What Really Drives HDB Resale Flat Prices in Singapore?
Even while cryptocurrency fortunes crumble at this instant, for many Singaporeans, their net worth remains safe (see here), as it is largely in the properties they stay in. And this remains true almost everywhere in the world. Although we devote countless pages to discussing how investing in stocks can help build your nest egg, the biggest asset market, and the single biggest source of wealth in the world, is property. In Singapore, the vast majority of residential property is the ubiquitous Housing and Development Board (HDB) flat. There is a large and fairly liquid market for resale HDB transactions, and the prices of these resale HDB flats indirectly set the prices of new Built-to-Order (BTO) flats sold by the HDB (see here). So what really drives HDB resale flat prices in Singapore?
And this is a question of some concern. While the HDB Resale Price Index is hitting all time highs, and hence raising the net worths of home owners, this is also helping to raise the prices of BTO flats sold to first-time home buyers. And what may be good for the parent may not be ideal for the child as the BTO flats may be just out of reach for these first-time home buyers. Admittedly, this is a problem in every major city, but the difference is that Singapore does not have sprawling suburbs or outlying towns, where first-time home buyers can find something cheaper within their budgets.
So let’s see if we can figure out what really drives HDB resale flat prices!
What Really Drives HDB Resale Flat Prices in Singapore?
What Factors May Drive House Prices?
Because a house or an apartment can be both a consumption and an investment good, there can be a number of factors which affect its price. Let’s run through some of these factors and why they can drive HDB flat prices.
A. The number of new flats built each year
In land scarce and labour constrained Singapore, and with a growing population on top of it, the supply of new HDB flats is likely to be a key driver of prices. This is especially because the demand for housing is relatively inelastic – people need a place to stay. And also because new HDB flats pricing are at a discount to resale HDB flats. As a result, if there are fewer new HDB flats built each year, the residual demand for housing will need to be satisfied through the resale market instead.
A second aspect of this is the number of new BTO HDB flats reaching the 5 year minimum occupancy period (MOP), after which the original owners can sell them. This represents additional supply to the resale HDB market, and may affect the prices as well.
B. Population growth and new family formation
Part of the reason why the demand for housing is so inelastic is because Singapore also has a growing population, due to inward migration. Hence the increase in the population can be a significant driver of HDB resale prices, especially when some HDB flats can be rented out as well. This means that higher HDB resale prices may be required to persuade HDB landlords to sell off their profitable investments!
Another driver of HDB resale prices may be the number of marriages. After all, new HDB flats are only available to family units, and the number of marriages is an indicator of the new demand for such flats, which in turn may spill over to the resale HDB market.
C. Private non-landed residential prices
To a certain extent, resale HDB units are imperfect substitutes for private condominiums. So, it is not unexpected that the prices in these two property markets may be linked. What is not clear, is whether they move in tandem because of a third factor driving both sets of prices, or if one “causes” the other, or vice versa, i.e. whether HDB resale prices affect private condominium prices, or vice versa.
D. GDP growth
Rising affluence may be yet another driver of resale HDB prices. In a fast growing country like Singapore, the incomes and purchasing power of the people may push resale HDB prices up as they compete for a limited pool of flats.
So there we have it, a list of possible factors which can drive HDB resale flat prices. And let’s go one step further to test these using historical data to see whether any of these factors really matter!
What are the drivers of HDB resale flat prices?
We start off with the following factors to test against the historically observed percentage changes in the HDB resale price index:
- Number of newly built HDB flats
- Number of BTO HDB flats reaching the 5 year MOP
- Increase in resident population (non-residents are barred from buying HDB flats)
- New marriages involving at least one resident
- Percentage changes in private non-landed residential prices
- Increase in GDP at current prices
A few assumptions when we look at this data:
- We use percentage changes when looking at the HDB resale price index, the private non-landed residential price index, and GDP growth, because these values can grow without an upper bound. That is, the dollar amount can become very large, and so we use percentages to normalise these factors
- We use the absolute change (in numbers) when looking at the number of new and MOP’ed HDB flats, population, and marriages, because in land-scarce Singapore, these factors cannot grow without any constraint, and given the finite number of HDB flats available, it is the absolute numbers which make a difference
- Even though the HDB resale market is fairly liquid, property transactions take time, both to decide, and to execute. As a result, we will look at the two year changes instead of annual changes. For example, we will consider the change in property prices, population, GDP etc. over a 24 month period e.g. Jan 2000-Dec 2001
- Finally, to ensure we have captured causality correctly, we will use past data to try to forecast future outcomes. For example, when looking at the effect of newly built HDB flats, we will consider the effect of new flats built in Jan 1998-Dec 1999 on the HDB resale prices in Jan 2000-Dec 2001
1. & 2. Number of new HDB flats built
In theory, when there are not enough new (and cheaper) flats built, the excess housing demand will have to be channeled to the resale market, which in turn pushes up the prices in the resale market. This is what it has looked like historically:
It’s obvious that the building of new HDB flats in Singapore goes through cycles – more in the late 1990s and mid 2010s, and much less in the other periods. And it is also obvious to see that HDB resale prices have risen when the building of new flats slowed, and flattened when the building picked up again. Of course, there is also a lag between when the new BTO flats are announced or launched (e.g. in late 2011), and when they were actually built (2014 to 2016). The correlation between the future 2 year percentage changes in the HDB resale price index and the past 2 year number of new flats built is -39.5%.
Now, we can also move the blue line ahead by 5 years to look at how the new BTO flats which finish their MOP affect the HDB resale prices. In this instance, the effect is smaller, and the correlation between the future 2 year percentage changes in the HDB resale price index and the past 2 year number of BTO flats reaching MOP is only -9.1%. So, it appears that the MOP effect of BTO flats hitting the resale market is small at best.
3. Population growth
The chart of the HDB Resale Price Index and the growth of the resident population looks like this:
While the resident population in Singapore grew strongly in the 1990s and 200s, there is little discernible trend between population growth and the change in the HDB resale prices. The correlation between the future 2 year percentage changes in the HDB resale price index and the past 2 year increase in the resident population is only 15.7%.
4. Marriages and New Family Formation
What about marriages and new family formation? Since most new families will be eligible for BTO HDB flats, if the number of marriages (with at least one resident) is higher, the demand for flats to set up a family will spill over to the HDB resale market, pushing up prices. This is how the historical chart of this looks like:
Here, we see that the number of new families being formed annually is relative constant at just below 50,000 every 2 years, compared to the gyrations of the HDB resale prices, and this lack of relationship shows up in the correlation between the future 2 year percentage changes in the HDB resale price index and the past 2 year increase in marriages of 0.8%.
5. Private Non-Landed Residential Prices
The Singapore property market is split into two related parts. The HDB market (which is 80% of the total units), and the private condominium market (which is around 20% of the total units). As a result, the private property market is far smaller than the HDB market, and at first glance, is unlikely to be the key driver of the HDB market prices, or at the most, both could be jointly under the influence of a third factor. But the historical chart of these two property price indices looks like this:
The correlation of the future 2 year percentage changes in the HDB resale price index and the past 2 year percentage changes in the private property price index is a very high 54.1%. To establish the direction of causality, we also ran the correlation of the future 2 year percentage changes in the private property price index and the past 2 year percentage changes in the HDB resale price index, which came out to only 8.2%. This means that private property prices lead or “cause” HDB resale prices, and not the other way around. This outcome is surprising, because it means that in Singapore, at least, the tail wags the dog, that is, the smaller private property market sets the benchmark for the much larger HDB market to follow!
Perhaps it is the psyche of the typical Singaporean home buyer which drives this, with owning private property being an aspiration or even obsession. And those who eventually must get priced out of the smaller private property market then purchase the most premium units in HDB resale market, and hence driving the prices there.
6. GDP Growth
Rising affluence can also explain the continual rise in the HDB resale prices. Here is what the history chart looks like:
GDP growth has been a pretty higgedly-piggedly process over the past 3 decades, and it is not surprising that a clear link between GDP growth and HDB resale prices cannot be discerned easily. But the correlation between the future 2 year percentage changes in the HDB resale price index and the past 2 year percentage changes in the GDP at current prices still is a very high 44.2%, so there may be more than meets the eye!
Putting the Key Drivers of HDB Resale Prices Together
Apart from the charts, the correlations between the future 2 year percentage changes in the HDB resale price index and the past 2 year changes in the various factors tell us which factors may the promising drivers of HDB resale prices.
|New HDB flats
|MOP HDB flats
|Private Property Prices
To see how these factors may work together, we can combine them in a linear regression. Right from the start, we drop the MOP HDB flats and the Marriages factors, since they do not look promising. Running through the linear regression, it also appears that Population Growth and GDP Growth also are not significant drivers of future changes in the HDB Resale Price Index. Which leaves us just two factors which can be the main drivers of HDB resale prices:
|Private Property Prices
|New HDB flats Built
A couple of things to note here:
- The R-squared of the regression is fairly weak at between 37% to 45%, depending on the length of the data used. But this is not surprising, as no contemporaneous factors are used, and instead we only use lagged factors
- Clearly, a better fit can be obtained if we had put in contemporaneous factors, like the property price index, which will have an even higher correlation of 77%. But this will create two problems – firstly of establishing causality, and secondly or forecasting the private property index itself!
- The contribution of the two drivers is also unstable, with the weights going down to 50-50 if a shorter period of data is used (i.e. after the Asian Financial Crisis of 1998)
So, overall, not a very satisfying statistical investigation of the key drivers of HDB resale price changes, although we can be fairly certain that the two factors identified are indeed the main ones which drive the market.
How well can we track HDB resale price changes?
But just as the proof of the pudding is in the eating, here, the proof of whether this investigation has any useful insights is in the predictions made, or forecasts. How well can our simplistic regression model above forecast future changes in the HDB resale prices?
Let’s start with the more difficult problem of forecasting the percentage changes in the HDB Resale Price Index. Here is what the Actual vs Forecasted percentage changes over time looks like:
Not too close, and it looks like the predicted percentage changes always happens a bit too early compared to the actual ones, except for the last few years. But note also that the peaks and valleys are pretty well matched, except for the period before Asian Financial Crisis back in 1994-1996. But things start to look better when we use these forecast percentage changes to map out the HDB resale price index:
Here, the model tends to over-predict how high HDB resale prices will go, with in 1997 and in 2013, although it captures the current peak in 2022 quite well. But the overall stagnation in HDB resale prices in 2000-2006 is captured quite well too, and so is the rise in prices from 2006-2012. Net, there is still some use in the model consisting of two drivers of HDB resale flat prices, the Number of New HDB flats built, and the percentage changes in the private property prices.
The inexorable rise in both BTO and resale HDB flat prices is a source of joy for owners and worry for first time home buyers in Singapore. Our limited statistical investigation shows that two key drivers of this rise in prices are the declining number of new HDB flats built, and the rise in the private property prices.
The effect of private property prices on the HDB prices justify, to some extent, the Government’s efforts in cooling the private property market, as this is the tail that turn wags the larger dog of the HDB resale market. But the in effectiveness of the calling measures means that all this has been for nought.
It is probably the inconsistency of the other factor, i.e. the need for the HDB to build new HDB flats at a constant rate, at around 25,000 per year, which has led to spiralling prices too. Given the relatively constant rate of new family formation through marriage and demand for new HDB flats, the see-saw pattern of new HDB flats built only serves to keep driving HDB resale prices up, especially in periods following years of low supply of BTO flats.