Are Asia-Pacific Housing Prices Too High for Comfort? Bank of Thailand Workshop Bangkok, Thailand Tientip Subhanij* Date 23 November 2007 * Joint research project of the BOT (Tientip Subhanij), BSP (Eloisa T. Glindro), HKMA (Jessica Szeto) and BIS 1 (Haibin Zhu) under the BIS-Asian Research Program. The views expressed herein are those of the authors and do not necessarily represent those of the BOT, BSP, HKMA or BIS.
Why housing prices matter to central banks? Housing is single largest investment by households. Bank financing with variable rate is the norm. House price downturns can cause the financial accelerator effect given the wide use of real estate as collaterals. 2
The housing environment Bursting of property bubbles in 1997 was associated with a long period of economic slowdown in Asia. During the past ten years, we have seen global housing price booms in developed countries, ending in the subprime mortgage woes in the US. Recent housing booms in Asian cities have renewed fears of another speculative bubble episode in Asia. 3
Post crisis, Asia-Pacific real average housing prices in most countries have started to go up lqrhp 4.2 4.4 4.6 4.8 5 5.2 1 9 9 3 q 1 1 9 9 6 q3 20 00 q 1 2 0 0 3 q 3 2 0 0 7 q 1 t Au stra lia H o n g Ko n g Ma la ysi a Ph ili p p in e s Th a ila n d C h in a K o re a N e w Ze a l an d S in g a p o re 4
Growth of Asia-Pacific real average housing prices 000001 -.2 -.1 0.1.2 1993q1 1996q3 20 00q1 2003q3 2007q1 t Australia Hong Kong Malaysia Philippines Thailand C hina K ore a N ew Zealand S ingapore 5
Motivation What factors determine the fundamental values and shortterm dynamics of house prices? What are the implications of housing market structure and housing finance system arrangements on house price movements? Is the recent boom in Asian housing market overdone? 6
Contribution of the Study Provides complementary insights to the literature of housing prices in Asia One of the first papers to investigate evidence of persistence and mean reversion parameters in the Asia-Pacific region. Analyze difference equation implied by serial correlation and mean reversion Explore the role of supply and institutional factors on the dynamics of housing prices across countries 7
Data Coverage ECONOMIES CITIES/MARKET SEGMENT China Hong Kong SAR Malaysia Philippines Singapore South Korea Thailand Australia Beijing, Chongqing, Guangzhou, Shenzen, Shanghai, Tianjin Average residential market High-end residential market Johor, Kuala Lumpur, Pahang, Perak, Penang High-end residential market: Kuala Lumpur Average residential market: 6 major cities in the NCR* High-end residential market: Manila Average residential market: public (HDB)* High-end residential market: private Busan, Daegu, Daejon, Gwangjiu, Incheon, Seoul, Ulsan Average residential market: Bangkok single-detached and townhouses High-end residential market: Bangkok luxury condominium National data New Zealand National data * NCR National Capital Region; HDB Housing Development Board 8
Asia-Pacific housing market structure and institutions 9
China YEAR 1988 1998 2003 2004 2005 REFORM Permitted transfer of land use rights 1/ 70 years residential 40 years commercial 50 years industrial/mixed use Abolished the public welfare system Restricted sale of additional land for villa use Banned the use of loans to purchase land PBoC put an end to preferential mortgage rates OBJECTIVE Re-allocate nationalized lands by market mechanism Introduce more market-oriented reforms in the property market as a means of making it an engine of economic growth. Contain land speculation and rapid housing price appreciation. 1/ Land use rights is the main form of property ownership applicable to foreign investors 10
Hong Kong SAR Land use rights system with 50-year leasehold period Highly active primary and secondary housing markets Application List System for land sales introduced after the Asian crisis Commercial banks are the main providers of loans. HKMC setup in 1997 to buy housing loans and issue MBS Government also runs low-cost housing and provide low rental housing for lower-income households 11
The Philippines Two types of tenure: freehold and leasehold system Freehold is open only to Filipino citizens or corporations with 60 % equity held by Filipinos Ownership of land by foreigners is not allowed Foreigners can purchase condominium units and lease land up to 75 years Rent control still in effect until December 2007 Mortgage financing: Commercial banks for middle-high income earners Contractual savings institutions such as HDMF, GSIS and SSS for low-middle income members 12
Singapore Land is sold through leasehold system up to 99 years 90% of lands are owned by government Extensive public housing system (84% of households) via HDB Highest homeownership rate in the world Mortgage financing : Central Provident Fund scheme administered by HDB Private financing 13
Thailand Possession of lands come under two types: Freehold and leasehold Foreigners generally can not obtain lands and can only buy condominium units Resale housing market is not active due to cultural bias in favour of new housing and banks tend to favour new housing due to relationships with property developers Mortgage loans is financed mainly by commercial banks, followed by Government Housing Bank which has strong role in housing finance of middle to low income group. 14
Malaysia Freehold and leasehold are the two types of tenure National housing policy promote affordable housing with active participation of both government and private sector Commercial banks are dominant mortgage lenders and offer both conventional housing loans and Islamic housing finance National Mortgage Corporation (Cagamas) was set up in 1988 to provide liquidity to mortgage lenders and is the main issuer of MBS 15
South Korea Three types of tenure are freehold, strata title and leasehold Housing market is characterized by shortage of urban housing and government control Since 1991, housing market is liberalized and price controls on new apartments abolished. Recently, Korean government re-introduced price caps on house prices and other restrictions in response to concern of possible house price bubble Public housing finance dominated the market. Market-based housing finance is a recent phenomenon. There is also Chonsei informal housing finance 16
.Different institutional settings largely explain heterogeneity in empirical findings 17
Mortgage Credit Conditions ECONOMIES Australia China Hong Kong SAR Japan Malaysia New Zealand Philippines Singapore Loan-to-Value 60-70 80 70 80 80 80-85 70 80 MORTGAGE CREDIT Mortgage Credit Rate 1/ variable rate variable rate variable rate variable rate variable rate variable rate variable rate variable rate Loan Term 10-15 (max 30) 20 20-30 30 10-20 30-35 South Korea 70 variable rate 3-20 Thailand 80 variable rate 10-20 (max 30) 1/ variable rate covers also loans with fixed rate for 1-2 years and floating thereafter Source: Global Property Guide 2007 18
Government Housing Finance System ECONOMIES GOVERNMENT HOUSING FINANCE With Formal Government Guarantee FEATURES Lending to HHs Issuance of Mortgage Backed- Securities China HPF Hong Kong SAR HKMC Japan GHLC Malaysia CAGAMAS Philippines HDMF Singapore HDB South Korea KHFC Thailand GHB Source: Davies, Gyntelberg, and Chan (2007) 19
Tenure Types, Price Control and Foreign Ownership Regulations ECONOMIES LAND USE RIGHTS FREEHOLD LEASEHOLD PRICE CONTROL RESTRICTION ON FOREIGN OWNERSHIP Australia 1/ China Hong Kong SAR Japan Malaysia New Zealand 1/ Philippines (rent control effective until Dec. 2007) Singapore 1/ South Korea Cap on selling price reinstituted Thailand 1/ While purchase of residential properties by foreigners in AU, NZ and SG requires prior approval from concerned government agency, the approval process is relatively flexible. The Philippines has constitutional limits but Thailand allows limited freehold interest in land, subject to certain conditions. 20 Source: Specific details can be found in Jones Lang LaSalle, 2005/2006.
Related Literature: Property prices and macroeconomic variables: Hofman (2004); Tsatsaronis and Zhu (2004); Green et al (2005); Capozza, et al (2002) Short-run dynamics due to changes in fundamental values: Wheaton (1999); Leung and Chen (2006); Davis and Zhu (2006) Property prices and bank lending: Herring and Watcher (1999); Hilbers et al (2001); Chen (2001); Gerlach and Peng (2005) Role of local market conditions: Green et al (2005); Garmaise and Moskowitz (2004); Malpezzi (1999) Role of housing finance systems: Warnock and Warnock (2007); Tsatsaronis and Zhu (2004); Egert and Mihaljek (2007); Peek and Wilcox (2006); McCarthy and Peach (2002); Estrella (2002) 21
Search for a framework. Simple enough for application to a large number of markets with reasonable data collection and analysis costs yet grounded in theory with some intuitive appeal (Malpezzi, 1999). In the long-run, housing prices are determined by fundamentals. In the short-run, price dynamics may indicate some overshooting and even bubbles. 22
Model Following Capozza et al (2002) and Tsatsaronis and Zhu (2004), it is assumed that there is a fundamental value for housing that is determined, in reduced form, by: P = * it f ( X it ) (1) where: * P it - log of real fundamental value in the country/city X it - vector of exogenous explanatory variables representing demand and supply factors affecting house prices 23
Model (cont) There will be short-run disequilibrium, which can be specified within an error correction framework P t = α P * * t 1 + β( Pt 1 Pt 1) + γ P t (2) Two important parameters: Persistence parameter (α) Measures the degree of inertia in house price movement Mean reversion parameter (β) Measures the speed of adjustment to fundamental value 24
25 Model (cont) * 1 * 1 1 ) ( t t t t t P P P P P + + = γ β α Recall: * 1 * 2 1 ) ( ) 1 ( + = + + t t t t t P P P P P γ β γ α β α We then analyze the corresponding characteristic equation of the difference equation in (3) given by: To examine the dynamics, the above equation can be re-written in second-order linear difference equation: 0 ) (1 2 = + + α β α b b (2) (3) (4)
Dynamics of house price movements Recall: b 2 (1+ α β ) b + α = 0 Case 1: two real and distinct roots (1 2 + α β ) 4α > 0 Case 2 : two real and identical roots 2 (1+ α β ) 4α = 0 Case 3: complex roots 2 (1+ α β ) 4α < 0 26
2 Case 1: real and distinct roots (1 + α β ) 4α > 0 Sub-case Time path of house prices Impact of higher α Impact of higher β 1.a α -β 1 Unstable/ Divergence Faster divergence Slower divergence 1.b α -β< 1, β 0 Unstable/ Divergence Faster divergence Slower divergence 1.c α -β< 1, β > 0 Convergence, no cycle Slower convergence Slower convergence 27
2 Case 2: real and identical roots (1+ α β ) 4α = 0 Sub-case Time path of house prices Impact of higher α Impact of higher β 2.a α -β< 1 Convergence, no cycle Faster divergence Slower convergence 2.b α -β 1 Unstable/ Divergence Faster divergence Slower divergence 28
2 Case 3: real and complex roots (1+ α β ) 4α < 0 Sub-case Time path of house prices Impact of higher α Impact of higher β 3.a α > 1 3.b α = 1 Explosive fluctuation Symmetric fluctuation Higher amplitude of cycles Higher frequency of cycles 3.c α <1 Damped fluctuation 29
Characterizing house price dynamics 0.8 higher frequency higher frequency 0.6 higher amplitude higher amplitude β 0.4 faster convergence Damped cycle Explosive cycle 0.2 Convergence 0 Divergence faster divergence 0.2 0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 α 30
Data Description VARIABLES Dependent variable (P it ) Housing price Independent variables: ( X it ) Real GDP Population Mortgage Credit-to-GDP Real Mortgage Rate Regulatory Indices Land Supply Index Real Construction Cost Stock Price Index Real Effective Exchange Rate Official statistics and JLL Residential house prices Official statistics and CEIC Official statistics and CEIC Official statistics and CEIC Official statistics and CEIC Heritage Foundation Official statistics: building permits building constructed China Official statistics: price index for construction materials Official statistics BIS website DATA SOURCES 31
Limitations of the Study Empirical exercise on fundamental housing price is limited to determining the linkages between residential property prices and macroeconomic variables. Data on city-specific regressors that are comparable across countries are not readily available. 32
Empirical methodology and results 33
Estimation Methodology 1. Estimate fundamental housing prices (reduced-form equation) P = * i, t f ( X i, t where: P - it residential house price or land price X it - vector of macroeconomic variables representing demand and supply factors affecting house prices (instrumented with own lags to avoid simultaneity bias) Demand variables: real GDP (+); population (+); real mortgage rate (-);real effective exchange rate (+) Supply variables: land supply index (-); mortgage credit-to-gdp (+); real construction cost (+) ) 34
Estimation Methodology (cont.) 2. Estimate dynamic equation using predicted housing price from first stage to explain 2 important parameters: P * * i, t = α Pi, t 1+ β ( Pi, t 1 Pi, t 1) + γ Pi, t Two important parameters: (a) Persistence parameter (α) Measures the degree of inertia in house price movement (b) Mean reversion parameter (β) Measures the speed of adjustment to fundamental value 35
36 Estimation Methodology (cont.) 3. Estimated endogenous adjustment equation by interacting the persistence parameter and mean reversion parameter with the contemporaneous change in economic variables and institutional variables. *, * 1, 1,, 1,,, ) )( ( ) ( t i t i t i t i i t i t i i t i P P P X P X P + +Σ +Σ = γ β α α
Determinants of fundamental house prices : Panel Regression Results Countries included: AU, CN, HK, KR, MY, NZ, PH, SG, TH Dependent variable: Average real house prices (log) Restricted Variable Coefficient t-value Real GDP Real mortgage rate Credit-to-GDP Land supply index Business freedom index Corruption index Real effective exchange rate Stock price index Adjusted R 2 0.10-0.02 0.35 0.12 0.17 0.10 0.38-0.09 0.82 2.01-3.62 16.30 9.40 4.74 5.27 5.08-4.639 Business freedom and corruption indices are sourced from the www.heritage.org. Higher index indicates higher rank. 37
Endogenous Adjustment Equation Using Predicted Fundamentals from Panel Regression Countries included: AU, CN, HK, KR, MY, NZ, PH, SG, TH Persistence parameter (α) Mean reversion parameter (β) Contemporaneous adjustment parameter (γ) α*(change in land supply index) α*(change in business freedom index) β *(change in mortgage rate) β *(change in business freedom index) Adjusted R 2 Parameters Coefficient 0.34-0.15 0.13-0.46 0.94-0.11 0.18 0.44 t-value 6.35-5.98 2.52-1.90 5.48-3.02 3.34 38
Endogenous Adjustment Equation Using Predicted Fundamentals from Country-Specific Regression Countries included: AU, CN, HK, KR, MY, NZ, PH, SG, TH Persistence parameter (α) Mean reversion parameter (β) Contemporaneous adjustment parameter (γ) α*(change in land supply index) α*(change in business freedom index) Adjusted R 2 Parameters Coefficient 0.31-0.23 0.27-0.44 0.88 0.51 t-value 6.13-8.67 6.83 6.09-2.04 39
Endogenous Adjustment Equation Using Predicted Fundamentals from city-level regressions Cities/markets included: 32 cities/markets in Asia excluding NZ and AU Parameters Persistence parameter (α) Mean reversion parameter (β) Contemporaneous adjustment parameter (γ) α*(change in business freedom index) α*(change real construction cost) β *(change in real GDP) β *(change in business freedom index) β *(dummy big cities) Adjusted R 2 Coefficient 0.11-0.29 0.22 0.77 4.77 2.32 0.086 0.14 0.32 t-value 4.01-12.59 8.69 11.42 2.89 3.19 2.23 5.07 40
Key findings Apart from income, ratio of real mortgage credit to GDP and real effective exchange rate appear to be the main factors driving Asian housing prices In the long-run, mortgage rate is less relevant than previously thought which maybe largely due to government subsidy in various forms, e.g. tax subsidy, provident fund scheme, chonsei. Institutional factors matter both in determining the level of housing prices and housing price dynamics In the short-run, institutional arrangement, supply factor and market segment are important in determining housing price cycles Liberalized environment tends to make the housing price dynamics more volatile 41
Recall: characterizing house price dynamics Restricted 0.8 higher frequency higher frequency 0.6 higher amplitude higher amplitude β 0.4 faster convergence Damped cycle Explosive cycle 0.2 Convergence 0 Divergence faster divergence 0.2 0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 α 42
House price dynamics in selected Asian countries Restricted Note: The results are based on a panel regression on the determinants of house 43 price fundamentals and a panel regression on the short-term dynamics
House price dynamics in selected Asian countries Restricted Note: The results are based on country-specific regression on the determinants of house price fundamentals and a panel regression on the short-term dynamics 44
Conclusion and Policy Implications Expectedly, there is high degree of heterogeneity in the determinants of fundamental house prices during the estimation period Largely due to the differences in housing regulatory framework, housing finance system, and housing market structure during the estimation period. The dynamics of housing prices point to a relatively benign housing market environment in Asia during the mid-90s until 2006. No country falls in the explosive region. There is no significant overvaluation or bubble in most countries in recent years, with the exception of some big cities/market segment 45
Conclusion and Policy Implications There is still the need to monitor housing markets as there could always be new sources of volatility as markets develop, especially in big cities/luxury market segment To mitigate house price overvaluation, policymakers should focus on measures aimed at reducing the magnitude and frequency of property price cycles. The government could affect housing price dynamics via supply, government support and housing finance system design There is the need for improvement in the quality and availability of sub-national level data for monitoring housing price dynamics at the city/provincial/state levels 46
Thank you 47