Empirical Results And Model Critique

Published Date: 02 Nov 2017

C:\Users\Jing\AppData\Roaming\Tencent\Users\554389572\QQ\WinTemp\RichOle\~VM5`5~PT]FV4[F{ZBLSC62.jpg

Detecting housing bubbles and analyses of government policies

in Beijing between 2000-2010 (GY 458 Project)

Candidate Number: 81263

Degree Programme: MSc in Real Estate Economics and Finance

London School of Economics and Political Science

January 2013

ABSTRACT

The primary objective of this paper is to detect the speculative bubbles in Beijing residential property from 2000-2010. This paper in particular, focuses on benefits that investors expected to achieve by observing past real estate data. This paper measures housing bubbles by using a self-established regression model of house prices with fundamental variables such as income and interest rate. In addition, the paper identifies the effects of government policies on property prices in Beijing. The paper finds out that despite the tremendous increase in housing price in Beijing, there is little evidence to conclude that speculative bubbles took place during 2000-2010 which also coincides with other research studies (Yu 2011, Chai and Dong 2012, Hou 2009).

GY 458 One-page Summary

The primary objective of this project is to investigate the speculative bubbles that occurred within the Beijing residential property market from 2000 to 2010. However, unlike studies undertaken in countries such as the USA, determining the history of speculative bubbles in China is challenging; accordingly models such as that used by Capozza (2004) cannot be applied. The problems centre on the fact that Chinese municipal governments and many third-party agents provide data only twice a year (or less) start from 2000, making it difficult to track prices on a month-by month basis to find the demand and supply of housing causes a large estimation error between predict and actual house prices.

This essay uses a self-established regression model in order to find housing bubbles during the past ten years, (results are) consistent with other models such as those applied by Hui and Yue (2003). After presenting the technique, this paper analyses government policies and their relationship with the non-fundamental growth rate of house prices. Limitations of this newly developed model will be highlighted and analysed in a later chapter.

Introduction

Real estate has become a major industry in China over the last two decades (Chai and Dong 2012). During the period 2000-2010, Beijing experienced dramatic increases in property prices. However, once house prices deviate largely from their fundamental value, this could result in real estate bubbles [1] or even financial crisis. The bursting of speculative bubbles in Japan and the USA in 1997 and 2009 respectively caused a worldwide economic downturn.

Some scholars suggest that different methodologies used in the detection of housing bubbles sometimes lead to conflicting conclusions. Ren and Wang (2010) applied multi-indicator analysis to housing price bubbles in Beijing/Shanghai and obtained inconsistent conclusions from different methods. The difference centres on the evidence obtained (Peng & Hudson-Wilson, 2002) and the method they used (Hou, 2009). Moreover, insufficient and unreliable data (especially in Beijing) amplify the probability of achieving a "wrong" or "unreliable" conclusion. Given the failings of such research, it is important to be able to measure housing bubbles effectively and accurately at the current stage of economic development in an insufficient data environment. Economists such as Lang (2011) [2] suggest that real estate is still a value-preserving asset while citizens on the other hand believe that these so-called "professional economists" spread rumours and push up house prices.

There are three basic views that are held by geographers and the public as to the cause of housing bubbles. The dominant view is supply-side economics, as proposed by Mark T (2004). Figure 1 shows that during the past ten years, the supply of housing has been relatively inelastic in Beijing, which was caused by the physical constraints of the city. These constraints amplified cyclicality of house prices which might cause speculative bubbles. The second view is Keynesian and it is suggested by the proponents of his behaviour economics. The bubble is often explained by the phrase "irrational exuberance" and psychological factors that affect people’s decision-making. The third view sees that bubbles consist of real and psychological changes which are caused by manipulations of monetary policies. But it is difficult to argue that the increase in house price triggers the reform of monetary policies or that the policies themselves lead to speculative bubbles.

The purpose of this paper is to introduce a model to detect the speculative bubbles in Beijing in terms of irrational exuberance. This article is constructed as follows: Section 2 introduces the current housing market in Beijing and contains a literature review; Section 3 discusses the model that will be applied for analysis. Section 4 examines empirical results and critiques. In Section 5, an analysis of government policies on housing prices is presented and finally, the conclusion follows in Section 6.

2. Literature review

Kindleberger (2008) defines a bubble as a sharp rise in price of an asset or a range of assets in a continuous process, with the initial rise generating expectations of further rises and attracting new buyers. Irrational investors, asymmetric information, as well as a non-transparent real estate market are the root of speculative bubbles.

In 2000, the average house price in Beijing was 4716RMB/m2 which increased to 22310 RMB/m2 in 2010. The dramatic increase in house prices could be explained by the fact that "housing bubbles" took place in the past or other fundamental factors such as the combination of increase in income and government policies that boost the real estate industry.

Figure 1 indicates that average house prices follow the trend of average wages of staff and workers [3] . By considering the generalised stable short term interest rate (5.5-6.9%), the primary conclusion as to the increase in house prices could be explained through reference to the dramatic increase in the average wage of staff and workers in Beijing.

Fig. 1 Average (nominal) wage during 1994-2010

Sources: Beijing Statistical Yearbook 1994-2010 (arranged by the study)

Ren and Wang (2010) estimated the Beijing housing market by using multi-index analysis. The index "housing price-to-income ratio" found that housing prices were irrational and that a housing bubble existed in the markets between 2000 and 2010, while other indices such the "housing price growth rate-to-GDP growth rate ratio" provide a general idea that speculative bubbles only happened in 2006 and 2007. Ning and Hoon (2012) add some improvements and regress some of these indices together with income and short-term interest rate. Their estimations illustrated that there was some overvaluation in Beijing in 2007 and 2009.

Hui and Yue (2006) analysed overall housing supply and demand with respect to housing prices, vacancy rates, income and GDP, using an econometric model in three Chinese cities. The model generated a reduced form equation as the proxy for market equilibrium. Their calculation suggests that no bubbles appeared in 2003. However, they failed to give a reason for speculative bubbles before 2003.

There are some issues in their estimations. First, the GDP could have a multicollinearity problem with income measure. Second, the data for house prices they took are not government published and may have led to unreliable conclusions. In addition, vacancy rate [4] measures differently in mainland China compare with the UK and the USA. He failed to modify the changes of these variables. Lastly, their regression model ignores interest rate and the effect of government policies which could largely affect supply and demand function.

3. The model

Although the definition of a bubble appears straightforward, testing for its existence can be difficult (Eddie and Yue 2008). This model focuses on the expectation benefits that investors are willing to achieve in the future.

The housing bubble in this model is measured as follows:

(1)

measures the non-fundamental growth rate of housing price at year t.

, depict the lag effect of the non-fundamental growth rate of housing price at year t-1, t-2 respectively.

, refer to the tendency of housing prices in the current period in respect of the incremental lagging effect in one/two years, Thus, the coefficients , reflect real estate bubbles.

The model analyses the formation of short-term speculative bubbles based on the definition of Kindleberger (2008). More specifically, under the weak form of efficient market hypothesis, investment decision is based on the observations of the past record. The larger the coefficients,, the more severe the speculative bubbles are.

According to the regression noted above, this article calculates the non-fundamental growth rate in each period in order to measure the real estate bubbles in the next step. The price of housing is represented as follows:

(2)

measures the house price in year t

represents the change in lagged-income in year t-1

is the change of one-year lagged-short term interest rate in Beijing in year t-1

stands for the change in lagged-supply of house (residential property) in year t-1

The model can be easily understood if the lagged-house price () is placed on the left-hand side. The equation will then produce the change in house price and is explained by the change in income, interest rate and supply of housing in the lagged-period. This in turn suggests that an investor’s decision to purchase a property is based on the fundamental factors that are notable in previous years.

This model assumes that the housing price in year t is related only to the income, interest, supply of housing and housing prices in year t-1. Income and short-term borrowing rate provide fundamental decision-making information on whether or not to buy a house while supply and demand [5] of housing represents the maximum number of properties available in the market. This model omits other factors which could influence the housing price such as government policies. Such inherent drawbacks with the model are analysed later within this work.

According to (1), the mathematical derivation is as follows:

(3)

(4)

Denote = and the same method applied to equation (4) in order to obtain

At this stage, the new housing price and eliminates the effect of interest, income and supply of housing.

The fundamental growth of the housing price can be therefore demonstrated as

(5)

Substituting the result in equation (5) into (1) as and the same method applied to ,. In terms of the formula , the model could obtain the coefficient as for the housing bubbles measurement.

4. Empirical results and model critique

Data source

The People’s Bank of China (the Chinese Central Bank) provides annual reports for one year short-term interest rates which are used in this article. With regard to income (Y), the Beijing Statistical Yearbook 2000-2010 offers non-biased data for the average wages of fully employed staff and workers in Beijing. Supply of housing (, in particular, will use "floor space of new building completed" in the Beijing Statistical Yearbook as well. Considering the price of housing, the model will use "the average price of residential premises in Beijing" from Beijing Real Estate Transaction Data Centre (BRETDC) (http://www.e-fdc.net). It is also worth noting that apart from the house price, all data are obtained from the government official reports which strengthen the accuracy of this model. Data provider BRETDC on the other hand, is supported by the State-owned Assets Supervision and Administration Commission of the People’s Government of Beijing Municipality.

Results

(2)

Parameters

Coefficients

T-statistics

Wage

0.09677

1.64

Interest Rate

-0.69823

-3.01

Supply of Housing

-0.00657

-0.1

Lagged House Prices

0.91032

28.04

Constant

0.71870

2.47

Adj =0.9798

Table 1. Regression results from equation (2)

Calculated from Stata

According to the estimated results above, all variables are statistically significant (except housing supply) and have the expected signs. The coefficient of Y is positive showing that nominal income has a positive relationship with housing prices as expected. Interest rate (I) result shows that an increase interest rate jeopardised investors’ borrowing while also increasing the cost of financing by construction companies. Supply of housing followed the predictions with the negative correlation with house prices.

Returning to equation (1), this paper finds out the coefficient of by equation (5)

From the results above, the non-fundamental growth rate of property prices --- the growth rate that eliminates the effect of increase in income, interest rates and supply of housing --- has a reduced relationship with the lagged-period growth rate. More significantly, the correlation between the current and two-period-lagged growth rate has the negative sign. Also, the main indicator is only 0.142, which means that 14.2% of the last year’s non-fundamental housing growth could be capitalised in the current growth of the housing price. This study concludes that during the period 2000-2010 as a whole, the real estate industry in Beijing did not encounter severe speculative bubbles.

Critique

The model used in this paper is not derived from previous models used in the literature. Capozza et al (2004), DiPasquale et al (1994) and other economists have applied more advanced and accurate models for detecting housing bubbles in the USA. They analysed the predicted price from supply and demand and compared it with the actual price in order to find speculative bubbles. The barrier to applying such models to the Beijing market is primarily as a result of inadequate data in Mainland China. The problems centre on the fact that Chinese municipal government and many third-party agents provide data only twice a year (or less) start from 2010, making it difficult to track price on a month-by month basis to find the demand and supply of housing causes a large estimation error between predict and actual house prices. Also, most third-party agents’ that provide monthly data for Mainland China are the ones that are predicted from the government report which cause a biased conclusion if used. This research has been unable to find literature that uses the methods common in analysis of the US housing market to detect speculative bubbles in Beijing. The model used in this paper, capitalised the supply and demand in equation (2) and somehow familiar with the past literatures. In applying the new model proposed in this paper, the conclusion may not be authoritative but provides an approximation of the real estate market in Beijing during the period specified, in terms of irrational exuberance

As has been mentioned, many factors may affect house prices which are not limited to income, interest rates and the supply of housing in equation (2). However, these three factors are considered to be the most important fundamental ones in influencing property prices.

Supply of housing in this model is described as "floor space of new building completed" which is somehow unreliable. Supply of housing is not only limited to new building completed in a year but again by, constraints of the official database.

The international standard figures for for the equation were unavailable. At this stage, the model cannot provide comparisons of data in other countries but 14.2% consider being a small number in an irrational exuberance situation (also is negative).

5. Extension

The table below summarises the government policies relevant to the real estate

Period

Effect(Direct) D, Indirect I.

From

Policies and results

07/1998

2D

State Council

File 23: Termination of Government housing distribution system. Introduction of "affordable housing" mechanism. Promotion of monetary housing system. Result: Boost of real estate industry.

06/2003

-1D

Central Bank

File 121: Limitation placed on luxury building construction, including villas. Increased down-payment ratio for borrowers who already have one property. No mortgage discount for real estate investment.

08/2003

0I

State Council

File 18: Confirmation that the Real Estate Industry has become the dominant component of GDP increment. Promotion of a sound real estate industry.

04/2004

-1I

Central Bank

Required reserved increase from 7% to 7.5%.

10/2004

-1I

Central Bank

One year base rate increase 0.27%.

05/2006

-2D

State Council

File Guo 6: Increase affordable housing for low/medium households.

01/2007

-1D

Beijing Municipal Government

Avoid speculative investments especially foreign investors.

01/2008

-2D

State Council

Severe punishments for land hoarding by real estate developers.

06/2008

-1.5I

Central Bank

Required reserve increase to 17.5%.

05/2009

-0.5I

National Development and Reform Commission

Introduction of Property tax.

06/2010

-3D

Beijing Municipal Government

File New Guo 10: Restrictions on property purchases for non-Beijing Hukou Holders.

09/2010

-2D

State Council

No mortgages for households who already have three or more properties. Restrictions on the number of properties people are allowed to buy

Table 1. Timeline of important government policies related to the real estate industry

Sources: Beijing Policy handbook, www.baidu.com

industry over the last ten years. The second vertical column "effect" measures the expected effect of each policy on housing prices as determined by www.baike.baidu.com as well as private and public judgement. Increased in reserve requirement in the past 10 years is mainly targeted on restrain (residential) real estate market.

The effect of government policies for the non-fundamental growth rate is summarised as below. Figure 2 illustrates several features of government policies in Beijing.

The non-fundamental growth rate has been volatile during the past ten years.

The effect of most government policies coincides with the expectations of the public and policy-makers.

Policies which have an indirect impact on house prices, such as increases in reserve requirements have less effect while policies which directly work on property have a larger effect on property prices like restrictions on purchase properties.

Monetary policies lagged by on in average 3 months while direct policies work immediately.

Figure 2. Government policies with non-fundamental growth

Calculated from Beijing statistical Yearbook

6. Conclusion

The empirical result in this paper reveals that serious speculative bubbles have not become a problem in Beijing, a finding that is in contrast to popular belief but consistent with those found in other literatures aiming at similar research such as Chai and Dong (2012), Hou (2009) and Eddie and Yue (2008). The paper assumes that property prices are only related to income, interest rate and supply of housing; by deducting the effect of these fundamental factors, the paper determines the non-fundamental growth of house prices and the coefficient of speculative bubbles. Although there are some drawbacks inherent to the model formation, the project attempts to find a new method to detect housing bubbles and could be useful for developing countries such as China and India to analyse speculative bubbles in an insufficient data environment.

The model also finds that monetary policies such as increase in interest rate and reserve requirement has less impact on property prices while causing significant impact in other industries and contribute a negative externality for the whole economy. In order to control property prices effectively, government and policy makers should impose direct regulations on the real estate industry. The project is also extensive from precise modification on model formation.