A Study on the Model Combination for Predictive Performance－Takes the House Price of Taiwan as an Example

• Main Authors: HUANG,CHIAO-YIN, 黃巧音
• Other Authors: CHENG,TING-YI
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/mexk47

碩士 === 僑光科技大學 === 財務金融研究所 === 107 === According to the Sinyi Realty house price index, the index rose by 187.2% to the fourth quarter of 2018 (equal to 287.2) based on the house price index in Taiwan in the first quarter of 2001 (equal to 100). In other words, the house price of Taiwan has risen for a long time. Recently, however, the house price has fallen step by step as a result of the Government's housing hoarding tax and the two-tax system. Therefore, for the investors , how to predict future house price, it has become an important issue. Based on this, this paper uses a random walk model and a fundamental model, to make a predictive performance test for the house price in Taiwan. Through empirical evidence, this paper concludes: 1. The accuracy of house price forecasting is as follows: random walking model and fundamental model, both are the same forecast accuracy. 2. On the forecast encompassing of the house price : the random walk model is superior to the fundamental model. 3. The price predictive performance of the random walk model is better than the fundamental model. 4.All variables of ADF unit root test are classified as the first-order integrated non-deterministic variables.5.House prices and gross domestic production, rent,the quantity of building sold, house price ratio, loan burden ratio, price level, as well as the current period of house prices and the lagging period of house prices have a co-integration relationship, and will not produce the spurious regression.