Bubbles, misspecification errors and nonstationary estimation: A study of the property market in Hong Kong

Abstract

This paper attempts to develop a modified signal extraction method for detecting the existence of bubbles. Since most of the economic series are non-stationary, the fully modified (FM) procedures are introduced in order to solve the spurious regression problems. To illustrate how our method works, we employ it to study the property markets of Hong Kong. The stock orthogonality test shows that the noise component incorporating misspecification errors and bubbles components is highly significant, while the flow orthogonality test suggests that the misspecification errors might not occur in the estimation. These two findings provide the evidence that the bubbles are likely to exist in the property markets of Hong Kong between 1985:Q1 and 1997:Q3.