A study of the determinants of the housing price in Taiwan

Transcription

1 A study of the determinants of the housing price in Taiwan National Chengchi University (NCCU) International Business Leo Chen Lisa Fu Lola Wu Annie Huang Gengin Wong Supervisor: Prof. Shieh Jun, 2014

2 I. Motivation Recently, Taiwan has been ranked as the fourth hottest housing market in the world by CNN News. Taiwanese have suffered the soaring housing price problem as many other countries in the world. According to the Ministry of the Interior of Taiwan, the house price to income ratio is defined as the ratio of the median housing price to median familial disposable incomes. It is also the basic affordability measure for housing in a given area. In the past ten years, it has risen to 9.2. It means that, without spending any money of his monthly salary, it would take almost 9 years for a person to own a house in Taiwan. However, in other countries, the ratio is 3-5 times in average. What s more, a recent report on the China Times shows that the house price to income ratio of Taipei city is the highest all over the world, and the New Taipei city is on the third place. There is a big concern that housing prices in Taiwan are already unaffordable for most first-time buyers. It is quite hard for those first-time buyers to own a house in urban areas. As purchasing houses is important for Taiwanese people, the soaring housing prices indeed becomes a serious problem for our society. When it comes to the soaring housing prices in Taiwan, many reasons could be discussed. First, we should take the uniqueness of real estate property in Taiwan into consideration. In Chinese traditional concept, as the old saying goes, Along with real estate comes about wealth. The saying means, for Taiwanese people, they not only view real estate property as a necessity but also expect purchasing real estate property as an investment as well. When the financial market is relatively unstable, people would seek to put their money into relatively inflation-proof and possible appreciation market, such as housing market, in order to hedge the uncertain risks. Second, compared to other

3 neighboring countries in Asia, the house tax and the land value tax in Taiwan are relatively lower. So the cost of owing houses in Taiwan are lower than in other neighboring countries. It motivates Taiwanese people to purchase houses as investments. Even though there are many reasons that have been discussed, we are still curious about what are the main determinants that influence the housing prices in Taiwan. Based on the articles we have read and the hypothesis we set, we conduct the statistical analysis, such as collinearity tests, autocorrelation, and the stepwise regression on some macroeconomic determinants. We found that M1a, Real GDP, Construction Stock Price Index, and Interest rate are the most significant determinants that affect housing prices in Taiwan for the past 20 years. II. Articles Review From the past studies, they all point out that different kinds of determinants are influencing the housing prices in Taiwan. These studies respectively choose different determinants to test the relationship with housing prices. However, most of them only focus on only two or three few determinants. We expect to figure out what the most important determinants affecting Taiwan's housing price are among all those important determinants. Based on articles we studied, we choose M1a, M1b, M2, real GDP, interest rate, foreign exchange rate, CPI, construction material CCI, TAIEX, and Construction Stock Price Index to study which are more related to housing prices. In the past studies, we found that money supply has an important influence on the housing prices in Taiwan. For example, Chien-Wen Peng, Vickey C.C.Lin and Ya-Ting Yang (2004) use the data from 1981Q1 to 2001Q4 and impulse

4 response analysis to explain structural changes of housing prices in Taipei City and Taipei County. The empirical results revealed that housing prices in both cities are correlated with money supply, stock market index and the allowed floor areas specified by the building permit. The study result reveals that stock market index has direct influence on housing prices while money supply has indirect influence on housing prices. When money supply changes, it will affect stock market and thus affect housing prices. Thus, we take money supply into consideration. Also, Chi-Wei Chen (2010) uses the data from 1994Q1 to 2010Q1 to indicate the dynamic relationship on stock price, housing price and exchange rate. In his article, he expects that there are abundant transactions in foreign exchange market in Taiwan. Thus, exchange can play a role in deciding housing prices. Also, the stock price has a positive relationship on housing prices. Next, in macroeconomic determinants, Dokko (1999) uses GDP as a determinant in his model. He expects that when GDP goes up apparently, the housing prices will go up with GDP. Therefore, GDP has a positive influence on housing prices. Yen-kuang Chen (2010) also analyzes the relationship between real estate cycles and the macroeconomic determinants. This study uses the data collected from Jan to Dec The indicators of real estate cycles adopted in the thesis are the trading volume of real estate and the housing price index, and the macroeconomics determinants include mortgage rates, the stock price index, money supply and the price index. Based on cointegration Test, they found there is a long-run equilibrium between real estate cycles and the macroeconomic determinants. There is a two-way feedback among the stock price index and the housing price index. Moreover, the price index affects the housing price index. The stock price index and the price index have less effect

5 on the housing price index. The price index is more accountable than the stock price index for the fluctuations of the housing price index. I-Chun Tsai and Ming-Chi Chen (2013) propose that asymmetric reactions of the stock and housing markets will result in asymmetric relationship between the two markets. Tsai and Chen use Sinyi Housing Price Index of Taipei city and the Taiwan Exchange Capitalization Weighted Stock Index from 1991Q3 to 2008Q3 for empirical analysis. They reveal that regarding the integration of housing prices and the stock price index, the correlation behavior of housing prices toward long-term equilibrium is more significant when demand increases significantly, but is not significant when demand declines. Chien-Wen Peng and Chin-Oh Chang (2000) say that from Taiwan has gone through three apparent cycles. In this article, the determinants are pre-construction real estate housing prices, vacant houses, and building permits. They use the data from 1981Q1 to 1999Q1. From their study, we 已註解 [u1]: 建造率面積 examine that in the first two booms, the insufficient house supply and increasing oil prices cause the inflation. People have expectation of housing prices maintaining so they increase the demand of house. Therefore, the supply is less than demand. The cause of third housing price peak is the economics grew too fast and the soaring stock market. Excessive money supply caused the third cycle. From their observation, we can examine that there is a business cycle every 7 year in Taiwan, and each cycle has its own important determinants affecting the housing price. However, the cycle has a structural change after Thus, we will focus on this change and use the quarterly data from 1994 to 2013, total 80 data, to find the causes that are affecting Taiwan s housing price in this cycle. Furthermore, we also want to examine and compare among all the determinants which are more influential to housing price.

6 III. Methodology 3.1 Research Framework Being curious about what determinants affect the housing prices, we come up with several important determinants and try to find relationships between each one of it and the housing prices. We will first introduce every variable we choose. Second, we will make a hypothesis on how the determinants are going to affect the housing price. Third, we will conduct the statistical analysis, and use methods such as collinearity tests and autocorrelation to reduce the error that may occur. Last, we will conduct the stepwise regression to eliminate the less-related determinants. 3.2 Determinants Dependent Variable (1) Sinyi Housing Price Index Sinyi Housing Price Index is the most common-used index in Taiwan, which is conducted by CNCCU- SINYI Research Center for Real Estate. The center applies the Hedonic Price Method and use the Laspeyres formula to conduct the index. To make our data more typical, we use the data of whole Taiwan, not just in Taipei or any other city. The graph 1 shows that the housing prices of Taiwan have been in a rise in the past two decades.

7 Graph 1: show the trend of different determinants on a timely basis. (Reference: CNCCU- SINYI Research Center for Real Estate) Independent Determinants (1) Money Supply(per million) M1a (X 1 ) M1A comprises the currency held by the public plus checking accounts and passbook deposits of enterprises and individuals. M1b (X 2 ) M1B comprises M1A and passbook savings deposits of individuals with other depository corporations M2 (X 3 ) M2 comprises M1B and quasi-money. Quasi-Money comprises time deposits, time savings deposits, foreign currency deposits (including demand deposits and time deposits), postal savings deposits (including giro accounts, passbook savings deposits and time savings deposits), repurchase agreements, non-resident NT dollar deposits and money market mutual funds of enterprises and individuals with other depository

8 corporations. (2) Macroeconomic Determinants Real GDP (X 4 ) Real GDP is defined as the market value of all officially recognized final goods and services produced within Taiwan each year, and we collect our data from Directorate-General of Budget, Accounting, and Statistics, Executive Yuan, R.O.C. Unlike nominal GDP, real GDP can account for changes in the price level, and provide a more accurate figure. In recent years, the real GDP has been rising continuously year by year. Though it rises and falls frequently for the time being, it has a smooth fluctuation. Interest Rate (X 5 ) Here we choose the data of the weighted average bank loan interest rate between local banks from TEJ. When interest rate drops, people can benefit from investing other than keeping your money in the bank. Also, the costs for borrowing money from others become lower. The downward curve means the interest rate in Taiwan has been in a decline in general in the past twenty years. Foreign Exchange Rate (X 6 ) Foreign Exchange Rate is the price of a nation s currency in terms of another currency. It is also an important index when we talk about macroeconomics. Here we use average foreign exchange rate in indirect quotation method as our data (NTD/$). As a result, when New Taiwan Dollar appreciates, the curve will go down instead of moving upward. (3) Indices CPI (X 7 )

9 CPI reflects people s living level of that time. As we can see the data from Directorate-General of Budget, Accounting, and Statistics, Executive Yuan, R.O.C, CPI has kept rising in the recent decades, which means the burdens on Taiwanese people are getting heavier and heavier. However, CPI only counts housing rents, housing prices are not included in CPI. But both of the two have been in a rise these years. so we want to see what is the difference between the trend of these two indices. Construction Material CCI (X 8 ) To know how cost factors affect the housing market, we choose construction material CCI as our variable. Construction Material CCI is an index conducted by Directorate-General of Budget, Accounting, and Statistics, Executive Yuan, R.O.C, which shows the price of materials in the producer market. In the past two decades, the price of construction material has risen a lot. TAIEX (X 9 ) TAIEX = ( Stock Price Volume ) 100 Base Value TAIEX is the abbreviation of Taiwan Capitalization Weighted Stock Index, with the base year of 1966.It is conducted by the Taiwan Stock Exchange Corporation, which is a corporation in charge of the stock market and securities market in Taiwan. TAIEX covers all of the listed stocks excluding preferred stocks, full-delivery stocks and newly listed stocks, which are listed for less than one calendar month. It is one of the most important index to see the economic condition of Taiwan, it also represents the trend of the overall stock market. Construction Stock Price Index (X 10 )

10 Construction Stock Price Index is an index which is also conducted by the Taiwan Stock Exchange Corporation. The formula of construction stock price index is same as the TAIEX one, but it especially take the data of construction stock. To see how stocks influence housing prices, we Construction Stock Price Index as our last variable. Table 2: show the statistical data of all the determinants mentioned above. Mean Maximum Minimum Standard Deviation Variance Skewness Kurtosis Sinyi Housing Price Index M1A (per million) 2,694, ,450, E M1B (per million) 6,868, ,764, E M2 (per million) 22,286, ,553, E Real GDP (per million) 3,021, ,063, E Interest Rate Exchange Rate (NTD/$) CPI Construction Material CCI TAIEX Construction Stock Price Index Table 3: Statistical data of each determinant (Source: Taiwan Economic Journal) (Source: Taiwan Economic Journal)

11 (Source: Taiwan Economic Journal) (Source: Directorate-General of Budget, Accounting, and Statistics, Executive Yuan, R.O.C.) (Source: Taiwan Economic Journal) (Source: Taiwan Economic Journal) (Source: Directorate-General of Budget, Accounting, and Statistics, Executive Yuan, R.O.C) (Source: Minister of the Interior, R.O.C.)

12 (Source: Taiwan Economic Journal) (Source: Taiwan Economic Journal) 3.3 Hypothesis We have made the hypothesis on the 8 determinants that we choose. These assumptions are based on previous articles that we have read and the current economic situation in Taiwan. Table 4 shows our hypothesis we made. Among the 8 determinants that we have chosen, we make a bold assumption that the money supply will be the main reason that affects housing price in Taiwan these years. We will clarify and explain the reason more below. Table 4: The hypothesis on these 8 determinants Factors Money supply Real GDP Exchange rate Interest rate CPI Construction CCI Construction Stock Price Index TAIEX Change In house price index (1) Money supply: The reason why we regard it as the most important determinants among

13 all of our determinants is because the financial crisis that occurred in It was known as the biggest financial event during the period that we observe from And at that time, American started to conduct the QE monetary policy. Hence, there are lots of U.S. dollar money being issued to the world, especially into emerging markets, including Taiwan. Because the hot money floods into Taiwan market, it forces the central bank to raise the same level money supply to prevent the soaring appreciation. So this event has heavily influenced the money supply in Taiwan. Thus, we assume that as the money supply increases, it will cause the inflation which results in the hedge in house market and even the higher housing prices. (2) GDP The GDP is highly related to the mortgage. Therefore, when the mortgage increase, it will reflect on the GDP. It also means a growing demand on house, leading the housing price to go up. (3) Exchange rate: We assume that when the exchange rate goes up, which means the new Taiwan dollar to depreciate. And it will cause the import building materials cost more. In the end, the house price will go up. (4) Interest rate The unit price of house is much higher than other products in the market. Thus, most of the purchasers need to borrow money from bank. Therefore, when the interest rate of the mortgage goes up, it will lower the willingness to buy the house and cause the housing price to drop. (5) CPI When purchasing power decreases, people will turn their currency into

14 inflation-proof asset in order to prevent the value of money from shrinking. So the demand of house will increase. Thus, when the CPI goes up, we expect the house price will go up. (6) Construction CCI Construction cost is a direct concern in pegging price. So we expect that the higher construction CCI will reflect on a higher housing prices. (7) TAIEX As for the TAIEX, we believe that the stock market and the housing market are two important investment markets. When the TAIEX goes up, it means the investors choose to join the stock market which will decrease their participant in housing market. And it will result in the decrease of housing price. (8) Construction stock price index Construction has been an indicate industry in Taiwan. Hence, we estimate that the construction stock price index can reflect the economy in Taiwan. So when the market is prosperous, in our opinion, we expect the purchase in house and the house price will also increase. 3.4 Multiple Regression Analysis Multiple regression would be the core method we conduct in this research. We mainly use this method to discuss about how the determinants we have chosen are related to the housing price index. With multiple regression, we could also investigate what factors may be the major causes that affect the current housing prices, which is currently too high in Taiwan.

15 Multiple Regression Formula Y = α + β 1 X 1 + β 2 X 2 + β 3 X 3 + β 4 X 4 + β 5 X 5 + β 6 X 6 + β 7 X 7 + β 8 X 8 + β 9 X 9 + β 10 X 10 + ε Y : Sinyi Housing Price Index X 1 : M1a X 2 : M1b X 3 : M2 X 4 : Real GDP X 5 : Interest Rate X 6 : Foreign Exchange Rate (NTD/$) X 7 : CPI X 8 : Construction Material CCI X 9 : TAIEX X 10 : Construction Stock Price Index α : Intercept β : Coefficient of each variable ε : Error Multiple Regression Results We ran our first regression using all the ten determinants. According to the result, M1b, interest rate, TAIEX and Construction Stock Index are the four significant determinants. This result has an adjusted R square of 0.98, which means the result is able to explain 98% of the data. Moreover, the coefficient of M1b is relatively small and positive, indicating that there is only slightly influences between the housing price and a positive influence. When the housing price rises, M1b also rises. The coefficient of interest rate is more than 12, which means that this determinant can dramatically influence the housing price, and when the housing price rises, the interest rate will rise, too. TAIEX only has on its coefficient, so the influence is not too dramatic. However, it has a negative effect. We expect the reason is that, when the

16 stock market is prosperous, people will invest in stock market more and at the same time invest less on the real estate market. The Construction Stock Index also influences the housing prices with a coefficient of 0.058, and has a positive effect. Table 5: The result of the multiple regression Money Supply Macroeconomic Variables Index Housing Price Index (Taiwan) Y coefficient t Statistics P-Value M1A X1 2.60E M1B X2 1.39E M2 X3 1.91E Real GDP X4-1.26E Interest Rate X E-10 Foreign Exchange Rate (NTD/$) X CPI X Construction CCI X TAIEX X Construction Stock Price Index X Intercept Adjusted R= Observations=80 Period= (quarterly) However, we have made this regression analysis by ignoring the collinearity problem and autocorrelation effect. Later on, we will conduct the stepwise regression to eliminate the errors that collinearity problem and autocorrelation effect could cause. Collinearity Test If the independent determinants are highly correlated with each other, that may cause the so-called collinearity problem. That way, we won t be able to proceed the regression process. One way to testify the existence of collinearity is to calculate the correlation coefficient. Normally, if the

17 correlation coefficient between two determinants are close to 1 (perfect correlated), that means it will be harder to use OLS (originary least squares) method to conduct the regression analysis. The results are shown below. Table 6: The collinearity test between each two determinants Sinyi Housing Price Index M1A M1B M2 Real GDP Sinyi Housing Price Index M1A M1B M Real GDP Interest Rate Foreign Exchange Rate (NTD/$) CPI Construction CCI TAIEX Construction Stock Price Index Interest Rate Foreign Exchange Rate (NTD/$) CPI Construction CCI TAIEX Construction Stock Price Index This table above indicates that 26 out of 45 correlation coefficient are over 0.8, and that is nearly 57%. So, first, we learn from the correlation coefficient that we cannot use all of these 8 determinants in the regression formula. Autocorrelation (lag=1) Next we calculate the autocorrelation and set the lag period for one quarter. We wonder if one single variable has its correlation itself when doing some time series tests. The results are demonstrated in table 7. Table 7: The autocorrelation result (lag=1) Sinyi Housing M1A M1B M2 Real GDP Interest Rate Price Index Foreign Exchange Rate (NTD/$) CPI Construction CCI TAIEX Construction Stock Price Index lag= From table 7, we can see that almost all determinants have correlation of more than 90%, which means that each variable itself has strong

18 relationship with its lag period. Both the results of correlation coefficient and the autocorrelation show the fact that some determinants must be eliminated, and in the meantime keep the more useful ones in the formula. In order to choose the useful determinants, we decide to conduct the Stepwise Regression. Stepwise Regression Since there are collinearity problems, some determinants must be eliminated, so that we are able to run multiple regression. When it comes to deciding which determinants should stay and which should eliminate, we conduct the stepwise regression as a way to keep the most valuable variable in the formula. In our theory, we got only general direction as to which of a pool of candidate determinants should be included in the regression model. After the auto-correlation& correlation test we have done above, we notice that we have to solve the collinearity problem. Thus, we have to select the final determinants set. First, we pursue the regression model to be as representative and realistic as possible. Second, we want to include as few determinants as possible because each irrelevant determinants decreases the precision of the estimated coefficients and predicted values. In order to reach a final model, we decide to use stepwise regression method as our way to select the determinants. Stepwise Regression Result We got the stepwise regression result below by using SPSS. Stepwise regression set two significance levels: 0.05 for adding determinants and

19 0.1 for removing determinants. Through this criteria, we select 4 main determinants as our determinants. Table 8: The stepwise regression result (part1) Model Determinants entered Method 1 M1A(million) Stepwise(Criteria:F-to-enter <=.050 possibility,f-to-remove >=.100 possibility) 2 Loan Interest rate of Taiwan Bank Stepwise(Criteria:F-to-enter <=.050 possibility,f-to-remove >=.100 possibility) 3 Real GDP(million) Stepwise(Criteria:F-to-enter <=.050 possibility,f-to-remove >=.100 possibility) 4 Construction stock price index Stepwise(Criteria:F-to-enter <=.050 possibility,f-to-remove >=.100 possibility) Determinants entered/removed (dependent determinants: house price index) We can also take a look at the table 9. From the data below, we can notice that the model 4 has biggest R, R square and adjusted R square value. It represents that model 4 has a higher explanatory than the other. So that is why we choose model 4 as our model in this paper. With the significance in ANOVA are all less than 0.05, we can be sure that there is at least more than one coefficient that is not equal to 0. Table 9: The stepwise regression result (part2) Model R R square Adjusted R square Standard Error of Estimate 1.936a b c d

20 Model Summary a. Predictors:(Constant) M1A (million) b. Predictors:(Constant) M1A (million), loan interest rate c. Predictors (Constant) M1A (million), loan interest rate, Real GDP(million) d. Predictors:(Constant) M1A (million), loan interest rate, Real GDP(million), construction stock price index e. Dependent: House price index Now, in order to explain the result, let s take a look the table 10. As we can see from this table, the multiple regression result of model 4 is listed. It shows that M1A has the highest correlation with house price index among all those factors we choose. Loan interest rate and the construction stock price also has a positive correlation, while the real GDP is negative correlated. However, there are big differences among all the unstandardized coefficients. M1A and GDP got comparative small number, while the others are normal numbers. In order to even the number of these coefficients, we are going to revise the regression model and change the arithmetic unit of M1A and GDP, to make the result easier to elaborate. Table 10: The stepwise regression result (part3) Unstandardized coefficients standardized coefficients Model B Std. error Beta T Sig. 1 (Constant) M1A(million) 4.266E (Constant) M1A(million) 7.116E Loan interest rate (Constant) M1A(million) 7.313E Loan interest rate Real GDP(million) E

21 4 (Constant) M1A(million) 6.965E E-50 Loan interest rate E-06 Real GDP(million) E E-14 Construction stock price Dependent: House price index Revised Multiple Regression (4 determinants) Y = α + β 1 X 1 + β 2 X 2 + β 3 X 3 + β 4 X 4 + ε Y : Sinyi Housing Price Index X 1 : M1a X 2 : Real GDP X 3 : Interest Rate X 4 : Construction Stock Price Index α : Intercept β : Coefficient of each variable ε : Error Revised Multiple Regression Result From table 10, we find that the M1A and real GDP has a much smaller unstandardized coefficient. In order to revise the coefficient to a more even number compared to the prior one. We change the arithmetic unit of M1A and real GDP. The sign of each determinant is still the same. And the Adjusted R square is equal to 0.98, which represents these determinants can explain nearly 98% of the determinants that affect housing price index. Table 11: The revised multiple regression result

22 F-Value= Adjusted R= Observations=80 Period= in quarter Residual Plots In order to validate our model, we check the residual plots. To see if the points in a residual plot are randomly dispersed around the horizontal axis. We can assess whether the observed error (residuals) is consistent with stochastic error. coefficient t Statistics P-Value M1a X E-50 Real GDP X E-06 Interest Rate X E-14 Construction Stock Price Index X Intercept Graph 12: The residual plot

23 From the four residual plots above, we can find out our determinants all get a random residual plot which means they don t fall into symmetrical pattern. And in the OLS context, random errors are assumed to produce residuals that are normally distributed. So we can believe the revised regression model is validate.

24 3.6 Conclusion Taiwan often shows a type of cyclical fluctuation. To explore this phenomenon, many papers dig into the relationship between 2-3 determinants and housing prices and many different conclusions has been proven. Thus, the main purpose of this study is to research which determinants affecting housing price in Taiwan efficiently in this housing price cycle. In our study, determinants are first selected from previous research and then chosen by stepwise procedure. So among all those determinants, we construct the indicators of this housing price cycle are M1A, loan interest rate, construction stock price and GDP. We discover that the housing price is positive correlated to M1A, loan interest rate and construction stock price, especially to M1A. On the other hand, it is negative correlated to GDP. However, as we can see, the interest rate and GDP is opposite from our hypothesis. These are very interesting phenomena. As to interest rate, this is in accordance with the conclusion as a foreign study (Summers 1981, Peiser & Smith 1985, Harris 1989). The study points out that the change in interest rate doesn t have an instant reaction. Thus, we think that the lag reaction of house price is the main reason why these two determinants have a positive correlation. For real GDP, in our opinion, we think that difference might stem from overvaluing housing prices and the uncertainty of the housing market which makes buying housing price bearing more and more pressure and risk. And in sum, we find that the money supply standing out as the most fundamental factors influencing Taiwan s housing price nowadays. Thus, after our empirical analysis, we expect housing investors or buyers should pay attention to the monetary policy and the money supply amount in Taiwan to

25 hedge the risk. Also, we suggest government should focus on the monetary policy to improve the housing market environment in Taiwan.

26 IV. References Chien-Wen Peng, Vickey C.C.Lin and Ya-Ting Yang An Analysis of Structural Changes in Housing Prices: Changes of Taipei City and Taipei. Chi-Wei, Chen The Dynamic Relationship On Stock Price, Housing Price and Exchange Rate. I-Chun Tsai and Ming-Chi Chen Asymmetric Correlation and Difference between the Volatility of Housing and Stock Price Indexes: Analysis Based on the Threshold Volatility and Cointegration Model. Journal of Financial Studies (TSSCI, EconLit), 21(4): Ming-Chi Chen and Kanak Patel An Empirical Analysis of Determinants of Housing Prices in the Taipei Area. Taiwan Economic Review, 30(4): Christopher Otrok, Marco E. Terrone House Prices, Interest Rates and Macroeconomic Fluctuation: International Evidence 彭建文 張金鶚 總體經濟對房地產景氣影響之研究 出自 人文及社會科學研究彙刊,Vol.10, No.3, pp (tssci) 陳彥光 2010 房地產景氣與總體經濟變數之關聯性分析- 價與量的對比分析 許易民 2009 台灣房地產業景氣循環之影響與對策, 綜合規劃研究 96 年及 97 年 出自 行政院經濟建設委員會, 林秋瑾 黃佩玲 1995 住宅價格與總體經濟變數關係之研究- 以向量自我迴歸模式 (VAR) 進行實證 出自 政大學報, 第 71 期,P 林秋瑾 王健安 張金鶚 1996 房地產業對總體經濟活動之影響之分析 出自 台銀季刊,47(1) p1-24 張金鶚 1998 台灣房地產景氣循環 理論與實証分析 (II) 出自 國科會八十七年度專題研究