Project dissertation submitted in partial fulfillment of the requirements for the Degree of B.Sc. (Real Estate)



ABSTRACT This study looks into the influence of underlying fundamentals and government policies on the dynamics of private housing price in Singapore using quarterly data from the first quarter of 1990 to the first quarter of 2009. Cointegration analysis is applied to an empirical model which shows that the long-run movement of real private housing price is explained by the real public resale housing price, the user cost of capital and the overall private housing stock during the investigation period. An error-correction model is estimated to capture the changes in real private housing price to account for the short-run deviations from the long-run equilibrium relationship. The volatility of real private house price changes appears to be significantly influenced by the changes in total private housing stock and the implementation of government policies. The error correction term shows that 3.3 per cent of the previous discrepancies between actual and desired long-run private housing prices are corrected in each quarter. This is a slow response rate compared to the range of 10 to 13 per cent reported for the Singapore housing market. Another important finding is the significant impact of the regulatory policies on private housing price such as the antispeculation measures, Government Land Sales programme, removal of Deferred Payment Scheme and deregulatory policies such as public resale flat credit policy, Central Provident Fund liberalization policy and other deregulatory measures for the public and private housing market. Given the tendency of the government to intervene in the property market to smooth out undesirable volatility from time to time, shocks from the implementation of such policies may have a lasting disequilibrating effect on private housing price. Thus, timing and coordination are important for any policy making. Keywords: Cointegration, Error Correction Model, Housing, Policy, Price Dynamics, Singapore

ACKNOWLEDGEMENTS I owe my deepest gratitude to my supervisor, Associate Professor Tu Yong, whose encouragement, guidance and support during the whole dissertation process enabled me to develop an understanding of the subject.

Lastly, I extend my appreciation and blessings to all of those who supported me in any aspect during the completion of this dissertation.


1 Pricing and Supply of New Public Housing 3.5 Property Market Cooling Measures 3.4 Economic Mechanisms and Price Dynamics of the Private Residential Property Market iii 26 28 30 30 31 35 36 38 39 .5 2.2 2.3 2.2 General Structure of the Singapore Housing Market 3. Speculations and Bubbles Government Intervention in Housing Markets Summary 9 9 10 13 19 25 CHAPTER THREE: MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET 3.3.1 1.3.3 Central Provident Fund Policies 3.3.4 2.1 Introduction 2.3 1.3 Spillover Effects from Government Policies and Interventions 3.6 Introduction The Broader Economy and Housing Markets Long-run Equilibrium in Housing Markets House Price Dynamics: Expectations.Table of Contents Abstract Acknowledgement Table of Contents List of Tables List of Figures List of Appendices CHAPTER ONE: INTRODUCTION 1.2 1.4 1.2 Changes in Eligibility Criteria for Public Housing 3.5 Background and Objectives Rationale and Significance of Study Hypothesis Statement Scope of Study Organisation of Study 1 5 6 7 8 i ii iii iv v vi CHAPTER TWO: LITERATURE REVIEW 2.4 Release of State Land for Private Housing Development 3.

6 Summary CHAPTER FOUR: RESEARCH METHODOLOGY 4.300 words) iv .3.1 5.4 5.5 5.4 Summary of Main Findings Research Implications and Contributions Limitations Recommendations for Future Research 87 90 92 93 B-1 A-1 Bibliography Appendices (12.7 Introduction Stationarity Test and Order of Integration Specification Test for Model Selection Johansen Cointegration Test Error Correction Modeling Error Correction Model Estimation Summary 63 64 67 68 74 82 86 CHAPTER SIX: CONCLUSION 6.7 Introduction Data Description Model Specification Unit Root Test Cointegration Test Concept of Error Correction Model Summary 40 44 47 49 50 50 53 55 58 60 62 CHAPTER FIVE: DATA ANALYSIS AND ESTIMATION RESULTS 5.6 5.2 5.3.5 4.4 4.1 6.2 4.5 Indicators of Macroeconomic Performance 3.3.2 6.3 4.3 6.1 4.5 Housing Commencements and Development Costs 3.6 4.3 5.5 Simple Model of the User Cost of Capital 3.

8 Table 5.2 Table 2.1 Table 2.7 Table 5.3 Table 2.1 Table 4.List of Tables Table 2.2 Table 5.3 Table 5.4 Table 3.6 Table 5. Speculations and Bubbles Supply Elasticities Local Studies Overview of DBSS Projects Determinants of Private Housing Price Augmented Dickey Fuller Tests for Unit Root Stationarity Phillips Perron Tests for Unit Root Stationarity Vector Auto Regression Lag Order Selection Criteria Selection of Appropriate Model by Pantula Principle Selection of Appropriate Model by Information Criterion Johansen Tests of Cointegration Phillips Perron Tests for Estimated Residuals Johansen Cointegration Equations Error Correction Model 15 16 17 22 31 51 65 66 67 68 68 69 69 70 75 v .1 Table 5.5 Table 5.4 Table 5.1 Table 5.9 Housing Modelling and Long-run Equilibrium Short-run Dynamics: Expectations.

2 Figure 3.1 Figure 3.9 Figure 3.3 Figure 3.5 Price Movements and Housing Commencements Structure of the Singapore Private Housing Market Pyramid Housing Structure of Singapore Price Movements of the Public and Private Housing Markets Proportion of HDB upgraders buying Private Property Government Land Sales and Growth in Real Private House Price Growth in Real Private House Price User Cost and its Key Components Performance of the Economy and the Property Market Unemployment Rate and the Property Market Construction Costs and Housing Commencements Impact of Policies on Real Private Housing Price Movement Cointegration Graph of Real Private Property Price Actual and Fitted Values of Model Modelling a Static Solution Modelling a Dynamic Solution 3 27 29 33 34 38 40 42 45 47 49 80 81 83 84 85 vi .4 Figure 3.3 Figure 5.8 Figure 3.2 Figure 5.10 Figure 5.7 Figure 3.6 Figure 3.1 Figure 3.4 Figure 5.List of Figures Figure 1.1 Figure 5.5 Figure 3.

List of Appendices Appendix One Appendix Two Visual Plot of Time Series of Variables Major Government Policies from period of 1990 to 2009 A-1 A-2 vii .

Background and Objectives “… Prices have been rising too fast too much.. Ong and Sing. The achievable idea is an orderly. 2002).A2) The Singapore housing market consists of the private housing and public housing sector. Local empirical studies have shown evidence of interactions among the two markets and the spillover effects of government interventions in the regulated public housing sector on the private housing sector (Phang and Wong. 2002. 2009). Developed within a complex institutional structure that comprises policies and eligibility regulations set by institutions such as the Housing Development Board (HDB) and the Central Provident Fund Board (CPF). Tu..” The Straits Times Editorial Team (2009.1. With the public housing sector playing a dominant role (82% of total housing stock) in the Singapore housing system (Department of Statistics. any studies on the local private housing market cannot be fully analyzed without taking . 2001. the latter subjected to a certain level of government intervention and regulations. Lum. pp. the Singapore housing system is characterized as one where a freely priced market co-exists with another market. mature market of supply-demand equilibrium and price stability.. 2004. 1997. Tu and Wong.1 Introduction CHAPTER ONE│INTRODUCTION 1. buying sentiment is unrealistic.

the simple application of partial equilibrium of housing demand and supply or real estate cycle analyses into the study of any housing market can only carried out if the market is free of government controls and intervention. Similar patterns can be observed from the period of time leading to the boom and bust of 1996 and 2007. . Expectations of continual asset inflation also drove up the excess demand pressure. this shows the significant impact of the private housing market on the overall economy.48 (Phang. There was remarkable real house price appreciation with the expansion of consumer credit and low interest rates. As noted by Phang and Wong (1997). From a macroeconomic point of view. Although the local private housing market is small compared with the public housing market (18% of total housing stock).1 shows the fluctuations in price movement and housing commencement in the Singapore private residential market since 1990.2 Introduction into account the role of public housing and the institutional structure of the overall housing system. Figure 1. 2001). it accounts for 52% of the total gross housing wealth while the ratio of the gross private residential wealth to GDP is about 1.

the overall residential price index for the private property market had declined by 14. 2009). the decline in real estate prices had contributed to Singapore experiencing a technical recession in the 4th quarter of 2008 after 2 consecutive quarters of negative growth. In particular.2% (URA.1 – Price Movements and Housing Commencements (Source: URA. . Private property prices began to fall in the 3rd quarter of 2008 after 17 quarters of growth (URA. 2009) The Singapore residential property market was severely affected by the global financial crisis in 2008.1% quarter-on-quarter (qoq).3 Introduction Figure 1. For the 1st quarter of 2009. This is a significant decrease as a similar situation was last observed in the 3rd quarter of 1998 following the Asian Financial crisis where the price index declined by 13. 2009).

a housing market will often be characterized by significant deviations from the market-clearing price which is determined by the long-run demand and supply for housing. the supply of state land for private housing developments.used as a policy tool to regulate the supply of new housing in the market . In order to empirically model any housing market. regulations are designed to shift supply and demand to an equilibrium point that is socially optimal and it often results in the increased value of a property. the long-run and short-run sides of the housing markets must be distinguished clearly. This study seeks to achieve several objectives as outlined below. . This study thus looks at the dynamics of Singapore’s private housing price by developing an empirical model to evaluate the fluctuations of house price and the impact caused by the implementation of various government policies. There is a wide range of government policies which have an impact on private housing prices such as public housing policies.4 Introduction The salient feature of the Singapore housing market is the pervasive intervention by the government to smooth out any undesirable volatility. as well as rules governing the use of CPF savings. According to Kenny (1999). the confirmed list of the Government Land Sales (GLS) programme . As noted by Malpezzi (1996) in his review paper on land use and regulation constraints. For example.was suspended during the end of 2008 to prevent property prices from going into the doldrums amid the poor market sentiments then.

To explain for the price dynamics of the private housing market in the short-run by using the Error Correction Model (ECM). To understand the relationship between the private residential market and the HDB resale sub-market. Rationale and Significance of Study As noted by Leung (2004). has wide implications for the whole economy. a healthy economy is beneficial to the real estate market as the financial commitments makes . As 90. home ownership constitutes the largest single source of personal wealth and plays a significant role in the national economy. though a part of an ordinary housing market cycle. A housing price downturn. Vice versa. 2009). 1.5 Introduction To examine if cointegration relationships exist between private housing price and selected macroeconomic and property specified variables through the application of an empirical model.2. “conventional housing economics and urban economics for its part virtually ignores interaction between housing markets and the macroeconomy” (pp. To evaluate the impact of the government polices on the price dynamics of the private housing market.250).1% of the households in Singapore own their dwelling units (Yearbook of Statistics.

1.3. Trends in previous house price movement. Hypothesis Statement The research hypotheses of this study are outlined below. expectations for future housing price appreciation and policy variables are more likely to affect the adjustment dynamics of price deviations from their long-run equilibrium values. . months after the supposed property downturn. this study highlights the interaction of the two housing sectors and prompts policymakers to rethink the role of public housing and its contribution towards the performance of the private market. residential construction and employment. A long-run contemporaneous relationship exists between demand and supply fundamentals and the price movement of the private property residential market.6 Introduction up a large form of expenditure for a homebuyer or investor’s portfolio. Policy makers should be concerned about housing prices as changes in home prices have a significant impact on the fluctuations in the macroeconomic activities such as private consumption. policy makers with a deeper understanding of the price dynamics can draft the necessary policies and coordinate the timing of their implementation. Furthermore. With an unexpected euphoria currently brewing in the market.

Scope of Study The investigated period for this study spans 77 quarters from 1990 Q1 to 2009 Q1.4. A qualitative analysis in the form of a market review will be undertaken of the private residential market to explore its linkages to the general economy and the relevant government policies.7 Introduction The private housing market is sensitive to changes in the public housing market and measures directed at the latter have significant implications for the former. 1. . Further on. Specified policies that are significant to the residential property market are identified and analyzed. Department of Statistics (DOS) and Building and Construction Authority (BCA). Housing and Development Board (HDB). Central Provident Fund (CPF). The data used for this study will be time series statistical data collected from the Urban and Redevelopment Authority (URA). the quantitative analysis portion will introduce cointegration modeling to model the private housing market.

Chapter Two reviews both local and international past research work on housing markets. Chapter Six concludes this study by summarizing its main findings and highlighting the research contributions made. Cointegration Modelling and the Error Correction Model (ECM) are developed to empirically understand the price dynamics of the local private housing market. the price dynamics of housing markets and the impact of government interventions. Organisation of Study This study is organized into six separate chapters. together with their implications.5. cointegration tests and the error correction model.8 Introduction 1. Chapter Three introduces the structure of the Singapore housing system and provides the background understanding of the local housing price dynamics. Chapter Five presents the empirical results and analysis of the unit root tests. Chapter Four outlines the research design and methodology. . Limitations and recommendations for further research studies are also addressed.

long swings in construction and price development were synchronized with the long swings in aggregate economic activity. 1976). attempts to look into the opposite . the broader economy and the housing markets.1 to 2. 2. Quigley (1999) on the other hand.2 The Broader Economy and the Housing Markets The real estate market is often assumed to display cyclical movements over time and studies of linkages between real estate prices and general economic conditions have an extensive history (Gottlieb. are discussed here. both locally and internationally.4. According to Gottlieb (1976). Table 2. Authors like DiPasquale and Wheaton (1994) and Case and Shiller (1989) build on this relationship and introduced models to discuss the response of housing price changes to economic shocks through the assumption of rational market expectations.1 Introduction The format of the literature review is structured according to the four major research areas.3 gives a summary of international works. namely. the long-run equilibrium in housing markets.9 Literature Review CHAPTER TWO│LITERATURE REVIEW 2. The salient findings from this extensive scope of research. dynamics of housing markets and government intervention in housing markets. Local works are presented in Table 2.

2. Wang (2001) examines the relationship between the property and the economy for the UK market and shows that the performance of the property is closely related to and explainable by some of the economic activities. or supply. Hence. Baffoe-Bonnie (1998) finds that the housing prices and stocks of houses do respond to economic fundamentals. which serves as the theoretical basis for the fundamental determinants of real estate construction and prices. The most prominent approach is the traditional stock-flow model.10 Literature Review line of causation – the effects of changes in property markets on the subsequent health of the economy (high price volatility in Asian property markets pose real consequences for national and regional economies during the late 1990s). although the results also suggest the possibility of speculative bubbles and overshooting of equilibrium values. of housing equals demand in equilibrium. According to Rosen (1974). where the market prices are set at a level that effectively balances supply and demand for housing at equilibrium. The change in supply. or the .3 Long-run Equilibrium in Housing Markets A significant number of empirical studies had been done on the housing markets from the perspectives of demand and supply. the stock-flow model also assumes a quick clearing process where the stock. the basic theory of house price determination holds that house prices result from a market clearing process.

1993. 1982. DiPasquale and Wheaton. This approach is useful in the analysis of housing dynamics and had been used extensively in the studies of housing markets in the British markets and US markets since the 1990s (Breedon and Joyce. economic growth rate. 1994). .11 Literature Review flow. Capozza and Helsey. Meen and Andrew. lagged real appreciation. inflation rate. 1984). non-housing consumption and savings over the course of a single period. 1990. The determinants of housing demand and prices typically include demographics. in addition to the housing market. 1997. of housing thus refers to the rate of new construction less depreciation of the existing stock. Green and Hendershott. income. supply of housing. Hendry. expectation and level of confidence. real after-tax interest rate. the model has been augmented as new theories were proposed. Another extension to the demand determinants of housing is the user cost of capital concept (Dougherty and Van Order. construction costs and other market-specified variables (Mankiw and Weil. 1984. 1989.1989. unemployment rate. Poterba. Meen (1996) finds that the user cost concept has significant influences on the aggregate demand and supply in the larger economy. the price of housing. the concept is used to study the allocation of resources between housing consumption. Over the years. Based on a utility maximization framework. 1992. stock price index. Meen. Muellbauer and Murphy. 1998). financing costs.

Therefore. a possible explanation for rapidly increasing house prices lies in inelastic supply. On the supply side. 1988. and regulations inhibiting new construction (Topel and Rosen. 2005). Topel and Rosen. Goodman. 1994. two broad approaches to modeling the housing markets were later broached in the review article of DiPasquale (1999): the investment/asset market approach and the urban spatial theory approach. Economic theory denotes that the impact of increased demand will be observed in house prices rather than in the quantity of housing supplied if the supply is inelastic. 1988. the volume of new construction is taken to be a function of housing prices and input prices which comprise the costs of labor. 1984. 1984). Malpezzi and Maclennan. . 2001. As such. the concept of supply elasticities are emphasized (Poterba. materials. DiPasquale and Wheaton.12 Literature Review To gain a deeper understanding of the supply side. financing. Poterba. DiPasquale and Wheaton (1994) proposes a more complete model that incorporates land as an input and assumes explicitly that land prices depend on the housing stock.

1989. a market may never reach the equilibrium level. In a study of ten structural models of the housing market. The unique characteristics of the real estate market such as asset heterogeneity. and Bubbles Mankiw and Weil (1989) discover evidence of sluggish reactions to shifting demand fundamentals that inhibit market clearing actions. bad policy can result when the differentiation between long-run movements and short-run dynamics of the market is not taken into consideration. 1994). Inefficiencies in the market are also identified in other studies (Case and Shiller. Further emphasized by Kennedy (1998). As Van Order (1990) points out. Speculation. Riddel (2004) further proves that when fundamental forces are constantly fluctuating. equilibrium models are too inflexible to explain dynamic adjustments and the specification of time lags in economies where fundamental values are often away from equilibrium. even if it appears to move so. down-payment requirements. DiPasquale and Wheaton.4 House Price Dynamics: Expectations. Fair (1972) finds that almost all fail to identify an appropriate adjustment process to explain for the shortrun disequilibrium. price expectations of market participants and supply lags are sufficient reasons to expect that the housing market . lack of information.13 Literature Review 2. DiPasquale and Wheaton (1994) established the importance of the price adjustment mechanism in contrast to traditional assumptions of quick market clearing process.

3 below illustrates a summary of the vast literature review done on housing markets and its dynamics. The subsequent investigation into short-run price dynamics prompted a number of researchers to look into speculative bubbles in housing markets. Table 2. the term speculative bubble is defined as “…if the reason that the price is high today is only because investors believe that the selling price will be high tommorw – when ‘fundamental factors do not seem to justify such a high price – then a bubble exists” (pp. As noted by Abraham and Hendershott (1994). .13). According to Stiglitz (1990).14 Literature Review is often characterized by deviations from this long-run long market clearing price in the short-run.1 to 2. the determinants of real house price can be divided into two categories: one that explains for changes in the equilibrium price and the other that accounts for the adjustment dynamics or changing deviations from the equilibrium price.

15 Literature Review .

16 Literature Review .

17 Literature Review .

18 Literature Review 1.1 .

The government can intervene in the housing markets in four possible ways (Brown and Jackson. Left to the market alone. 1983. 1996). the market system is unlikely to be efficient (Brown and Jackson. 1996). The government can also adopt a stabilization role by implementing specific policies or measures to stabilize the market or take up a regulatory role to further maintain the proper functioning of the market. 1996). As housing is deemed as an important social good. . For example.5 Government Intervention in Housing Markets In a pluralistic society. the needs and wants of a diverse community are aplenty and individual rationality is not equal to society’s rationality. For example. or a distributive role to ensure the entire society can enjoy equal benefits. Regulations are often put in place to reduce negative effects or channel positive effects in the market. the government can adopt an allocative role to achieve efficiency in the market. governments in most countries see the need to intervene in markets to correct the market failure or to introduce policies to compensate its effects (Dunkerley. Anas and Choo (1988) examine the effects of institutional regulation in the Swedish mixed housing market and finds that the Swedish institutional schemes and policy instruments creates a number of counterintuitive and unintended effects in the housing markets.19 Literature Review 2. Brown and Jackson.

Others like Malpezzi (1996) and Jud and Winkler (2002) implement fixed-effect coefficients in their housing models and find that variables such as the availability of local land and growth management policies have a positive impact on housing prices.” Although a clear connection can be drawn between high housing prices and constraints.20 Literature Review Land use zoning. According to Rothenberg et al.48). monetary policies and other local fiscal policies are able to influence housing prices. Furthermore. property taxation. environmental regulations. A positive interaction between the role of the government in housing and urban development. Some authors such as Segal and Srinivasan (1985) study the impact of government restrictions on land prices and find that percentage of developed land removed by regulations has a positive influence on the long-run home prices. the design and scale of any of these government programs thus leave an impact on the market equilibrium. In Malpezzi’s (1996) review paper on land use and regulations. and the processes of capital accumulation in housing programs in Hong Kong and Singapore was found by Castells et al (1990). (1991): “…Housing market events and government policy initiatives which impact one submarket will have their primary effects in that submarket. limited studies have been done on the impact of regulations on housing price changes. with secondary effects appearing in other submarkets to the extent those submarkets are linked in substitution possibilities with the original target submarket (pp. he points out .

a number of local empirical studies had been done to examine the linkages between government policies and the housing market (Phang and Wong. Tu. 2002. This emphasis is largely due to the huge proportion of public homeowners (82% as at end 2008) and the spillover impact of public housing policies onto the private housing market. Ong and Sing. Recognizing the unique institutional character of the residential market in Singapore which consists of the private housing sector and the heavily regulated public housing sector. although it is difficult to identify the source of the price increase (demand or supply effect). 2004). 2002. 2001. 1997.21 Literature Review that regulations are designed to shift supply and demand to a socially optimal equilibrium point and finds that regulations often raise the value of a property. Lum. Tu and Wong. .

22 Literature Review .

23 Literature Review .

24 Literature Review .

6 Summary A detailed discussion of the extensive literature regarding housing markets and its dynamics was carried out in this section. while local authors emphasized on the empirical front and covered heavily on the existence of the heavily regulated public sector and its spillover effects onto the private housing sector.25 Literature Review 2. . It is observed that the international authors focused on the theoretical front.

1 Introduction The stock-flow model serves as a theoretical basis for the fundamental determinants of real estate prices and constructions. This chapter offers a detailed discussion of the structure of the Singapore private housing market within such a theoretical framework (refer to Figure 3.26 Market Review of Singapore Private Residential Market CHAPTER THREE│MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET 3. .1). The urban spatial framework by DiPasquale and Wheaton (1994. The role of the government in the housing market and the interactions between the public and private housing sectors are also emphasized. 1996) enhanced the traditional stock-flow model with a price adjustment mechanism and also take into consideration the interactions between the real estate space and asset markets.

1 – Structure of the Singapore Private Housing Market (Source: Author’s Compilation) .27 Market Review of Singapore Private Residential Market Figure 3.

the new units are priced at subsidized rates – with below market mortgage rates and various housing grants. strict eligibility criteria and exit conditions exist such as a minimum income ceiling. Furthermore. The statistics tell of a successful housing programme. Till date. making it an active owner-occupied housing sector compared to the new public housing .573 units in 2008 (Department of Statistics.2 General Structure of the Singapore Housing Market Due to the manifestation of the government’s commitment to provide decent housing for its population.28 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET 3. 2009). 2009). The HDB resale sub-market on the other hand. the local residential real estate market is dominated by the public housing sector. The prices of the resale units are largely determined by market forces. The government does not participate directly in the real estate market but rather through a combination of a strict supervisory role and the implementation of polices. 80% of the resident population owns their own HDB flats under the Home Ownership for the People Scheme which was introduced in 1964(Department of Statistics. family nucleus size and maximum occupancy period. an estimated 82% of the resident population is living in HDB flats. the total number of public housing units built and managed by the HDB since its inception amounted to 990. To encourage home ownership. However. requires minimal restrictions.

2 illustrates a simple housing pyramid structure of the Singapore housing system in Singapore as of 2008.29 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET segment. Figure 3. 2009) Figure 3. where the units are sold at heavily subsidized prices (Ong and Sing. Above that in ascending order are the non-landed private housing properties and finally the landed properties. 2002). By targeting the high income households (who did not satisfy the minimum income criteria set by the HDB). the image of the private housing sector is one of an elusive and prestigious lifestyle. The private housing market is another active owner-occupied housing sector where the prices are freely determined. . the expatriate community and foreign investors.2 – Pyramid Housing Structure of Singapore (as of 2008) (Source: Department of Statistics. The two largest stratums comprise households who live in public housing units.

the prices of the different types of housing increases up the pyramid structure. suggesting that the property prices in each stratum is supported by the property prices in the stratum below it. the transaction volume for DBSS have slowed (as shown in Table 3. A set of major government policies concerning the property market is listed and explained in Appendix Two. attracting a significant proportion of homeowners to upgrade to private housing. However. Build and Sell Scheme (DBSS) projects were introduced by the HDB in 2005 (HDB.3 Spillover effects from Government Policies and Interventions Three broad areas of involvement from the government includes the implementation of policies concerning public housing.1).3.30 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET As noted by Tu (2004). 3. the objective of becoming private home-owners has become a national phenomenon. the use of CPF for housing and the release of state land for private housing developments. . Therefore. especially for the projects that command a higher level of pricing.1 Pricing and Supply of new HDB flats To improve its quality of public housing and provide Singaporeans with another choice to meet their needs and aspirations. the Design. 3. 2009).

some of the restrictions for homeowners were eventually waivered off. As a result.3. 2009) 3. the pricing for DBSS projects are generally set at a high level. For example. the 5-room flats are the only category left unsold in all the DBSS projects due to the high pricing. the pricing of these new HDB flats can also cause a corresponding change in HDB resale units and subsequently the private housing units. For example. despite the potential homeowners being subjected to eligibility conditions such as the $8.000 income ceiling and the 5 year minimum occupation period. under the HDB Liberalisation of Finance . (Source: Cheam. Such high pricing and restrictions may push the demand towards the HDB resale sub-market and the private housing mass market instead.2 Changes in eligibility criteria for public housing As the public housing program started to take off and was met with success.31 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Furthermore.

the price appreciation in the public resale housing may enhance the owners’ affordability level to upgrade to private property. opening the market to permanent residents and private property owners who had to occupy their HDB properties (Phang and Wong. 1997). the income ceiling restriction was removed for HDB resale flats. For example.32 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Terms scheme. such flexible changes in the eligibility criteria for resale units further contribute to the dynamics of the resale housing market and subsequently a spillover effect on the private housing market. For example. In addition. HDB resale flat buyers were able to obtain higher mortgage loans (up to 80 per cent of the market value or resale price. The lending rates offered to resale flat buyers were also at much better discounts compared with the commercial mortgages offered in the market (Lum. As the prices of resale HDB units are largely determined by market forces. whichever is lower) than the previous stipulation of a much lower quantum. 2002). the minimal eligibility criteria for entry to the HDB resale sub-market were made less stringent over the years. The housing needs of single person households were also consistently looked into with the implementation of the HDB Single Citizens Scheme in 1991 and a subsequent revision in 2005. This is .



reflective of the upgrading hypothesis covered in previous empirical studies (Tu and Wong 2001; Ong and Sing, 2002).

As further supported by Choo (2000), ceteris paribus, private housing becomes more affordable to public home owners when the gap between private housing prices and public resale prices becomes narrower. This is reflected by Figure 3.3 which illustrates the movement of the price indices of the public resale housing sector and the private sector since 1990 Q1.

Figure 3.3 – Price Movements of the Public and Private Housing Markets (Source: URA, HDB; 2009)

From Figure 3.3, the price index of HDB resale flats in the public sector grew throughout 2008, and started to fall only in Q1 2009 - by a mere 0.8



per cent quarter-on-quarter (qoq) after nine quarters of growth. There is a further increase of 1.4 per cent qoq for the second quarter. It is clear from the graph that the performance of the private residential sector is largely affected by the market conditions in contrast to the HDB resale sector. The price gap between the two sectors had narrowed; largely due to the pricing strategies of private condominium developers to attract the pentup demand from the rising proportion of HDB upgraders (refer to Figure 3.4).

Figure 3.4 – Proportion of HDB upgraders buying Private Property (Source: URA, 2009)



3.3.3 Central Provident Fund Policies

The rules governing the use of CPF funds were extended to the payment of mortgages for private housing in 1981 (the Approved Residential Properties Scheme) and subsequent easing of CPF regulations were announced for the purchase of both private and public property and other investments. For example, the implementation of the CPF Liberalisation policy in 1993 allows buyers of public housing units to pay only the deposit (20 % of purchase price), freeing the remaining balance of their CPF for the monthly mortgage payments and other financial investments. It was also possible for HDB flat-owners who have fulfilled the minimum occupancy period to use their excess CPF savings for investments in private residential properties since late 1991 (Lum, 2002).

The extensive use of CPF savings on housing purchases however started to result in the increasing but worryingly trend of asset-rich, cash-poor situation for many home owners by their age of retirement (Tu and Wong, 2002). The reduction of usage of CPF savings scheme was implemented in September 2002 to solve this problem, which implies that homebuyers have to pay out more cash for the remaining months when their CPF drawings reach their limit.

3. can contribute to a rapid appreciation of house prices. through its sale of sites programme makes available state land via auction or tender to the private sector for the development of new private residential developments.5 illustrates the timing and number of land sales (measured in square metres) and the growth rate of private real house . This move makes the policy aligns with that for the purchase of HDB resale flats by non-related singles jointly using their CPF savings. CPF members can use their CPF savings to purchase private residential properties with remaining leases of 30 to 60 years instead of the original limit of 60 years. This change ensures that the lease can last the average life expectancy of home buyers (CPF. Rationing the land market has an impact on the ability of the private market to adjust the supply of housing units to accommodate demand changes.36 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET From 2005. 2005). whereas too much supply can contribute to a slump in housing prices if sentiments are weak. A too restrictive control over the release of land however. The government. Another change is to allow non-related CPF members to jointly purchase private residential properties. Figure 3.4 Release of State Land for Private Housing Development Land in Singapore is subject to a high degree of institutional control.3.

More recently. the proper timing of the land sales programme certainly play a pivotal role in providing flexibility for the real estate market to adjust supply in accordance with the current economic conditions. The timing of these land sales is arranged such that the general economic situation is taken in consideration. sent the private residential market into the slumps (Sing. The low / (high) level of land sales seem to be followed by an increase / (decrease) in real house prices with a one to two quarters lag. In 2001 however. where it appears to be a correlation between the two. the weak market sentiment. . the land sales programme (confirmed list) was suspended on October 2008 and extended till the second half of 2009 amid the poor sentiments. For example. Therefore.37 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET price changes.000 units to 7-8.000 units to ease the price pressure created by strong market demand (refer to appendix Two). coupled with the oversupply situation. 2001). one of the effective antispeculation measures in May 1996 was to step up the land sale program and increase the supply of private housing from 6.

the Tightening of HDB mortgage financing scheme was implemented in .5 Property Market Cooling Measures Given that public housing had become a highly speculative commodity due to the heavily subsidized prices for new units and subsidized mortgage rates (Tu and Wong.5 – Government Land Sales and Growth in Real Private House Price (Source: URA. 2002). property gain tax and increment of loan-to-value limit.3. In addition. 2009) 3. increase in release of housing units. the anti-speculation policy were implemented in 1996 for both private and public housing markets which included measures such as stamp duty.38 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Figure 3.

but rather gradually. which also tend to be persistent. DiPasquale and Wheaton. In October 2007. this also indicates that house prices depend directly on economic and property specified variables. The changes in real prices appear to be persistent and reflective of inefficiencies in the market (Case and Shiller. the government put a halt to the interest absorption scheme (IAS). . In addition. such as employment growth or changes in personal income. interest only loans and reinstated the GLS confirmed list to prevent property asset-price inflation in a market that had barely recovered from the property downturn in 2008. 1994).4 Economic Mechanisms and Price Dynamics of the Private Residential Property Market Figure 3. 3. the Deferred Payment Scheme was removed to prevent overheating in the buoyant market. It reflects that the market takes times to clear or prices and expectations of future price changes are set in a backward looking manner (myopic expectations).6 illustrates the changes in real house price for the private residential sector. In September 2009. 1989. Such persistency means that the positive price changes tend to be followed by more positive price changes and vice versa for negative price changes. asset prices do not adjust immediately to reflect new information about fundamental value.39 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET April 1997 to further tighten the eligibility criteria for subsidized mortgage loans to homebuyers. Hence.

6 –Growth in Real Private Housing Price (Source: URA. 2005).40 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Figure 3. Malpezzi and Wachter. 2005).4. expectations of future price changes are set in a backward looking manner (myopic expectations) as noted in past empirical literature which finds evidence for these myopic expectations (Ott. In the model adopted by this study. Yi and Yoshida. Riddiough. 2000.1 Simple Model of the User Cost Real estate price and rent growth expectations are central to the efficient pricing of real estate (Malpezzi and Wachter. 2009) 3. .

property tax rates. it will benefit the owner when he sells the house (Krainer. 1984). the user cost is a marginal rate of substitution). . 2002). Expected appreciation serves to lower the user cost.41 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Expectations of future price changes play a pivotal role in the typical specification of the user cost which also includes the mortgage rates. the user cost measures the willingness of a household to trade off housing consumption for nonhousing consumption over the course of a single period (i. because. Decreases in user cost are believed to accompany increases in the demand for housing and vice versa. 2004). Technically. Empirical studies conducted of the Singapore private housing market in the 1990s show the importance of real mortgage rates and the expectation of future housing price appreciation in the dynamics of the user cost of housing capital (Tu. a maintenance costs and a deduction for depreciation.e. if that appreciation eventually comes to pass.. Many international studies on housing demand recognize the significance of the user cost of housing capital as an important determinant of financing costs (Poterba.

7).7 –User Cost and its Key Components (Source: Author’s Compilation. any accelerating house price appreciation therefore reduces the real cost of housing and may stimulate property investment as buyers increase their consumption of housing services in anticipation of future capital gains. 2009) Judging from their inverse relationship (refer to Figure 3. Indeed. It was a similar situation leading to the boom period of 2007 where .42 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Figure 3. Tu (2004) finds that fast growing real housing prices in the first half of the 1990s significantly enhanced expectations for future housing price appreciations and resulted in negative user costs in the mid 1990s.

The implementation of the anti-speculation measures .299-300). Such speculation leads to real estate cycles even when there is no cyclicality in the underlying demand and supply fundamentals (Malpezzi and Wachter. 2005). Such a situation corresponds to the definition of a bubble as defined by Stiglitz (1990) and mentioned in the literature review. and whether there is sufficient confidence in such expectations to motivate action” (pp. The property market started to overheat in 1996 as private home prices soared beyond their valuation price. whether these expectations are salient enough to generate anxieties among potential homebuyers. basing price expectations on extrapolation of past price increases will likely lead to classic speculative bubbles. investors are speculating on a continuation of past high rates of price appreciation. Property buyers were borrowing beyond their means to enter into the market based on myopic expectations. Further reinforced by Case and Shiller (2003). “The basic questions that still must be answered are whether expectations of large future price increases are sustaining the market.43 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET strong expectations for future price appreciation play a role in the strong demand for housing. According to Malpezzi and Wachter (2005).

There are reasons to believe that many Singapore households and potential homebuyers revised their expectations of lifetime labour income based on the economic outlook. which in turn depend on the economic well being of the country (Ong and Teck. 3.44 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET coupled with the Asian Finanical Crisis hit many players hard as they were burnt by the high mortgage debts and user costs. The booming period of 2007 was reminiscent of the events in 1995. thus leaving an impact on housing prices.4. A prosperous economic outlook can then boost the potential investor or homebuyer’s confidence in the private property market. . 1996).2 Indicators of Macroeconomic Performance Income is an important determinant of residential price movements. The government is quick to act this time around just as private residential prices are starting to soar again.

45 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Figure 3. As noted by Sing (2001). real (inflation-adjusted) house price changes became negative with GDP growth during both recessions (1997 and 2008). The macroeconomic performance is also reflected by the contribution rates towards the CPF.8 –Performance of the Economy and the Property Market (Source: Department of Statistics. It can cause demand for private housing to .8 shows. the private housing market activities are found to be highly dependent on the lagged period economic performance. the performance of the stock market and the general rate of unemployment. 2009) As Figure 3. Unemployment rates are regarded as another indicator of the long-term macroeconomic performance.

46 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET drop significantly. price declines will only increase. slowing down the price appreciation. . An overheated market with price appreciation that had increased way beyond fundamental growth will force marginal buyers and badly burnt players out of the market. Coupled with job growth slowdown or increasing unemployment in a recession situation. while at the same time. This was observed during the current downturn period where job growth slowdown and unemployment rates were increasing amid a recession (refer to Figure 3. demand falls off sharply. With their exit out of the market. reduce household incomes and increasing the supply of housing on the market by desperate sellers (current homeowners).9).

increases in the housing prices often serve as a signal to developers about the current market sentiments.47 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Figure 3. 2009) 3. As seen from Figure 3.9 –Performance of the Economy and the Property Market (Source: Department of Statistics.5 Housing Commencements and Development Costs Rosen (1979) considers the residential property developer as a passive player who reacts to the market in a logical way.10.undertaking a project when demand is high and reducing construction when demand is poor and development costs are high. and the developers will therefore increase the supply of housing to meet the demand. the number of housing units under the building commencement . Thus.

Since the last downturn in 1997. the rising prices of private residential properties have increased the number of new housing units granted for commencements. In addition to the housing prices. As can be observed during the last boom period of 2007. wages and financing are expected to have a direct impact on the supply of new housing as higher development costs and financing costs may reduce new housing construction. .48 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET stage is strongly correlated to the housing prices. the development costs such as construction costs. As the global financial crisis developed in late 2008. the market sentiments were deeply affected and demand for housing dropped. developers started to increase the number of development projects in respond to the appreciating prices. This was reflected by the falling housing prices as well as the construction costs.

public residential price and the spillover effects from government policies. development costs. gives a basic understanding of the interactions between the two local housing sectors in Singapore and the influence of certain macroeconomic and property-specified determinants on the private residential price. .6 Summary The discussion covered in this chapter. to name a few.10 – Construction Costs and Housing Commencements (Source: Author’s Own) 3. income growth. user cost of capital. employment growth. macroeconomic performance.49 MARKET REVIEW OF SINGAPORE PRIVATE RESIDENTIAL MARKET Figure 3. The determinants are. coupled with the literature review.

4. are identified and shown in Table 4. The principle task of adopting these methodologies is to establish a long-run relationship and short-run dynamics between the macroeconomic and propertyspecified variables and real private residential price. cointegration test and error correction model.1. although not extensive.1 Introduction This chapter illustrates the various empirical methodologies used in this study. several important macro-economic determinants of house prices. .2 Data Description Based on the literature and the private residential market dynamics reviewed in the previous chapters.Research Methodology 50 CHAPTER FOUR │ RESEARCH METHODOLOGY 4. which comprises the unit root test.

Research Methodology 51 .

Research Methodology 52 .

Research Methodology 53 Not all the available data series is in consistent quarterly form. where appropriate. CPI and unemployment rate. prices adjust to equate. The real GDP growth rate is calculated on a year-on-year (y-o-y) basis as this method removes the trend and seasonal components. Seasonality is considered for selected variables such as GDP. facilitating the examination of the data characteristic. Public policies will be introduced in the error correction stage as dummy variables to analyse their impact on real house price changes (with 1 denoting their implementation from the respective quarter onwards and 0 denoting prior to their implementation). Similarly. The two quarterly property price indices are deflated by CPI to account for inflation. the monthly data is converted into quarterly data and interpolations are made between annual points. The natural log transformation is applied to selected variables to linearize the exponential trend and enable the interpretation of elasticities for their resultant estimates. A single model for house . Following the work of Tu and Wong (2002). the stock-flow model assumes that the housing market clears quickly. 4. the various mortgage rates are also adjusted. the user cost of capital is determined using the difference between Finance Company Housing Loans for 15 Years and the price expectation of nominal PPI for 4 quarters. and at any time t.3 Model Specification Traditionally.

the existence of a long-run . Nsi.t.t* = Qdi.3) to get the single.t) (4.t . because price is a determinate of both supply and demand. Isi.Research Methodology 54 price can be reduced from two parallel equations for supply (Qs) and demand (Qd).1) Qd i. Thus. which are shown for any market i at time t as: Qs i. (4. Idi.t. Is and Id are sets of exogenous economic variables respectively influencing supply or demand. t*. Zdi ) = f(Xi) (4.t. Ndi. or Qsi. Zdi.Zsi.Qdi) = f(Isi. and there is an equilibrium.4) There is substantial empirical evidence from authors such as DiPaquale and Wheaton (1994. Following the basic theory.2) into (4.1) and (4. Ndi.t* = f(Qsi . Idi. 1996) showing that it may take a long time for the housing market to adjust to the equilibrium. whereby supply equals demand. Ns and Nd are sets of demographic variables.t* which allows us to substitute equation (4.t.3) there is also an equilibrium price.2) where P is the current housing price.t.t = f(Pi. Pi. Nsi.t = f(Pi. reduced-form model for house prices at their long-run equilibrium: Pi.t) (4.t*. and Zs and Zd are sets of other explanatory variables. Zsi.

4 Concept of Stationarity and Unit Root A key concept underlying the development of time series model is the concept of stationarity. a time series Xt that was differenced d times is therefore. The question of whether a time series is stationary depends on whether it has a unit root. the independent variables are required to be tested for the presence of stationarity and integrated of the same order. 1991). As non-stationary time series will result in incorrect conclusions. 4. Prior to the cointegration test. are trended and non-stationary. As such. The transformation of a non-stationary time series can then be carried out by differencing the time series d times and subsequently. 1987) and denoted by Xt ~ I(d). A stationary time series is one where the mean and autocovariances do not depend on time . it is necessary to transform it into a stationary time series as proposed by Box and Jekins (1970). integrated of order d (Engle and Granger. the order of integration of the time series can also be determined.the underlying stochastic process of the time series is assumed to be time-invariant (Pindyck and Rubinfeld. The .Research Methodology 55 equilibrium relationship between housing prices and fundamental variables can be determined by cointegration analysis. Subsequently. Most macroeconomic time series however. the speed of price adjustment to this equilibrium together with other short-run dynamics of housing price movements can be analyzed based on an error correction model.

1995). The optimal lag length can be selected by basing it on the smallest Akaike Information Criterion (AIC) values (Akaike. 1973) or Schwarz Information Criterion (SIC) values.Research Methodology 56 Augmented Dicky-Fuller test (ADF) (1979) and Pillips Perron (PP) (1998) unit root tests are the formal approaches employed in this study. Three regression equations are considered to test for the presence of a unit root (Enders. . the data has to be further examined for the presence of a unit root in their first difference form. Another approach is to estimate a model using a large lag length and reducing it down by re-estimating the model by the usual t-test or F-test until the lag length is significant at the specified critical value (Enders. The ADF test is sensitive to the choice of lag length and sufficient lags are required to be included so that the error term can be assumed white noise in relatively short lags. where the null hypothesis is the presence of a unit root (H0: ψ=0) against the one-sided alternative of no presence of a unit root (H1: ψ<0). 1995). If the null hypothesis of a unit root in the series cannot be rejected at 5% significance level.

k is the number of lags and a0. The null hypothesis testing is similar to that of the ADF. ψ and θ are the regression parameters. . The data has to be further examined for the presence of a unit root in their first difference form. if the null hypothesis of a unit root in the series cannot be rejected at 5% significance level. The test is carried out using the PP procedure shown in the following equation. a2.Research Methodology 57 Where Δ represents the differencing operator. Yt = a0 + α(t.T/2) + ψ Yt-1 + εt Where (t-T/2) is a time trend. εt is a independently and identically distributed error term. with T being the sample size and εt the error term). The PP test uses a non-parametric adjustment to the Dickey-Fuller test statistic which allows for dependence and heterogeneity in the error term.

1988. where the two well known techniques are the Engle and Granger ordinary least squares (OLS) method (Engle and Granger. uncertainty in the diagnostics that normally arises in OLS estimators as with the Engle and Granger approach is avoided. The Johansen approach is most preferred in many empirical studies as it has better properties than other approaches including the Engle and Granger approach. The hypothesis testing with respect to the cointegration vector is conducted using the standard. Although these variables may in the short-run diverge. Consider the following vector autoregressive model of order p: Xt = A1 Xt-1 + A2 Xt-2 + … +Ap Xt-p+ εt . 1987) and the Johansen Maximum likelihood method (Johansen. asymptotic chi-square tests. the economic forces will eventually bring them together in the long-run (Granger. 1986). 1989). It uses the maximum likelihood based test which determines the number of cointegrating vectors in a vector autoregressive (VAR) process.Research Methodology 58 4.introduced by Granger (1981) and Engle and Granger (1987) holds that certain economic variables should not diverge from each other by too far a distance.5 Concept of Cointegration The concept of cointegration . Thus. There is a fair extensive formal literature developed on cointegration methodology.

However. In the case of rank (π) being equal to 1. . 1995). If π is of rank n. it may be transformed into a more usable form (Enders. there is a single cointegrating vector and πXt-1 is the error-correction term. and it shall be noted that the (λmax) test is most preferred for identifying the number of cointegrating vectors (Enders. the matrix is null and the equation is the usual VAR model in first differences. 1995). For the purpose of this study. When rank (π) is equal to 0. Where and The rank of π is equal to the number of independent cointegrating vectors.Research Methodology 59 Where Xt represents a k-vector of non-stationary I(1) variables and εt denotes a independently and identically distributed error term. The number of distinct cointegrating vectors can be obtained by using the trace test (λtrace) or maximum eigenvalue test (λmax) to check the significance of the characterisitic roots of π. the vector process is stationary. the results of these two tests can conflict.4 to provide firm evidence on the variables which are significant in explaining the long-term movement of house prices. the Johansen multivariate cointegration test is applied to equation 4.

In accordance with the study’s objectives. an ECM representation among the variables concerned will always exist. Following equation 4. At any time period t. the Pt* long-term equilibrium value and ∆ Pt. house prices are represented as Pt = Pt* + ∆ Pt. Subsequently. It is stated in the theorem that if a set of I (1) variables is cointegrated.4 where Pt* = α Xt With α being a vector of coefficients and Xt is a vector of independent variables. the short term deviation. with Pt being the actual house price.6 Concept of Error Correction Model (ECM) Granger (1981) first introduced the relationship between Error Correction Mechanism (ECM) and cointegration in the Granger representation theorem.Research Methodology 60 4.5) . the time series can experience short-run disequilibrium and move around the long-term equilibrium path. a long-run equilibrium relationship between the real private housing price and the independent variables is first approximated. The short term dynamics around the long-term trend can be expressed as: ΔPt = βΔYt -πECMt-1 (4.

4 and 4. ΔYt is the first difference of the vector of explanatory variables which are postulated to influence the house price in the short term.Research Methodology 61 Where ECMt-1 = Pt-1 – Pt*-1 is the ECM representation. The relationship between Pt and Xt in its first differences form and the incorporation of an ECM specification can be expressed as: ΔPt = βΔYt -πECMt-1 + Фt (4.7) . It shall be noted that the independent variables captured by the vector. (postulated to influence the house price in the longterm) need not be the same as Yt as the time lagged effects of the variables on the house prices need to be taken into consideration (Munro and Tu. Xt. 1996).6) As the error term еt is stationary. this implies any deviations of Pt from its long run path Pt* are also stationary. the empirical estimation of equation 4.5 is approximated as follows: Pt = α Xt + еt (4. and the coefficient π representing the speed of adjustment in cases of disequilibrium. As Pt* cannot be observed in practice. The ECM term should be negative to ensure that the dynamic of house price adjustment is reverted back towards the long-run equilibrium trend.

7 Summary A two-stage approach is used in the application of a Cointegration ECM . Equations 4.6 and 4.7 are used to model the Singapore housing market in order to investigate the long-run price trend and short-run dynamics respectively.the first to estimate equilibrium price as a regression of house price on fundamental factors and the second to regress house price changes as a function of the current distance from the equilibrium price estimated from the first equation.Research Methodology 62 4. .

1 for description of variables).1) denote the percentage change in long-run equilibrium private house price in relation to the percentage changes of any one of the independent variables. The coefficients in the long-run model (Equation 5.Data Analysis and Estimation Results 63 CHAPTER FIVE│DATA ANALYSIS AND ESTIMATION RESULTS 5. The long-run and short-term dynamics equations of the Singapore private housing model are as follows: (5.1) Where α0 represents the intercept. The analysis covered in this chapter is applicable only to the selected variables (see Table 4. not all the variables discussed in the previous chapter are included in the empirical specifications.1 Introduction Due to data unavailability and the need to maintain parsimony. ceteris paribus. Unit root tests. ln the natural log of continuous variables and t is the given time period. model fit and cointegration tests of the other variables are available at request. .

As house price continue on its equilibrium path at time t-1. ECMt-1 the error correction mechanism. ΔLNPPIt-1 the serial correlation term. The subsequent test is conducted in first difference form with intercepts only. the coefficients in the short-run model (Equation 5.2) denotes the short-term percentage change of house price in response to any percentage changes of the independent variables. the coefficient π representing the speed of adjustment. following any shock to the housing market. 5. . Appendix one shows the visual plot of the time series variables which confirmed the assumptions.2 Stationarity Test and Order of Integration The Augmented Dickey-Fuller (ADF) test and the Phillip-Perron (PP) test are performed to test for the presence of unit root and verify the order of integration for all the variables prior to the cointegration test. and D1 to D4 being the policy variables.2) Where Δ denotes the difference operator.Data Analysis and Estimation Results 64 (5. at time t. The tests are carried out for all variables with an intercept and a time trend to account for the positive drift in their growth levels in the level form. except for unemployment rate and user cost of capital. and other variables held constant from t-1 to t.

(Source: Author) .Data Analysis and Estimation Results 65 Lag lengths are important as most housing markets are inefficient markets and time is needed for the market to adjust to all relevant information. However. the KPSS test for stationarity was conducted and concluded that the series is non-stationary in its level form and stationary after first-difference form.2 report the results of the ADF tests and PP tests for all the variables in their level and first differenced forms. the number of lags was not specified so as not to over-restrict the test. All the series are integrated of order 1 – I(1). Hence. the lag length presented in the results is automatically selected by using the Akaike Information Criterion (AIC) over a maximum lag length of 4. As the PP test for the LNPPI series contradicts the ADF test.1 and 5. Table 5.

Data Analysis and Estimation Results 66 (Source: Author) .

Model two is denoted as the appropriate model.4. Following Johansen (1995) where a maximum of two lag length is appropriate for macroeconomic data. (Source: Author) The Pantula principle (Johansen. the lag length of two is selected instead. The Schwarz Information Criterion test indicated a lag length of three. which is reaffirmed by the AIC and SIC . It involves the estimation of the three models as shown below and the selection is based on concluding at the model where the null hypothesis of no cointegration will be rejected for the first time. The highest number of lags considered was nine and the lowest was zero (refer to Table 5. 1992) is next applied to select the appropriate model (refer to table 5.3).3 Specification Tests for Model Selection An unrestricted vector auto-regression (VAR) is run to establish the maximum lag. with the other tests indicating lag lengths of larger than three.).Data Analysis and Estimation Results 67 5.

In addition. (Source: Author) (Source: Author) 5.).5). The trace eigenvalue test statistic and the max eigenvalue test statistic indicates the presence of three cointegrating vectors for the relationship of the group of six variables over the full period from year 1990Q1 to 2009Q1. .4 Johansen Cointegration Test The selection of model 2 produced the following results (refer to Table 5.6. the model specified in this study includes a constant term and model two therefore seems a reasonable model to be used.Data Analysis and Estimation Results 68 (refer to Table 5.

7.Data Analysis and Estimation Results 69 (Source: Author) The results of the PP unit root tests for the corresponding estimated residuals from this equilibrium equation are reported in table 5. This implies that the residuals from the equilibrium regression are stationary and the variables are cointegrated of order one. the null hypothesis of a unit root in the estimated residual sequence can be rejected. Since all the test statistics are beyond the 5% critical value. (Source: Author) .

a new cointegrating vector that excludes one of the series can be obtained. The second and third cointegrating vector does not contain any new information about the long term relationship.8) have the right coefficient signs according to real estate economic theory. The long-term relationships can be interpreted as partial equilibria. All three cointegrating equation are normalized with respect to LNPPI. restricting the insignificant unemployment rate variable will derive cointegrating equation two where the relationship between real house prices and the three other variables is consistent with cointegrating equation one.Data Analysis and Estimation Results 70 All the variables in the three identified cointegrating vectors (Table 5. since a linear combination of the cointegrating vectors is also a cointegrating vector. (Source: Author) . For example. which reflects the contemporaneous relationship between real private housing price and the independent variables. However. The disappearance of one variable naturally changes the elasticities of the remaining four. It is useful to eliminate one of the variables between the first two eigenvectors and test for a new relationship in order to verify if the relation obtained previously is robust.

which indicates little ‘noise’ in the estimates. Although this finding is different from Tu (2004). 2002). The coefficient is statistically significant. This finding is especially reflected during the first quarter of the year where the price of the public resale flats had increased in contrast to the negative performance of the private sector (refer to chapter three). All the coefficients are statistically significant.0770%.Data Analysis and Estimation Results 71 The results show that the long-run movements of real private housing price is better explained by the real public resale housing price. where this variable is found to be significant only in the short-term price dynamics. total private housing stock and the user cost of capital (Cointegrating Equation 3) at the 5 per cent significance level. a 1% decrease in the user cost of capital will boost real house prices by 0.9078%. The long-run parameters suggest that a 1% increase in the pricing of HDB resale flats will boost real private house price by 0. The magnitude of standard errors for these variables are small. it is still reinforced by the earlier section (Chapter Three). less than two. its influence is . This reflects the upgrading hypothesis where home-owners in a land-scarce economy with a dominant public housing market are able to upgrade to private property as a result of capital appreciation on their public flats (Ong and Sing. Although the coefficient is statistically significant. which points out that upgraders from the public home-owner group form a major part of the demand for private housing. Further on.

any further increases in the housing stock serves to ease the demand for private housing. 2004). The small coefficient of 0.2786% drop in real housing prices. The total housing stock variable provides information for the supply side of the housing market. However.29 for total private housing stock. as increasing user cost of capital makes it more costly for potential home buyer.2786 is much smaller than European estimates which indicates an elasticity estimate of between 2 and 3 per cent (Meen and Andrew. With a rapid development in the private housing market. 1998). Tu (2004) finds this variable to be more significant in explaining the short-run dynamics of private housing price. It is postulated to possess a negative sign which is consistent with the empirical findings.Data Analysis and Estimation Results 72 likely to be small. It is thus a major barrier for potential homebuyers to enter the local private housing market as it has been significantly higher than the user cost for public housing owners (Tu. The estimated equation one suggests that a 1% increase in the y-o-y growth rate of real GDP increases real housing prices by 0. The small influence can be attributed to the large influx of the expat community and the en-bloc fever which serves to continuously boost up the demand during the booming period of 2007. As suggested by the model. This coefficient is .012%. The user cost of capital has a negative impact on real house prices. a 1% increase in housing stock may result in a 0. the estimate from this variable is found to be small – Tu (2004) indicates a coefficient of 5.

a better indicator for this study could be the average monthly earnings of the upper tier of the population (eg: the professionals) who is often the target market of the private residential sector. For every 1% increase in the unemployment rate. unemployment rate is statistically insignificant in equation one and subsequently removed.569 for real GDP. on the other hand. This is in contrast to other local studies . However. the real GDP coefficient for equation two is not significantly different from zero. gives an indication of the broader economic performance. Tu. A possible explanation is that the private residential market is largely catered to the higher end market and only comprises less than 12% of the total local housing market.94 for real GDP per capita and Ho and Curevo finds a coefficient of 0.Data Analysis and Estimation Results 73 significant only at the 10% level and unlikely to have a large impact on real housing price. 2001. . As 82% of the population lives in public housing. the real housing prices are expected to decrease by 0.0147%. the negative coefficient is consistent to economic theory. 1999). 2004. Although considered as an explanatory variable of real housing price (Sing. Ho and Curevo. the general unemployment rate may not be an adequate indicator for the demand of the private housing market.Tu (2004) finds that a coefficient of 3. In addition. Finally. The y-o-y real GDP growth.

5 Error Correction Modeling The Johansen’s cointegration test estimates of the cointegration relationship imply that the long run equilibrium does in fact exist. The ECM term derived from the cointegration vector is included in a restricted VAR regression model to derive the short-run model.2. The current changes in real housing price are estimated using the error correction term. The speed of adjustment is expected to have a negative sign. 1996) and can include lagged and dummy variables. The specification of the short-run model for real private housing price is based on equation 5. two lagged values of the change in real housing price and four dummy variables as the exogenous shocks. .Data Analysis and Estimation Results 74 5. two lagged values of the change in all the underlying explanatory variables. This model is not required to have the same explanatory variables (Tu and Munro. the error correction mechanism will be triggered automatically. When the actual housing prices deviate from the specified variables.

Data Analysis and Estimation Results 75 (Source: Author) The error correction equation residuals are evaluated using standard diagnostic tests. The White heteroscedasticity test (no cross terms) and the Arch test indicate the absence of heteroscedasticity at . The results of the final model are presented in table 5. The serial correlation LM test displays no indication of residual autocorrelation at the 5% significance level.9 with the statistical diagnostics which indicates a satisfactory performance from the estimated model.

Hort. a yearly estimate of -0. 1993. This estimate is comparable to the estimate of -0.05 was reported by Abraham and Hendershott. In comparison with international findings.028. The estimated standard error of regression is very low at 0.47 to . For the US housing markets however.3% of the deviation of real house prices from its fundamental value last quarter is removed in the current period. .034 reported by Lum (2002) but lower than the estimate of -0. Munro and Tu.107 and -0. the yearly error correction term estimates for the European housing markets fall in the range of -0. this model can explain changes in the private residential property prices accurately. 2000) reported for the Australian housing market. 1998). This coefficient is an indication of the speed of adjustment by real housing price towards their longrun equilibrium level following a shock to the market.0.13 reported by Tu (2004) and Ho and Cuervo (1998) respectively for the Singapore private housing market. In addition. 1984. Therefore.033) is robust and significant at the 5 per cent level. The Ramsey RESET test is unable to reject the null hypothesis at the 5% significance level that the correct specification is linear and therefore the estimated model passes the Ramsey RESET test of misspecification. the high value of adjusted R2 at 0.87 (Hendry. it implies that only 3.Data Analysis and Estimation Results 76 the 5% significance level. The negative and statistically significant coefficient of the error correction term (-0. It is also lower than the estimate of -0.132 (Tu. This implies that within the sample. Drake. 1996.693 further indicates that this specification has a strong explanatory power for deviations in the private residential property price.

Further on. a subsequent Wald test on the null hypothesis that past price change is not a significant determinant of the short-run price adjustment process was rejected. house prices would react instantly to economic shocks and the degree of serial correlation would be zero. any comparison between the resultant estimates has to be careful (Tu.88) is significant and less than unity. However. et al. In perfectly efficient real estate markets.34 obtained by Tu (2004) and 0.25 to 0. Although the estimate of the serial correlation term reported by this model is unusually high (0. Case and Shiller (1989) estimates the annual serial correlation in their sample of four cities in the range of 0. there is a positive autocorrelation among the real private housing price. academic research has shown that this term is positive and economically and statistically significant (Capozza. with a computed test statistic of χ2 (2) distribution being 78.44 by Lum (2002).000).85).5. 2004). 2004). With the estimate being less than unity. The coefficient estimate of the serial correlation term (0. Abraham and Hendershott (1993) obtain an estimate of 0. this indicates that the market experiences a gradual price adjustment process to converge to the .33.Data Analysis and Estimation Results 77 There is such a wide range of estimates of the coefficient as the exact specifications of variables are different in each country or city-based studies and thus.590(p=0. Consistent findings were also reported by local studies with an estimate of 0.4 on a panel of 29 cities and Capozza et al (2004) reports a comparable estimate of 0.

708 (p=0. Although it is not significantly different from zero and may explain little of the short-run variations in the real housing price (0. with a χ2 (6) distribution of 5. The coefficient estimate of the total private housing stock is also significant and further indicates that increasing the amount of private housing stock has the desired impact of cooling the private residential market. unemployment rate and HDB resale price are not significantly influencing the price dynamics. The coefficient estimate on the one lagged change in the user cost of capital is significant at the 5% significance level. This suggests that in the short run. the GDP growth. the fact remains that the user cost of capital is a concern for potential homebuyers who desire to enter into the private residential market.Data Analysis and Estimation Results 78 equilibrium price during the study period. unemployment rate and HDB resale price was unable to reject the null hypothesis that they are not needed in the ECM specification at the 1% and 5% significance level.008).457). As mentioned before. This is consistent with the stock-flow theoretical model by DiPasquale and Wheaton (1994) who argued that the housing market is characterized by a slow market clearing process. The Wald test done on GDP growth. the average earnings of high income professionals could be a better explanatory such variable as GDP than general and macroeconomic performance indicators growth unemployment rate in explaining the short-run dynamics of housing prices as this target group makes up a dominant portion of the local private residential .

taxing gains and restricting home loans to 80 per cent of the property’s price – introduced in May 1996 and the removal of the deferred payment scheme (DPS) in October 2007 (captured by D1 and D4 respectively) has a negative influence on the short-run housing price (-5.0001).466 (p =0. This indicates that the policies introduced by the government have a significant impact on the shortrun price dynamics.Data Analysis and Estimation Results 79 market. The computed test statistic with a χ2 (4) distribution was 24. An unusual finding is the insignificance of the HDB resale prices. Similarly. which leads to a positive spillover effect (6. which is postulated to have a significant positive impact on the real private housing prices as the capital gain for HDB flat owners are able to aid them in upgrading to the private residential sector. The implementation of the anti-speculation measures – which included stamp duty on property sales. the implementation of a number of favorable policy changes in 2005 .15% and -4.12%). On the other hand.27%) in the private housing market in the short-run. A Wald test conducted on the null hypothesis that these policies are not significant determinants in influencing the house price adjustments to equilibrium levels was rejected. the implementation of the HDB resale flat credit policy and CPF liberalization policy in 1993 (D2) is an example of the government initiative to deregulate the public housing market. The coefficients on the government policies are all statistically significant and bear the correct signs as postulated.

Data Analysis and Estimation Results 80 (D3) – which included further easing of CPF rules.1. Impact of Policies on Private Housing Price Movement (Source: Author’s Compilation) . amendments to foreign purchases and ownership of properties and increasing home loans to 90 per cent of the property’s price – have the desire effect of boosting the property market in the short-run. The impact of these policies on the private housing price movement is illustrated in Figure 5.1. 1993 HDB Resale Credit and CPF Liberalization Policy 2005 MND Policy Changes 2007 Removal of DPS 1996 AntiSpeculation Policy Figure 5.

2). government intervention through the implementation of policies over the years often leave a long and sustained impact on private housing price. Cointegration Graph of LNPPI (Source: Author) As observed from the cointegrating graph of the empirical model (Figure 5.2. In addition. Favorable policies on the other hand.3%.Data Analysis and Estimation Results 81 Figure 5. It shall be noted that the implementation of these four policies cause the real private housing to further deviate from their adjustment path. this substantial price inertia may also be attributed to the adjustment speed to equilibrium level of 3. may destabilize the market by increasing the house price deviations from their adjustment path. yet its effects may be felt beyond the implementation period. Regulatory policies aid to smooth out any undesirable volatility in the market. .

the residuals appear to be relatively homoscedastic.Data Analysis and Estimation Results 82 Figure 5. From Figure 5.2 further implies that real house price has reached the equilibrium level in the first quarter of 2009 and is moving towards an upward trend.5 shows the performance of the model to forecast more than one period into the future. This empirical finding is further supported by the increased market activity in the second quarter of 2009 and the concerns by property analysts of yet another property bubble looming.3 gives the graph of the actual and fitted values of the resultant error correction model. 5.4. yet general trends of the model are still evident. with few outliers visible.6 Error Correction Model Estimation Figure 5. Furthermore. which shows a close fit. The government is quick to intervene with its preemptive measures (refer to Appendix Two) aimed at curbing any speculation just when the housing price is about to rise above its fundamentals. The sample period began from 1995 over the next 14 years. . small deviations from the means were observed and the model is observed to perform well in a one period forecast. It is then forecasted to test if it could predict one period ahead of actual values. Figure 5. Substantial deviations can be observed in this dynamic modeling.

Data Analysis and Estimation Results 83 Figure 5.3. Actual and Fitted Values of Model (Source: Author) .

Data Analysis and Estimation Results 84 Figure 5. Modelling a Static Solution (Source: Author) .4.

5.Data Analysis and Estimation Results 85 Figure 5. Modelling a Dynamic Solution (Source: Author) .

3 per cent of the previous discrepancies between actual and desired long-run private housing prices are corrected in each quarter. The Johansen methodology is carried out. the one lagged change in total private housing stock. one lagged change in real private housing price and government policies are significant in explaining for the changes in private housing prices. In the short-run. Further on. the ADF test and the PP test are conducted to ensure that the variables are all stationary after the first differences and are integrated of order one.Data Analysis and Estimation Results 86 5. which reports a long-term cointegrating relationship among real private housing prices. real public housing price.7 Summary Prior to the cointegration test. . user cost of capital and total private housing stock. the error correction term shows that 3.

A long-term contemporaneous relationship existed between real private housing price and real public housing resale price. . the one lagged change in total private housing stock and the implementation of government policies. The deviations in real private housing price in the short run is attributable to its own lagged value. The profits generated from the sale of public resale flats enabled the owners to upgrade to private housing.1 Summary of Main Findings The major findings from this study are summarized as follows. The number of HDB upgraders who made up the demand for private residential had been rising steadily since the fourth quarter of 2008. user cost of capital and the total private housing stock.87 Conclusion CHAPTER SIX│CONCLUSION 6. The long run cointegrating relationship between real public housing resale price and real private housing price indicates an interaction between the two markets as postulated. making this group of buyers a key target group for mass market condominium developments.



The price appreciation of public housing units raise the flat-owners’ affordability to enter into the private housing market. The demand for private property is thus increasing, enhancing the price level in the process. This is evident from the narrowing price gap between the two markets since the first quarter of 2009.

With an increasing number of HDB upgraders moving to the private sector, there is a transfer of public housing subsidies into the private housing market. This is because the HDB upgraders can finance their new private housing units through the profits from selling their subsidized HDB flats.

As the user cost of capital is measured by the difference between real mortgage rate and myopic expectation of nominal housing price inflation, its significance implies that future housing price appreciation is a crucial factor in explaining the long-run movement of private housing price. This finding is contrary to the a priori expectations of this study, where it should be significant in explaining the short–run deviations of real private housing price. It however indicates that, buyers and investors tend to enter the property market based on expectations of future capital gain.

In the short-run, the derived ECM is able to explain 69.2 % of the deviations in real private residential price and the resultant error correction term of -



0.033 is significant in adjusting the deviation in price back to the long-run equilibrium level.

Consistent with a priori expectations, the serial correlation term of the real private housing price is present in the short-run model. The rate of 0.88 implies that the Singapore private housing market does not clear itself immediately. However, this is not a gradual process, compared to international and local findings, which typically put this serial correlation term in the range of 0.25 to 0.5.

Government policies are found to be significant in explaining the dynamics of real private housing price. The policies are namely, the anti-speculation measures such as Stamp Duties and GLS program, favorable policy changes such as HDB resale flat credit policy, CPF liberalization policy, increment of loan-to-value (LTV) requirement, amendment of foreign ownership and the removal of the DPS.

Due to the slow adjustment speed of 3.3% for each quarter, any deviations from the fundamental housing values due to any shocks imposed by the government policies can persist for quite some time. As such, the impact of these government policies may linger on the market for a longer duration than expected. For example, the anti-speculation measures in 1996 managed to prevent investors from excess speculation and yet have a larger than



expected negative impact on the market, sending it into slumps as the period coincided with the 1997 Asian Financial Crisis.

The significance of the total private housing stock variable for both the longrun and short-run model indicates its importance in cooling the private residential market. This explains the intention of the government to reinstate the GLS confirmed list following the unexpected euphoria in the property market in the second quarter of 2009.

6.2 Research Implications and Contributions

The research findings have strong implications for policy makers. In the formulation of any policies to smooth out undesirable volatility, policy makers have to consider the resultant impact on the market and the timing of their implementation. This is largely due to the substantial price inertia characteristic of the local market as any policy implementation can therefore cause a sustained persistence in the deviations of housing price.

It is especially important in the current context as the government recently put a halt to the interest absorption scheme (IAS), interest only loans and reinstate the GLS confirmed list in a pre-emptive move to prevent property asset-price inflation in a market that has barely recovered from a recession. The current strategy of the government is totally different from the past - when the anti-

For the developers and investors. it may be time to rethink about the allotment of subsidies for public resale housing. With the narrowing gap between public resale and private housing price movement (Figure 3. As accelerating house price appreciation can reduce the user cost of capital and stimulate property investment. buyers tend to increase their consumption of housing services in anticipation of future capital gains.91 Conclusion speculation measures and removal of DPS was implemented months after the housing price is above its fundamentals. . This is typical of homebuyers who see their homes as an investment tool instead of consumption good. they need to treat the current euphoria in the private property market with the right attitude. a deeper understanding of the price dynamics of the private housing market enable them to respond to any changing economic conditions or policy changes in the market at the investment and development planning stage. A vicious cycle will start again as these buyers begin to speculate on a continuation of past high rates of price appreciation.3) – to the point that they have crossed each other. The pre-emptive move is timed to prevent any speculative bubble from forming at the very initial stage. This is essential due to the spill over effects onto the private sector. For potential homebuyers.

3 Limitations A number of limitations noted in the study are summarized here. which would have illustrated a more precise picture of the price dynamics in Singapore. 2008 global credit crisis). To maintain parsimony. Some of the empirical findings are contrary to the a priori expectations of this study (i.: the insignificance of HDB resale price in the short-run). 2001 terrorist attacks. 1997 Asian Financial Crisis. A number of structural breaks may exist in the study period (1996 boom.e. . Data availability limited the initial intention of this study to model the demand and supply sides of the local residential market. 2003 Sars epidemic.92 Conclusion 6. the constraints of this study limited the use of Gregory-Hansen cointegration approach in view of these possible structural breaks. the variables included in the empirical model are not extensive and may not have fully captured the entire characteristics of the market. However.

4 Recommendations for Future Research Future studies can look into the structural modeling of the demand and supply sides of the market to give a clearer picture of the underlying mechanisms. The linkages between the public and private housing markets can also be further studied via the Granger Causality approach. future studies can also follow up on the anti-speculation policies implemented by the government in September 2009 and analyse the impact on the property market. In addition. 6. . A more appropriate method would be to construct the price indices using the hedonic method and using it for the study.93 Conclusion The property price indices maintained by the URA and HDB are not quality controlled indices. the Gregory-Hansen cointegration approach can be applied to identify the structural breaks that would be evident in any study period. Finally.

Box. G. 7 . Goh. 191-207. Bank of England Quarterly Bulletin. Jenkins. & Shiller.. (1988). J. Journal of Housing Research. London: Pion. forecasting and control. P. 441-470 . Real Estate Economics. Breedon. & Mack. 32(1) . Time series analysis. 173-179. Baffoe-Bonnie. K.. F. Applied econometrics: a modern approach using Eviews and Microfit. & Cho. The Dynamic Impact of Macroeconomic Aggregates on Housing Prices and Stock of Houses: A National and Regional Analysis. September 4-5 . Paper Presented at the HDB International HousingConference. May . D. K. Bubbles in metropolitan housing markets.. Capozza. P. Journal of Urban Economics. 79(1) .. The efficiency of the market for single family homes. J. 26. & Jackson. market fundamentals and housing price volatility. D. & Kwok. Real Estate Economics. Anas.. Brown. New York: Palgrave Macmillan. Clayton. A. Public and private housing market interface: the Singapore experience. The Sunday Times . C. arrears and possessions. (2009. 26 . (1993). American Economic Review. (1996). L. Cheam.. Castells.B-1 Bibliography Bibliography Abraham. J. May. (1998). & Joyce. San Francisco: Holden-Day Inc.. Asteriou. 18(2). S. Public Sector Economics.. p. A Dynamic Policy Oriented Model of the Regulated Housing Model.E. (2000). An anatomy of price dynamics in illiquid markets: analysis and evidence from local housing markets. & Helsley. Case. Is there a bubble in the Housing Market? Brookings Panel on Economic Activity. 295-306. J. 1-32. (1996). (1989). Regional Science and Urban Economics . R. P. (2003). & Hendershott. M. 179-197. (2006). House prices. The fundamentals of land prices and urban growth. 17 (2) . (1990). 7(2) .201-231. (1970). July 26). Rational Expectations. & Shiller. The Shek Kip Mei Syndrome. (1996). Hendershott.. Choo. (1989).. R... Capozza. R. Case. J. Oxford: Blackwell. and G. C. Slower sales for HDB's condo-style projects. 125-137. Journal of Real Estate Finance and Economics. (2004). M. Singapore. R. D.

1-27. 49. Dunkerley. housing costs and the consumer price index.. (1981). 190-199. Republic of Singapore. from http://www. Long Swings in Urban Development. Applied Economics.. (1999). Oxford: Oxford University Press. Dougherty. (2004). Why don’t we know more about housing supply? Journal of Real Estate Finance and Economics.. W. (1983). H. (1987). Retrieved July 25. & Wheaton. Journal of the American Statistical Association. (1982). (2009). Urban Economics and Real Estate Market. Denise. W. Enders. (1979). Journal of Finance. 207-221. L. Modelling UK house prices using cointegration: an application of the Johansen technique. 154-165. (1976). 1225-1238. Distribution of the estimates for autoregressive time series with a unit root. R. D. DiPasquale. M. New York: Columbia University Press. . (1979). 1057-72. Fair.gov. Disequilibrium in housing models. Inc.B-2 Bibliography Department of Statistics. D. Engle. USA: John Wiley & Sons. 55 . D.singstat. C. R. (1996). DiPasquale. 2009 Yearbook of Statistics Singapore. Applied Economics. Likelihood ratio statistics for autoregressive time series with a unit root. Ministry of Trade & Industry. Gottlieh. R. W. Journal of Urban Economics. Applied Econometric Time Series 2nd Ed. Dickey. Housing market dynamics and the future of house prices. 9-24. and Fuller. D. J. 35(1) . Land Economics. and Fuller. 42731. estimation and testing. & Granger. Drake. & Van Order. Follain. 25 . 74. Econometrica. Inflation. Urban Land Policy: Issues and Opportunities.pdf Dickey. (1972). W. 55 .sg/pubn/reference/yos09/yos2009. (1993). The American Economic Review 72 (March) . 2009.. DiPasquale. (1994). 18(1). The price elasticity of the long run supply of new housing construction. Cointegration and error correction:representation. & Wheaton. 251-76. A. W. New Jersey: Prentice Hall. 27 .

Eligibility Scheme. 539–569. 2009 from http://www. NBER Working Paper No. National Bureau of Economic Research. 7 .nsf/WPDis/HDB%20Housing%20Market% 20StatisticsOverview?OpenDocument Housing Development Board (2009). The determinants of urban house price fluctuations in Sweden 19681994. 231-254 Johansen. Retrieved August 30. 35-60. (2005). pp. 93-129.nsf/WPDis/Buying%20A%20Resale%20Fl atEligibility%20Scheme%20%20Public%20Scheme?OpenDocument&SubMenu=Eligibility_Schemes Johansen. (1999). K. 2009 from http://www. Hendershott. C.hdb.hdb. and P. K. The other side of eight mile: suburban population and housing supply. J. Good riddance to property froth. 52 . 169-209 . Maximum likelihood estimation and inference on cointegration with application to the demand for money. (1981). Statistical analysis of cointegration vectors. 16. Inc. Real Estate Economics. Granger. Journal of Housing Economics. Green. (1993). R. (1988). (2009. September 16). Demographic Factors and Real House Prices. K. 17(1) . & Juselius. Ho. Hendry. 211-252). Econometric modelling of house prices in the United Kingdom. K. Housing Development Board (2009).A22.sg/fi10/fi10201p. A cointegration approach to the price dynamics of private housing.J. Retrieved August 30.. Journal of Property Investment and Finance. HDB Housing Market Statistics. 2009 from http://www.hdb.sg/fi10/fi10201p.sg/fi10/fi10201p.W. S.. (1984). 4332.. 33. The Straits Times. D.B-3 Bibliography Goodman. 121-30. CPF Housing Grant Scheme. Allen C.gov. H. Oxford Bulletin of Economics and Statistics. S. (1990). (1998). Some properties of time series data and their use in econometric modelspecification. In Econometrics and Quantitative Economics (pp.gov. Journal of Economic Dynamics and Control. 12. Oxford: Basil Blackwell.gov.nsf/WPDis/Buying%20A%20Resale%20Fl atCPF%20Housing%20Grant%20Scheme%20%20Overview?OpenDocument&SubMenu=CPF_Housing_Grant_Scheme Housing Development Board (2009). & Cuervo. Hort. Retrieved August 30. Journal of Econometrics.

T. the baby bust. (2001). 249-267. Malpezzi. Kennedy. D. Cambridge. FRBSF Economic Letter. Maclennan. A Guide to Econometrics. and Regulation in U. (2002). 159-178. S. and Shin. P. Lum. A simple error correction model of housing prices. public policy and private gain: house price dynamics in Singapore. 27-62.. Metropolitan Areas. Journal of Housing Economics. & Maclennan. Charles (2004) Macroeconomics and housing: a review of the literature. Market fundamentals. 389-409. The Dynamics of Metropolitan Housing Prices. (1996). taxes and subsidies on housing supply in Malaysia.S. 73(3) . Likelihood-Based Inference in Cointegrated Vector Autoregressive Models. Modeling the demand and supply sides of the housing market: evidence from Ireland.. and the housing market.. Journal of Housing Economics. 7(2). 54. N. MA: The MIT Press. & Wachter. Malpezzi. Kenny. and D. 10(3) . D. (2002). Journal of Real Estate Research. (2005). (1994). J. 19 . S. 143-164. (1992). 16. Journal of Real Estate Literature. S. S. G. Malpezzi. (2002). Journal of Housing Economics. (1998). Kwiatkowski. Housing Prices. S. Regional Science and Urban Economics. 8 . Y. Mankiw. 235-58.. Journal of Property Research. (1999). The role of speculation in real estate cycles. .. 2002(13) . Joseph Rowntree Foundation Malpezzi. (1989). G. (1999). Externalities. Land Economics. Krainer. & Weil. 209-242. D.B-4 Bibliography Johansen. S. Jud. Winkler. & Mayo. 23(Jan-April). 4th ed. Phillips. Journal of Housing Research. Journal of Econometrics. Getting housing incentives right: a case study of the effects of regulation. S. P. D. Economic Modelling. 372-391. A Competitive UK Economy: The Challenges for Housing Policy. S. Malpezzi. The long run price elasticity of supply of new construction in the United States and the United Kingdom. (1997). (1995). 19(2) . P.Oxford University Press. Leung. House Price Dynamics and the Business Cycle. 278-306. 121-143. D. 29-45. Testing the Null Hypothesis of Stationarity Against the Alternative of a Unit Root. 13 (4).. The baby boom. Schmidt. 13(2) . S..

186-201.. H. Riddiough. 34(11) . Ong. The dynamics of UK national and regional house prices. 3rd ed.. 16(4) . (1997). Modelling regional house prices: a review of the literature.B-5 Bibliography Mayer. 30(6) . A report prepared for the Department of the Environment. (2001). Phang. S. New York: McGraw-Hill. (1998). S. Econometric models and economic forecasts. S.UK. Testing the present value relation for housing prices: Should i leave my house in San Fransico? Journal of Urban Economics. Meen. & Wallace. 14(1). 35 . Price Discovery between private and public housing markets. (2000) On demand: CrossCountry Evidence from Commercial Real Estate Markets. 107 . Journal of Property Valuation & Investment.. (1988). Meen. Y. Munro. The removal of mortgage market constraints and the implications for econometric modelling of UK house prices. Yi. P. monetary policy and housing market stability. Singapore residential market. (1996). P.C. A. Housing policy. (2000). P. T. P. Land use regulation and new construction? Regional Science and Urban Economics. S. Inc.. and Rubinfeld. & Tu. 335-346. G. 443-459. Testing for a unit root in time series regression. Processed. R. Oxford Bulletin of Economics and Statistics. 1819-1829. G. M. C. R. & Somerville. 1701-1727. The housing user cost of capital.. Meen. G. Pindyck. (1991). Working paper. The Economic Journal. 245266. and Yoshida. (1997).C. J. I. Meese.. M. Phang. . & Murphy... H. Muellbauer. Urban Studies . 75. Ott. D. and Teck. Booms and busts in the UK housing market. 2000. T. (1996). & Andrew. 639-662. Ong. wealth formation and the singapore economy.. T. J. W. J. P. 57-67. Transport and the Regions. The University of Reading. and Perron. Biometrika.52(1). 50. Housing Studies. (1994). (1996).. Phillips. & Wong. N. S. L. Review of Urban and Regional Development Studies 8(2) . (1990).E. Department of Economics. S. C. Government policies and private housing prices in Singapore. (2002). 1-24. Urban Studies. & Sing.B.

(2000). Butler og. The Quarterly Journal of Economics. S. 17(4) . vers. Urban Studies. EViews. Pitkin. G. International Real Estate Review. Stiglitz. (1991). S. Contemporary Economics.. (1990). Journal of Economic Perspectives 4 (2). Joseph E.. House price dynamics: the role of tax policy and demography.. (1984). The Dynamics of the Singapore Private Housing Market.R. 4th ed. (2009a). Housing-market disequillibrium: an examination of housing-market price and stock dynamics 1967-1998. R. Riddel. Y. Tu..S. R. Journal of Housing Economics. 4(1) . Journal of Property Research. Housing investment in the United States. Sing. 135-158. housing finance and socioeconomic development in Singapore.13. & Srinivasan. 34-55. Rosen. Worth Publishers. (1974). M. CA. Segmentation of Australian Housing Market. 143-203. 2 . (1999) Public homeownership. (2004). 11(2). Chicago: University of Chicago Press. J. Quantitative Micro Software. 729-752. The impact of suburban growth restrictions on the U. Tu. 6(1). Urban Geography . Irvine. M.. (2004). 14-26. 82(1). 718-740. J. 311-329. Dynamics of the Condominium Market in Singapore. Poterba. Journal of Political Economy. (1985). Y.B-6 Bibliography Poterba. Centre for Real . Segal. CA. Topel. Y. 13–18.. M. D. 41(3) . (1991). Symposium on bubbles.V. Quantitative Micro Software. Journal of Political Economy. Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition. Galster. Review of Urban and Regional Development Studies. Inc. 120-135.0. Tu. EViews 6 User's Guide. (1980). The Maze of Urban Housing. 605-619. Spencer. (1988). Rothenberg. Tax subsidies to owner-occupied housing: an asset market approach. Irvine. housing price inflation.. (2009b). Understanding the Dynamics of Singapore Urban Housing Markets: A Policy-Modeling Approach. Y. 100– 114. Staff Research Paper. T. Brookings Papers on Economic Activity. 96 . & Rosen. November . 6. M. & Wong (2001). Tu. (2001). P.

(1983). Retrieved July 30.sg/pr/text/2009/pr09-09. School of Design and Environment.html Urban Redevelopment Authority.ura.html Wang. P. Retrieved July 30. 85. Scandinavian Journal of Economics . 2009 from http://www. URA releases flash 2nd quarter 2009 private residential property price index. Applied Economics .gov. Property and the Economy in the Short-term and the Long-run. National University of Singapore. Weibull.gov. A dynamic model of trade frictions and disequilibrium in the housing market.sg/pr/text/2009/pr09-35. April 1). (2001). (2009. 33.ura. Urban Redevelopment Authority.327-337.B-7 Bibliography Estate Studies. . 2009 from http://www. (2009. Department of Real Estate. URA releases flash 1st quarter 2009 private residential property price index. July 1). J.

Appendices A-1 Appendix One Visual Plot of Time Series Variables (Level Form) .

Appendices A-2 Appendix Two List of Government Policies .

Appendices A-3 .

Appendices A-4 .

Appendices A-5 .