Residential Market Predictions 2021: Are property prices rising faster than GDP?

The performance of the private residential market in 2020 appears to be nothing short of a miracle. Demand and prices behaved in a manner that was diametrically opposed to the performance of the economy, leaving seasoned market observers bewildered even after the year had ended. In this blog, we shall seek an explanation as to why the market performed as it did and, using those explanations, build a plausible logic trail as to how the market may meander through 2021. The importance of understanding why 2020 (and in fact for previous years as well) had turned out so counterintuitively cannot be overstated because a failure to comprehend the market dynamics would result in blind market forecasts backed by specious reasons.

In our Q1/2020, Q2/2020 and Q3/2020 residential briefings, we laid down some foundational blocks that we will revisit to build up our understanding of the market behaviour in 2020. Let us start by looking at what we wrote in Q1/2020. In that briefing note, we mentioned that the slew of forces building up was increasing in numbers and it would be difficult to overcome our pre-pandemic forecast of a 3% YoY price increase for the market as a whole (please refer to Graph 5 in our Q1/2020 Residential Sales briefing). That said, prices fell QoQ in Q1/2020 until the middle of June 2020, as the URA price index flash estimate recorded a decline of 1.1% QoQ for Q2/2020. However, when the final numbers were released, prices for Q2/2020 did a stunning turnaround from -1.1% to a 0.3% increase. Thereafter, prices not only continued rising but had also accelerated. In Q4/2020, the flash estimates of the URA Private Property Price Index (PPI) showed it rising 2.1% QoQ, up from the 0.8% QoQ measure for the previous quarter. Demand was also strong with developers’ sales volume exceeding 2019 numbers. In our Q1/2020 briefing, we had also opined that revenge buying may rear its head once the pandemic (here) winds down.

Most would intuitively attribute all this to the high levels of liquidity in the system as well as a low interest rates environment. But is that really the case or, to rephrase, is that directly influencing demand and prices? Table 1 shows that these two and other intuitive variables are not significant in influencing developers’ new sales for the period 2004 to 2020.

The implications of the findings shown in Table 1 cannot be overemphasized because it throws a spanner in the works of commonly held beliefs that economic growth, interest rates and those variables that we often encounter, though sounding intuitively logical, are not the raison d'être explaining new sale demand! Using a univariable correlation, which matched over 90% of actual sales of uncompleted developments to developers launches, we worked out a simple forecast. For 2021, given fewer large project launches, sales would come in at about 6,800 units, down 30.8% YoY. (Please refer to Graph 1).

We also found that getting the rate of private residential price changes to synch up with GDP growth would then be a tall order because, although the correlation is significant, the fit had been ‘loose’. In short, to get the URA PPI to move in close tandem with GDP is akin to pushing on a string. Graphs 2 and 3 show that, while one can model the directional movement of URA PPI, the modelled URA PPI, being a function of GDP change, tends to stray about the actual PPI. This makes it very difficult to control because there could be a multitude of other factors that are causing a loss in trackability.

Therefore, to manage the URA PPI, one must find other factors that are significant in nudging the index to close up on GDP growth at each time frame. These are factors that either force the URA PPI to decelerate when it overspeeds or accelerate when it lags GDP growth. As it stands now, the behavior of the change of URA PPI to GDP is already quite good. Graph 2 illustrates that, after a slate of measures, private residential prices track real GDP changes reasonably well, much like a dog being held on a leash when its owner takes it for a walk. This straying can be described as the tracking error* and, for the period 2000 to 2020, it was 7.34% and 5.32% from 2013 to 2020. Graph 3 shows that in nominal GDP terms, the tracking error is even lower, at 6.92% for the period 2000 to 2020 and 6.43% for the years 2013 to 2020 (the nominal GDP for 2020 was estimated).

*: Tracking error = Standard deviation <URA PPI Change – GDP Change>

To expect zero or much lower tracking errors is perhaps asking too much because the two are measuring different things. The URA PPI measures asset values whilst GDP changes is an output (income) measure. However, it is not that one cannot further reduce tracking error, it is just that to match the performance of the two requires a lot of resources and constant tweaking/addition of measures. For private residential properties, one must un-earth other factors that influenced prices to deviate from the underlying economic growth. For instance, over the course of the decades, we believe that demand has moved from economic fundamentals to market psychology. Aside from the psychological ones, other factors that we believe have been significant in moving private prices are:

• HDB Resale prices;
• Land bid prices one year ago which is the major component of developers breakeven cost;
• Number of sites sold under the GLS program and in the Collective Sales market a year ago i.e. land supply cycle;
• Alternative investment channels;
• Immigration policy towards skilled workers and high net worth individuals;
• Policy intervention;
• Technology e.g. the use of webinars to raise awareness of property launches

The number of factors we believe can influence private residential prices are thus plentiful. Looking at things from the financial industry, the authorities are acting like fund managers, but whose role is reversed in that they try to reduce tracking error by minimizing the outperformance of residential prices versus GDP.

However, though policy measures have reduced the magnitude of deviations, we believe that there are limits on the use of policy interventions as a tool to manage prices. This is because hard coded rules often do not self-correct when markets reverse course. To look for better ways to get the URA PPI to synch with GDP, we must look for analogies and one would be treating price or demand as an object moving through a viscous medium. Policy measures will be the resistive force acting on the object. The faster the object moves, the greater the resistance. Draconian measures can lead to over or undershooting events and distort the market. For example, from 2012 to 2017, property prices undershot GDP growth due to the slate of cooling measures that were sequentially added which took effect during those years. The disadvantages of hard coded measures are:

1. Adaptive expectations can rear end policies. The more measures are enacted, the more the market will try to pre-empt when the next set of measures will come. This plays into the hands of those who try to drum up market sentiments using the fear of missing out mantra.

2. Policy makers will be on edge each time a launch performs unexpectedly well or poorly. This applies also to players and buyers in the private residential market who, when they see a outstanding performance from a launch, may jump the gun trying to pre-empt potential cooling measures and vice versa.

3. Raising the hurdle to join the club of private home-owners can have the reverse psychology effect. They make people pine for the product. For example, when someone buys a more expensive house, they are likely to see that experience as having a better outcome. Their thinking is: if I spend more money on this item, I’m less likely to have regret it. (Source: Here’s the psychology of why we like expensive things – Dec 11, 2018, Business Insider)

Some ways to reduce tracking-error;

There’s no need to resort to sophisticated analysis because lenders can move in steps of five percentage points and the authorities can then assess how the market reacts in the next two quarters. To prevent the market from second-guessing LTVs, the rations would not be published.

2. Valuation

An IF-THEN condition could be built into mortgage valuations. To illustrate:

IF

URA PPI change for previous quarter(s) > GDP growth

THEN

Mortgage Value = GDP growth

For now, there should not be an ELSE condition because if residential prices underperform GDP, such a condition will limit them to catch up with GDP growth. However, that option remains open because a situation may arise whereby an ELSE condition is required.

3. Changing societal behaviour towards residential real estate

Each time the world finds a new growth industry, Singapore reacts by trying to attract such companies to come here to set up shop. This means that the constant refreshing of our economic landscape in a world where the rewards for getting it right in the new industries are becoming ever larger, it would be a constant battle to rein in excessive house price inflation. If that policy remains in place, then what is left is to amend societal behaviour towards residential properties and/or offering alternative avenues for investing with decent ‘safe’ returns. Unfortunately, this is difficult to implement because the societal positive beliefs towards property is so ingrained across generations and it may also upset the social-economic ecosystem developed over the decades.

Benchmarking the URA PPI to GDP

Aligning house price changes with GDP has some attractive features. The pros are:

• It prevents a further deterioration (or improvement) of a modified Gini coefficient for asset prices.
• It reduces (or increases) the likelihood of overconsumption of residential real estate vis-à-vis the economy.
• It matches changes in house prices with productivity changes.
• Indirect means to marry household debt levels with economic fundamentals.

Though it is intuitively an attractive proposition, there are also downsides to this measure of benchmarking. Some of them are highlighted below.

1. The 376,040 units of completed private residential properties in Q3/2020 represent about 27% of our total housing stock. On a S$psf basis, private residential properties fall to the right of the median price of our total residential stock. GDP growth is the average of all the sectors in the economy. To pair GDP (average growth) with private housing (upper 27th percentile) is not ideal.

2. Different sectors of the economy are expanding or contracting at different speeds. Employees in sunset industries are likely to earn less than those on the opposite end of the spectrum. In current times, companies in new growth sectors have astronomical valuations and so are the payoffs to the founders and staff. Income wise, the pressure for the Gini coefficient to increase is higher than before. By attracting companies in the hot growth sectors e.g. technology, e-gaming, socio media, hedge funds, private equity etc, to set up here, the earning power of employees in these sectors will pull away from the rest. This right tail would probably be easily mappable to private housing because the latter is in the upper 27th percentile our housing price spectrum. To constrain private residential properties to average GDP growth would distort the housing market by placing undue weightage on the underperforming sectors.

3. For foreigners, Singapore has a policy of importing skilled workers and welcoming high net worth individuals. These high-income permanent residents are either working in the better performing industries or for entrepreneurs and have generated their wealth in their home countries. For instance, if someone made their fortune from a global e-gaming hit and subsequently bought a Singapore home, their contribution to our GDP is minimal.

4. For an economy where 90% of Singaporeans own their home, and with structural unemployment coming earlier in one’s economic life, having house prices move counter cyclically to GDP may not be that bad. When the overall economy is not performing well, if home prices fall, it aggravates the pain of those who need to sell their properties to alleviate their financial problems. It will not be a bad thing for those who need to raise capital and are able to do so by selling their homes at a good price during such dire times but who otherwise will be doubly hit by both financial problems brought on by a poor economy and getting a less than optimal price for the house. What we mean is this: if GDP is growing normally about the long-term trend line, then property prices can grow in synch with GDP. However, if GDP contracts, then the limit of property price changes is zero (i.e remains flat). This would lessen the financial burden for those in need. (Please refer to Graph 4.)

Long-term behaviour of private property prices

After years of piling on cooling measures to the residential market, private residential prices have on average been tracking GDP changes well. Whilst hard coded measures have been effective in taming over exuberant prices, they are also like sledgehammers. Each time a measure is enacted or ‘fine-tuned’, sales activity drops precipitously. Prices flatline for a period and thereafter, it rises again. The analogy is like a patient suffering a relapse after getting a clean bill of health post treatment. There are several reasons why private property prices tend to inflate faster than economic growth and, lately, have for the first time been counter cyclical!

Firstly, there are ingrained expectations that residential property prices will only go up because land is ‘scarce’ in Singapore. This has been built into the psyche of multi-generations of Singaporeans and even foreigners who are familiar with our real estate market. Such social behavior has great inertia and not easy to change course. Secondly, Singapore has been successful in attracting new growth industries and, in turn, companies in such industries pay salaries that are in line with global benchmarks. Private residential properties, being a high-end asset class, map right into these high-income brackets.

The third reason is that liquid assets amongst Singapore households have been growing at a CAGR of over 3.8% pa (2013 to 2020), whereas nominal GDP has been rising at about 2.6% pa. (Please refer to Graph 5.) The store of equity has been gaining ground faster than GDP and if this trend continues, it just means that additional measures will have to be enacted in the future to prevent property demand, and thus prices, from beating the economy. Unfortunately, that will not be good because more measures lead to market distortions and attract even greater demand.

The fourth reason pertains to demographics and this source of demand runs counter to future economic growth. Graph 6 shows that the number of Singapore residents, comprising citizens and PRs, aged 50 to 69 living in private residential properties has grown 2.5x from 87,503 in 2000 to 221,274 in 2020.

The characteristics of these age groups are:

• Likely to have either fully or substantially paid off most of their mortgages;
• Finding that their homes are becoming too large to maintain due to their age or having that empty nest feeling when their children have moved out;
• Need to free up capital for consumption or medical expenses.

For this age group, there is a likelihood that they will downgrade to smaller homes, most likely to be public housing. By moving from a more expensive home to a significantly lower cost one, the weight of money would:

• Drive up prices of cheaper replacement homes;
• Vest the aging downgraders with equity for disbursement to their children;
• Given an inverted population pyramid (please see Graph 7), the equity realized through this process will be divided over fewer heads, that is each will be endowed more.

If the downgraders have two children (since they are of the Stop at Two generation), then out of one downgrade, an incremental demand for 2 private properties can arise. The demand for private residential properties would be further raised if the policy to counter the aging population is to import ‘talent’. Immigration of economically active people who later become permanent residents or citizens leads to additional sources of buying demand.

Conclusion

The hard coded cooling measures have been created to lower demand by reducing affordability. This is similar to enacting physical barriers slowing down the liquidity rush. In the short-run it penalizes those at the margin who couldn’t clamber over these measures until they charge up their financial resources, leaving the well-to-do the freedom to buy top end properties. However, that would not stop the aspiration of those who can have been temporary barred for after a period, they would return to buy and buy something that may still stretch their finances. That means that if one goes on just enacting measures to cool the market, the market will expect this to become the norm and, in turn, it will encourage more to buy. If a private housing Gini coefficient can be constructed, it will be kinked or distorted over time because of such measures.

Because the economy is constantly being refreshed with new industries taking root here and because of the importation of skilled workers means that they also earn better salaries than the average Singaporean, private property prices, being in the upper end of the housing spectrum, should rise faster than average house prices (both public and private). Market observers have been proffering different solutions to curb an overconsumption of private residential properties. The fact that there is a plethora of views shows that there are various camps out there who see the problem differently. This is not a good development because it means that the issue at hand is blurred rather than crystal clear. We offer some suggestions on how instead of further rigid measures being implemented, the approach could be taken from the lending and valuation side of the housing ecosystem.

In the meantime, notwithstanding the increasing risk of structural unemployment, heightened levels of corporate downsizing and restructurings, private residential house prices are still expected to increase this year. Owing to the anti-pandemic measures in 2020, the economy in 2021 should rebound at a rate exceeding that of the URA PPI. For 2021, we are anticipating private residential prices to rise by 0% to 2% this year. In that case, the tracking error for 2021 should be lower. However, if we use the historical tracking error between real GDP and URA PPI changes, then should the former rise 5.5%, the latter may rise by 12.6% YoY. While that remains a risk, we think it is low. But then, who knows?