model which compares the housing markets between China and 15 OECD countries over the period of 2000 to 2010 shows that the growth of a particular macroeconomic factor such as interest rate or disposable income has different effect on housing price inflation in the two objects of this study. Most remarkably, this paper suggests that the measures of adjusting interest rate to control the inflation of housing price in China have not achieved the expected results. Furthermore, the city level model of 35 major cities in China during the period of 2006 to 2009 indicates that the heavy burden of houses purchasing not only exists in the most developed cities but also extends to the less developed ones. Finally, some political implications such as interest rate should be marketized and that the welfare housing system should be promoted are provided.
JEL classification codes:
I38, R20, R30, R31, R38, R50
Keywords:
- 2 -
1. Introduction
Since the abolishment of the welfare housing system in 1998, the Chinese housing market had been gradually transformed into market-oriented along with a significant increase in demand. Besides, a huge rush of moving into cities and a high rate of baby-born raised urbanization from 30% to about 47% over the latest decade, which exacerbated the imbalance between housing demand and supply. Moreover, a continuous fast growing of GDP with an average of over 10% annual growth rate hiked the disposable income from RMB 6,860 to RMB 19,100 over the decade from 2001 to 2010 and contributed to the booming of housing price. The nominal housing price went from RMB 2850/m2 to RMB 6830/m2, suggesting a 12% annual growth rate. During this period, although the government remains the owner of land, the right to decide how to develop the land had been passed to developers. The privatization of housing market has accelerated its fast growing. Along with the rapid growth, related financial development intended to stimulate housing market such as lower interest rates, longer mortgage maturities and lower down-payment should be noticed. The mortgage-to-GDP ratio in China, although still lower than that of 50% in Western developed countries, also experienced a significant take off from the late 1990s.
Since 1997, the fast development of housing market has promoted other industries including machinery, steel, electronics, chemical products and architectures. Through interactions and relationships, these different sectors formed an "economy cycle" where the housing market is in the center (Fung et al., 2010). The stimulation of a rapidly developing housing market is meaningful for the economy in China where the consumption only constitutes to 40 percent of the whole economy in comparison with 60 to 80 percent in the Western developed economies.
Thus, the housing price to disposable income ratio which has always been taken as a measure of housing affordability increased significantly during the last decade (IMF, 2010). According to Shen (2012), housing price to income ratios of China are 8.3 and 9.1, respectively, in year 2008 and 2009. In addition, according to the Chinese Statistics Yearbook of 2010 (2011), average housing price in China was RMB 6,830/ m2 while disposable personal income was RMB 19,110. Take a three-member-family which owns a 100 m2 house as an example, the housing price to income ratio is 12.0. In contrast, according to Demographia International (2010), the ratios are 2.9, 5.1 and 6.8, respectively, in US, UK and Australia. The existence of a huge ratio gap which suggests a less purchasing power of residents together with the fact of unprecedented housing price in recent years which imposes a heavy burden for residents drives more and more people to think about whether housing price is growing too fast in China.
Figure 1. China House Price Source:China Year Book 2001-2010
- 4 -
worrying along with the unprecedented increasing housing price show necessity and importance to make a deeper study on housing price in China.
The development trend of housing market in China which shows both a dominant position in the macro economy and an increasing risk to the macro economy coincides with the trends of many other countries, indicating a gradual integration and interrelation of housing prices in China and worldwide (Fung et al., 2010). Motivated by this, in this paper, I will dig deeper into both the similarities and differences of housing markets in China and other countries. Specifically, I will choose 15 OECD countries which have the most mature housing markets worldwide as a benchmark. The 15 OECD countries that I will study in this paper include the United States, Japan, Germany, France, Italy, Spain, Canada, Australia, Denmark, Finland, Sweden, the Netherlands, Ireland, New Zealand and the United Kingdom. By comparing the housing prices of the two objects of study, this paper gives a deeper understanding of the current housing market situation in China and also provides some related policy suggestions.
There is abundant previous literature which studies the determinants of housing price and its main findings are summarized in Appendix I. This paper combines the significant determinants suggested by these research. Specifically, a multi-factor econometric model adopted by many scholars (Schunre, 2005; Nagahata et al., 2004; Annett, 2005; Terrones and Otrok, 2004) to determine the dynamic of housing price is applied in this paper. In this model, factors from both demand side such as GDP, disposable income, rent, interest rate, inflation and population and supply side such as land cost and construction cost are taken as the determinants of housing price. Firstly, the model is applied to 15 OECD countries to get the coefficients of each determinant factor. Secondly, the model is applied to China, thus the corresponding coefficients of China are calculated. Finally, the different responsiveness of each factor in the Chinese housing market and the Western mature housing market is analyzed. In line with the study of Tsounta (2009), an error-correction mechanism will be embedded in the theoretical housing price determinant model automatically, if long-run data are applied. Thus, data from 2000 to 2010 at a yearly frequency is used to examine the housing price movement in this paper.
their effectiveness since they are rarely linked to the determinants of housing prices. Miles and Pillonca (2008) shed some light on this problem in their study of the measures to control housing price volatility in Europe by suggesting a new type of mortgage contract, which is inspired by their empirical findings that the mortgage rate shows the highest explanatory power of housing price movement. This gives me some insight and in this paper I will provide some useful suggestions to policy makers based on the importance of housing price determinants in China.
In addition, another potential drawback of existing literature lies in the fact that the studies of the housing market in China either focus on a country level through comparing with other countries or on a city level by comparing among different cities in China. However, both of the methods have some shortcomings. For the first one, it will cause potential discrepancies of estimation since the economy in China is marked with significant imbalance across regions. For instance, housing price and monthly disposable income are RMB 17,151/m2 and RMB 58,140; RMB 3,807/m2 and RMB 27,370 respectively in Beijing and Shijiazhuang which lies in the middle of Hebei Province. Therefore, the overwhelming housing prices of metropolitans to medium and small sized cities imply that a national average housing price may not be a proper representative. As for the second method, although it will not cause any deviation, it tends to ignore the Western housing markets which can provide some useful guidance for the development of the Chinese housing market. Thus, this paper will examine the housing price movement in China from the two levels.
Studying the housing price movement in China has many practical meanings:
Social issues: For the reasons that housing has grown to be less and less affordable for
the Chinese citizens and that housing is a necessity for every household, a proper control of housing price movement which will benefit both the citizens and the whole economy is a critical issue for the Chinese government.
Financial issues: Since most of the housing purchasers are financed by mortgage,
mortgage lending has already become one of the most important income sources for financial institutions. The development of housing market will have a potential impact on the healthy and stable operation of financial institutions.
Policy issues: Understanding the similarities and differences in the explanatory power
- 6 -
Specifically, this paper will shed some light on the following questions:
(1)What are the differences between the explanatory power of housing price determinants in China and the 15 OECD countries?
(2)What are the differences among the explanatory power of housing price determinants in 35 different cities of China?
(3)What are the political implications from the results of the previous two research questions?
The remainder of the paper will be organized as follows. Section 2 focuses on the development of housing market in China. Firstly, it offers a brief introduction of the development history and then it summaries related policies enacted by the government to control housing price as well as important previous studies about the determinants of housing price. In addition, a brief description of the housing markets in the 15 OECD countries is also included. Finally, two hypotheses that are regarded to the research questions are formulated. Section 3 provides a data description and describes the empirical model that is used to examine the potential different responsiveness of housing price movements to housing price determinants between China and the 15 OECD countries. Besides, this section also extends the model to a city level by comparing responsiveness of housing price movements to their determinants among 35 cities in China. Section 4 discusses the results of the empirical study and analyzes the drivers for the potential inconsistent results across different countries as well as different cities. Section 5 digs further into the regression model by testing whether the model still holds when certain variables are changed. Section 6 summaries the main conclusions. Moreover, based on the findings of the paper, this section also provides recommendations for policy reform to better control housing prices.
2. Literature Review
Understanding the Development of the Chinese Housing Market 1. Market Development of China from 1949 to 2002
house and the land of which the house is built on are considered as an entirety in China, therefore at that time the houses also belong to the government. Additionally, since the real estate industry was not marketized then, residents were not allowed to purchase houses privately. Thus, houses were firstly distributed to the state-owned enterprises (SOEs) and then allocated as basic welfare at very low cost to citizens who just had the right to use but without private ownerships (Zhang et al., 2012).
During the second stage, from 1978 to early 1987, reforms of state ownership housing systems ("Welfare housing System'') and rent systems appeared, which allowed for a personal involvement in the housing market. Specifically, government, the state-owned enterprises (SOEs) and individuals each accounted for one-third of the construction and maintaining costs of a newly-built house.
In the third stage, from late 1987 to 1991, a marketization was initiated, marked with the appearance of selling land-use right in some cities and provinces during the second half of 1987. Later, in April 1988, government showed the determination to encourage the development of housing market by amending the constitution to legalize the transfer of land-use right. In addition, a financial aid – housing Public Accumulation Fund (PAF) for employers to purchase houses was provided nationwide.
The fourth stage, from 1992 to 1997, saw the full-speed development of the housing market. The launching of economic development zone which was suggested by Deng Xiaoping relaxed the regulation restriction on transferring land-use rights. Besides, the enactment of Urban Real Estate Law, which is the fundamental legislation of the Chinese housing market, indicated the direction of development. Furthermore, bank loans, stimulation for residents to purchase houses, became available.
- 8 -
Attributed to these related policies implemented to stimulate housing market, housing prices experienced a huge hike from RMB 2,380/m2 in 1998 to RMB 2,850/m2 in 2002.
2. Policies to Constrain Housing Price from 2003 to 2011
The over-speed increasing of housing price led to growing worrying about the occurrence of a housing bubble. Therefore, the government enacted a series of policies which include interest rate adjustment, tax adjustment, and land-use transaction reform to regulate housing market and constrain the increase of housing price. Besides, more and more social welfare houses are provided to low-income families from mid-2003 onwards (Wang, 2011).
In March 2004, the government tightened the restriction on land allocation. Later in October, the People's Bank of China (central bank, PBC) raised interest rate by 0.27% to 5.58%, which was the first time that the PBC raised interest rate during the past decade. In March 2005, the State Council published the notice "Old National Eight Item" to control the increasing housing price and it was the first time that the government enacted a housing price control policy. According to the notice, government focused on optimizes the housing supply structure which increased the land supply for residential buildings and social welfare housings. In addition, the PBC announced the abolishment of preferential policies for housing loans and raised the down payment from 20% to 30% of the housing price in the fast developing markets such as Beijing and Shanghai.
In April 2006, the PBC raised the mortgage interest rate from 6.12% to 6.39%. Later, in May 2006, premier Wen declared the implementation of a country level control for housing price – "New National Six Items" – which further deepens the tight constraint on land supply. Later, a new tax regulation on second-hand house transactions was announced in June. Moreover, in July, government published a provision which limited foreign direct investment in housing market.
In 2008, during the first three quarters, government continued with the tightening policy to control housing price and raised the required reserve rate three times from 14.5% to 16%. However, later housing price experienced a significant decrease due to the severely negative effect of the financial crisis. Faced with this, government quickly responded by decreasing required reserve rate twice from 16% to 15% and lowering interest rate three times to 4.05%. Moreover, some preferential tax policies were enacted, including reducing housing property deed tax from 1.5% to 1% and waiving stamp duty and land value-added tax for the first-time house buyers.
In the first five months of 2009, loose monetary policies to boost housing market were still in implementation. However, the growing incentive for speculation along with the surging housing price induces the enactment of tight policies to cool down the housing market. From June onwards, various tax preferential policies had been abolished.
Led by the rapid recovering of housing market, the growth rate of GDP in China surpassed 11% in the first quarter of 2010. Then government announced the most tightening policies in history to control the housing market in mid-April. These policies are mainly focused on administrative tools that include raising down payment from 20% to 30% of the housing price for the first-time house purchasers, prohibiting purchasing a second or more houses for homeowner, restricting requirements for purchasing houses in the non-resident cities unless a five year tax payment proof was provided by the purchasers and increasing the supply of social welfare houses for the low-income families.
In 2011, government continued to comply with the tightening policy which raised the down payment for the second-time house purchaser to 30% of the housing price. Besides, no more mortgage loans were available for the third-time house purchasers. Furthermore, the extension of property tax for ordinary commercial dwellings was fast and Chongqing, Shanghai got the first trials to implement the property tax.
3. Key Determinants of Housing Price
- 10 -
Hence, many previous studies divide the main determinants of housing prices into two categories: demand and supply. GDP, disposable income, population, interest rate and rent are the most important determinants from demand side while land cost and construction cost play vital roles in determining housing price from supply side (Tsatsaronis & Zhu, 2004). In the following, I will have a brief review of the factors mentioned above.
1) Demand Factors GDP
A lot of existing literature focuses on the interrelationship between housing price and GDP growth rate. According to Otrok and Terrones (2005), a 1% fall of housing price will have a significant effect on GDP growth rate. Tsatsaronis and Zhu (2004) suggest that GDP has a comprehensive explanatory power of housing price movement which summaries the information contained in other macroeconomic factors of wages and unemployment rate. In addition, Égert and Mihaljek (2007) summarize that the fast growth of GDP is a precondition for a significant development of housing market.
Disposable Income
Population
Since the housing price is influenced by the supply-demand relationship, the growth of population appears to show little interaction with the movement of housing price in a long run (Terrones & Otrok, 2004). However, since the 1990s, the fast growth of the population accompanied by the huge increase of housing price has indicated a potential relationship between the two factors. Moreover, the movement of the population aged 25-39, which indicates the household formation rate, is considered as a measurement for demographic pressure (André, 2010).
Rent
Controls and related policies that are targeted at maintaining a stable rent price will impose a potential constraint on housing price increase because the rental market will alleviate the demand of housing purchase to a certain degree (Mikhed & Zemčík, 2009). However, the comparable slower growth rate of rent is not sufficient to be seen as an indicator that an increasing number of people will choose renting rather than buying a house, as some other factors such as social and individual preferences are also involved here (Shen & Liu, 2004).
Interest Rate
According to Adams and Füss (2010), interest rate has a significant and negative influence on housing price as the opportunity cost of capital will influence potential buyer's expectation of housing price in the future. Since the rapid increase of housing investment in China, it has already become an important component of the Chinese economy. Policies such as the adjustment of interest rate which is intended to affect investment will result in a housing price movement (Zhang, et al., 2012; Schnure, 2005).
- 12 -
Inflation
According to Tsatsaronis and Zhu (2004), the inflation rate is the most important determinant of housing price variation. Considering the investment function of a house, Tsatsaronis and Zhu conclude that people tend to purchase houses to hedge against the potential losses of their savings due to inflation. In addition, purchasing a house which has a significant high growth rate net inflation rate will avoid the rising uncertainty caused by inflation when choosing other types of investment instead of houses. In addition, according to Andrews (2010), a higher inflation rate tends to be an attractive sign for potential housing purchasers, since it will erode part of real mortgage debt in a long run.
2) Supply Factors Land Supply Policy
In China, government is the sole owner of land and therefore it owns the right to decide upon the allocation of land. The land supply system under the Chinese planned economy means the quota administration of land allocated by the municipal government for housing market development and urban construction (Zhang, 2008). Since 2002, the Chinese government has carried out a series of macro-economic policies such as the regulation on quota of land supply to control the fast development of real estate market. Based on two economic models, Peng and Wheaton (1994) conclude that land supply restriction will lead to a higher expected rent and ultimately a higher housing price. Hui (2004) completes the theory by suggesting that housing price will go lower if a higher supply of land is provided in the previous years whereas no significant relationship between the two factors is observed in a one year period. Based on previous studies, Zhang (2008) digs further into the relationship between housing price and land policy and concludes that an increase in housing price will result in a growth of land supply in the short run, which will in turn generate a decrease of housing price in the long run.
local governments. Meanwhile, the proportion of land cost to housing price has risen quickly since 2008. According to Wu et al. (2012), the ratio of land cost to housing price for Beijing reached to 60% in 2010.
Construction Cost
An increase of construction cost such as a rise of construction material cost, labor cost or financing cost will result in a decrease of housing supply and thus a decrease of housing stock. The gradual decrease of housing stock later will lead to an increase of rent and will finally lead to a growth of the housing market (Derger & Zhang, 2010). Besides, according to Capozza & Helsley (1989) and Case & Shiller (1990), construction cost is positively related to housing price, since it will slow down the response speed of housing suppliers to the increasing demand of housing purchasers.
4. Country Specific
Zhu (2006) suggests that a lot of differences exist among various housing markets. In addition, the differences in housing finance markets and economic arrangements can be explained by the country-driven factors. Besides, according to Sanchez and Johansson (2011), differences across countries result from both policy factors and non-policy factors. For instance, land use planning which may result in different responsiveness to housing demand is an illustration of a policy factor while geographical and demographic conditions which may have significant influences on responsiveness to housing demand are instances of non-political factors.
- 14 -
housing markets in France, Japan and Italy have indicated significant correlations with business cycles since the 1980s.
However, except for the obvious differences, the OECD housing markets share a similarity that all of them had witnessed a soaring housing price during the last decade. According to André (2010), due to the long-term low interest rate, loose lending conditions and the mortgage equity withdrawal, housing prices reached historical highs in many OECD countries during the period from 1995 to 2006, especially in UK, New Zealand, Spain, France, Sweden, Denmark, the Netherlands, Finland and Australia, where the housing prices had been more than doubled. Although the price increase rates in Italy, Canada and the United States were relatively low, they still reach to 50%. Among the 15 OECD countries, only three countries including Japan, Korea and Germany have shown a continuous declining movement of housing price since the late 1990s. However, the growing trend which was considered to last forever halted firstly in the United States at the end of 2006 and then quickly spread to other OECD countries.
The development of housing markets in the 15 OECD countries during the latest decade differed from that during 1970-1990 in three aspects: Firstly, the annual growth rate of housing prices averaged to 120%, highly surpassed that of 45% in 1990; Secondly, the upward period of housing prices had lasted for more than 10 years on average, which significantly outpaced that of 20 years ago with an average of 6 years expansion phase; Thirdly, more than three fourths of the 15 countries experienced an annual housing price growth rate of over 25% while only a third of those countries in the past reached to a 25% of the annual housing price growth rate (André, 2010).
Based on the analysis above, two hypotheses regarding to the three research questions come up:
H0: The coefficients of the corresponding housing price determinants are the same in
OECD countries and China, which can be expressed as: βt = βt^;
H1: The coefficients of the corresponding housing price determinants are not all the
same in OECD countries and China, which can be expressed as: βt≠ βt^.
Where βt and βt^ denote the coefficients of housing price determinants in OECD
3. Empirical Study
In this paper, an OLS regression model which was adopted in many previous studies (Schunre, 2005; Nagahata et al., 2004; Annett, 2005; Terrones and Otrok, 2004) is applied to implement all the determinants which reflect consensus in the reviewed literature. Specifically, this approach can be divided into three steps. Firstly, the study uses a panel data model for 15 OECD countries including the United States, Japan, Germany, France, Italy, Spain, Canada, Australia, Denmark, Finland, Sweden, the Netherlands, Ireland, New Zealand and the United Kingdom during the period 2000 to 2010, whereby common coefficients βt of the corresponding housing price
determinants for each country are derived. Secondly, the OLS model is applied to the housing market in China during the same sample period and therefore βt^ which
denote coefficients of housing price determinants in China are calculated. Then a comparison between βt and βt^ follows. Thirdly, after a slight adjustment of housing
price determinants, a new model which uses a panel data from 2006 to 2009 is applied to 35 major cities in China.
There are also some limitations in this model. Firstly, since “housing market” is a very complex aggregation, countless factors will influence housing price. However, from a practical viewpoint, only some of the most common used factors based on previous studies can be included. Therefore, the explanatory power of all these factors for the movement of housing price is low due to the inter-subjectiveness and incompleteness of choosing proper indicators. Secondly, compared with the 15 OECD countries, China has a short history of housing market since the abolishment of the welfare housing system in 1998 and thus the related data is limited (Yan et al., 2007). Thirdly, due to the lack of related data, some important factors, for instance land supply information, couldn't be incorporated in this model. Besides, the model failed to take the constant changing elasticity of supply and demand into consideration, which may result in instability of the model (OECD, 2005).
1. Country Level Model 1) Regression Model
The model used to analyze housing price determinants in both the 15 OECD countries and China is as follows:
- 16 -
hpi,tα0β1GDPi,tβ2DisIncomei,tβ3Renti,tβ4IntRatei,tβ5Infi,tβ6ConCosti,tβ7POPi,ti,t
(Eq 1) where hp is annually growth rate of housing price for the commercial residential buildings, DisIncome is annually growth rate of personal disposable income, Rent is annually growth rate of housing rent, IntRate is annually interest rate growth rate and
Inf is annually growth rate of CPI which shows the degree of inflation. In this
equation, ConCost denotes annually growth rate of construction cost and POP is the proxy for annually nationwide growth rate of the whole population. One thing needed to be stressed in this model is that all the variables except for Inf (inflation rate) are expressed by year on year growth rates rather than absolute values. The most important reason for this is that this paper intends to dig deeper into the question about which factor's growth (decrease) will have a significant influence on the increase (decrease) of housing price.
Then, by applying the model (Eq 1) to China, potential differences of the corresponding coefficients between the 15 OECD countries and China which are expressed by βt and βt^ respectively are detected. However, there are only 11 data
points at annual frequency in the Chinese housing market. In order to alleviate the potential deviations resulting from the small sample, a city level regression model is applied to 35 Chinese major cities in later section.
2) Data Sources and Description of Country Level Model a. Data Sources
Based on related previous studies and in accordance with the determinants described in the literature review, Table 1 lists the following explanatory variables for housing price deviation applied in the empirical study model. Besides, the expected signs of regression coefficients which are set in square brackets are retrieved from the existing literature. Details including their sources are illustrated in Appendix I.
Disposable Income growth rate, Rent growth rate, Interest Rate growth rate (3-month lending interest rate), Inflation rate, Construction Cost growth rate and Population growth rate) and the dependent variable of Housing Price growth rate.
Table 1. Expected signs of regression coefficients from related previous studies
Determinants Expected signs of coefficients and related previous studies
GDP
(GDP)
[+] The growth rate of GDP is closely related to housing price movement (Otrok and Terrones, 2005; Égert and Mihaljek, 2007)
Disposable Income
(DisIncome)
[+] Case and Shiller (1990), Mikhed, Zemčík (2009), Wang et al. (2011), Andrews (2010), Schunre (2005), Nagahata et al. (2004), Annett (2005), OECD (2004), Terrones and Otrok (2004)
Rent (Rent) [+] Mikhed and Zemčík (2009)
Interest Rate
(IntRate)
[-] Schunre (2005), Nagahata et al. (2004), Annett (2005), OECD (2004), Terrones and Otrok (2004)
Inflation
(Inf)
[+] Inflation accounts for more than 50% explanatory power of housing price fluctuation (Tsatsaronis and Zhu, 2004).
Construction Cost
(ConCost) [+] Case and Shiller (1990)
Population
(POP)
[+] Case and Shiller (1990), Mikhed and Zemčík (2009), Terrones and Otrok (2004)
Notes: The symbols for the determinants expressed in the regression model are reported in parenthesis.
The data of all variables cover the range of 16 countries and last for 11 years from the period of 2000 to 2010. Considering the short period of study and the imperfection of housing market, this study should be defined as a short-run effects study. Table 2
summaries the definitions and sources of the variables for the regression model in this study. Besides, Appendix IIlists the detail information for those data that come from Datastream.
As it is shown in the table below, the data from Datastream of each variable for each country have different sources. Consequently, potential subtle variations may exist in the definitions of house price among the observations. Therefore, one pitfall related appears that the potential of not unified data may affect accuracy of the estimation results.
- 18 -
the sample period. It can be observed that the housing prices had experienced a significant increase in the majority of the 15 OECD countries (e.g. Spain, U.K., Canada, France) during 2000-2010, except for Ireland, Germany and Japan where the housing prices remain rather stable or even show a slight decrease.
Table 2. Variable Definition for OECD countries
Variable Definition Source
GDP
(GDP)
An aggregate measure of production and equals to the sum of the gross values added of all resident
institutional units engaged in production (OECD).
OECD database
Disposable Income
(DisIncome)
Derived by adding all current transfers, except social transfers in kind, receivable by that unit or sector and subtracting all current transfers, except social transfers in kind, payable by that unit or sector (OECD).
OECD database
Rent
(Rent)
Expressed by Consumer Price Index (CPI) of Housing
Rents. Datastream
Interest Rate
(IntRate)
Interest rate is the cost or price of borrowing, or the gain from lending, normally expressed as an annual percentage amount (OECD).
IMF database
Inflation
(Inf)
Expressed by Consumer Price Index (CPI) which measures changes over time in the general level of prices of goods and services that a reference population acquires, uses or pays for consumption (OECD).
IMF database
Construction
Cost (ConCost) Expressed by Cost of Construction Index for Residents. Datastream
Population
(POP)
The base population refers to the number of people in a given area (e.g. a nation, province, city, etc.) to which a specific vital rate applies, that is, the denominator of the crude birth rate or death rate; that population
determined by a census (OECD).
IMF database
House Price
(hp) Expressed by Consumer Price Index for Housing Price. OECD
Notes: The symbols for the variables expressed in the regression model are reported in parenthesis.
b. Data Description
indicating a taking off of the Chinese economy. As for the growth of construction cost, the rate in OECD countries significantly surpasses that in China. This suggests a potential relationship between the increasing housing price and construction cost, which will be examined in this study. One interesting finding is that China has an identical inflation rate with the average of the 15 OECD countries. When it comes to the change in population growth rate of the two observations, the comparable lower rate of China certifies the effectiveness of the birth control policy. Surprisingly, the average growth rate of rent inflation in China is negative. After checking the yearly data of this determinant, the negative mean value of China can be explained by the decreasing growth rate of rent at the beginning of the sample period.
Figure 2. OECD Countries House Price Source:OECD (2010-2011)
Table 3. Data Description of 15 OECD countries and China
Notes: The variables reported in the table are expressed as the symbols in the regression model.
15 OECD countries China
Variables Mean Std. Dev. Mean Std. Dev.
- 20 -
2. City Level Model
This paper contributes to the blank which has been mentioned in several previous studies that existing literature seldom studies housing prices from both country-level and city-level (Wang, 2012). In order to build up this gap, in this section, I classify the 35 major cities in China which have the corresponding official data of housing prices from the National Bureau of Statistics into 3 subgroups (the locations of the 35 major cities are shown in Figure 3). To do so, I group these cities according to their relative housing prices which mean the ratios of regional housing prices to the average nationwide housing price.
Figure 3. The locations of 35 cities. Source: Wu et al. (2012)
has remained in a relative lower level in terms of housing price but also has shown insufficient fast growing trend in the near future.
After classifying the 35 cities into three subgroups, this section focuses on finding out whether a responsiveness discrepancy of housing price to its determinants exists among these groups. Specifically, first I will apply the panel data which consist of housing prices and their determinants of the three subgroups to the regression model to derive the coefficient of each regressor individually. Then by comparing the coefficients at the regional level to those at the country level, it can be tested whether there is a potential deviation. Moreover, which determinant has the most significant different explanatory power among different cities is also to be found out.
Table 4. City groups for 35 major cities in China
City Groups City Lists (35 cities)
Fast Group
(10 cities)
Beijing, Fuzhou, Guangzhou, Hangzhou, Ningbo, Nanjing, Shanghai, Shenzhen, Tianjin, Xiamen
Average Group
(15 cities)
Changchun, Changsha, Chengdu, Chongqing, Dalian, Hefei,
Harbin, Haikou, Jinan, Qingdao, Shenyang, Taiyuan, Wuhan, Xi'an, Zhengzhou
Slow Group
(10 cities)
Guiyang, Hohhot, Kunming, Lanzhou, Nanchang, Nanning, Shijiazhuang, Urumqi, Xining, Yinchuan
1) Regression Model
The second step of the empirical model is to study the housing price determinants at a city level. An empirical model is established to study housing prices and their determinants by applying panel data from 2006 to 2009 for 35 major cities in China. The difference between the city level model and the previous country level model is that the independent variable of interest rate in the country level model is excluded from the city level model, which can be explained by the fact that the national unified interest rate is set by the PBC. Meanwhile, land cost, which according to previous studies (Wu et al., 2012; Abraham & Hendershott, 1993) is closely related to housing price in China, is also included in this model.
- 22 -
model. Besides, LandCost is a proxy for the new added independent variable of Land Cost growth rate in this model. Therefore, the city level model yields the following expression:
hpi,tα0β1GDPi,tβ2DisIncomei,tβ3Renti,tβ4Infi,tβ5ConCosti,tβ6POPi,tβ7LandCosti,t
(Eq 2) In addition, the effect of land cost on housing price volatility suggested by a previous study (Wang et al., 2011) is exhibited in the following Table 5.
Table 5. Expected sign of regression coefficient from related previous studies
Determinants Expected signs of coefficients and related previous studies
LandCost [+] Wang et al. (2011)
2) Data Sources and Description of Country Level Model a. Data Sources
As it is shown below, Table 6 summaries the variables of China in both the country level and the city level models. Almost all the data from Table 6 are retrieved from the National Bureau of Statistics which has been suffering from widely criticisms in recent years for failing to reflect real situations in China. Among all the variables, housing price is deemed as the one with the largest deviation from the real value. However, there is no alternative source that can provide cross cities data during the sample period. Thus, the problems related to the inaccurate data are drawbacks of the study, which are left for future improvement.
b. Data Description
Table 6. Variable source for China
Variable Data Source
GDP Derived from China City Statistical Yearbook and Statistic Yearbook
(2001-2011).
Disposable Income
Expressed by Disposable Income China per Capita from the National Bureau of Statistics of China.
Rent Expressed by Rent Building Rental Price Index from the National
Bureau of Statistics of China.
Interest Rate (3-month lending interest rate)Derived from People's Bank of China.
Inflation Expressed by Consumer Price Index from the National Bureau of
Statistics of China.
Construction Cost
Expressed by Retail Prices Index for building materials, hardware and electric materials from the National Bureau of Statistics of China.
Population Expressed by Population Index (year-end figures ) from the National
Bureau of Statistics of China.
Land Cost Expressed by land Transactions Price Index for Residential Use from
the National Bureau of Statistic of China.
House Price Expressed by Housing Price Index for Newly built residential
buildings from the National Bureau of Statistics of China.
Notes: The variables listed in the table are symbolized as GDP, DisIncome, Rent, Intrate, Inf, ConCost, POP, LandCost and hp accordingly in the regression model.
Table 7. Data Description for 35 cities in China
35 cities Group 1 Group 2 Group 3
Mean Std. Dev. Mean Std.Dev. Mean Std.Dev. Mean Std.Dev. HP 0.010 0.090 0.022 0.125 0.002 0.066 0.008 0.079 INF 0.002 0.039 0.000 0.035 0.003 0.040 0.004 0.041 POP 0.015 0.045 0.020 0.051 0.011 0.045 0.015 0.039 RENT 0.001 0.024 -0.003 0.023 0.001 0.025 0.004 0.022 GDP 0.180 0.054 0.158 0.042 0.192 0.057 0.183 0.057 LANDCOST 0.012 0.129 0.042 0.205 -0.002 0.083 0.005 0.076 CONCOST 0.001 0.060 0.003 0.055 -0.002 0.060 0.005 0.066 DISINCOME 0.127 0.061 0.111 0.069 0.132 0.050 0.137 0.067 Obs 140 40 60 40
Notes: Included regressors are INF (inflation rate), POP (population), RENT (rent of house), GDP, LANDCOST
(land cost of China), CONCOST (construction cost), DISINCOME (disposable income), while HP (house price) is the dependent variable.
4. Empirical Results and Analysis
1. Country Level Model
- 24 -
observations (across 15 countries and last for 11 years). Firstly, I estimate the model with the approach of a simple pooled regression which implies that neither fixed nor random effects are considered. The results are shown in Appendix III. One thing to notify is that the pooled regression model assumes that the intercepts are the same for each country and for each year. However, this assumption can't hold in reality and it is also contrary to related previous studies. As a consequence, the pooled regression model is not a proper choice in this study. Moreover, the Redundant Fixed Effects- Likelihood Ratio Test which is a useful tool in Eviews can be applied to determine whether a fixed effects model is necessary or not. The output which is reported in
Appendix IV indicates that a fixed effects model is appropriate. Secondly, as for the fixed effects model, it can be classified into two categories: cross-section fixed model and period fixed model. The two kinds of model assume a country-related interception and a period-related interception, respectively. Thirdly, as for the random effects model which is an alternative to the fixed effects model, the intercepts of each cross-section unit are different across sections but are constant over time just as those of the fixed effects model. However, the difference between the two types of models is that the intercepts in a random effects model are assumed to come from a common global intercept term "", which indicating a loose relationship between individual intercept and each entity.
Table 8. Comparison of results between OECD countries and China
Variable 15 OECD countries China
Constant -0.013 (-1.893) * 0.766 (2.989) * CONCOST 0.096 (1.547) -3.122 (-2.369) * DISINCOME 0.143 (1.521 ) -4.379 (-4.782) ** GDP 0.469 (4.924) *** 4.670 (3.054) * INF 0.579 (2.634) *** -3.406 (-2.342)1 * INTRATE 0.002 (0.659) 0.442 (2.938) * POP 0.651 (0.788) -1.156(-2.407) * RENT 0.237 (2.429) ** -1.621 (-1.764) Adjusted 2 R 0.408 0.773 Obs 165 11
Notes: (1) Included regressors are CONCOST (construction cost), DISINCOME (disposable income), GDP, INF
(inflation rate), INTRATE (interest rate), POP (population), Rent (rent of house). (2) T-statistics are reported in parenthesis.
(3) Significance at the 1%, 5%, and 10% levels is denoted as ***, **, and *, respectively.
proper choice. Consequently, in this paper a cross-section fixed effects model is applied and it is also adopted by almost all the scholars (Zhang, 2008; Hilbers, 2008) who study in this field. Table 8 reports the estimation result from the country level model (Eq 1), showing the effects of the changes in macroeconomic factors on housing price deviations in 15 OECD countries and China during the sample period.
The regression results are summarized as follows:
Construction Cost The result shows different effects of change in construction cost on
change in housing prices in OECD countries and China. The effect of change in construction cost on change in housing price in China is negative and significant (at 10% level). Specifically, the coefficient is -3 which means that a 1% growth of construction cost inflation is associated with a 3% decrease of housing price inflation. This can be explained by the fact that although both of the two variables had experienced a significant increase during the sample period in China, the growth of the continuous soaring housing price is faster than that of the construction cost. However, the relationship between the two variables is not significant in the 15 OECD countries, which can be explained by the fact that the construction cost in OECD countries accounts for a small percent of the total housing price. Moreover, since the study focuses on a short run effect and it is possible that the effect of change in construction cost has not been expressed in a change in housing price.
Disposable Income A lot of previous studies indicate that the increasing disposable
- 26 -
mature and most importantly, the demand for housing is more stable. Therefore, the decrease of disposable income growth rate coincides with the increase of housing price inflation rate in China while the relationship of the two variables is not significant in the sample OECD countries.
GDP Similar to previous studies that indicate GDP is significantly related to housing
prices, the results show significant relationships of the growth rates of the two variables in both OECD countries (at 1% level) and China (at 10% level). Moreover, the coefficient in China (5) is ten times than that in OECD countries (0.5), which indicate every 1% growth of GDP increase rate will cause a rising of housing price increase rate in China ten times as high as that in OECD countries. GDP growth rate is a sign of the whole economy development. With an average of 10% growth rate of GDP which is five times as high as that in OECD countries (2%) during the sample period (see Table 3), China shows a rapid development in both the whole economy and the housing market. Moreover, the two factors are closely interacted with each other: except for the promotion effect of GDP on housing markets, housing markets also cause a significant increase of the whole economy. Unlike China, the 15 OECD countries are in rather stable GDP development stages which are expressed by the low GDP growth rates. Hence, we can easily infer that the housing markets are not that closely related to the macroeconomic factors in OECD countries.
Inflation Rise of housing price inflation rates in 15 OECD countries shows significant
Interest Contrary to the expectation that housing price is negatively related to interest
rate, the relationship between the change in interest rate and the change in housing price is not significant in OECD countries whereas a positive and significant (at 10% level) relationship is observed in China. The coefficient of change in interest rate on change in housing price is 0.4 which indicates that a 10% increase of interest growth rate will lead to a 4% growth of housing price inflation. This is related to the fact that interest rate in China is not marketized, it is a policy rate set by the government. Moreover, government always realizes the purpose to control housing price growth through adjusting interest rate. For example, since 2004 when the PBC raised interest rate for the first time, raising interest rate has been a frequently used tool for the government to control the rising housing price, especially when the housing price grows rapidly. For instance, in 2007 when the growth rate of housing price reached to 10% which peaked at a historical high, government raised interest rate 5 times from 6.39% to 7.29%. The reasoning behind raising interest rate to control the housing price is that an increasing interest rate that will enhance the mortgage rate in China causes a higher monthly payment of mortgage debt to bank and therefore inhibits the over-speed of housing purchase demand. As for the relationship in 15 sample OECD countries, changes in interest rates have little effects on housing price deviations, which is in line with the previous study by Arnnett (2005). Moreover, recent studies show mixture relationships of the two variables. For instance, the increase of interest rates have negative influences on housing prices especially in those countries with a high competition in the bank sector (Andrews, 2010). In contrast, housing prices of the countries with less competition banking sectors show little interactions with the changing of interest rates, as these banking sectors tend to pass fewer declines in policy rates to mortgage rates. Besides, the change of interest rate is observed to have a positive relationship with housing price deviation, e.g. interest rate may response to innovations in housing markets, interest rate and housing price will change simultaneously with macroeconomics.
Population Previous studies indicate that population has a positive effect on housing
- 28 -
housing price growth is not significant. This result can be partly explained by many previous studies (Andrews, 2010) which suggest that the increase of housing price is related to net migration growth and youth population growth and the two categories of population growth lead to a potential increase of household formation. Thus, the whole population growth may not be a proper determinant of increasing housing price for OECD countries.
Rent Present studies evidence that the interaction between the change in rent growth
rate and the change in house price inflation is positive in OECD countries but not significant in China. The mixture of results may be explained by the different development stages that the house renting markets of OECD countries and China are positioned in. Compared to the mature house renting markets in OECD countries, house renting market in China has just started to prosper in recent years. Besides, Chinese tend to endow rich meaning to a house which relates to a shelter or a harbor of oneself. Consequently, Chinese prefer to own their houses rather than just live in someplace belonging to others. Nevertheless, this kind of situation changed a lot in recent years especially among the young generation and more and more young people started to rent a house.
In sum, the regression model suggests the rejection of null hypothesis of the coefficients of the corresponding housing price determinants are the same in OECD countries and China. Except for the positive effects of changes in GDP growth rate on the changes in housing price deviation which are significant at 1% level and 10% level respectively in OECD countries and China, other regressors show a mixture of effects on housing price deviations in OECD countries and China. The increase of interest rate growth rate is positively associated with the increase of housing price inflation in China whereas the relationship of the two variables is not significant in OECD countries. Besides, the increase of population growth rate and the increase of disposable income are negatively related to the growth of housing price inflation in China while the relationships between the two pairs of variables are not significant in OECD countries. In addition, the change in inflation rate is negatively (at 10% level) related to the change in housing price inflation in China while a positive relationship (at 1% level) is observed in OECD countries.
account for two fifths of the total deviation of housing prices in OECD countries, compared to more than three fourths of the explanatory power of the macroeconomic factors for the housing price volatility in China. The results confirm previous studies (Andrews, 2010) on housing prices determinants of OECD countries which suggest that except for the macroeconomic factors, structural factors and policy factors also play key roles in determining housing prices.
2. City Level Model
The city level estimation employs a panel data model which covers 35 major cities in China during the sample period of 2006 to 2009. Similar to estimation 1, a cross-section fixed effects model is applied, which is useful for comparing the results of the two models. Table 9 below reports the estimation results from the city level model (Eq 2), indicating the effect of the change in macroeconomic factors on change in housing price deviations of 35 major cities in China during the sample period.
Table 9 provides the respective results for testing the relationships between housing prices and determinants of 35 major Chinese cities which are classified into 3 groups according to their relative housing prices over the period of 2006 to 2009.
Table 9. Comparison of results among 35 major cities in China
Variable 35 cities Group 1 Group 2 Group 3
Constant -0.010 (-0.305) 0.028 (0.229) -0.006 (-0.145) -0.006 (-0.126) CONCOST 0.003 (0.029) -0.221 (-0.752) 0.106 (0.825) -0.023 (-0.126) RENT -0.267 (-1.017) 0.075 (0.109) -0.427 (-1.342) -0.478 (-0.798) POP -0.160 (-6.068)*** -1.683 (-3.871) *** -0.174 (-0.892) -0.118 (-0.417) LANDCOST 0.095 (1.837) * -0.021 (-0.228) 0.191 (1.697) * 0.246 (1.444) INF 1.152 (6.367) *** 2.234 (3.609) *** 0.797 (3.217) *** 1.063 (3.622) *** GDP 0.310 (2.039) ** 0.353 (0.561) 0.235 (1.152) 0.430 (1.891) * DISINCOME -0.280 (-2.350) ** -0.247 (-0.736) -0.273 (-1.720) * -0.486 (-2.651) ** Adjusted R 2 0.430 0.571 0.322 0.405 Obs 140 40 60 40
Notes: (1) Included regressors are CONCOST (construction cost), Rent (rent of house), POP (population),
LANDCOST (land cost of China), INF (inflation rate), GDP, DISINCOME (disposable income). (2) T-statistics are reported in parenthesis.
(3) Significance at the 1%, 5%, and 10% levels is denoted as ***, **, and *, respectively.
- 30 -
Construction Cost On the supply side, the increase in construction cost growth rate
has no significant effect on the housing price volatility on the aggregate of 35 cities and three sub groups, which is not consistent with the result of China in model 1. The sample period of 11 years for estimation at country level differs from that of 6 years for estimation at city level, which can be the reason for the mixture of results. The short run effect of the change in construction cost growth rate may not be incorporated in the deviation of housing price.
Rent Not in line with previous studies that suggest a positive relationship between
housing price growth and rent growth, this study finds no significant effect of the increase of rent growth rate on the increase of housing price growth rate in all of the four models. This might be explained by the fact that compared to the fast developing housing market during the sample period, house renting market is rather stable which is expressed by a small mean value of growth rate (Table 6) at the aggregate of 35 cities level and three subgroups. Therefore, the change in rent growth rate has little explanatory power on change in housing price inflation.
POP For the relationship between the change in population growth rate and the
change in housing price growth rate, the result of 35 cities on aggregate is in accordance with that of China. Moreover, group 1 with a coefficient rounded to -2 at 1% significant level shows that a 1% decrease of population growth rate will result in a 2% growth of housing price inflation. The possible reason for this relationship can be similar to that of China in the previous estimation, the gradual decrease of population growth rate in China which shows the effective outcome of the birth control policy coincides with the increase of housing price inflation. Nevertheless, the associations between the two variables in group 2 and group3 are not significant, which can be explained by the rather stable population growth rate with a equal (group 2) or smaller (group 3) standard deviation compared to that of the aggregate of 35 cities (see Table 6).
Land Cost The change of land cost, as a new added independent variable on supply
coefficient of 0.2 doubling that of the aggregate 35 of cities. However, the result doesn't indicate significant relationships between the two variables in both group 1 and group 3. For group 1, the result can be explained by the fact that land cost does not account for a dominant portion of the total housing price volatility. Group 2 has similar mean values of the variables for growth rate of housing price and growth rate of land cost, which implies a potential relationship between the two variables. As for group 3, the relatively small increase of land cost growth rate which is less than 50% of the mean value for the aggregate of 35 cities and 20% of the value for group 1 suggests that land cost may not be a proper explanatory factor for the growth of housing price inflation (see Table 6).
Inflation In line with the expectation, change in inflation growth rate is observed to
have a significant positive effect on change in housing price growth rate in all of the four models including the aggregate of 35 cities and the 3 division groups. However, the relationships of the four models (all significant at 1%) have different signs with that of China in the previous estimation which shows significant and negative relationship at 10% level. This can be explained by the fact that the aggregate of 35 major cities and three subgroups have relatively lower increase rate of inflation growth than that of housing price growth except for group 2 where the two figures are close to each other. In contrast, the increase rate of inflation growth is five times than of the housing price growth in China. Hence, the regressor of inflation has more significant explanatory power in the city level model compared to the country level model.
GDP In line with the relationship between change in GDP growth rate and change in
housing price inflation of China, the relationships of the two variables are positive and significant at 5% and 10% respectively in the aggregate of 35 cities and group 3. Moreover, the coefficient of the two variables in group 3 surpasses that in the aggregate of 35 cites. In contrast, the relationships between the two variables in both group 1 and group 2 are not significant, which can be explained by the fact that the increase of GDP growth rate in the two groups with the most developed economies in China is not in accordance with the soaring housing price.
Disposable Income Contrary to the expectation, changes in disposable income growth
- 32 -
significant effect on change in housing price growth rate of group 1 can be observed. Considering the mean values of disposable income growth rate among the three sub groups, group 3 ranks first, followed by group 2 and group 1 is in the last position (see Table 6). The negative relationships between the two variables might be explained by the fact that the fast increase of disposable income growth rates in group 2 and group 3 coincides with the even faster soaring housing price during the sample period. Nevertheless, group 1 which denotes the most developed sub economies in China has already gone through the fastest developing period that the other two groups are experiencing now.
To sum up, the result of the OLS regression model suggests that each regressor has a different effect on housing price volatility in China, except for the change in inflation growth rate which significantly (at 1% level) results to the growth of housing price appreciation in all of the three groups. For group 1 with the most developed economies in China, change in population growth rate negatively relates to change in housing price appreciation at 1% significant level. For group 2, change in land cost growth rate is positively associated with the increase of housing price inflation whereas change in disposable income growth rate results in a decrease of housing price inflation. As for group 3, increase in GDP growth rate is related to growth of housing price inflation rate while a decrease in disposable income growth rate leads to growth of housing price inflation rate.
positive relationship between change in land cost growth rate and change in housing price inflation is only observed in group 2, from which we can infer that the land cost accounts for a proper portion of housing price deviation.
5. Discussion
The estimation results of this paper add some new findings to the related field of study. Some of the interesting results are worthy of a deeper study. Firstly, the increase of interest growth rate in China associates with the increase of housing price inflation whereas the relationship of the two variables is not significant in OECD countries. Most remarkably, the result suggests that the measure of interest rate changes which is adopted by the Chinese government to control housing price growth may not be an effective one. Secondly, the negative effect of change in disposable income growth rate on change in housing price inflation, which is observed in the country level model and the city level model except for the first-tier cities in China, indicates that the growth rate of increasing disposable income is slower than the growth rate of housing price inflation. This result confirms the fact that the soaring housing price puts heavy burdens to the citizens with relatively less and less purchasing power. Thirdly, the mixed effects of the increase of regressors on the growth of housing prices inflation among the three subgroups suggest that the second-tier and the third-tier cities are experiencing a fast development process of the housing market. Moreover, the heavy burden of house purchasing not only exists in the most developed cities. E.g. Capital, Shanghai and Southeastern coastal cities but also occurs in the less developed cities in China.
In the following, I want to test the robustness of the regression model. For the first approach, in order to check whether the model will still hold if some less significant explanatory variables are deducted, a distinction between the significant explanatory variables and the less significant ones is necessary. According to the results of the empirical study (Table 8 and Table 9), the independent variables of Construction cost and Rent show less explanatory power on housing price movement, especially in 35 major cities of China. Hence, a new model is built as follows:
hpi,t α0β1GDPi,tβ2DisIncomei,tβ3Infi,tβ4POPi,tβ5LandCosti,t
- 34 -
Disposable Income growth rate (DisIncome), Inflation rate (Inf), Population growth
rate (POP) and Land Cost growth rate (LandCost) are as defined in the previous models. The result of the comparison between the original models and the new models for 35 major cities in China is reported in Table 10.
Table 10. A comparison of the regression results for 35 major city in China
Notes: (1) Included regressors are CONCOST (construction cost), Rent (rent of house), POP (population),
LANDCOST (land cost of China), INF (inflation rate), GDP, DISINCOME (disposable income). (2) T-statistics are reported in parenthesis.
(3) Significance at the 1%, 5%, and 10% levels is denoted as ***, **, and *, respectively.
As it is shown in Table 10, in all of the four cases (the aggregate of 35 cities and group 1-3), no significant change occurs in the regression results when the two independent variables of construction cost and rent are deducted. Specifically, there is
Variable 35 cities 35 cities Group 1 Group 1
Approach Concost / rent included Concost / rent excluded Concost / rent included Concost / rent excluded Constant -0.010 (-0.305) -0.012 (-0.391) 0.028 (0.229) 0.075 (0.741) CONCOST 0.003 (0.029) -0.221 (-0.752) RENT -0.267 (-1.017) 0.075 (0.109) POP -0.160 (-6.068)*** -0.156 (-6.013) *** -1.683 (-3.871) *** -1.780 (-4.439) *** LANDCOST 0.095 (1.837) * 0.095 (1.842) * -0.021 (-0.228) -0.012 (-0.140) INF 1.152 (6.367) *** 1.132 (6.421) *** 2.234 (3.609) *** 2.350 (4.388) *** GDP 0.310 (2.039) ** 0.322 (2.130) ** 0.353 (0.561) 0.103 (0.200) DISINCOME -0.280 (-2.350) ** -0.279 (-2.407) ** -0.247 (-0.736) -0.308 (-0.981) Adjusted 2 R 0.430 0.435 0.571 0.596 Obs 140 140 40 40
Variable Group 2 Group 2 Group 3 Group 3
