“The role of news in the fluctuations of housing price”

The main purpose of this paper is to evaluate the impact of the news on the housing price volatility in Iran. To do so, symmetric and asymmetric models such as GARCH, T-ARCH, EGARCH and APGARCH are applied by using annual data for the period 1971–2013. The empirical results confirm the asymmetric and leverage effects of news in Iran housing market. Also the impact of shocks indicates that negative news affect the housing price fluctuations further more than positive news with the same size.


INTRODUCTION
Fluctuations on financial time series clustering and random perturbations are accompanied by more substantial fluctuations in the small amplitude fluctuations associated with smaller amplitude fluctuations. Early volatility model requires random disturbance with variance can not explain this phenomenon, Engle (1982) [1] proposed the ARCH model, Bollerslev (1986) [2]to promote the formation of GARCH model. Since the ARCH and GARCH model, the impact of the variance on the different directions symmetrically to react [3], because only the square of the impact conditional variance mapped to the information contained in the result on a change in price of symbols to be lost. (Li and Wang, 2013).
Engle's theory suggests the asymmetric effects of shocks on price fluctuations in financial asset markets and the positive news than negative news lead to higher returns and less volatility with the same size. Black (1976) noted that the negative impact on the positive impact than the same degree of volatility is higher, the first time he uses the term "leverage effect" to describe this phenomenon, referring to the stock price movements and volatility negatively correlated with the same intensity bad news than good news led to greater market volatility (Li et al., 2013).
Housing is one of the most important sectors in Iran's economy. It has a large share in household's expenditures, on the one hand, and has substantial share in GDP, on the other hand. Due to the fact that financial markets are under development in Iran, housing, gold, currency and stock markets attract people's savings. Therefore, the housing demand is affected by another asset demand. So, it is natural that news of future changes in the housing market and the country's macro and micro variables are thereby affecting the housing price. According to World Bank reports, about 40 percent of investment in Iran is devoted to housing market. The ratio of investment to GDP is 8 percent and in average it grows by about 1.12 percent every year. 30 percent of household's expenditures are dedicated to housing. Housing sector relates to 134 fields of occupational activity in Iran.
There is no empirical study assessing the relationship between housing news and the fluctuations of housing price according to Iran economy's data. So the aim of this study is to examine how the shocks of the housing news impact on the prices. The paper proceeds as follows: literature review is presented in section 1, section 2 introduces the theoretical model, section 3 turns to the results and last section concludes.

LITERATURE REVIEW
In recent decades, the evidence for predictability has led to variety of approaches, some of which are theoretically motivated, while others are simply empirical suggestions. The most interesting of these approaches are asymmetric and leverage volatility models, in which good news and bad news have different predictability for future volatility. These models are motivated by the empirical work of Black (1976), Christie (1982), French et al. (1987), Nelson (1990), Schwert (1990), Engle et al. (1993). Asymmetry of news for asset price has been confirmed in many studies. Henry (1995), Engle and NG (1993), Henry (1995), Friedman et al. (2002) find the asymmetric effects of shocks on price fluctuations in financial asset markets and that the negative shocks introduce more volatility than positive shocks. Li et al. (2013, p. 306) study the representative industries, such as industrial, financial, real estate, medical and health industries to learn more about the changes in the characteristics of different industries volatility. The empirical results indicate that the GARCH (1,1) model can explain the fluctuations in theAs mentioned before It has been copied from Li, et al., 2013 so i'm not going to change it TARCH (1,1) and EGARCH (1,1) models examined the impact of fluctuations in the various sectors of the leverage effect and information asymmetry, the results show that the negative impact generates greater volatility than the same amount of positive impact in various industries. Heidari et al. (2012) investigate the relationship between inflation uncertainty and economic growth through GARCH_M model for Iranian economy during the period 1988-2007 by using quarterly data. Their empirical results show that inflation uncertainty does not affect the level of growth rate and positive inflationary shocks have more effect than negative ones. Tsai and Chen (2011), in their paper, reviewed the fluctuations in housing market prices in the two markets of total houses of England and new houses using non-conditional variance models ARCH and GARCH and for seasonal data for the period 1995-2005. The results indicate that the size of the fluctuations over the price is 4.89 times bigger than the fluctuations under the price for the total housing market and 2.78 times bigger than the fluctuations under the price in the new houses market. Miles (2008) using ARCH model and given seasonal prices for the period from 1979 to 2006 for American States, showed that in over half the states (18 states), the effects of clustered fluctuations of housing prices were revealed. Therefore, the application of the GARCH models is useful. In addition, the study has found evidence of asymmetric effects on the U.S. housing market so that the sensitiveness to negative news about the housing price is more than positive news shocks. Andersen et al. (2002) explore the relationship between macroeconomic news and the U.S. dollar exchange rate against six major currencies. They confirm that macroeconomic news generally have a statistically significant correlation with intra-day movements of the U.S. dollar, with "bad" news, for example, data indicating weaker-than-expected growth having larger impact than "good" news.  prove the impact of the strategies of the market participants on the price volatility, which is reflected in the fact that different expectation of the two types of investors on the future prices will cause price volatility. The results show that the change of fundamentalists' expectation on the house price will influence the frequency of the house price volatility, while the change of chartists' expectations, which increases with the acceleration of beliefs in evolution, will influence the range of the volatility.  study the impact of information disclosure on housing market efficiency. The results show that as information disclosure increases, the volatility of housing price reduces significantly. The likelihood of experiencing a residential pricing bubble was reduced by 57.4%, as information disclosure increased. Therefore, information disclosure and price transparency in residential real estate are necessary to stabilize the housing market. As in the literature, the impact of news on fluctuations of housing prices has been less studied. Considering that housing is regarded as an important asset in Iran and because of speculative motives, the present study examines the impact of good and bad news on housing price fluctuations.

The news shock in the model
In traditional econometric models, constant variance of the residual terms has always been the main and classical econometric assumption to account for. Robert Engle laid a new approach called ARCH to escape this restrictive assumption. One of the reasons to use ARCH models are big and small predicting errors in economic clusters in a way that the mentioned series may have reacted differently over the years, in other words, in some years with a lot of fluctuations and in some with a little. Under such circumstances, it is expected that the variance is not fixed over the random time series and is a function of error term's behavior. In fact, the advantage of the ARCH model is that it can explain the conditional variance's trend according to its past information.
Let us suppose that housing returns over the period 1 t − to , t and the information available to investors during the time that they make their decisions is 1 .
t − Ω Expected return and return fluctuations conditioned on available information at time , will be symbolized as follows: Unpredictable return at time , t t ε is the result of reducing real turnover from the expected return: According to Engle and NG, unpredictable return ( ) t ε can be introduced as a measure for news, is introduced as a measure for good news (excess return greater than expected return). According to Engle (1982), ARCH (q) is defined as a function with lag from , t ε which is the predictable fluctuations of a function of past news, meaning: This means that predictable instabilities are a function of previous news. According to Engle (1982), older news have less impact on instability than newer ones. After Engle, Bollerslev introduces the generalized autoregressive conditionally heteroscedastic GARCH (p, q) model as follows: For positivity of conditional variance, it is necessary to have: The process of GARCH (p, q) would be a weak stationary process.
If 11 1, pq ij ij a β = = +< ∑∑ then most shocks are not stationary in the model. GARCH (p, q) models implicitly imply the symmetric effect of the news on fluctuations meaning that good and bad news (of similar size) have symmetric effects on fluctuations. This implicit assumption of simple GARCH models is not aligning with the assumption of this research that asymmetric effect exists on fluctuations of housing prices. According to Nelson, simple GARCH models can't consider such roles, since they suppose that only the value and not the sign of past returns are involved in the future fluctuations.

EGARCH model
To control the asymmetric effect of the news on fluctuations, Nelson (1991) defines the EGARCH model in which the effect of asymmetric news is as follows: ( ) ( ) 22 11 1 11 log .
The left side of 5 is brought in a logarithmic way and guarantees that the variance is a positive condition and there is no need to generate constrains in coefficients. The asymmetric effect is tested by assumption 0. γ < If 0, γ ≠ then the effect of the news is asymmetric. The model presented by Nelson is a special case of the given model below:

T-ARCH model
The GARCH model is the threshold of another asymmetric model, which was introduced by Zakoian (1994) and Glosten et al. (1993) as below: αγ + Also, if 0, γ ≠ the effect of news on fluctuations will be asymmetric, meaning that good and bad news will have the same asymmetric effect on fluctuations. GARCH model is a special case of T-ARCH in which 0. γ =   Table 1 gives an abstract of the implied constraints on APGARCH and production of different models. Like previous models, in case 0, i γ ≠ then asymmetric effects will apply (Abunouri et al., 2009).

RESULTS
The data used in this research are annual and related to period 1971-2013 that has been obtained from Statistical Yearbook of the Central Bank of Iran and building Office of Ministry of Housing. The study is limited to 2013 due to the fact that since then, to control speculative activities, the government has been preparing a housing market policy package and application of the system of real estate transactions registration. The price index of housing has been 43,723, 41,417, 41,760, 39,546 (thousands Rials) respectively, during the period 2014-2017. Therefore, housing market hasn't faced the price fluctuations.

Estimation and hypothesis testing
For using ARCH family models, it is required to use a conditional mean equation for housing prices index. It is necessary to examine the stability of housing price index before time series modeling with

GJR-ARCH
Note: There is no limitation for the coefficient.
Box -Jenkins method. Results based on generalized Dickey-Fuller and Philips-Perron methods are presented in summary in Table 2. As Table 2 shows, the difference of housing price index is stationary.
To determine p, m, q in ARIMA (p, m, q), various models are estimated using OLS method and then according to determination measures of AIC, SBC and R 2 , the best model is recognized. Results show that ARIMA (2, 1, 3) model has the least AIC and SBC statistics and the most R 2 statistics among other models.
The state of DLHP series is analyzed in Table 3 to survey the defining statistical measures.
It can be inferred from Table 3 that DLHP time series have negative skewness or the left and it shows that most of data mass is piled on the right side of the mean and stating that the probability of positive return taking place in housing market is more than the negative. In other words, likelihood of price increases this year and more is than the last year, unlike the occurrence, it's complimentarily, i.e., market information shows market tendency to prices increase.
DLHP kurtosis is 12.51, which is a high kurtosis compared with normal distribution, which is 3. Therefore, compared with normal distribution of housing returns, extreme values of the range are more likely to happen. Bera and Jarque statistic reject the normality assumption of DLHP. Accordingly, series graph's kurtosis is a prove that cluster phenomenon exists and there are skewness coefficient points to leverage effects of positive and negative news in conditional mean model of DLHP.  Table 4 shows the results of ARCH-LM test for time series conditional mean model of DLHP, ARIMA (2, 1, 3). According to the statistics, it is   confirmed that there is conditional heteroscedasticity of 5% level of significance, which makes it necessary to use ARCH models.

Leverage effects of the news
After surveys to determine the best delays for asymmetric models using diagnostic measures AIC and SBC and significance of model components, the best models were GARCH (1, 1), T-ARCH (1, 1, 2), EGARCH (0, 1, 2) and PARCH (2, 1, 0). Estimation results and test statistics of news effects on housing prices' fluctuations is summarized in Table 5 for these models.
According to Table 5, ( ) , αβ in estimated GARCH, EGARCH and APGARCH models is of high significant level and reconfirms cluster effects in the series. The asymmetric coefficient in APGARCH and EGARCH is significant, which implies that shock effects on housing prices resulting from news have been asymmetric in Iran.   Table 5, LM-ARCH test was again taken for error term of all the 4 models, which indicate removal of ARCH effects and confirmation of models' rectitude. The statistics of kurtosis, skewness and Bera and Jarque show good fitness and a normal distribution of error term. This can also be seen in Figure 3, which shows quartiles related to the four models.
In Figure 4, actual and estimated and residual values of the EGARCH model are drawn besides each other for intuitive understanding of the fitness of the model. 1 Integrated GARCH (IGARCH).

Convergence or divergence test of the estimated parameters from non-conditional dissimilar variance models
To analyze the subject that whether the implied EGARCH model to investigate news effects is cumulative in variance or a IGARCH 1 type model and in order to survey the convergence of estimated parameters that show news effects, Wald test is used to the following assumptions:  Table 6 indicate that the assumption of cumulativeness of EGARCH model is rejected.
This shows that variance returns to its mean value in the long term.   Table 7 shows news effects for GARCH (1, 1), EGARCH (0, 1, 2), T-ARCH (1, 1, 2) and APGARCH (2, 1, 0) models for different values of ing from asymmetric T-ARCH, EGARCH and APGARCH models have significant differences in negative and positive news shocks. However, GARCH model is not sensitive to sign and value of the shocks, this issue indicates that GARCH model has symmetric news effects and is not suitable for measuring news effects in housing market in Iran.

CONCLUDING REMARKS
This research's main objective is to describe the theoretical model of how fluctuations in housing prices are affected from the news that spread through market, according to asymmetric and non-conditional heteroscedasticity variance models over the period 1971-2013, because these models have considerable talents to estimate positive and negative shocks that result from the presented news in economic clusters and have also been used in recent years in Iran to evaluate the influence of the news on variables such as exchange rates, stock market indices and stocks. Among the conditioned dissimilar variance models, most important asymmetric and symmetric models, including GARCH, T-ARCH, EGARCH and APGARCH, were used to estimate news effects on housing price fluctuations. The results from this research indicate the existence of asymmetric effect of the news during the studied period and show that the effects by shocks from bad news (negative) are more than good news (positive) on housing price fluctuations. This result is in line with Engle and NG (1993), Henry (1995), Friedman et al. (2002) and Miels (2008). Also the EGARCH model is the best when fitted to the data during the period 1971-2013 to explain news effects on housing price fluctuations in Iran.