Bayesian analysis of Hong Kong's housing price dynamics

DOIhttp://doi.org/10.1111/1468-0106.12232
Date01 August 2017
AuthorMichael Cheng,Ken Wong,Tommy Wu
Published date01 August 2017
bs_bs_banner
SPECIAL ISSUE ARTICLE
Bayesian analysis of Hong Kong's housing price
dynamics
Tommy Wu | Michael Cheng | Ken Wong
Research Department, Hong Kong
Monetary Authority, Hong Kong
Correspondence
Ken Wong, Research Department,
Hong Kong Monetary Authority, 55/F,
Two International Finance Centre, 8
Finance Street, Central, Hong Kong.
Email: kcswong@hkma.gov.hk
Abstract
This paper uses a Bayesian vectorautoregressive model
with sign restrictions to estimate the underlying drivers
of Hong Kong's housing price dynamics in the short
run. While existing studies are useful in analysing
housing valuation, little attention has been paid to the
shortrun dynamics. In contrast, the present paper
identifies shortrun drivers of housing prices using
structural identification with theoretical underpin-
nings. We find that among the shocks that we have
identified, bank lending shock and housing supply
shock were the main factors affecting Hong Kong's
housing prices. Low mortgage rates were another key
factor that led to the significant increase in housing
prices after the global financial crisis.
1|INTRODUCTION
Many economies experienced housing booms in the postglobal financial crisis (GFC) period.
Some prime examples are the real estate markets in London, Vancouver, Sydney and Hong Kong.
The surge in housing prices has attracted the attention of and created concerns among academics,
market participants and policymakers given the overheating risks in the housing market. As
such, many economies, particularly those in Asia, including Singapore and Hong Kong, have
introduced macroprudential measures to safeguard banking stability against risks stemming
from the housing market. As boombust cycles in housing prices would pose risks not only to
financial stability but also to the price stability of an economy, it is important to have a good grasp
of housing price dynamics to facilitate preemptive policymaking.
In this paper, we focus on the drivers of Hong Kong's housing price dynamics. The case of
Hong Kong is of particular interest due to the city's free market economy and full capital account
convertibility. Many studies have analyzed Hong Kong's property market, in light of its salient
Received: 26 January 2017 Accepted: 20 April 2017
DOI: 10.1111/1468-0106.12232
312 © 2017 John Wiley & Sons Australia, Ltd Pac Econ Rev. 2017;22:312331.wileyonlinelibrary.com/journal/paer
features and importance to the economy (e.g. Hui, Cheung, & Pang, 2010; Chang, Chen, &
Leung, 2013; Leung and Tang, 2015a).
1
Nevertheless, most of these studies focus either on the
transmission of financial market fluctuations to the housing market, or on the micro features
using transactionlevel data. The present paper aims to fill the void by focusing on the macro
drivers of housing prices.
It is common to analyze housing price dynamics using the error correction framework
(e.g. Malpezzi, 1999; Capozza, Hendershott, & Mack, 2004; Beenstock & Felsenstein, 2010;
Leung, 2014).
2
In the case of Hong Kong, several studies (e.g. Leung, Chow, & Han, 2008; Craig
and Hua; 2011; Chung, 2012) have estimated error correction models (ECM) with a focus on the
longrun determinants of housing prices, such as demographics, housing supply and other
fundamental variables. While useful in analysing housing valuation, the literature has paid
little attention to revealing the shortrun drivers of housing prices. The present paper adds to
the previous studies by identifying the role of different factors in affecting housing prices in the
short run.
We use a Bayesian vector autoregression (BVAR) model with sign restrictions to study the Hong
Kong housing market. The approach we take is similar to that in Towbin and Weber (2015), and we
modify the model specifications to suit the case of Hong Kong. We find that among the shocks that
we have identified, bank lending shocks and housing supply shocks were the main drivers of the
shortrun housing price dynamics in our sample period between 1996 and 2016. Low mortgage
rates were another key factor leading to the significant increase in housing prices after the GFC.
The rest of the paper is organized as follows. Section 2 discusses Hong Kong's housing market
development since the Asian financial crisis. Section 3 presents the BVAR model and discusses
the sign restrictions for shock identification. Section 4 presents the empirical results. Section 5
concludes and discusses policy implications.
2|HONG KONG'S HOUSING MARKET DEVELOPMENT
After a downturn between the Asian financial crisis in 1998 and the SARS outbreak in 2003,
Hong Kong's housing market has been on a long rally of more than 12 years. Hong Kong's
housing prices have increased nearly fourfold between its trough in 2003 and its peak in 2015,
despite occasional declines (Figure 1).
Several factors are likely to have contributed to the robust growth of housing prices in
Hong Kong. On the demand side, the solid growth in household income over the past decade
and the steady growth in the number of households have continued to drive housing need
(Figures 2 and 3).
1
Leung and Tang (2015a) find that the initial public offerings of Chinese firms in the Hong Kong stock market and Hong
Kong's housing market can improve the prediction of each other, pointing to the role of market sentiment as the driving
force. The regimeswitching models of Chang et al. (2013) indicate that the unexpected shock of US stock returns had the
most significant effect on HK asset returns and GDP. Hui et al. (2010) use a hierarchical Bayesian approach to value Hong
Kong's residential properties with reference to its micro features, and find that their model can outperform other valua-
tion methods that are based on average pricepersquarefeet or expert assessments.
2
Malpezzi (1999) proposes an ECM featuring housing pricetoincome ratio and finds that the model can match the state
level of US data well. Capozza et al. (2004) proposes an ECM featuring some longrun equilibrium housing prices and
find support for their model from US data. Leung (2014) builds a DSGE model that can produce reducedform dynamics
consistent with the ECM proposed by Malpezzi (1999) and Capozza et al. (2004). Beenstock and Felsenstein (2010) intro-
duce the spatial element into ECM of regional housing prices in Israel.
WU ET AL.313
bs_bs_banner

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT