HDB Flats: Prices, Bubbles, and Policy

Introduction

Transforming Singapore into a ‘Smart Nation’ has been a key aspiration of the Singapore government since it was rolled out as an initiative in 2014. Harnessing technology to enhance citizens’ lives and to promote opportunities for growth has already borne fruit in several sectors such as transportation (Parking.sg) and mobile banking (PayNow), with many other schemes still in the pipeline.

One underlooked aspect of this initiative is the public release of some government datasets as part of a movement towards open data. Datasets from various ministries and agencies are made available on a consolidated repository, Data.gov.sg, which makes it easier for the public to access and download such data. Many cities around the world have adopted a similar approach, with New York, London, and Berlin being prime examples.

Beyond transparency and accountability, another compelling reason to make more data open to the public is that it promotes civic engagement on issues that matter to all Singaporeans. Having access to rich datasets on important topics, be it housing, healthcare, or transportation, would encourage more citizens to explore the data themselves and become genuine partners in the policymaking process.

In that spirit, this post explores the HDB resale flat prices dataset, released by HDB and available online here. For reproducibility reasons, the R code used to process the data is available on my GitHub page here, and all Tableau charts are available on my Tableau Public page.

Unfortunately, due to the Tableau dashboards and tables presented here (and my lack of HTML skills), this post is not mobile-friendly and is best viewed on a computer instead.

Exploring the data

The HDB resale flat prices dataset provides data on every HDB resale flat transaction since 1990, along with details such as the resale price, the flat type, flat size, and neighbourhood. The main processing step I took was to adjust prices for inflation, using 2014 prices as the benchmark. Data on the yearly Consumer Price Index was obtained from SingStat. I also removed all transactions of flats in Lim Chu Kang, since there were less than 50 entries and all occurring before 2000. It’s also helpful to note that some transactions may refer to flats that no longer exist (quite a few of them actually!) so you may not be able to find all the addresses on Google Maps anymore.

Looking back to the mid-1990s

The line graph below provides an overview of how HDB resale flat prices have changed since the 1990s, differentiated by flat type. Recall again that prices are adjusted to correct for inflation, which are indexed at 2014 prices – this gives us a more accurate comparison across time.

The most immediate observation here is the two peaks in the graph: one around 1996 Q3 and another around 2012 Q4. There is a trough in the middle, caused by a consistent slide or stabilisation in resale flat prices for 7 full years (from 2000 to 2006). In contrast to the relative volatility in the earlier two decades, prices have remained reasonably stable over the past 5 years.

Admittedly, I was quite surprised to see a housing price bubble in the mid-1990s. The post-recession rise in housing prices was a hot-button issue in the 2011 General Elections, but there was little talk about the earlier housing bubble just fifteen years ago. Resale flat prices nearly tripled between 1993 and 1996 across all flat types, a rate not nearly matched by the 2007 to 2011 increase in housing prices.

After a Google search and a quick browse through some papers, I found a section from this paper (Phang & Kim, 2013) particularly illuminating:

In his memoirs, Mr Lee Kuan Yew (who remained in the Cabinet as Minister Mentor to Goh’s government) recalled this as an event in which the government yielded to popular pressure: “I should have known that it does not pay to yield to popular pressure beyond our capacity to deliver. Yet I was party to a similar mistake in the early 1990s. As property prices rose, everybody wanted to make a profit on the sale of their old flat and then upgrade to a new one, the biggest they could afford. Instead of choking off demand by charging a levy to reduce their windfall profits, I agreed that we accommodate the voters by increasing the number of flats built. That aggravated the real estate bubble and made it more painful when the currency crisis struck in 1997. Had we choked off the demand earlier, in 1995, we would have been immensely better off (Lee, 2000, p. 121).”

With this in mind, I revised the original graph to include some additional information. In the graph below, the blue shaded regions represent periods of time when the government introduced property cooling measures (information obtained from the same paper above) while the red shaded regions are the quarters when Singapore experienced a negative GDP growth rate. Singapore’s real GDP growth across time is also visualised in the chart below.

Cooling measures were first introduced on 15 May 1996 and had a nearly immediate effect in containing the spike in resale flat prices that had been rising rapidly since 1995 Q1. By the end of 1996, the prices plateaued for all flat types and began falling for the more expensive flat types, suggesting that the cooling measures had worked in deflating the housing bubble.

Unfortunately, the first waves of the 1997 Asian financial crisis hit Thailand and Indonesia on May and July 1997 respectively, and we can observe a one-quarter lagged impact on Singapore’s real GDP growth rate in 1997 Q4 and 1998 Q1. This caused housing prices to fall even further as Singapore slipped into a recession for almost a year. Despite the subsequent recovery in prices in 1999, prices slid continually for seven years for 5 room and executive HDB flats and rose only slightly for 3 and 4 room HDB flats. The Dotcom bubble in 2001 and the SARS crisis in 2003 did not help much in that respect.

With the benefit of hindsight, Lee Kuan Yew makes a fair point about the timing of the cooling measures. If they were implemented in early 1995 instead, the housing market may have been stabilised before the 1997 Asian financial crisis, preventing the extended freefall of resale flat prices from 1996 Q4 to 1999 Q1. Perhaps this lesson motivated the government’s decision to begin implementing cooling measures in 2009, even though the economy had just begun recovering from the country’s longest recession in two decades.

The late 2000s

All of the above focused very keenly on the mid-1990s housing bubble, but there are also some interesting questions to ask about the more recent increases in resale flat prices. For one, why were the cooling measures from 2009 to 2012 unable to stabilise prices more rapidly?

To answer this, we have to look a bit further back to the early 2000s. In his collection of commentaries, ‘Reflections on Housing a Nation’ (available online here), former Minister of National Development Mah Bow Tan writes:

At the height of the property boom in the mid-90s, there were as many as 150,000 buyers in the queue, and the wait for a flat was as long as seven years. However, when the Asian Financial Crisis struck in 1997, the queue vanished, virtually overnight. HDB ended up with 31,000 unsold flats, which took more than five years to clear.

Data on the number of HDB flats built every year is available online as well, so here is a graph which makes essentially the same point.

The number of HDB flats built yearly hit a peak in 1998, two years after the peak in HDB resale flat prices and right in the middle of Singapore’s first recession in the decade. The subsequent drop in flats built every year is a sensible response to the immense surplus of flats, especially since these HDB flats were viewed as retirement nest eggs and the government could not afford for these prices to spiral downwards further. Another paper (Lum, 2011) points out that government sales of confirmed land sites were suspended from late 2001 to early 2007 – mainly to offload the huge surplus of HDB flats and to revive the housing market. This tracks what is being depicted in the graph above, with the grey region highlighting the government’s efforts at curbing the oversupply.

While public housing supply was significantly reduced from 2001 to 2007, the demand for HDB flats had begun to rise from 2005 onwards. Measures to promote market demand, such as reducing stamp duty rates and lowering the minimum occupation period to one year, coincided with a strengthening global economy from 2005 to 2007. Moreover, a rise in foreign capital flows into Singapore kept domestic interest rates low, making the speculative flipping of resale flats a highly profitable endeavour.

Yet, if this was the full story, then prices ought to have fallen when the Global Financial Crisis began to hit in 2008 Q2. Singapore was hit with a GDP contraction of -8.36% in 2009 Q1, but resale flat prices only dropped slightly or remained stable during that period. In contrast, Singapore’s worst quarter during the Asian Financial Crisis saw a -5.29% change in real GDP, and prices fell across all flat types.

We have to look elsewhere for the answer: Singapore’s population growth.

While the government suspended land sales and the construction of HDB flats, population growth continued apace except for the brief interruption in 2003 caused by the SARS crisis. In the ten years between 1999 to 2008, the residential population grew by over 400,000 amidst the lowest construction rates for HDB flats since the 1990s. What is missing from the visualisation is the fact that non-resident population growth also surged during those years (also identified by Lum, 2011), adding further to the strain on the supply of public housing by encouraging existing flat owners to rent rather than sell their flats.

As such, resale prices did not drop much during the recession and picked up immediately again once Singapore started to recover in 2009 Q2 and Q3. There was simply overwhelming demand for a limited supply of HDB flats, which was enough to keep resale prices from crashing even after the 2009 Global Financial Crisis recession and the cooling measures. In contrast, prices fell nearly immediately and kept spiralling downwards immediately after the cooling measures and the 1997 Asian Financial Crisis.

To the government’s credit, they identified this fairly early and started implementing cooling measures within a year of the country’s recovery from the recession. Unfortunately, HDB flats take time to build. Lum (2011) also points out that the government restarted land sales as early as 2007, but the effects can only be seen two to three years later with the uptick in housing units constructed in 2009 and 2010. The pent-up demand meant that prices continued to increase in the interim period while more units were being built, and they only stabilised after a bumper crop of over 27,000 units were built in 2014.

Conclusions

Singapore’s public housing model is often praised for providing high-quality housing at affordable prices and at such a large scale, and rightfully so. In countries like the United Kingdom or the United States, public housing has a bad reputation for shoddy construction and poor maintenance, and in some cases have even cost lives. To many, it is inconceivable that public housing could house a majority of the country and even rival private properties in terms of quality.

One of the main challenges in discussing public housing issues in Singapore is how intertwined HDB flats are to various aspects of our lives - housing, retirement, integration, and mobility are some that come to mind. Housing policy is also affected by a confluence of many factors, such as economic development, population policy, infrastructural needs, and environmental concerns. The squeeze in resale prices in the late 2000s was partly due to a lack of synchronisation across these different areas, with population and economic growth continually marching ahead while infrastructure development slid behind. Understanding the complexity behind the issue of public housing is necessary to having more constructive discussions about how it should look like as Singapore continues to grow and develop.

Finally, if LinkedIn is any indication, data analytics and data science is rapidly growing in prominence in Singapore. As it does, more students will be looking for rich datasets to work with for their projects to demonstrate their capabilities. To my knowledge, HDB’s resale flat prices dataset is one of the few publicly available datasets which are fine-grained enough for more complex analyses, and I have learnt so much over the past few weeks while working on this article. It would be a great idea to release more datasets on other issues that matter to everyone. What better way is there to encourage more civic engagement with tough public policy issues than to allow Singaporeans to analyse Singapore’s data themselves and to contribute constructively to our national discourse?