Housing research in different countries also yields contradictory results regarding housing’s inflation hedge capabilities. Mixed and varied empirical results have been demonstrated in Switzerland (Hamelink and Hosli, 1996), Canada (Newell, 1995),
Hong Kong (Ganesan and Chiang, 1998), Singapore (Sing and Low, 2000) and China
(Chu and Sing, 2004). Chen and Sing (2006) examined the inflation-hedging ability of five international housing markets (namely Hong Kong, Tokyo, Singapore, Taipei and
London). Their results show that the inflation-hedging features of housing vary significantly across different markets. This finding also highlights the importance of international evidence on the inflation-hedging effectiveness of housing.
In Malaysia, the inflation-hedging effectiveness of residential property has been largely ignored, although there are some studies in housing portfolio management (Hui, 2010;
Lee and Ting, 2011). In addition, no study has been undertaken to compare the inflation-hedging attributes of different types of residential property. Given the unique features of the Malaysian housing market, a dedicated study in this market is critical.

Inflation-hedging characteristics 65

4. Data and methodology
Data
To assess the inflation-hedging features of Malaysian residential property, the quarterly Malaysian housing indices from the Valuation and Property Services
Department, Malaysia over 1999:Q1-2012:Q1 were obtained. The Malaysian All House
Price Index was used to represent the performance of all houses in Malaysia. Moreover, the Malaysian terrace, semi-detached, detached and high-rise house price indices were also collected to measure the performances of different types of residential property[2].
Consumer price index (CPI) was also obtained to gauge the inflation rates in Malaysia.
All data were downloaded from DataStream.
Following Gatzlaff (1994) and Hoesli et al. (2008), an autoregressive integrated moving average (ARIMA) model was utilised to estimate the expected and unexpected inflation indices[3]. The validity of using the ARIMA model as the proxy of expected inflation was also assessed by regressing the actual inflation rate on the expected inflation. Results show that the beta coefficient is significantly different from zero[4]. Therefore, the ARIMA model was found to be an effective proxy of expected inflation. To eliminate potential scaling effects, all indices were set at 1.00 for the first month of 1990 (Hoesli et al., 2008).
Table I exhibits the summary statistics of the overall Malaysian housing market and four different types of residential property. All types of residential property have
Statistics

Mean

SD

Max.

Min.

Inflation rate
Expected inflation rate
Unexpected inflation rate
All house returns
Terrace house returns
High-rise unit returns
Semi-detached house returns
Detached house returns

0.618
0.625
0.001
1.330
1.334
1.240
1.496
1.411

0.917
0.189
0.904
1.756
1.882
3.644
2.960
3.904

5.057
1.382
4.066
6.424
7.120
15.496
8.148
13.651

2 2.023
2 0.036
2 3.404
2 4.283
2 2.265
2 8.471
2 4.550
2 7.584

Note: The values are in percentages

Table I.
Summary statistics: quarterly returns/changes