IN a push to make property indices more reflective of market changes, the Housing & Development Board (HDB) - the first among official agencies here to change the way it computes its price index - has unveiled details on how its revised approach will be more comprehensive and robust. (see infographic)
HDB said on Tuesday that it was expanding the coverage of its resale price index (RPI) to all towns and flat models, having excluded 12 towns in the past. It is also adopting a new method to compute the RPI called the "stratified hedonic regression" method, which it says is better able to screen out "noises" in measuring price changes.
Other government agencies that track property prices, namely the Urban Redevelopment Authority (URA) and JTC, could soon be making changes to their respective property price indices too.
A JTC spokeswoman told BT that it is working on a revised method to compute its industrial price and rental indices to "better reflect market trends" and will reveal more details next month. URA disclosed last week that it was reviewing the need to revise its private residential property price index (PPI). Some market watchers believe that URA is also reviewing price indices for commercial properties.
HDB's move to change its computation of the quarterly RPI is itself a response to significant market changes that have taken place since the last revision in 2002. In recent years, a wider range of flats have been transacted, more resales took place in newer towns and there is now greater age variance across resale flats, Minister for National Development Khaw Boon Wan has flagged.
Starting from the fourth quarter, HDB will include all towns and flat models in the RPI. It had previously excluded newer towns such as Sengkang, Punggol and Sembawang.
Market watchers were, however, surprised yesterday that even popular towns such as Bishan, Queenstown, Marine Parade and Clementi were previously excluded in the RPI.
"Obviously, including all the towns will provide a fairer and better representation of HDB resale market," said PropNex chief executive Mohamed Ismail.
Mature estates such as Bishan, Queenstown and Marine Parade tend to have greater demand, fetching the highest cash-over-valuations in the past, he pointed out. A few HDB executive maisonettes were transacted at a record S$1 million in Bishan and Queenstown.
According to HDB, the 12 excluded towns made up less than 30 per cent of all resale transactions in recent years; market watchers felt however that this is significant especially when market volumes come down.
But a back-testing of the new approach by HDB found no deviation in the overall price trend compared to the previous method, though resale prices would have fallen more over the past three quarters based on the revised approach. In the third quarter, the revised RPI would have fallen 1.8 per cent, compared to 1.7 per cent using the older approach.
Property consultants believe that this is due to newer towns being included, as they typically face weaker demand. Now, even the popular towns such as Bishan where flats are more pricey have been hit by the lending curb on home buyers, Mr Ismail said.
To compute the index, HDB is switching to "stratified hedonic regression", which it says allows for quality control at the broader level compared to the current "stratification" method. As a general principle, housing units are heterogeneous; to measure actual price changes over time, their quality differences have to be controlled to make fair "apple- to-apple" comparisons.
Previously, HDB sorted resale transactions by flat types, models and regions and derived the index by aggregating the average prices of the segments, each weighted based on a 12-quarter moving average of transaction volumes. But this means that within each segment, effects on price changes due to differences in flat attributes, such as age and floor level, were not stripped out.
The new method sorts resale transactions by flat types (three-room, four-room etc) and computes their average price changes over time using regression analysis. These price changes are then aggregated using fixed weights (based on five quarters of transacted values) to get the overall index. This hedonic regression is able to control for the differences in housing attributes such as location, proximity to facilities or amenities, age or floor level. Even factors such as distance to MRT stations and primary schools are accounted for.
To avoid user confusion, past values of the resale price index will not be re-calculated using the new method, HDB said. But the RPI's base period will be changed from Q4 1998 to Q1 2009, the index's most recent trough.
ERA Realty key executive officer Eugene Lim noted that for the academics and analysts, this revised index will be a more accurate representation of the market, "just like water has been made more pure".
"But to the layman, it does not really make a difference," he said. "They will look at the index for the general market direction. For pricing indication, most of them will be looking at the recent transaction information that HDB publishes on its website, which is more relevant when it comes to pricing a flat to be sold or making an offer on a flat to be purchased."
Century 21 Singapore CEO Ku Swee Yong felt, however, that home buyers in towns such as Bishan and Marine Parade might have felt that they were making decisions on data that was not complete. "We need at least next five quarters to see if the revised index is good and reflecting market situation," he said.
Some property consultants have hoped to see sub-indices for mature and non-mature HDB estates, similar to how URA slices sub-indices for different regions.
HDB explained that for simplicity, it publishes one composite price index. "There are individual flat transactions publicised daily in the HDB InfoWEB to give a good sense of the current prices," it said.
A team from HDB have worked on the index review over the past one year with a consultant from NUS Department of Real Estate, associate professor Lum Sau Kim. She had earlier led the launch of the NUS Singapore Residential Price Index that also uses hedonic regression.
Globally, many private and official agencies have started reviewing their property price indices amid efforts by Eurostat, the statistical office of the European Union, to harmonise official house price indices across jurisdictions.
The hedonic regression method has increasingly been used to compute national level indices, with France, Finland and Japan among countries that have adopted this method. Official statistical agencies in Australia and the UK are also developing such indices.