Applications of artificial neural networks in the identification of real estate cycles: Evidence from China

Abstract

Deep understandings of the cyclical changes in real estate market have significant meanings for market participants to make appropriate investment decisions. This paper innovatively applied artificial neural networks to identify real estate cycles in China, and accurately predicted its development phases with a welltrained artificial neural network based on 1993-2008 historical training samples. The results indicate that, China's real estate market has oscillational characteristics and the performance of the artificial neural networks reaches high accuracy. In the context of continuously deepening governmental interventions, the volatility in real estate cycles has become more evident since 2008, when the market reached its peak in 2009, but quickly plunged into recession in 2010, and then approached to its trough in 2011. Therefore, a series of governmental macro-control policies since 2008 have tremendous impacts on the duration and frequency of China's real estate cycles, by adjusting the expansion speed of real estate business. © Springer-Verlag Berlin Heidelberg 2014.