Page 47 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Science
Department of Physics

Housing prices in Hong Kong are worth studying as they are closely related to the well-being of the society. Efforts have
been made previously on analysing the characteristics of the housing market and predicting the housing prices so as to
identify potential bubbles and issue warnings. In this project, analysis of the periodicity of the housing market from 1996
to 2010 and its driving force using cross correlation is conducted. In our previous work, it was shown that there is a 30
month periodicity in the housing price of Hong Kong. The work on the periodicity is continued and the reason for the
periodicity is investigated. There are evidences showing that the interest rates of Hong Kong are closely correlated to
housing prices and therefore interest rates could probably be the driving force of the periodicity of housing prices.

Supervisor: WONG Michael Kwok Yee / PHYS
Student: MA Ka Long / PHYS-IRE Course: UROP1100, Fall

At the very beginning of this project, predicting housing prices and looking for precursors to bursts of bubbles from
existing price data have always been one of the main goals. After housing price data for all 18 districts in Hong Kong
were collected, attempts at predicting housing prices, with different methods and techniques, of some or all districts
were made recently. Results with satisfactory performance (in both interpolation and extrapolation within a short time
interval) were obtained by applying the Gaussian process with a suitable kernel. In this report, some of the major work
and findings, including those by applying the Gaussian process (GP), will be discussed.

Supervisor: WONG Michael Kwok Yee / PHYS
Student: WANG Juntao / PHYS Course: UROP2100, Fall

Housing markets not only have close relations to residence livelihood but could affect the overall economy of a whole
country, so researchers invented several methods to predict the housing price of different countries. Many scientists are
seeking the non-stationary behavior in housing price time series to detect the rapid growth, which is considered as asset
price bubbles. Thus, statistical techniques are used to confirm the existence of explosive behaviors, such as root test,
augmented Dickey Fuller test and PSY method. However, though these methods are plausible in some special case, they
are still far from practical, since they lack predictive power. Thus, in this paper, we would apply Gaussian Process (GP)
with appropriate kernels to predict the changes of Hong Kong housing price.

Supervisor: WONG Michael Kwok Yee / PHYS
Student: YAM Wun Kwan / PHYS-IRE Course: UROP2100, Fall
UROP3100, Spring
UROP4100, Summer

In this report, we attempt to find a model for Hong Kong housing price and try to find the cause of periodicity in the
Hong Kong housing market. In the first part of the report, we analyze the correlation between price and interest rate in
the Hong Kong housing market. Periodicity is found in price change and interest change, and it seems that price is
correlated with changing interest. In the second part of the report, we use information from Fourier components to
construct a model for housing price. The calculated model somewhat matches real values and suggests that the Hong
Kong housing market resembles an oscillating system.

Dynamics of Traffic Congestions

Supervisor: WONG Michael Kwok Yee / PHYS
Student: KWOK Tsz Hong / SSCI Course: UROP1100, Summer

37