Property Prices Spiraling Out of Reach? And Pitfalls of Property Price Index

Here I am back in Singapore. The effect of the property boom is clearly evident from the numerous construction sites and cranes dotting around downtown areas. The news media reports property market news almost every day. The classified ads screams at you - "Buy now before prices go up!", "Sure En Bloc, it's a matter of time!", "Don't miss the last chance for district 10!" - property agents appealing to both the greed and fear in you. Greed to make a pile from the current boom. And fear that your dream home will forever escape you if you do not act now. The latest URA data showed that property prices rose almost 8 per cent in the second quarter, on the back of a 6 per cent increase in the first.

But let's play a simple number game here. Imagine there are 100 units of property at Bishan, and another 100 units at Newton. They are of the same size, and they have the same starting price. Now, suppose that 10 units of Bishan units are transacted, all suffering a 10 per cent loss, while 10 units of Newton units are transacted with 10 per cent gain. What does the property price index show? If the property price index is based on the volume of transactions, it will show no movement. The gain in Newton is cancelled out by the loss in Bishan.

Now let's rewind and start again. There are 10 units transacted in Bishan, with a 10 per cent loss on each unit. But there are now 20 units transacted in Newton, all with 10 per cent rise. Look and behold. Weighted by the transaction volume, the index trends up, because Newton is now given more weight than Bishan. Come another quarter, if there are even more transactions in Newton, the price index will show even more upward increase, never mind the fact that all Newton units are selling only 10 per cent more - exactly the same as the first scenario. Before you know it, every one watching the price index becomes convinced that prices are shooting up and we have a frenzy in the making. And as they transact, their higher purchase price enters into the weighted index, further fuelling the impression that overall prices are rising.

Here, I thank an anonymous reader for pointing out to me URA's methods, which is something of a hybrid. Based on recorded transactions, the median for a location and housing type is picked. These then enter into the overall index using a value weight. Suppose again Bishan and Newton units are all sold at the same price initially. The transacted price for Bishan then falls 10 per cent, and the transacted price for Newton rises 10 per cent. Weighted by the moving value of the transactions, Newton will have a larger influence on the future direction of the index because it now has a higher value than Bishan. Again, there is a feedback effect where areas with rising prices are given an increasing weight in the index.

Of course, it is never easy to construct a summary statistic that can fully capture the overall price trends. This is even more so since housing property is highly heterogeneous (different) in locations, types and so many other attributes. It is often difficult to interpret the index given the thicket of statistical manipulation that is employed to create it.

But since the idea of this index is to reflect in some sense the affordability of housing space in Singapore, perhaps a good way is to use gross floor area as the weights - that is, Bishan and Newton be given the appropriate weights that reflect their total living space - thereby stripping away any components that might be endogenous in nature. Maybe this might give the lay person a much clearer picture just how affordable housing space in Singapore is.