V ECTOR A UTOREGRESSION M ODEL

According to the results of unit root verification, all variables rate of change is I(0) series.

For analysis of relationship between GDP, consumption and other variables, the VAR model proposed by Sims(1980) was adopted to estimated direct effect relationship between the variables of intertemporal, provided that each variable should be I(0) series.

According to estimated coefficients of VAR model (as shown in Table 4-2). In the consumption equation, HPI, SPI and consumption are significantly, implying that under the direct effect, current CS is influence by the first lag of HPI, SPI and consumption. First, consumption (rate of change) of first lag had a positive significant influence on current consumer spending (rate of change), although under the intertemporal choice of consumption, the increase of previous consumption would substitute for the current consumer spending; in other words, the current consumer spending would decrease as the previous consumption increased, and the reduction of previous consumer spending would probably increase the current consumption. However, our consumption to GDP is 60%, when the lag period of consumption increase its means our domestic consumption demand growth, the country could avoid the environment of deflation, to cause future consumption (rate of change) positive growing.

The HPI (rate of change) of first lag has a negatively and significantly effect on current consumption (rate of change). Most of previous literatures thought that the real estate could bring wealth effect and further affect economic development through the influence of employment, saving, investment, and labor productivity. The real estate investment was considered contributor to economic growth. However, the study showed that house prices negatively affected consumer spending, which caused economic depression. The reason was that the real estate played different roles in different conditions, such as research time, research methods, and research regions. Since early 1980s, the economy as well as income in Taiwan grew rapidly, and most people bought houses for their actual residence. House prices,

economy, and income grew at the same time, and the wealth effect came accordingly, which was consistence with past literatures. However, currently, the real estate market in Taiwan faced the low-rate and high loan-to-value, low carrying cost of real estate (land tax, house tax), and low capital gains tax (land value increment tax) caused by the easing monetary policy, which indirectly encouraged investors to enter the real estate market with low financing cost and stimulated the demand for investment, making the house price increase significantly. At that time, the real estate did not bring direct productivity to consumers; they were only investment goods. When the real estate market could not increase the actual national productivity, it was like pouring money into the bottomless hole. For the nation and consumers, such investing behaviors were inadequate types of consumer spending.

The SPI (rate of change) of first lag had a positively significant influence on consumption (rate of change), showing that the stock index during the study brought wealth effects for consumer spending. The stock market and real estate market varied significantly.

Compared to real estates, stocks had small prices, central market, transparent information, a number of products, high homogeneity, and flexible short-term supply, which were reasons for higher liquidity and trading frequency of stock than real estates. They also caused stocks to have more significant wealth effect on consumer spending. Moreover, development of stock market could enable companies and vendors to increase production, employment and income, promote economic growth, and issue dividends to stockholders, which benefited the public, companies, and countries. Thus, compared to the real estate market, the stock market had a steady contribution to consumer spending.

In the study, the insignificant variables in consumption were IR and CPI. The interest rate continues decline under the easy monetary policy in recent years. This study showing that it was hard to encourage civic consumer spending through the easy monetary policy in Taiwan. Besides, according to Fisher equation (MV=PQ), money supply and consumption rate varied reversely. If the government intended to increase civic consumer spending through

the monetary policy, it should first increase economic growth or CPI in order to increase the consumption rate. However, after the impact of U.S. subprime mortgage crisis in 2008 on global economy and financial market alleviated, Taiwan promoted the easing monetary policy to encourage economic growth. So far, economic growth in Taiwan still growing slow, while the prices of assets increased, especially the price of real estates, which significantly has crowding-out effect on consumption and brought economic growing slow.

The first lag of INCOME (rate of change) had a positive insignificant influence on current consumer spending (rate of change), which was inconsistent with traditional consumption theory proposed by Friedman and past literatures. The reason was that during the study “the real national income” increased by 1.6% on average annually, showing that the growth of “real income” was nearly zero; it would possibly lead to the reduction of consumer spending. However, the effect was still insignificant. The growth of income would have a positive influence on consumption, but the zero growth or decline of income could possibly inhibit consumer spending.

In the consumption equation, we can learn that the first lag of HPI and SPI (rate of change) has positive significant influence on current HPI (rate of change), and when the first lag of consumption increasing, has negative influence on current HPI under the direct effect.

Therefore, there has crowding-out effect between consumption and HPI. And the coefficients of current HPI is strongly affected by the first lag of IR, showing that when increase interest rate of mortgage, which can lead to house prices dropping down.

Table 4-2 VAR Result on CS and HPI

CS HPI

Variables Coefficient T-sta Coefficient T-sta

CS(-1) 0.386718** [2.35] -1.206595*** [-2.77]

Note:1. *, **and***denote significance at the 10%,5% and 1% level respectively.

2.The optimal lag length is selected by Akaike information criterion (AIC) and Likelihood ratio (LR) test statistic. The first lag is suitable for these two VAR models.

In the GDP equation, we can observe the first lag of HPI is not significantly affect current GDP (as shown in Table 4-3), though house prices has crowding-out effect on consumption(60% of GDP) under the direct effect, since the constituent elements of GDP also contain government spending, investment, and net exports, this study did not observe house prices on other variables. Under the average of direct effect in 15 years research period, the lag of house prices is not significantly affect current GDP. It is hard to stimulate consumption or to make a significant positive impact on the overall economy. Hence, it difficult to consider real estate market as a "locomotive industry" under the easy monetary policy.

Furthermore, the first lag of SPI has positive significant influence on current GDP. The growth of SPI can stimulate consumption directly on the discussion above, for the other constituent elements, the development of the stock market can bring direct productivity to industries, attract domestic and foreign investment, raise government spending, and produce related products to enhance the net exports. Thus, compared to the real estate market, the stock market had a steady contribution to economic growth on the average of effects.

Moreover, the first lag of CPI has significant negative influence on the current GDP, consistent with the Fisher equation. The high CPI will cause inflation, distort market prices operational mechanism, affect the efficiency of resource allocation. Lead to some companies easily obtain high profits, and cause people who had private wage income in dilemma. Hence, central bank had to ensure the market price mechanism to operate effectively, and stable price level, so that wage earners can enjoy equitable distribution for the whole social wealth, and increase standard of living gradually.

Table 4-3 VAR Result on GDP and HPI

GDP HPI

Variables Coefficient T-sta Coefficient T-sta

GDP(-1) 0.438934 [1.50] -0.403192 [-0.72]

HPI(-1) 0.003596 [0.05] 0.343571*** [2.55]

SPI(-1) 0.069641*** [3.29] 0.043823 [1.09]

INCOME(-1)

0.002931 [0.01] 0.261442 [0.59]

CPI(-1) -0.834289*** [-3.30] -0.420333 [-0.87]

IR(-1) -0.001218 [-0.13] -0.059177*** [-3.40]

C 0.026980* [1.95] 0.111794*** [4.25]

Note:1. *, **and***denote significance at the 10%,5% and 1% level respectively.