Date issued : 09 Dec 2014
The Resale Price Index (RPI) provides the general price trend of resale HDB flats. The index computation methodology was last revised in 2002. Since then, the variety of resale flats, e.g. design, age, location, which has been transacted in the resale market has increased. It is therefore timely to update the computation methodology of the RPI. From the 4th quarter 2014, HDB will update the computation methodology of the RPI by:
- i) Adopting the Stratified Hedonic Regression Method;
ii) Switching to 5-quarter fixed weights; and
iii) Adopting a new base period of 1Q2009.
New Stratified Hedonic Regression Method
2Currently, the RPI is computed using the stratification method, with a representative basket of towns and flat models. Resale prices are stratified into segments based on flat types, models and regions. The average prices for each segment are then aggregated using 12-quarter moving average weights to derive the index.
3With effect from the 4th quarter 2014, HDB will adopt the stratified hedonic regression method to compute the RPI. This method will control for variations in flat attributes, such as proximity to amenities, age and floor level, through a hedonic regression, to derive the general price movements in each segment. These are then aggregated using 5-quarter capital value fixed weights to derive the aggregate price change. To better reflect prevailing market structure, the weights will be updated once every three years. Please see Annex 
(PDF 218KB) for the graphical illustration.
(PDF 218KB) for the graphical illustration.
Updated Base Period
4Along with these changes, the base period will be updated to 1Q2009, from 4Q1998. This means that the RPI for 1Q2009 will be at 100. The current RPI series from 1Q1990 to 3Q2014 will be re-scaled to the new base period of 1Q2009. This adjustment will only impact the absolute levels of the index (see Table 1 and Chart 1), and the quarterly percentage changes will remain unchanged.
Table 1: Comparison of Current and Rebased RPIs
Quarter
|
Current RPI
(4Q1998=100) |
Current RPI
(Rebased to 1Q2009=100) |
Previous base 4Q1998
|
100
|
72.3
|
New base 1Q2009
|
138.3
|
100
|
2Q2009
|
140.2
|
101.4
|
3Q2009
|
145.2
|
105.0
|
4Q2009
|
150.8
|
109.0
|
1Q2010
|
155.0
|
112.1
|
2Q2010
|
161.3
|
116.6
|
3Q2010
|
167.8
|
121.3
|
4Q2010
|
172.0
|
124.4
|
1Q2011
|
174.8
|
126.4
|
2Q2011
|
180.3
|
130.4
|
3Q2011
|
187.2
|
135.4
|
4Q2011
|
190.4
|
137.7
|
1Q2012
|
191.6
|
138.5
|
2Q2012
|
194.0
|
140.3
|
3Q2012
|
197.9
|
143.1
|
4Q2012
|
202.9
|
146.7
|
1Q2013
|
205.5
|
148.6
|
2Q2013
|
206.6
|
149.4
|
3Q2013
|
204.8
|
148.1
|
4Q2013
|
201.7
|
145.8
|
1Q2014
|
198.5
|
143.5
|
2Q2014
|
195.7
|
141.5
|
3Q2014
|
192.4
|
139.1
|
Note: The re-scaling uses a factor of 100 (new index in 1Q2009) / 138.3 (original index in 1Q2009) multiplied on the original index level to derive the re-based index level for the respective quarters. Indices from 1Q1990 to 4Q2008 will be similarly re-scaled using the same factor. Due to rounding, there could be some differences in the quarterly price change compared to the RPI series before re-scaling.
Chart 1: Rebased RPI (1Q2009=100)
(JPG 1152KB)
5Back-testing of the index using the new method on 2014 data shows that there is no change in trend, and the quarterly changes follow closely to those computed using the current method (Table 2).
Table 2: Back-testing of RPI Based on Current Method and Revised Method
Time Period
|
Current RPI
|
Revised RPI
|
1Q2014
|
-1.6%
|
-1.9%
|
2Q2014
|
-1.4%
|
-1.5%
|
3Q2014
|
-1.7%
|
-1.8%
|
6The stratified hedonic regression methodology will be adopted from the next index release, i.e. the release of the flash estimate of 4Q2014 RPI in Jan 2015.
- wong chee tat :)
