The relationship between house prices index and stock prices index based on China market : an empirical analysis

Master Thesis

The Relationship between House Prices Index and Stock

Prices Index based on China Market: An Empirical Analysis

by

Yichen Ran

Under the Supervision of Professor Liang Zou

Quantitative Finance, MSc Finance

Amsterdam Business School, University of Amsterdam

Statement of Originality

This document is written by Student Yichen Ran who declares to take

full responsibility for the contents of this document.

I declare that the text and the work presented in this document are

original and that no sources other than those mentioned in the text and its

references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the

supervision of completion of the work, not for the contents.

Abstract

The relationship between stock market and housing market is tight, and some investors may be concerned that the stock market crash may increase housing market uncertainty, even result in a similar crash in house prices, in the meantime, other investors may also be unsure that if the collapse in housing market would pose influence on the stock market. Therefore, it is essential to analyze the relationship between the housing prices index and the stock prices index in order to understand the impact of the stock market to the housing market and vice versa.

Based on China market, this thesis employs an econometrics approach to examine the relationship between stock prices index and house prices index. Shanghai Composite Index was chosen as the stock prices index, and monthly housing prices index data was collected from the National Bureau of Statistics of China. According to this thesis, empirical results suggest that the relationship between the housing market and stock market is relatively weak, although the stock market does impact on the housing market, the influence is relatively small.

Firstly, this thesis starts with the background of China’s housing market and stock market. Secondly, the relevant literatures are reviewed. Thirdly, the data collection and test procedure are outlined, and finally, the results of tests and analysis of results are presented.

CONTENTS

1．INTRODUCTION ... 1

2. LITERATURE REVIEW ... 2

3. DATA ... 4

4. METHODOLOGY ... 7

5. EMPIRICAL ANALYSIS ... 9

5.4IMPULSE RESPONSE ANALYSIS ... 14

5.5MODEL CONSTRUCTION ... 16

6. ROBUSTNESS ... 18

7. CONCLUSION ... 21

1.

Introduction

The housing market and the stock market are both vital parts in the structure of China economy, the housing market constitutes an essential part of China’s GDP, and the stock market has become a key to China’s financial markets and a barometer of economic performance.

In the past a few years, housing demand fabricated a boom in house prices in China, leading to a rising supply from tremendous constructions by houses developers, and in the meantime, the stock market also experienced some unprecedent fluctuations, further to influence China economy.

Recently, because of the worldwide stock market collapse, many investors are worried that the falling stock market will increase the uncertainty of the housing market, even induce a similar housing prices collapse. Thus, researches on the relationship between the stock market and the housing market need to be carried out in order to understand the degree of impacts of stock market changes on the housing market. Actually, some scholars have already managed to figure out the empirical relationship between stock market and housing market, however, they reached conflicting results. There are analyses concluding positive causality, negative causality or no significant causality. Moreover, China’s domestic empirical studies relevant to this topic are rare. As a result, this thesis hopes to resolve the controversy based on China market.

2. Literature Review

Brent W. Ambrose, Esther Ancel, and Mark D. Griffiths 20 (1992) derived a long-term cointegration relationship between the real estate market and the stock market based on the real estate investment trust income and the S&P 500 market index from 1962 to 1989.

David Ling and Andy Naranjo18 (1999) used the multi-factor asset pricing model, and found that the cointegration of the real estate market and the stock market intensified in the 1990s.

Quan and Titman (1999) examined data from seventeen different nations and regions over fourteen years and concluded that there is a strong correlation between stock prices and commercial housing prices.

John Okunev, Patrick Wilson et al. (2000) used Granger causality analysis and based on the S&P 500 index and the monthly housing market in the United States between 1972 and 1998. The conclusion was that the impact from the housing market to the stock market is not significant, and the nonlinear Granger causality test does not show a strong practical significance.

Raymond Tse (2001) obtained a conclusion that the prices of the stock market and the housing market in Hong Kong have a stable and balanced relationship through the study of the data from 1974 to 1998, and concluded that the fluctuation of stock prices is because of the housing market.

the housing prices and the stock prices from 1980 to 1998 using the ECM model. The Swedish asset market information cannot reflect asset prices in a timely manner, therefore, the stock price has guiding significance for real estate investment opportunities.

Particularly, among China’s findings, Li Xiaohuan and Feng Xiujuan 12 (2010) selected the quarterly data of the real estate sales price index from 1998 to 2009. Through the Engle-Granger two-step cointegration analysis, the VAR model was established to verify both the wealth and the substitution effects between stock prices and real estate prices. Effects and Granger causality test are used to analyze the interaction between the two indexes, and results show that the real estate price and the stock price are Granger causality.

According to the empirical analysis of the stock market and real estate data from 1996 to 2007, Zhao Dahua and Zhao Yuhua 11 (2010) found that the real estate market has a strong correlation with the stock market. At the same time, the current stock price has a positive correlation with the housing prices in the current period, and negative correlation in the next period.

Huang Yi 4 (2014) intercepted the data from January 1998 to December 2012 for empirical analysis, finding that there is a close dynamic relationship between the stock market and the real estate market in China, and the dynamic relevance of the two markets has obvious time changes characteristics. But overall, the two markets exhibit a dynamic positive correlation.

Overall, the foreign literature analyzes the situation in different regions at different times and reached conflicting results. As for China’s domestic literature, analyzing of different time periods, concludes that the two markets are definitely relevant, but also reached reversed results.

As a result, this thesis aims to resolve the controversy, and finding out the relationship between stock market index and housing market index based on China market.

3. Data

Representatives of housing prices in China mainly include the China Real Estate Index System and the National Housing Climate Index. Real Estate Index System in China is a set of indicators and systems in the form of price indices that reflects the status and development trends of real estate markets in major cities across the country, and the national housing climate index is an authoritative data developed by the National Bureau of Statistics to depict the development of the housing market in China, which is more representative in China market, so, this thesis chooses the National Housing Climate Index (hereinafter referred to as GF) as the housing prices indicator. In addition, Shanghai is one of the largest cities in China, and it is one of the financial centers in Asia, which makes the Shanghai Composite Index a powerful representative of the stock market in China. Therefore, this thesis uses the Shanghai Composite Index

(hereinafter referred to as SZ) for analysis.

The sample ranges from March 2001 to March 2018. Since the National Housing Climate Index was missed from January 2004 to January 2016 and August 2014, the missing data will use the average of the month and the next two months as an alternative. The calculations in this thesis were performed on the Stata and Eviews platform. The Akaike Information Criterion was used to determine the lag period of the test.

The relationship between the Shanghai Composite Index and the National Housing Climate Index is shown in Figure 3a. It can be seen from the figure that from the beginning of 2001 to the end of 2012, except for the sudden rise of the National Housing Climate Index in early 2003 and 2006, the two indexes basically showed the same fluctuations trends, presenting a trend of first rising and then falling. At the beginning of 2013, the two indexes began to deviate from each other, especially in the end of 2014 to mid-2015, the stock market flourished, while the housing market was shrinking.

The analysis of the two trends showing that the National Housing Climate Index and the Shanghai Stock Index almost fluctuate in the same trend at most of time, with only a small part of the time divergence. Therefore, in the empirical analysis section, this thesis mainly considers the long-term relationship between the two indexes, and ignores short-term changes. At the same time, taking the sharp fluctuations in housing prices into account, in order to make the Shanghai Composite Index and the National Housing Climate Index trends smoother and eliminate sequence heteroskedasticity, both data will be taken the natural logarithm, and be converted into LNSZ and LNGF.

Figure 3a: Relationship between Shanghai Composite Index and National Housing Climate Index

This figure shows the tendency of Shanghai Composite Index and National Housing Climate Index during the testing period, excluding the sudden rise of the national housing climate index in early 2003 and 2006 Outside the big bull market of 2007, the two basically showed the same trend fluctuations.

Table 3b: Descriptive Statistical Analysis for National Housing Climate Index (GF) and Shanghai Composite Index (SZ)

This table presents the descriptive statistics for National Housing Climate Index (GF) (in column (1)) and Shanghai Composite Index (SZ) (in column (2)), including observations, mean, median, maximum, minimum, standard deviation, skewness,

kurtosis, JB statistics and P-value. National Housing Climate Index （GF） (1) Shanghai Composite Index（SZ） (2) Observations 205 205 Mean 100.8229 2374.478 Median 102.0300 2218.030 Maximum 109.1400 5954.765 Minimum 92.4300 1060.738 Standard deviation 4.631616 937.7548 Skewness -0.319173 1.110105 Kurtosis 1.701550 4.492980 JB Statistics 16348598 56.37178 P-Value 0.000263 0.000000

4. Methodology

When using time-series, it is prerequisite to test the stationarity before analyzing the correlation, in this thesis, stationarity of LNSZ and LNGF was verified by ADF.

If the result shows that the time series is stationary, a linear model can be constructed to analyze the causality and influence degree of variables. If the data is not stationary,

test the first-order difference for stationarity. If the first-order differences are stationary, then there may be a long-term stable equilibrium relationship between LNSZ and LNGF. Then use the Engle-Granger two-step method to continue to examine the association, which is to do a regression on the variables, generate residual sequences and test the stationarity of the residual sequences. If the residual items are stationary, then the sequence has a cointegration relationship, if not, examine if there is a cointegration with more lags included.

If cointegration relationship holds in the sequence, an Error Correction Model (ECM) will be applied. Examine the Likelihood ratio (LR), Final Prediction Error (FPE), Akaike Information Criterion (AIC), Schwarz Information Criterion (SC) and Hannan-Quinn Information Criterion for time-series data to find the best lag order, and then form the regression model. The Error Correction Model (ECM) regression equation can be shown as below.

To analyze the influence from Shanghai Composite Index (SZ) to National Housing Climate Index (GF):

∆𝐿𝑁𝐺𝐹_𝑡= 𝛼₀+ 𝜇₁∆𝐿𝑁𝐺𝐹_𝑡−1+ 𝜇₂∆𝐿𝑁𝐺𝐹_𝑡−2+ ⋯ + 𝜇_𝑝∆𝐿𝑁𝐺𝐹_𝑡−𝑝+ 𝛽₁∆𝐿𝑁𝑆𝑍_𝑡−1 + 𝛽₂∆𝐿𝑁𝑆𝑍_𝑡−2+ ⋯ + 𝛽_𝑝∆𝐿𝑁𝑆𝑍_𝑡−𝑝+ 𝑢_1𝑡

To analyze the influence from National Housing Climate Index (GF) to Shanghai Composite Index (SZ):

∆𝐿𝑁𝑆𝑍_𝑡= γ₀+ 𝛿₁∆𝐿𝑁𝑆𝑍_𝑡−1+ 𝛿₂∆𝐿𝑁𝑆𝑍_𝑡−2+ ⋯ + 𝛿_𝑝∆𝐿𝑁𝑆𝑍_𝑡−𝑝+ 𝜑₁∆𝐿𝑁𝐺𝐹_𝑡−1 + 𝜑2∆𝐿𝑁𝐺𝐹𝑡−2+ ⋯ + 𝜑𝑝∆𝐿𝑁𝐺𝐹𝑡−𝑝+ 𝑢2𝑡

Housing Climate Index, 𝛼 and 𝛾 are both constants, 𝜇, 𝛽 𝑎𝑛𝑑 𝜑 are matrices of lag coefficients that are up to some lag lengths p, and 𝑢_1𝑡 and 𝑢_2𝑡 are both vectors of error terms.

5. Empirical analysis

5.1 Stationary test

To test the stationarity before performing the cointegration test to analyze the correlation, this thesis uses ADF to examine the stationarity of LNSZ and LNGF, the result has been presented in Table 5a.

Table 5a: ADF test results for LNSZ and LNGF

This table presents the ADF test results for LNSZ (in column (1)) and LNGF (in column (2)), using t-value to examine the stationarity. Since both t-value for LNSZ and LNGF are above the 5% significance level, which means that the null hypothesis cannot be rejected, and imply that both series are not stationary.

LNSZ (1)

LNGF (2)

t-value -2.532632 -1.783843

result Non-stationary Non-stationary

Table 5b: ADF test results for DLNSZ and DLNGF

This table presents the ADF test results for the first difference for LNSZ(∆LNSZ) (in column (1))and LNGF(∆ LNGF) (in column (2)), using t-value to examine the stationarity. Since both t-value for ∆LNSZ and ∆LNGF are below the 5% significance level, which means that the null hypothesis need to be rejected, and imply that both series are stationary.

∆LNSZ

(1)

∆LNGF

(2)

t-value -5.073772 -5.376883

5% significance level value -2.878015 -2.877012

result Stationary Stationary

According to tables above, the original time-series of the variable accepts the null hypothesis at a significant level of 5%, which means that the time series is non-stationary, whereas the ∆LNSZ and ∆LNGF are obtained after a first-order differential, rejecting the null hypothesis at a significant level of 5%, which means that the time series is stationary. It is inferred that all variables are first-order single integers, and there may be a long-term stable equilibrium relationship between LNSZ and LNGF. Combining the considerations of de-volatility and de-heteroscedasticity, the empirical analysis using ∆LNSZ and ∆LNSZ is carried out below.

Theoretically, the National Housing Climate Index and the Shanghai Composite Index do exhibit a stable and balanced relationship, namely, the wealth effect. It can be understood that when the stock prices rises, the wealth of consumers or investors changes, and this part of the funds may flow into the housing market, then the price of houses may rise. In addition, stocks are a barometer of the economy, the rise in the stock index may increase the confidence of the entire market to improve the tendency of investment and consumption.

The same is true for China’s national conditions. In fact, the overwhelming majority of Chinese’s thoughts are relatively conservative, many Chinese are eager to buy a house in pursuit of stability. This is one of the reasons for the high housing share in China.

5.2 Cointegration test

This thesis uses EG two-step method to cointegrate the two variables, that is, to do a regression on the variables, generate residual sequences and test the stationarity of residual sequences. If the residual sequences are stationary, then a cointegration relationship depicts. The cointegration test result has been presented in Table 5c.

Table 5c: Cointegration test results

This table presents the EG two-step test result, using t-value to examine the stationarity. Since the t-value for Res1 (residual sequence obtained by regression using LNGF as an

explanatory variable) (in column (1)) and Res2 (residual sequence obtained by regression using LNSZ as an explanatory variable) (in column (2)) are both below 5% significance level, then the null hypothesis should be rejected, and both series are stationary. Res1 (1) Res2 (2) t-value -2.425106 -3.296808

5% significance level value -1.942555 -1.942594

result Stationary Stationary

Table 5c shows that both Res1 (residual sequence obtained by regression using LNGF as an explanatory variable) and Res2 (residual sequence obtained by regression using LNSZ as an explanatory variable) are stable. That is, cointegration relationship holds between the two datasets, and there is a stable relationship between the two datasets as well.

Judging from China’s national conditions, there is a certain substitution effect between the housing market and the stock market. As in the bull market, some aggressive investors will mortgage houses or even sell houses to increase capital leverage or funds, at the same time, although the market is prosperous for housing investment, if there is a more profitable investment method emerges, like the stock market, then capital will soon be shifted into the stock market, which means that a lot of hot money will flow from the housing market into the stock market, and makes the

stock market more and more prosperous, while the real estate market is sluggish.

5.3 Lag order determination

To determine the proper lag orders, examining the Likelihood ratio (LR), Final Prediction Error (FPE), Akaike Information Criterion (AIC), Schwarz Information Criterion (SC) and Hannan-Quinn Information Criterion for model to find the best lag order.

Table 5d: 5 different methods results for finding proper lag order

This table presents the Likelihood ratio (LR) (in column (1)), Final Prediction Error (FPE) (in column (2)), Akaike Information Criterion (AIC) (in column (3)), Schwarz Information Criterion (SC) (in column (4)) and Hannan-Quinn Information Criterion(in column (5)) test results, with * represents the best result in one column. Taking every result in 5 columns into consideration, AR(1,1) would be the best model to be employed in this thesis. Lag order LR (1) FPE (2) AIC (3) SC (4) HQ (5) 0 N/A 3.48e-07 -9.196612 -9.161135 -9.182228 1 30.91761 3.05e-07* -9.326844* -9.220412* -9.283690* 2 5.115002 3.10e-07 -9.311628 -9.134241 -9.239705

3 6.357286 3.12e-07 -9.303931 -9.055590 -9.203239 4 10.79273 3.06e-07 -9.322602 -9.003306 -9.193141 5 4.552523 3.12e-07 -9.305095 -8.914845 -9.146866 6 4.279245 3.18e-07 -9.286275 -8.825070 -9.099276 7 9.549791* 3.14e-07 -9.299708 -8.767549 -9.083940 8 6.857739 3.15e-07 -9.297336 -8.694222 -9.052799

5.4 Impulse response analysis

In order to study further about the influence between the two variables and their influence on themselves, an impulse response function can be established. The impulse response function is a sequence of dynamic changes after impact from a random perturbation factor to describe the dynamic response of the system to impact.

Figure 5e: Impulse response analysis results for ∆𝑳𝑵𝑺𝒁 itself, ∆𝑳𝑵𝑮𝑭 itself, of ∆𝑳𝑵𝑺𝒁 from ∆𝑳𝑵𝑮𝑭 and of ∆𝑳𝑵𝑮𝑭 from ∆𝑳𝑵𝑺𝒁.

This figure presents the impulse response curve, showing a convergence trend with the passage of the period, that is, changes in endogenous variables due to the impact become less important.

The upper left graph is the response curve of ∆𝐿𝑁𝑆𝑍 to its own fluctuation. It can be seen that the error impact of a standard deviation in the current period will bring about 8% impact in the later period, and then decay in the next period. Then there is basically no impact from the fourth period on.

The upper right graph shows the response curve of ∆𝐿𝑁𝑆𝑍 from ∆𝐿𝑁𝐺𝐹 fluctuation. It can be seen that ∆𝐿𝑁𝑆𝑍 has little effect on the ∆𝐿𝑁𝐺𝐹 except a weak positive response of the second period.

The lower left graph shows the response curve of ∆𝐿𝑁𝐺𝐹 from ∆𝐿𝑁𝑆𝑍 fluctuation. It can be seen that ∆𝐿𝑁𝐺𝐹 has a positive response to ∆𝐿𝑁𝑆𝑍 from the 1st to the 5th phase and reaches the peak value of 0.1% in the 2nd phase. There is no effect from the 5th phase on. However, after comparing with the upper right graph, it

can be concluded that the real estate market price is more sensitive to the fluctuation of the stock market price and is basically a positive relationship.

The lower right graph shows the response curve of ∆𝐿𝑁𝐺𝐹 to its own fluctuation. It can be seen that the error impact of one standard deviation of the current period will bring about 0.6% of the impact of the latter period, and it will be attenuated in the subsequent periods. Comparing to the upper left graph, the fluctuation is more gentle, and with no impact from the sixth period on.

5.5 Model construction

According to previews analysis, construct regression models using the Error Correction Model (ECM).

To analyze the impact from Shanghai Composite Index (SZ) to National Housing Climate Index (GF), form the regression model:

∆𝐿𝑁𝐺𝐹_𝑡 = 𝛼₀+ 𝜇∆𝐿𝑁𝐺𝐹_𝑡−1+ 𝛽∆𝐿𝑁𝑆𝑍_𝑡−1+ 𝑢_1𝑡

To analyze the impact from National Housing Climate Index (GF) to Shanghai Composite Index (SZ), form the regression model:

∆𝐿𝑁𝑆𝑍_𝑡= 𝛼₀+ 𝛿∆𝐿𝑁𝑆𝑍_𝑡−1+ 𝜑𝐿𝑁𝐺𝐹_𝑡−1+ 𝑢_2𝑡

Table 5f: Error Correction model regression results for ∆𝑳𝑵𝑮𝑭𝒕 and ∆𝑳𝑵𝑺𝒁𝒕

This table presents the results for Error Correction Model (ECM), and shows the impact level from Shanghai Composite Index (SZ) to National Housing Climate Index (GF)

(in column (1)) and vice versa (in column (2)). *, ** and *** indicate significance at 10%, 5% and 1%, respectively. ∆𝑳𝑵𝑮𝑭_𝒕 (1) ∆𝑳𝑵𝑺𝒁_𝒕 (2) ∆𝑳𝑵𝑮𝑭_𝒕−𝟏 0.3322*** (4.86) 0.0822 (1.11) ∆𝑳𝑵𝑺𝒁_𝒕−𝟏 0.0004** (2.05) 4.4730 (0.17) Constant term -0.0494 (-1.01) 0.0822 (1.11) Adjusted R-squared 0.1376 -0.0037 N 205 205

The t-value from ∆𝑳𝑵𝑺𝒁_𝒕−𝟏 to ∆𝑳𝑵𝑮𝑭_𝒕 is above 1.96, which means that it is statistically different from zero at a 5% significance level, implying that the change in Shanghai Composite Index (SZ) will pose positive influence on the National Housing Climate Index (GF), however, the coefficient is relatively small (0.0004), which means that the impact from Shanghai Composite Index (SZ) to National Housing Climate Index (GF) is so small that even can be ignored. The Error Correction Model can be shown like this:

∆𝐿𝑁𝐺𝐹_𝑡 = −0.0494 + 0.3322∆𝐿𝑁𝐺𝐹_𝑡−1+ 0.0004∆𝐿𝑁𝑆𝑍_𝑡−1

that the National Housing Climate Index (GF) pose no impact on the Shanghai Composite Index (SZ), then no econometrics model can be reached.

6. Robustness

To perform the robustness check, using the statistics from the stock market and the housing market in Hong Kong to fit into the Error Correction Model (ECM).

The data employed there will be yearly data from 1st _{of January 2010 to 31}st _of

December 2017. For the stock prices index, Hang Seng Index (hereinafter referred as SPI) will be used, and calculate the average of the index to get the yearly data. For the housing prices index, the yearly housing price index in Hong Kong (hereinafter referred as HPI) can be acquired from the official website of Census and Statistics Department in Hong Kong.

Table 6a: Descriptive Statistical Analysis for Hong Kong Housing Prices Index （HPI）and Hang Seng Index (SPI)

This table presents the descriptive statistics for Hong Kong Housing Prices Index (HPI) (in column (1)) and Stock Prices Index (SPI) (in column (2)), including observations, mean, maximum, minimum, standard deviation, skewness and kurtosis.

Hong Kong Housing Prices Index

Stock Prices Index (SPI)

（HPI） (1) (2) Observations 96 96 Mean 244.3958 22686.22 Maximum 352.9 29919.15 Minimum 138.3 17592.41 Standard deviation 59.0611 2418.405 Skewness 0.6911 0.0085 Kurtosis 0.0000 0.1340

To achieve robustness check, first construct a similar Error Correction Model for Housing Prices Index (HPI) and Stock Prices Index (SPI).

To analyze the impact from Stock Prices Index (SPI) to Housing Prices Index (HPI), form the regression model:

∆𝐻𝑃𝐼_𝑡 = 𝛼₀+ 𝜇∆𝐻𝑃𝐼_𝑡−1+ 𝛽∆𝑆𝑃𝐼_𝑡−1+ 𝑢_1𝑡

To analyze the impact from Housing Prices Index (HPI) to Stock Prices Index (SPI), form the regression model:

∆𝑆𝑃𝐼_𝑡 = 𝛼₀ + 𝛿∆𝑆𝑃𝐼_𝑡−1+ 𝜑∆𝐻𝑃𝐼_𝑡−1+ 𝑢_2𝑡

Table 6b: Error Correction model regression results for ∆𝑯𝑷𝑰𝒕 and ∆𝑺𝑷𝑰𝒕

This table presents the results for Error Correction Model (ECM), and shows the impact level from Stock Prices Index (SPI) to Housing Prices Index (HPI) (in column (1)) and

vice versa (in column (2)). *, ** and *** indicate significance at 10%, 5% and 1%, respectively. ∆𝑯𝑷𝑰_𝒕 (1) ∆𝑺𝑷𝑰_𝒕 (2) ∆𝑯𝑷𝑰_𝒕−𝟏 0.6535*** (8.54) 22.2656 (0.65) ∆𝑺𝑷𝑰_𝒕−𝟏 0.0005** (2.25) -0.0212 (-0.2) Constant term 0.7542 (2.46) 51.6079 (0.37) Adjusted R-squared 0.4731 -0.0172 N 96 96

The t-value from ∆𝑆𝑃𝐼_𝒕−𝟏 to ∆𝐻𝑃𝐼_𝒕 is above 1.96, which means that it is statistically different from zero at a 5% significance level, implying that the change in Stock Prices Index (SPI) will pose positive influence on the Housing Prices Index (HPI), however, the coefficient is relatively small (0.0005), which means that the impact from Stock Prices Index (SPI) will influence the Housing Prices Index (HPI) is so small that even can be ignored. The Error Correction Model can be shown like this:

∆𝐻𝑃𝐼_𝑡= 0.7542 + 0.6535∆𝐻𝑃𝐼_𝑡−1+ 0.0005∆𝑆𝑃𝐼_𝑡−1

For the ECM for ∆𝑆𝑃𝐼_𝒕, all statistics results show no significance, which means that the Housing Prices Index (HPI) pose no influence on Stock Prices Index (SPI), then no

econometrics model can be reached.

Results based on Hong Kong market are very similar to China’s mainland market, showing that the stock market tends to pose positive influence on the housing market, however, the impact is so small that even can be ignored. Thus, robustness can be achieved.

7. Conclusion

The housing market and the stock market both constitute vital parts of China’s economy. The former makes China’s economic structure slightly misshapen, while the latter is the most important part of the China financial market.

This thesis mainly uses empirical analysis to study the causality between National Housing Climate Index and Shanghai Composite Index. Learn from some scholars, this thesis brought in China’s mainland statistics, specifically, National Housing Climate Index represents China’s housing market, and Shanghai Composite Index represents China’s stock market.

From empirical results shown above, the housing market seems to have no significant influence on the stock market. To illustrate this situation, on the one hand, when the housing market is prosperous, instead of influencing the stock market, this good signal would arise people’s eagerness to purchase, which means that the hot money will continue to flow into the housing market, and induces the housing prices to climb higher.

On the other hand, if the housing market experiences a sluggish, then the housing market may froze, which means that the houses transaction will slow down, and people can only hold those houses in hand, however, this terrible situation may still cannot affect the stock market.

According to the results that coefficients are statistically different from zero, the stock market should have positive impacts on the housing market. Liquidity may contribute to the discrepancy, on the one hand, when the stock market is prosperous, people can enter into the stock market and participate in transactions without efforts, furthermore, after earning a large amount of money in the stock market, some people may consider to move a part of money from the stock market to the housing market to achieve the currency preservation, which will improve the housing market. On the other hand, when the stock market crashes, people tend to hold negative expectations toward China’s economy, and choose not to invest money into the housing market, either, since in most people’s mind, the two markets connect tightly.

However, although significant, those coefficients are nearly zero, which can even be ignored, implying that, actually, the stock market will pose nearly no influence on the housing market. To illustrate this, firstly, only minority people have the ability and risk preference to move money from the housing market to the stock market, since the stock market always experiences much higher risks than the housing market. Secondly, few people possess commercial real estates that can be sold out for money to gambling for high yield. Thirdly, some investors still view the stock market and the housing market as two separate markets, and they will not tend to move money from one market to the

other although one of the markets fluctuates. To make it more clear, when the housing market fluctuates, people tend to hold still their investment position, and when the stock market fluctuates, investors may choose to move out money and wait for the next investment opportunity.

In conclusion, based on statistics from China market, the changes in stock market tend to pose positive influence on the housing market, however, the impact is so small that even can be ignored, and the housing market fluctuations seem to have no impact on the stock market.

References

Okunev, J., Wilson, P., & Zurbruegg, R. (2000). The Causal Relationship between Real Estate and Stock Markets. The Journal of Real Estate Finance and Economics, 21(3), 251- 261

Khaled I. Batayneh & Abdullah M. Al-Malki (2015). The Relationship between house prices and stock prices in Saudi Arabia: An Empirical Analysis. International Journal of Economics and Finance, 156-167

Ghulam Ali & Khalid Zaman (2017). Do house prices influence stock prices? Empirical investigation from the panel of selected European Union countries. Economic Research, Vol. 30, 1840-1849

Nannan Yuan, Shigeyuki Hamori & Wang Chen (2014). House Prices and Stock Prices: Evidence from a Dynamic Heterogeneous Panel in China. Discussion Pape, 1428

Asli Yuksel. The relationship between stock and real estate prices in Turkey: Evidence around the global financial crisis[J]. Central Bank Review, 2016,16(1).

Mansor H. Ibrahim. House price-stock price relations in Thailand: an empirical analysis[J]. International Journal of Housing Markets and Analysis, 2010, 3(1). Zan Yang. Co-integration of housing prices and property stock prices: evidence from

the Swedish market [J]. Journal of Property Research, 2005, 22(1): 1 -17.

Raymond Y. C. Tse. Impact of Property Prices on Stock Prices in Hong Kong[J]. Review of Picnic Basin Financial Markets and Policies, 2001, 4(1): 29- 43.

John Okunev, Patrick Wilson, Ralf Zurbruegg. The causal relationship between real estate and stock markets[J]. Journal of Real Estate Finance and Economics, 2000, 21 (3):251-261.

David C. Ling, Andy Naranjo. The Integration of Commercial Real Estate Markets and Stock Marketsf[J]. Real Estate Economics, 1999, 27(3): 483- 515.

Daniel C. Quan, Sheridan Titman. Do Real Estate Prices and Stock Prices Move Together an International Analysis[J]. Real Estate Economics, 1999, 27(2):183-207. Brent W. Ambrose, Esther Ancel, Mark D. Griffiths. The Fractal Structure of Real Estate Investment Trust Returns: The Search for Evidence of Market Segmentation and Nonlinear Dependency[J]. Journal of the American Real Estate and Urban Economics Association, 1992, 20: 25-54.

Zhang Nan. An empirical study on the correlation between China's real estate market and stock market[D]. Dongbei University of Finance and Economics, 2016.

Cai Zengchang. Empirical research on the time-varying correlation between China's stock market and real estate market returns [D]. Dongbei University of Finance and Economics, 2015.

Wang Wenfei. Research on the linkage between China's stock market price and real estate market price [D]. Jinan University, 2014.

Estate Market[J]. Statistics and Decision, 2014, (02): 127-131.

Yang Boli, Liu Boyyu. Research on the Dynamic Correlation between Real Estate Stocks and Related Assets in China [J]. China Real Estate, 2013, (18): 3-17.

Liu Zhiping. Research on volatility and correlation of real estate market and stock market based on ARCH model family [D]. Central South University, 2012.

Yu Rui. Empirical study on the correlation between China's stock market and real estate market [D]. Fudan University, 2012.

Guo Shiping, Ji Jie. Research on the Relationship between Real Estate Price and Stock Price Index in China[J]. Journal of Shenzhen University (Humanities and Social Sciences), 2012,(05):106-112.

Peng Xingting. Research on the Correlation between the Fluctuation of China's Real Estate Market and Stock Market[J]. Journal of Beijing Institute of Technology (Social Sciences Edition), 2011, (05):39-43.

Guo Dexian, Sheng Lifeng. Research on the Dynamic Relationship between Real Estate Market and Stock Market in China [J]. Economic Forum, 2011, (02): 69-72.

Cheng Dazhao, Zhao Yuhua. Research on the Correlation Degree between China's Housing Market and Stock Market[J]. Economic Theory and Business Management, 2010, (08): 38-44.

Li Xiaohuan, Feng Xiujuan. Correlation Analysis between Stock Market and Real Estate Market [J]. Chinese Business Community (second half), 2010, (03): 4-5.