Identifying Turning Points of Real Estate Cycles in Taiwan
Chun-Chang, Lee, Chih-Min, Liang, Hsing-Jung, Chou, Journal of Real Estate Portfolio Management
Executive Summary. This paper applies the bivariate Markov-switching autoregressive model (MS-ARX) to identify the turning points of real estate cycles. The model with the best fit is L (1)-MSIH (2)-AR (8), of which, there 8 lags and both the intercepts and variances are subject to the influence of unobservable variables. This model confirms that the Composite Leading Index is ahead of the reference cycles by one quarter. Generally speaking, the Composite Leading Index has been producing rather accurate identifications of the four troughs and three peaks announced by the Taiwan Real Estate Research Center.
The identification of real estate cycles has always been an important issue in the study of real estate. The turning points of business cycles are, in particular, an important reference for the government and private sectors in their economic decisions. However, the Taiwanese government currently determines the dates of peaks and troughs of real estate cycles from an ex-post perspective. This paper constructs its own historical series of reference cycles in order to identify the turning points of real estate cycles and hence the expansion and contraction periods in the real estate cycles of Taiwan.
In order to grasp the turning points of real estate cycles, both government agencies and academic institutions compile various indexes, such as the Real Estate Indicator, the Countermeasure Signal and Consumer (Manufacturing Business) Confidence Index, hoping to gain insight into the changes of real estate cycles in the future. Among all the current indicators, the Composite Leading Index has drawn the most attention. The Composite Leading Index is compiled by the Architecture and Building Research Institute, Ministry of the Interior and Taiwan Real Estate Research Center, National Chengchi University, by incorporating investment, production, transactions, and utilization in the real estate market. Currently, the Composite Leading Index consists of GDP, M2, and the Construction Index on the stock market, Changes in the Outstanding Balance of Loans to Construction and Consumer Price Index.
Considerable research has engaged in the relevance analysis regarding whether leading indicators help to determine the changes of business cycles. Diebold and Rudebusch (1991) and Estrella and Mishkin (1998) used linear vector autoregression, (VAR) models and probit models, respectively, and found that leading indicators are not helpful in forecasting out-of-sample performance when it comes to the turning points of business cycles. However, Filardo (1994), Hamilton and PerezQuiros (1996), and Camacho and Perez-Quiros (2002) suggested that if the leading indicators are applied to MS models, they are beneficial to the determination of the turning points in business cycles. According to the empirical study by Camacho and Perez-Quiros (2002), the combined forecasts formed by MS models and non-parametric models can derive the best out-of-sample forecasting performance. Birchenhall, Jessen, Osborn, and Simpson (1999) utilized the logistic classification method, and argued that leading indicators are able to accurately determine the turning points of business cycles. Therefore, leading indicators are highly relevant to the determination of changes in future business cycles and model set-ups. In terms of the studies on the turning points of real estate cycles, Lahiri and Wang (1994) used the two-state MS model to estimate how leading indicators in commerce are used to forecast the turning points of business cycles. They found that the predictability of leading indicators is rather good. Krystalogianni, Matysiak, and Tsolacos (2004) used leading indicators as the dependent variable in the Probit model to predict the probabilities of declines or increases in the U.K. capital value. The findings showed that leading indicators are a useful decision tool in real estate investments. These studies suggest that leading indicators have good predictability in pointing out the turning points of business cycles. …