Housing Price Gradient Changes between Macau and Hong Kong

Transcription

1 Housing Price Gradien Changes beween Macau and Hong Kong C.Y. Yiu Deparmen of Real Esae and Consrucion, The Universiy of Hong Kong Submied o AsRES 2007 Conference, Macau 15 June 2007 Absrac: This paper applies a wo-workplace residenial locaion choice model o sudy he housing price gradien changes beween Macau and Hong Kong. The classical land marke model assumes a one-workplace scenario. However, wih he globalizaion rend, ravel beween wo workplaces is becoming more and more common. This paper found ha housing price gradien from Macau o Hong Kong is flaened when more non-residen workers raveled from Hong Kong o Macau, ceeris paribus. The resuls have imporan implicaions on polycenric ciy models, and provide a novel mehod o sudy neighboring ciy effecs on housing price. JEL Classificaion: R21 Housing demand Keywords: workplaces, housing demand, price gradien, commuing mileage, raveling frequency *Please send commens o: C.Y. Yiu, Assisan Professor, Deparmen of Real Esae and Consrucion, The Universiy of Hong Kong, Pokfulam, Hong Kong. ecyyiu@hkucc.hku.hk Tel.: (852) , Fax: (852)

2 1. Inroducion There have been very few sudies on he real esae markes of Macau, probably because of he lack of daa. Wih he esablishmen of he Direcção dos Serviços de Esaísica e Censos (The Saisics and Census Service, DSEC) by he Macao SAR Governmen since 1999, sysemaic daa series are now available from is websie (hp:// This paper is probably he firs academic sudy on he housing markes of Macau. However, wih he grea changes of socio-economic environmen of Macau since he hand-over of sovereign in 1998, a direc and conrolled sudy on he price changes of housing in Macau is no ye echnically feasible. This paper herefore aims o sudy he housing price gradien changes beween Macau and Hong Kong, which reduces he explanaory variables o a manageable scale. Beer sill, i can sudy he effec of Hong Kong, which is a neighboring ciy, on he housing price in Macau. Daa of macroeconomic series and real esae of Hong Kong are readily available a he Census and Saisics Deparmen (CSD), Hong Kong SAR Governmen a hp:// and he Raing and Valuaion Deparmen (RVD), Hong Kong SAR Governmen a hp:// 2. Background of Macau and Hong Kong Macau and Hong Kong share a lo of commonaliies. They are special adminisraion regions of China, wih he sovereignies handed over o China from Porgual and Briain in 1998 and 1997 respecively. Boh are locaed on he souheasern coas of China (as shown in Figure 1) wih very high densiy of populaion. The oal land areas of Macau and Hong Kong 2

3 are 21 km 2 and 1,108 km 2, respecively; bu here are 513,000 and 7,041,000 people in he end of 2006, which make hem he 2 nd and he 3 rd highes densiy ciies hroughou he world. Boh governmens adop free marke policy, which makes virually no aemp o influence prices on boh resources and produc markes. There are no ariffs or quoas on impored goods and he ax raes are among he lowes ones worldwide. The GDP per capia per annum of Macau and Hong Kong are reasonably high a US$ 27,727 and US$26,808 respecively 1. Figure 1 The Geographical Relaionship beween Macau and Hong Kong (Source: middle_eas_and_asia/) 1 The nominal GDP per capia and he GDP in PPP per capia of Hong Kong were ranked respecively he 27 h and he 6 h among all he counries (or ciies) sampled by he Inernaional Moneary Fund in 2006 (Wikipedia, 2007) 3

4 However, he economies of Macau and Hong Kong are very differen, Macau is he only one ciy of China where casino gambling is legal, whereas Hong Kong plays is role as an inernaional financial cenre. People-flow beween he wo ciies is remendous. The inflow of visiors from Hong Kong o Macau reached 650,000 per monh in he end of 2006, which exceeds he oal number of residens in Macau. This figure amouns o 32% of he oal inflow of people ino Macau. In addiion, here were 14,118 Hong Kong workers per monh working in Macau in he end of This figure also amouns o 20% of he oal non-residen workers in Macau. Currenly, ravelers rely heavily on ferry o ravel beween Macau and Hong Kong. Wih he heavy raffic demand, a bridge is planned o be buil o connec Hong Kong, Macau and Zhuhai in The macroeconomic parameers of he wo ciies from 2000Q1 o 2007Q1 are ploed in he following figures. Figure 2 shows he GDP per quarer of he wo ciies. Boh ciies show very srong growh in he pas few years, especially afer he SARS in Figure 2 shows he rapidly declining raes of unemploymen, which is a consequence of he economic boom. Similarly, Figure 4 shows heir composie consumer price indices, which reflec a srong rebound since Figure 5 shows heir housing price indices (averages of all classes) by normalizing heir values a 2007Q1=247. I shows ha Macau has undergone a very srong increase in housing price, whereas Hong Kong go a rebound afer Figure 6 shows a modes growh of heir populaions. Lasly, Figure 7 shows he people flows beween he wo ciies, he solid line represens he number of visiors from Hong Kong o Macau per quarer, 4

5 whereas he doed line is he number of workers from Hong Kong o Macau per quarer. The srong growh of people flow is unprecedened Macau GDP per quarer (lef axis) Hong Kong GDP per quarer (righ axis) Figure 2 GDP per quarer of Macau and Hong Kong, Macau Unemploymen Rae (%) Hong Kong Unemploymen (%) Figure 3 Unemploymen Raes of Macau and Hong Kong,

6 Macau Consumer Price Index ( =100) Hong Kong Consumer Price Index ( =100) Figure 4 Composie Consumer Price Indices of Macau and Hong Kong, Macau Housing Price Index (2007Q1=247) Hong Kong Housing Price Index (2007Q1=247) Figure 5 Housing Price Indices of Macau and Hong Kong,

7 Macau Populaion ('000) Hong Kong Populaion ('000) Figure 6 Populaions of Macau and Hong Kong, Macau Visiors from Hong Kong per quarer (lef axis) Macau Non-residen Workers from Hong Kong per quarer (righ axis) Figure 7 Visiors and Workers from Hong Kong o Macau,

8 3. Changes of he Price Gradien beween Macau and Hong Kong Figure 8 shows a very srong correlaion (0.9331) beween he housing prices of he wo ciies, which indicaes a srong link beween heir socio-economic relaionships beween hem in he ligh of heir close proximiy. Thus, independen macroeconomic analysis of housing price in any one ciy would resul in he omission of exernal facors exered from he neighborhood ciy. This paper herefore pu forward a housing price gradien approach o analyze he cross-boundary housing price changes HK_HPI MA_HPI Figure 8 Scaer Plo of he relaionship beween housing price indices (HPI) beween Macau (MA) and Hong Kong (HK),

9 Price gradien beween Macau and Hong Kong a ime, G is defined as he raio of he price levels of housing of he wo ciies as shown in he following equaion: G = (1) φ η where φ, η are he price levels of housing in Macau and Hong Kong a ime, respecively. For he ease of comparison, heir price gradien a 2007Q1 is normalized o 1. Hence, ifg > 1( G < 1), hen he uni housing price in Macau is higher (lower) han ha of Hong Kong, which, in urn, indicaes ha he housing price curve is negaively (posiively) sloping from Macau o Hong Kong. Figure 9 shows he gradien series from 2002 o 2007, which shows a general increasing rend (i.e. G < 1), reflecing he recen srong growh of housing price in Macau G() Figure 9 Housing Price Gradien beween Macau and Hong Kong,

10 Housing price sudy involves a large number of explanaory variables, including no only socio-economic, demographic and ransporaion facors, bu also ax srucure, povery rae, crime rae, school qualiy, neighborhood qualiy, elecommunicaion nework qualiy and even culure and ehniciy. Unforunaely, hese variables are difficul o quanify and heir daa in ime series are ofen no readily available. Ye he omission of hese variables would resul in serious specificaion bias. For example, Clapp, e al. (2001) incorporaed a large number of socioeconomic, demographic, zoning and ransporaion variables ino heir esimaion model, which renders he esimaion highly difficul. Forunaely, he differences in many of he explanaory variables beween wo ciies, such as ax srucure, school qualiy, and culural difference, are raher fixed over ime or evolving ogeher a he same pace. This paper, herefore, circumvens his esimaion problem by analyzing he emporal change of he price gradien insead, which can be regarded as a emporal-difference of he spaial-difference (emporal price gradien changes) of housing price. This mehod of emporal differencing has been raised by Bailey, e al. (1963), in heir developmen of he repea-sales mehod in price index consrucion. The mehod has successfully reduced he large number of variables in hedonic pricing models o jus a few dummy variables of ime. This mehod has received grea aenion and has been adoped inensively in real esae research. Numerous refinemens of he model have also been developed. The underlying principle of he mehod is he assumpion ha mos of he explanaory variables do no change over ime. An exension of he principle o a 10

11 wo-dimensional differencing will be equally advanageous in urban spaial srucure research. The spaial-emporal differencing mehod (i.e., he emporal change of price gradien analysis) has been successfully developed by Yiu and Tam (2007) in heir inra-ciy sudies. By means of his wo-dimensional differencing, mos of he explanaory variables can be eliminaed in he esimaion process. Mahemaically, if housing price P, a locaion L and a ime can be explained by a L funcion of housing characerisics H,, socio-economic facor S,, demographic facor D,, L L L ransporaion facor T L,, qualiy of life Q L,, and culural facor C L,, ec., hen i can be represened as equaion (2), by assuming he Cobb-Douglas funcion: P ln P = f ( H, S, D, T, Q, C,...) = α ln H + β ln S + γ ln D = H α + λ lnt S β D γ T λ + θ lnq Q θ C ϕ... + ϕ lnc +... (2) Afer accouning for he implici value of housing characerisics by he consrucion of a housing price index, he general log price levels ln P ln PL, α ln H = are obained. Then, he spaial-difference of he housing price levels beween locaions A and B, i.e., he price gradien, becomes: ln G = ln + θ PA, ln PB, = α + β ( ln S A, ln S B, ) + γ ( ln DA, ln DB, ) + λ( lnta, lntb, ) ( ln Q ln Q ) + ϕ( ln C ln C ) +... A, B, A, B, (3) where α is a consan erm. Those variables which are he same in he wo locaions will cancel each oher ou in his spaial-difference process. A emporal-difference of he spaial-difference will furher eliminae a large number of explanaory variables because many changes are regional and conagious. Alhough qualiy of life and culural facors, for example, are no he 11

12 same in he wo ciies, he emporal-difference of he spaial-difference is likely o be unchanged for a long period of ime. Mahemaically, when he variables such as qualiy of life and culure are eliminaed, he emporal-difference of he spaial-difference of housing prices beween locaions A and B from ime o τ is: ln G τ ln G = = ( ln PA, τ ln PB, τ ) ( ln PA, ln PB, ) ατ α + β[ ( ln S A, τ ln S B, τ ) ( ln S A, ln S B, )] [( ln DA, τ ln DB, τ ) ( ln DA, ln DB, )] + ( lnt lnt ) ( lnt lnt ) γ λ[ ]. A, τ B, τ A, B, +. (4) Using quarerly daa, and aking he emporal differences beween one quarer, i.e. τ = +1, equaion (4) can be simplified o: Δg = Δ ( p p ) = ( ) + β[ Δ( s s )] + γ [ Δ( d d )] + λ Δ( ) α [ ].. (5) A, B, A, B, A, B, A, B, where variables in small leer represen he logarihm of he corresponding variables of he same capial leer. The explanaory variables can, herefore, be confined o socio-economic, demographic and ransporaion changes. In our sudy, we consider he number of non-residen workers from Hong Kong o Macau as a proxy of T. Figure 7 above plos he quarerly cross-boundary headcoun of non-residen workers over he sample period. The figure shows very clearly ha he commuing frequency from Hong Kong o Macau increases exponenially. 4. Lieraure Review Since von Thünen s (1826) isolaed sae model, Alonso (1964) and Muh (1969) furher developed heir heories of urban spaial srucure in he assumpion of a monocenric and isolaed ciy. However, he impracical assumpion of a monocenric cener has been 12

13 challenged by Papageorgiou and Casei (1971) and Romanos (1977). Polycenric urban spaial srucure heory is developing in recen decades. For example, Sivianidou (1997), and McMillen and McDonald (1998) have documened he decenralizaion of U.S. economic aciviy, while Anas, e al. (1998) explained polycenric urban spaial srucure by spaial heerogeneiies, scale economies, and imperfec compeiion. Clapp, e al. (2001) found a flaening of he land value gradien due o he decenralizaion process. They explained he decenralizaion rend by seven reasons, including he fac ha wih he rapid developmen in he echnology of communicaion and ransporaion, he cohesiveness of any single business cener is diminishing; businesses can sele a places oher han he cener, and hen employmen and business aciviies become decenralized. A more comprehensive lieraure review on polycenric models can be found in Yiu and Tam (2004). Ye, he previous polycenric cenre sudies are sill based on a single workplace assumpion, i.e., each household ravels o one workplace only. Bu, here are a leas hree causes of muliple workplaces per household, namely: (1) muliple earners households, (2) job locaion changes in he fuure, and, (3) muliple workplaces per earner. Madden (1980), Curran, e al. (1982), Timmermans, e al. (1991), Hochkiss and Whie (1993), Kim (1995), Freedman and Kern (1997), van Ommeren, e al. (1998) and Whie (1999) invesigaed he firs cause: residenial locaion choice in wo-earner households. Crane (1995) sudied he second: effecs of uncerain job locaion on urban form. Ye, he sudy of he hird cause remains relaively unexplored, due primarily o he fac ha he rend has been developing only since he lae 13

14 1990s. I is very common nowadays for employers and employees o work a wo or more locaions and ravel frequenly. For example, raveling beween headquarers and a branch office in a neighboring ciy is a rouine for personnel of an inernaional company. Hong Kong-Macau, Hong Kong-Shen Zhen and Johor Bahru-Singapore are some ypical examples. We conend ha his mobilizaion rend affecs he equilibrium of housing demand and he housing price gradien beween workplaces. In oher words, his paper aemps o sudy he labor mobilizaion effec from a neighboring ciy on he housing price gradien. A heoreical model on his wo-workplace residenial choice model has been developed in Yiu and Tam (2007). Inuiively, i is more convenien for hese mobile workers, who need o aend differen workplaces regularly, o live closer o he workplace wih he highes visiing frequency. In his sense, he classical monocenric models give a good descripion of he housing price gradien from ha busies locaion: sloping downward from he cener. However, he monocenric models canno predic a household s choice of residenial locaion when he visiing frequencies of workplaces vary. The analyical model exends Alonso s (1964) and Muh s (1969) urban spaial models for wo workplaces per household wih exogenous commuing frequency. The purposes of his paper are (1) o apply he wo-workplace model on he housing price gradien changes beween Macau and Hong Kong o es i empirically and (2) o sudy he effec of a neighboring ciy on he housing price gradien. 14

15 5. Daa and Empirical Resuls This sudy applies Yiu and Tam s (2007) wo-workplace heoreical model on Macau-Hong Kong housing price gradien. The inuiive inerpreaion of he model is ha he increase in commuing frequency resuls in a flaer price gradien beween he ciies. From 2002Q1 o 2007Q1, privae housing price indices in he wo ciies are colleced from heir corresponding governmen census and saisics deparmens. The dwelling aribues in hose housing samples shall have been aken ino accoun by heir index consrucion mehod. Explanaory variables on demographics and socio-economic facors, such as populaion size (POP), gross domesic producs (GDP), inflaion measured by consumer price index (CPI), unemploymen rae (UNE) of he wo ciies are colleced. All variables are aken logarihms excep he unemploymen rae because i is already a percenage figure. Prefixes MA_ and HK_ sands for Macau s and Hong Kong s figures respecively. Lasly, he number of non-residen workers raveled from Hong Kong o Macau (MA_WOR) is included as one of he explanaory variables. Table 1 shows he summary saisics of he variables. Table 1 Summary Saisics of he Variables Mean Sd. Dev. Minimum Maximum Δln(MA_HPI/HK_HPI) Δln(MA_POP/HK_POP) Δln(MA_GDP/HK_GDP) Δln(MA_CPI/HK_CPI) Δ(MA_UNE/HK_UNE) Δln(MA_WOR,2)

16 Uni roo ess on he ime series reveal ha hey are srongly saionary series a he firs differences, excep ha he number of non-residen workers series is saionary a he second difference. Table 2 shows he uni roo es resuls. Table 2 Uni Roo Tes Resuls of he Variables Uni Roo Tes wih rend and Level erm Firs difference inercep -sa. Prob. -sa. Prob. ln(ma_hpi/hk_hpi) ln(ma_pop/hk_pop) ln(ma_gdp/hk_gdp) ln(ma_cpi/hk_cpi) (MA_UNE/HK_UNE) Δln(MA_WOR) The residuals of he esimaion of equaion (5) also reveal srong auocorrelaion a he second lag in correlogram-q saisics. Therefore, second-order auoregression for he residuals is esimaed. The esimaed resuls of he Two-Workplace Model (eq. 5) wih auoregressive AR(2) processes are summarized in Table 3: All of he esimaed coefficiens of equaion (5) are significan a he 5% level, excep inflaion (CPI) and unemploymen (UNE). The esimaed coefficien of he log difference of non-residen workers headcoun is saisically significan and posiive, which indicaes ha he log difference of price raio is increasing wih he increase of he log difference of commuing frequency raio beween he wo ciies. More specifically, he price gradien from Macau o Hong Kong became flaer when he log difference of commuing frequency raio beween he wo ciies increased during he sample period. This suppors he analyical resul of he wo-workplace model. The resuls of oher variables are also reasonable and as expeced. For 16

17 example, populaion growh gradien and inflaion growh gradien beween he ciies impose a posiive effec on heir housing price gradien, alhough he laer one is no saisically significan. I reflecs he local demand effecs. Similarly, unemploymen growh gradien is found o impose a negaive impac on heir housing price gradien, alhough i is no saisically significan. I shows he local affordabiliy impacs. Ineresingly, GDP growh gradien is found o exer a negaive and significan effec on housing price gradien beween he ciies. I explains, among ohers, he srong link beween he wo housing price indices, and reflecs he grea invesmen fund flows beween he ciies. Table 3 Empirical Resuls of he Two-Workplace Model Dependen Variable: Δln(MA_HPI/HK_HPI) Mehod: Leas Squares Included observaions: 12 afer adjusmens Convergence achieved afer 10 ieraions Whie Heeroskedasiciy-Consisen Sandard Errors & Covariance Variable Coefficien Sd. Error -Saisic Prob. C Δln(MA_POP/HK_POP) Δln(MA_GDP/HK_GDP) Δln(MA_CPI/HK_CPI) Δ(MA_UNE/HK_UNE) Δln(MA_WOR,2) AR(2) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood F-saisic Durbin-Wason sa Prob(F-saisic)

18 6. Conclusions This sudy provides an empirical evidence of he Two-Workplace Model on choice of residenial locaion by sudying he housing price gradien changes over ime beween Macau and Hong Kong. I shows ha, in wo linear bicenric ciies, he housing price gradien beween wo workplaces will become flaened when he visiing frequency o he workplace increases relaively. In his sudy, oher socio-economic variables, including GDP, unemploymen rae, inflaion rae and populaion size are incorporaed in he esimaion, which is an improvemen from Yiu and Tam (2007). However, he sudy is sill limied by he shor ime series of daa in Macau, furher sudies on longer daa series shall be performed. This paper is probably he firs aemp o explore he housing markes of Macau empirically, and is also he firs sudy of he effec of a neighboring ciy on he housing price gradien beween Macau and Hong Kong. I does no only consider a wo-cenre urban spaial srucure, bu i also invesigaes he labor mobilizaion impac on he housing price difference beween he wo ciies. I is a novel bu of paramoun imporance concep in sudying neighboring effec on real esae in he rend of globalizaion. Acknowledgmens This paper is fully suppored by he funding from he Research Grans Council of he Hong Kong Special Adminisraive Region, China (Projec No. HKU1230/06E). 18

19 References Alonso, W. (1964) Locaion and Land Use, Harvard Universiy Press, Cambridge, MA. Anas, A., R. Arno, and K. Small (1998) Urban Spaial Srucure, Journal of Economic Lieraure 36(3), Bailey, M.J., R.F. Muh and H.O. Nourse (1963) A Regression Model for Real Esae Price Index Consrucion, Journal of he American Saisical Associaion 58(304), Clapp, J.M., M. Rodriguez and R.K. Pace (2001) Residenial Land Values and he Decenralizaion of Jobs, Journal of Real Esae Finance and Economics 22(1), Crane, R. (1995) The influence of uncerain job locaion on urban form and he journey o work, Journal of Urban Economics 39, Curran, C., L. Carlson and D. Ford (1982) A heory of residenial locaion decisions of wo-worker households, Journal of Urban Economics 12, Freedman, O. and C.R. Kern (1997) A model of workplace and residence choice in wo-worker households, Regional Science and Urban Economics 27, Hochkiss, D. and M.J. Whie (1993) A simulaion model of a decenralized meropolian area wih wo-worker, radiional and female-headed households, Journal of Urban Economics 34, Kim, S. (1995) Excess commuing for wo-worker households in he Los Angeles meropolian area, Journal of Urban Economics 38,

20 Madden, J.F. (1980) Urban land use and he growh in wo-earner households, American Economic Review, Papers and Proceedings 70, McMillen, D.P. and J.F. McDonald (1998) Suburb and subceners and employmen densiy in meropolian Chicago, Journal of Urban Economics 43(2), Muh, R.F. (1969) Ciies and Housing, Universiy of Chicago Press, Chicago. Papageorgiou, G.J. and E. Casei (1971) Spaial equilibrium residenial land values in a mulicenric seing, Journal of Regional Science 11, Romanos, M.C. (1977) Household locaion in a linear muli-cener meropolian area, Regional Science and Urban Economics 7, Sivianidou, R. (1997) Are cener access advanages weakening? The case of office-commercial markes, Journal of Urban Economics 42(1), Timmermans, H., A. Borgers, J., van Dijk and H. Oppewal (1991) Residenial choice behaviour of dual earner households: a decomposiional join choice model, Environmen and Planning A 24, van Ommeren, J.N., P. Rieveld and P. Nijkamp (1998) Spaial moving behaviour of wo-earner households, Journal of Regional Science 38(1), Von Thünen, H. J. (1826) Der isoliere Saa in Beziehung auf Naionalökonomie und Landwirschaf (The Isolaed Sae wih respec o Agriculure and Poliical Economy), Gusav Fisher, Sugar. Whie, M.J. (1999) Urban Areas wih Decenralized Employmen: Theory and Empirical Work, 20

21 in P. Cheshire and E. Mills, eds., Handbook of Regional and Urban Economics Vol. 3, Norh-Holland, Amserdam, Yiu, C.Y. and C.S. Tam (2004) A review of recen empirical sudies of propery price gradiens, Journal of Real Esae Lieraure 12(3), Yiu, C.Y. and C.S. Tam (2007) Housing Price Gradien wih Two Workplaces an Empirical Sudy in Hong Kong, Regional Science and Urban Economics 37(3),