What are the most significant geographical factors affecting the price of public housing in singapore?

This paper aims to determine the most significant geographical factors affecting the price of public housing in Singapore. Understanding the importance of location as a determination of value lends a greater comprehension of consumer decision-making and property pricing in urban areas. This paper explores the significance of geographical factors at different scales through a combination of primary and secondary research.

Singapore, a city-state in Southeast Asia, is well-known for its high standard of public housing. In 2020, 80% of the Singaporean population lived in housing provided by the Housing and Development Board (HDB), a government-owned corporation (Statista, 2020). HDB flats are allocated to individuals based on requirements including age and marital status. Flats are heavily subsidised by the Singaporean government, encouraging the purchasing of flats over renting.

 

Singapore is divided into 26 HDB towns, totalling 1.09 million housing units (Statista, 2021). The towns have a mean size of 7.54 km2 and mean population of 125,000 (data.gov 2020).

In order to answer the research question, a combination of primary and secondary data was collected at two scales: HDB towns and individual HDB estates. Data from the larger HDB towns/areas indicate large-scale trends in housing prices e.g. distance to the CBD, whereas the smaller-scale HDB estates indicate more local factors e.g. distance to a mall.

 

To determine the HDB estates that were investigated, a combination of stratified and random sampling was used. 26 HDB estates were chosen, one in each HDB town. This was to ensure a geographical spread of data was used. South-East Asia’s leading real-estate platform, PropertyGuru, was used to select a 3-bedroom, for-sale, HDB flat in each town. PropertyGuru provided recent transaction prices allowing the mean to be taken. Figure 2.1 shows the distribution of the 26 estates used in the study.

A 2018 study determined the following as factors affecting housing prices in Dalian, China (Yang et al, 2018): Distance to the airport, distance to the railway station, distance to the nearest park, the number of medical facilities within 1000 metres, transport accessibility, age, distance to education, distance to the coastline and a series of building attributes. These factors were adjusted to Singapore’s context and assigned to either the HDB Towns data set or the HDB estate data set based on the nature of the factor. If the factor referred to the distance to a less common place, like the airport, it was analysed under the HDB Town data set. However, frequently occurring variables, like malls, were analysed under the HDB Estate data set as there may be more than one mall in each HDB Town.

 

The distance to the closest hospital was measured using Google Earth in reference to the 14 major hospitals in Singapore (MOH). These hospitals are shown in Figure 3.1.

The distance to the closest MRT station and the MRT station density were found using the 130 MRT stations in Singapore (LTA).

The density of bus stops was calculated using information from the map in Figure 3.3.

In an informal interview with two HDB owners, the following was determined as qualitative factors affecting the value of an HDB: building condition, general housekeeping, greenery, facilities, aesthetic appeal, air quality, and safety. These factors were cumulated into an Environmental Quality (EQ) survey conducted at all 26 HDB estates. Each factor was given a score out of 10 based on the criteria below. A pilot study was conducted to determine the standards for each category.

 

The pilot study was conducted on June 24, 2022, at 13:50. The chosen site was an HDB estate at Redhill Close in the South of Singapore. The estate scored extremely low on building condition (3), general housekeeping (2), facilities (3), and safety (3). This was due to it being a relatively old estate (built in 1955) and a lack of maintenance (PropertyGuru). The estate scored highly in greenery (7), accessibility (7), and air quality (9). Older HDB estates tend to be less dense than recently built estates. This is due to rising concerns surrounding Singapore’s ability to house its population, leading to more dense HDB estates being built. These tend to have less communal and green spaces. Since the estate at Redhill is older, there are much larger open areas, leading to better air quality and more greenery. Furthermore, a survey among HDB residents found that 86% have a ‘grow in place’ mindset, this means older HDBs will tend to have older residents (Channel News Asia). As the residents of an HDB estate age, its town council adapts to provide for the needs of the elderly, making it more accessible.

Qualitative factors were cumulated into an Environmental Quality (EQ) survey conducted at all 26 HDB estates. Each factor was given a score out of 10 based on the criteria in Table 1 (Appendix).

Multiple Spearman’s Rank tests were done to mathematically compare the significance of factors affecting the price of HDBs. This eliminates the issue of normal distribution in data sets. The Spearman’s Rank test was conducted for all factors against price. The outcome (-1 < r < 1) must then be assessed for statistical significance. The Spearman’s Rank test is conducted using the formula below.

 

\(p=1\frac{6∑d^2}{n(n^2-1)}\)

 

Where ⍴ is the Spearman’s Rank Correlation Coefficient,

 

d is the difference between the ranks of a pair of data,

 

and n is the number of pairs of data.

A series of figures demonstrating the distribution of the different factors across Singapore.

The most statistically significant factor, determined by the Spearman’s Rank test, is VCBD. There is a strong negative correlation between price and VCBD. Housing in this area is of much higher value for a number of reasons. 12.9% of Singapore’s resident workforce worked in the CBD in 2021, decreasing the land available for real estate and increasing its demand, leading to higher prices (The Straits Times). The CBD is relatively new with major developments occurring in the 1980s, thus HDB blocks tend to be relatively new, averaging 35 years old (PropertyGuru). Newer HDBs tend to have better value as they use more advanced technology, have better layouts, and are more aesthetically pleasing. In addition, since the CBD tends to be a very publicised area, it is in the best interest of the government to display the highest quality and most modern HDBs nearby, driving up construction prices.

 

This trend is explained in bid-rent theory, which states that land prices are higher closer to a city’s centre, see Figure 4.7 (Trussell, 2010). The trend in Singapore follows the bid-rent theory extremely well, with the initial steep negative gradient becoming a gentle negative gradient. According to figure 4.6, the initial steep gradient begins to become less steep after 2500 metres from the CBD, demonstrating a large difference in value between living in the CBD versus living 2500 metres away from it. This is likely due to the ability to walk to the city centre compared to taking a form of transportation as 2500 metres is a walkable distance. From about 7500 metres and above, the difference between commuting 7500 metres and 10,000 metres becomes negligible and thus prices of land decrease at a lesser rate.

The other statistically significant factor was VPD. There is a strong negative correlation between VPD and price. However, when comparing VPD to VCBD, a positive correlation was found, see Figure 4.8. It can be concluded that this is again due to bid-rent theory. As VCBD increases, price decreases, increasing the demand for HDBs, thus there is a higher density of HDB residents.

Figure 4.8 demonstrates a strong correlation between VCBD and the VPD, indicating that there are more HDB residents in the more rural areas of Singapore. If all areas (including non-residential areas) in Singapore were plotted on this graph, the expected outcome would be a downward curving parabola. As VCBD continues to increase, the large distance to the city centre will eventually outweigh the benefit of lower housing costs. In between, there will be high HDB density as the distance to the CBD will not be large and prices will not be high. Furthermore, in areas that are furthest away from the CBD, land use tends to be more industrial and agricultural, decreasing VPD.

 

It can be concluded that the negative correlation between price and HDB resident density is likely due to the positive correlation between VPD and VCBD.

 

Both VBus and VMRT Density had a slight positive correlation to price. VMRT Density had a slightly higher Spearman’s Rank correlation coefficient. Although transportation is likely a key factor in determining HDB prices, due to Singapore’s urban planning and efficient public transportation, it may be a smaller facto r than in other cities. In 2013, a land-use plan was published by the Ministry of National Development, this stated that 19% (139km2 ) of the land in Singapore would be used for reservoirs, nature reserves, utility plants, ports, and airports, the other 81% (591km2 ) would be used for remaining needs (Ministry of National Development, 2013). Since Singapore has 5000 bus stops, it can be calculated that each bus stop has a range of 191 metres. Thus VBus has minimal effect on the price as there is a bus stop within 191 metres of HDB blocks on average. Furthermore, bus stops are placed at high-demand areas like housing estates, increasing the likelihood of bus stops within walking distance of HDBs.

 

Interestingly, the Spearman’s Rank correlation coefficient dictates that VMRT Density has only a slightly stronger correlation to price than bus stop density. A strong positive correlation is expected as HDBs near MRT stations (a preferred form of transport) tend to be more accessible. However, while MRTs are an extremely efficient method of transportation, they are also loud and attract large amounts of pedestrian traffic, making living next to one slightly less desirable. Thus, a high density of MRT stations may not increase the price of HDBs in the HDB town.

Although in many cities a coastal premium exists, in which coastal properties have increased value, the coastal premium cannot be seen in this data set. Firstly, the data set captures HDBs in a town, although the town may be on the coast, not all the HDBs are on the coast. Secondly, since Singapore is a small island, the distance to the coast at all times is no more than 10 kilometres. In other cities, coastlines provide valuable economic, social, and environmental resources. This does not apply to Singapore as 10 kilometres is a commutable distance considering Singapore’s infrastructure and transportation systems.

Although a negative correlation is expected and likely exists in larger cities, due to Singapore’s small size, the furthest town from the main international airport in Singapore is Jurong West, 31590 metres away. This distance is insignificant compared to the likely total travel time of the individual going to the airport. The slight positive correlation shown in Figure 4.10 may be due to noise levels. Singapore’s Changi Airport has received over 86,000 aircraft in the past year (Changi Airport, 2022), this constant air traffic is noisy, reducing the desirability of housing nearby. Furthermore, due to the location of Changi Airport in the extreme East of Singapore, the distances from Northern towns and the CBD are very similar (see figure 4.5), and as previously established, there is a strong negative correlation between the distance to the CBD and price.

 

A final statistical test was done to determine whether there is a statistically significant correlation between all factors and price. This was done by taking a weighted mean of a towns rank for each factor. The weighting was done using the Spearman’s Correlation coefficient, this adjusts for the significance of the factor in determining price.

The conclusion can be made that the geographical factors presented in this data set have a statistically significant effect on the price of HDBs in Singapore.

A series of figures demonstrating the distribution of the different factors across Singapore.

 

The EQ Survey was conducted to assess factors that cannot be quantified from a remote location. The survey was designed so that the higher the score the estate received, the greater its perceived value. Thus, a positive correlation was expected between VEQ total and price.

Figure 4.20 establishes the positive correlation between VEQ and price. The Spearman’s Rank correlation coefficient dictates a likely non-statistically significant correlation between the two variables. This is likely due to other factors influencing the price of the HDB estate. Since only 26 HDBs were assessed in this study, one outlier in this data set has a larger impact on the final correlation. Consider data point (39, 666). This data point is the biggest outlier in the data set. This HDB estate is on 7 Commonwealth Avenue in central Singapore. This point being an outlier indicates the significance of other factors affecting the price at this location.

Figure 4.21 demonstrates a weak positive correlation between VNoise and price. Although a high noise level may be bothersome and thus reduce the value of the property, it may also be indicative of other factors. For example, transportational accessibility. Properties nearby MRT stations and major highways may be louder, however, they have an increased value due to their proximity to these infrastructure. It is also important to note the source of the noise. Noise levels may be a sign of a community or nature, potentially value-adding factors. Noise levels collected in this study have both positive and negative effects on price, indicating a net negligible effect on price.

Although V_Primary may be an important factor for young couples in Singapore, it is of little importance to older buyers. Furthermore, Singapore has an ageing population, demonstrating the lack of importance of living near a primary school. Figure 4.23 demonstrates the population bulge between 45 years old and 64 years old and the very skinny base, indicating few young children. Singapore’s fertility rate has also decreased from 1.82 live births per woman in 1980 to 1.12 in 2020 (Singstat, 2021), further explaining the reduction in importance to live near a primary school.

The expected outcome was a strong negative correlation between V_Hospital and price, Figure 4.24 demonstrates a slight negative correlation. Living nearby a hospital allows for easy access to general and emergency healthcare, thus an expected increase in value near hospitals. However, the correlation may be less strong than expected due to a number of factors. Firstly, hospitals tend to be extremely busy, leading to unwanted noise, road congestion, and commotion. Secondly, culture plays an important role in living near hospitals. Living near a hospital is considered bad ‘Fengshui’, this affects the choices of many people who practise Fengshui. Lastly, the value of living near a hospital may greatly differ from individual to individual. A vulnerable individual may find much greater value in living near a hospital than a non-vulnerable individual, however, this study does not take this into account.

Interestingly, V_Mall showed the greatest correlation to price in this data set as shown in Figure 4.25. Malls are an extremely popular location in Singapore, VivoCity, located in Southern Singapore, receives 50 million visitors a year (Mapletree, 2014). This popularity is due to the convenience of malls as well as their well-designed communal spaces and air-conditioning. Malls can also be transportation hubs, for example, Plaza Singapura, a mall in central Singapore, is located at the intersection of 3 MRT lines, making it an essential transportation hub. This increases the benefits of living near malls. The location of malls also indicate important commercial areas, thereby increasing the value of the land. Although the proximity to malls may not directly affect the price of the estate, it is an indication of other factors that do affect the price of the estate.

The expected relationship between V_Park and price is a strong negative correlation. Parks not only provide recreational areas, they also increase a sense of community and are pockets of nature, two very important things in Singapore’s society. Thus a negative relationship is expected. However, Figure 4.26 shows a slight positive correlation between the distance to a park and the price of the estate. Parks in Singapore tend to be small, the intention is to make the city look green and provide communal areas. Singapore has over 400 parks spread across the country (NParks, 2022), which means the distance to the closest park is relatively small. This can be demonstrated by the range of data: 104 - 1120 metres, with 19 out of 26 of the studied estates being within 600 metres of a park. Because the range is small in comparison to the actual size of Singapore, the difference between living 100 metres from a park and living 600 metres from a park is negligible. It can reasonably be concluded that the data set misrepresents the relationship between VPark and the price at the estate.

MRTs are extremely efficient forms of transport, in 2019, 3.1 million people a day used the MRT (Statista, 2021). Thus, a strong negative relationship was predicted between VMRT and the price at the HDB estate. However, Figure 4.27 demonstrates a slight positive relationship. This may be due to noise pollution from the MRT, however, the majority of MRT lines run underground, reducing the noise pollution created. Furthermore, this study doesn’t take into account the significance of different MRT stations. Most stations cater to one MRT line, the value of the estates near the station can depend on the other stations on the line, for example, a station on the East-West line or the North-South line may have a greater impact on the price of land nearby as these lines go to the CBD. In addition, a few stations have two or three lines passing through, making it more valuable as a transportation hub. These stations also tend to be near commercial centres. This data set demonstrates little to no correlation between VMRTT and the price of HDB estates nearby, however, it may be beneficial to look at the distance to more significant MRT stations and interchanges.

The expected relationship between V_Age and the price at the HDB estate is a strong negative correlation. Newer HDBs tend to be in higher demand due to better designs, facilities, technology, and innovation. Older HDBs tend to be out of date, run-down and less equipped. Figure 4.28 demonstrates a weak negative correlation between age and price. This may be because older HDBs tend to be larger than new ones, both in communal areas and in units. 12 of the estates in this study were built before the 1990s; Singapore’s population in 1990 was just over 3 million, with just under double that number in 2022, population density has become an issue. This means newer HDBs tend to be denser and have smaller units, making them slightly less desirable compared to larger, older estates. The process on page 19 was repeated with this data set. The results are as follows.

Based on the statistical significance of this test, the conclusion can be made that the geographical factors presented at this scale do have an effect on the price of units at HDB estates.

This investigation took place at two scales to ensure a variety of factors were considered. On a large scale, the distance to the V_CBD and the V_PD were determined as significant factors affecting the average price of HDBs with a certainty of > 99.9% and > 95% respectively. Upon further analysis, it was concluded that the V_PD is greatly affected by V_CBD, indicating that V_CBD is the greatest factor affecting HDB prices in the 24 studied HDB towns. V_{MRT density} had a positive effect on the price, however, this correlation was not statistically significant. Interestingly, a negative correlation was found when comparing V_Bus and HDB prices, it was concluded that due to the already extremely high density of bus stops (bus stop present within 191 metres of any given location on average, busrouter.sg) in Singapore, day-to-day life would not be significantly affected based on slight changes in bus stop density. V_Coastline did not significantly affect the price of HDBs, it was concluded that this is likely because Singapore is a small island, thus one is always somewhat near the coast. V_Airport did not have a significant impact on the price either. It was found that this is due to the location on the East coast of Singapore and the low frequency of visits to this facility as HDB residents. Looking at individual HDB estates yielded some more conclusions. V_EQ had a slight positive correlation to price as expected. Interestingly, V_Noise had a slight positive correlation to price, this was likely due to the unknown source of the noise and other, potentially value-adding, factors that are associated with noise, including close proximity to MRT stations and highways. The V_Primary and V_Hospital had slight positive and negative correlations respectively. It was determined that the lack of significance of the proximity to primary schools is likely due to an ageing population and fewer kids among HDB residents. A negative correlation between the V_Hospital and prices was expected, however, the correlation was relatively weak, this was explained using cultural factors and crowdedness. V_Mall was determined as the most significant factor mathematically. This was justified by the assets associated with a mall and convenience. The V_Psrk had a positive correlation with price. This can be explained to some extent by the frequency of parks in Singapore, however, the positive correlation is likely due to the estates chosen and is therefore not representative of all estates. The V_MRT was found to have a very weak positive correlation to price. Although noise pollution may play a small role in decreasing the value of the nearby property, it was found that more specific detail was required as different MRT stations have different values: large interchanges versus one-train MRT stations. While a strong negative correlation between V_Age and price was expected, experimentally, a weak positive correlation was found. This was found to be due to increasing population density and therefore decreasing size of HDB communal areas and units. This likely offset the value-added in technology and building quality in newer HDBs.

This investigation had a series of strengths and weaknesses. The main strengths came from secondary data sources and accuracy in primary data collection. Data collection for the average price of HDBs in the 26 HDB towns came directly from a Singapore government source: data.gov. This ensured that the most accurate data available was used. In addition, primary data collection for distances from point to point was collected using google earth measurements, this was the most accurate method available for this investigation.

Ho, Grace. "Downtown Core draws the largest share of S'pore resident workers: Population census." The Straits Times, [Singapore], 18 June 2021.

 

Housing & Development Board. "HDB Towns, Your Home." Housing & Development Board (HDB), 1 Apr. 2021, www.hdb.gov.sg/about-us/history/hdb-towns-your-home. Accessed 10 July 2022.

 

LTA. Land Transport Authority (LTA), 8 Dec. 2021, www.lta.gov.sg/content/ltagov/en/getting_around/public_transport/rail_network.

 

Mapletree. "VIVOCITY." Mapletree - Home, 2022, www.mapletree.com.sg. Accessed 21 Aug. 2022.

 

Ministry of Health. "Health Facilities." Ministry of Health, www.moh.gov.sg/resources-statistics/singapore-health-facts/health-facilities. Accessed 21 Aug. 2022.

 

PopulationPyramid. PopulationPyramid, Accessed 21 Aug. 2022.

 

Roots. "The Central Business District on the South Bank of the Singapore River." Roots, 4 Apr. 2021, www.roots.gov.sg/Collection-Landing/listing/1183807. Accessed 7 July 2022.

 

Singapore. Ministry of National Development. A High Quality Living Environment for All Singaporeans: Land Use Plan to Support Singapore's Future Population. 2013, Accessed 19 Aug. 2022.

 

"Singapore: Population Living in Public Housing 2020." Statista, 26 Apr. 2021, www.statista.com/statistics/966747/population-living-in-public-housing-singapore. Accessed 30 July 2022.

 

Statista. "Singapore: Daily MRT Ridership 2021." Statista, 7 Apr. 2022, www.statista.com/statistics/1006216/singapore-daily-mrt-system-ridership/. Accessed 21 Aug. 2022.

 

"Total Fertility Rate." Base, 9 May 2022, www.singstat.gov.sg/modules/infographics/total-fertility-rate. Accessed 21 Aug. 2022. "Traffic Statistics | Changi Airport Group." Welcome To Singapore Changi Airport, 31 July 2022, www.changiairport.com/corporate/our-expertise/air-hub/traffic-statistics.html. Accessed 21 Aug. 2022.

 

Trussell, Benjamin. The Bid Rent Gradient Theory In Eugene, Oregon: An Empirical Investigation. 2010. U Oregon, MS thesis.

 

Yang, Jun, et al. "Impact of Accessibility on Housing Prices in Dalian City of China Based on a Geographically Weighted Regression Model." Chinese Geographical Science, vol. 28, no. 3, 2018, SpringerLink. link.springer.com/. Accessed 8 Aug. 2022.