A Study Based On State Space Model Of Real Estate "Sentiment Index"

In recent years, the real estate industry has got so much attention, most scholars use a lot of different ways to evaluate real estate prices and estimate the real estate market bubble.But these traditional prediction methods failed so many times.So it shows us that these kind of methods have huge flaws. Based on behavioral finance theory, we know that market participants'psychological factors and the price both can determine the price movement, and this kind of psychological factors has become an extremely important role on the price trend. So this paper will use the real estate related industries stock data to evaluate real estate participants' sentiment index comprehensively.Here we assume that participants' sentiment index in the real estate market is hidden,it is a kind of latent variable hiding behind the real estate price trends. So we want to use a practical and data reliable way to estimate this potential variable. The core thought is establishing the state space equation,giving parameter initial values, applying Kalman algorithm to get initial potential variable,then bringing them back state space model, applying the EM algorithm to update parameter.That means in each update process we need use EM algorithm and Kalman algorithm, So in this way,we can estimate the parameters of the equations and we can gain latent variables at the same time.