The Nexus between Labor Wages and Property Rents n the Greater Chna Area by Terence Ta-Leung Chong, Kenny Ch-Wa Shu and Vvan Wong Workng Paper No. 17 July 2013 Insttute of Global Economcs and Fnance The Chnese Unversty of Hong Kong 13/F, Cheng Yu Tung Buldng, 12 Chak Cheung Street, Shatn, Hong Kong
Acknowledgements The Insttute of Global Economcs and Fnance s grateful to the followng ndvduals and organzatons for ther generous donatons and sponsorshp (n alphabetcal order): Donors Johnson Cha BCT Fnancal Lmted Vncent H.C. Cheng Henderson Land Development Co. Ltd. Fred Hu Zulu Lau Chor Tak Foundaton Lmted Lawrence J. Lau Sun Hung Ka Propertes Ltd. The Bank of East Asa, Lmted The Hongkong and Shangha Bankng Corporaton Lmted Programme Supporters C.K. Chow Bangkok Bank Publc Co Ltd Alvn Chua Bank of Chna Lmted - Phnom Penh Branch Fang Fang Be Shan Tang Foundaton Vctor K. Fung Chna Development Bank We Bo L Chna Soft Captal K.L. Wong HOPU Investment Management Co Ltd Industral and Commercal Bank of Chna - Phnom Penh Branch Kng Lnk Holdng Lmted Sun Wah Group
The Nexus between Labor Wages and Property Rents n the Greater Chna Area Terence Ta-Leung Chong 1, Kenny Ch-Wa Shu Department of Economcs The Chnese Unversty of Hong Kong and Vvan Wong Insttute of Global Economcs and Fnance The Chnese Unversty of Hong Kong 4/7/2013 Abstract Tse and Chan (2003) nvestgated the relatonshp between property sales prce and value of commutng tme. However, property sales prce s subject to the nherent lmtaton that t ncludes speculatve elements. A better measure to be used for such study should be the rent pad by the genune end-user of the property. Ths paper examnes how equlbrum rents n dfferent locatons n the Greater Chna area are determned by the tme value, or the shadow wage, of an ndvdual. Usng the rental nformaton, we provde a frst estmate of the rato of the tme values of ndvduals n Hong Kong, Shangha and Tape. Our result shows that the shadow wage rato of the households n Hong Kong, Shangha, and Tape s about 2.25: 1: 1.61. Keywords: Shadow wage; Property rental prce; Central busness dstrct. JEL Classfcatons: J31; R30. 1 We would lke to thank Charles Leung, Travs Ng and Tommy Leung for helpful comments. Correspondng Author: Terence Ta-Leung Chong, Department of Economcs, The Chnese Unversty of Hong Kong, Shatn, N.T., Hong Kong. Emal: chong2064@cuhk.edu.hk. Webpage: http://www.cuhk.edu.hk/eco/staff/tlchong/tlchong3.htm. 1
1: Introducton Locaton s undoubtedly a key factor n determnng the prce of a property. The prce of property n dfferent locatons hnges upon the perceved value of travelng tme between home and the central busness dstrct (CBD), ceters parbus. In partcular, property value s negatvely related to the dstance from the CBD, as predcted n the monocentrc cty model developed by Alonso (1964), Muth (1969) and Mlls (1972). 2 Gven the rental dfferentals across dfferent dstrcts, one should be able to retreve the tme value assocated wth commutng tme, or the shadow wage, of an ndvdual. The man contrbuton of ths paper to the lterature s to deal wth the nexus between property value and the shadow wage. Both property value and shadow wages have long been studed n the lterature. For example, Stegman (1969) and Henderson (1977) examned envronmental qualty and locaton accessblty, as they are deemed to affect property prce. Nelson (1978); So, Tse, and Ganesan (1997) examned the effect of transportaton on property prce. Mahan, Polasky, and Adams (2000) studed the relaton of property prce to urban ar qualty and wetlands separately. 3 Heckman (1974) examned the observed wage rate of women and the shadow prce of tme. However, the relaton between shadow wage and property rental prce has seldom been examned. Ths paper attempts to brng the two ssues together and provde a new perspectve to recover the shadow wage from the property rental value n three major ctes n the Greater Chna area, namely, Hong Kong, Shangha and Tape. The rental value of a property depends manly on two types of varables. prce = f ( X, t) (1) where X s a vector representng the characterstcs of the property such as property age, sze, and floor number. Informaton on the property can be extracted drectly 2 Accordng to McMllen (2006), although ctes are becomng ncreasngly polycentrc, modern urban areas stll tend to be domnated by the tradtonal CBD. The monocentrc model stll has predctve value, even though the rate of declne n property values wth dstance from the CBD has fallen over tme. 3 The hedonc regresson method and factor analyss are often appled to estmate the ndvdual factor effect on property prce (Kan and Qugley, 1970; Bajar and Kahn, 2008). 2
from the webstes of real estate companes. Meanwhle, t s the tme cost varable reflectng the CBD proxmty effect on property. Smlar to Tse and Chan (2003), travelng tme from a property to the CBD, nstead of dstance, s used n ths paper as a measurement of ts effect on property prce. The estmated coeffcent assocated wth the varable can be used to derve ratos of tme values among dfferent ctes. The rest of the paper s organzed as follows: Secton 2 descrbes the data and provdes the detals of the measurement methods. Secton 3 ntroduces the models and methodology used n the emprcal research. Secton 4 reports the results of dfferent models. Fnally, Secton 5 concludes the paper and suggests future research drectons. 2: Data The resdental property rental markets (referred to as the property rental market thereafter) of the Greater Chna regon, ncludng Hong Kong, Shangha, and Tape, are examned and compared. Unlke McMllen and Sngell (1992), who studed seven ctes 4 n Amerca, ths paper focuses on three Greater Chna ctes for two reasons: Frstly, they are densely populated; secondly, publc transport systems,.e., the ralways and buses, are the major means of travel to the workplace. Snce nformaton on travelng tme and fare of the ralway systems s avalable on publc transportaton webstes, a precse measurement of commutng tme can be obtaned easly. An addtonal reason for usng the three Chnese ctes s that cultural factors that affect the property market can be controlled for. The CBDs n Hong Kong, Shangha, and Tape are examned n ths paper ndvdually. 5 To lmt the complexty of ths research, one sngle leadng CBD was 4 McMllen and Sngell (1992) studed Cleveland, Columbus, Dayton, Detrot, Indanapols, Phladelpha and Pttsburgh n Amerca. 5 All the three ctes n our sample have a sngle CBD. Although the Hong Kong government plans to turn Kowloon East nto the second CBD but the most mportant busness dstrct wll reman n Central. (http://www.scmp.com/artcle/981836/planners-thnk-bg-kowloon-east). In addton, a number of Hong Kong s landmark buldngs are located n Central such as IFC, whle the offces of Hong Kong Monetary Authorty and Hong Kong Exchanges and Clearng Lmted are located n Central as well. For Shangha, accordng to the webste of Pudong New Area Government, Lujazu s the only natonal-level development zone named wth fnance and trade. (http://englsh.pudong.gov.cn/html/pden/pden_busness_dz/info/detal_73178.htm). In addton, a number of Shangha s landmark buldngs are located n Lujazu such as Shangha IFC and Jn Mao 3
selected n each cty (Central n Hong Kong, Lujazu n Shangha, and Xny n Tape) based on the government's offcal recognton and the study of real estate consultants. A varety of the cty s landmark buldngs located n the selected dstrct serves as a further proof of ts leadng role as a commercal dstrct. The property market s composed of the rental market and sales market. Property buyers can be end users or nvestors, whle tenants are most lkely genune end users. Therefore, unlke Tse and Chan (2003), who focused on the prvate property sales market and ts prce, ths paper uses nformaton only from the prvate property rental market, n order to exclude the nfluences of speculaton and nvestment n the property market. It s reasonable to assume that most people travel to the CBD by publc transportaton n Hong Kong and Shangha, snce prvate motor vehcle ownershp n these two ctes are at a relatvely low level due to expensve and lmted parkng n the CBD, hgh gasolne taxes and mport dutes on motor vehcles. 6 In Tape, prvate vehcle ownershp s relatvely hgh, wth one out of four people ownng a prvate car; and one out of two people, a motor cycle. Nevertheless, average daly rdershp n the mass transt ralway n all three ctes s hgh. 7 Compared wth alternatve publc transports systems, the ralway has an advantage to provde more accurate and relable nformaton on commutng tme. Propertes atop or adjacent to the ralway statons are sampled for a precse measure of commutng tme, whereas those beyond walkng dstance from the ralway statons are excluded from our study due to measurement dffcultes. Snce our sample resdents are those who are self-selected to lve near the ralway statons, t s not unreasonable to assume that these people choose to lve near the ralway statons for easy access to the CBD. Thus, our sample group has a much hgher chance of workng at CBD compared to others from other Tower. The Shangha headquarters of the People s Bank of Chna and the offce of Shangha Stock Exchange are located n Lujazu as well. For Tape, Xny wll contan at least 50% of the total Grade A offce stock n Tape, and wll reman as the leadng commercal center of Tape. (http://www.prweb.com/releases/2007/07/prweb542482.htm). 6 In Hong Kong, there are only 63 prvate cars lcensed per 1000 people n 2012. In Shangha, there were only 51 prvate cars owned per 1000 people n 2011 (Data sources: Shangha Statstcal Yearbook 2012, Hong Kong Transport Department) 7 Average daly rdershp per capta n 2012 was around 0.68, 0.36 and 0.67 n Hong Kong, Shangha and Tape respectvely. (Data sources: Hong Kong MTR, Xnmn News, Tape Rapd Transt Corporaton) 4
parts of the cty. Ths self-selected sample also elmnates the potental bas of people travelng by prvate vehcle, snce people stayng very close to ralway statons are more lkely to use publc transport nstead of drvng. As mentoned n Secton 1, the nformaton on vector X representng the characterstcs of the property such as property age, sze, and floor number s obtaned from the webstes of varous real estate companes 8, whereas the tme cost varable or the necessary commutng tme s obtaned from the webstes of the Metro Systems 9 n the three ctes. A total of 1,086, 1,741, and 893 observatons 10 n the property rental markets of Hong Kong, Shangha, and Tape are collected respectvely. 11 The sample wndow s January 2011 to March 2011. The observatons are classfed nto three data sets accordng to property sze n order to measure the household stuaton n dfferent prvate propertes. The frst data set, referred to as all property data set, covers all observatons n whch prvate propertes of all szes are ncluded. Ths data set can be used to measure the general condtons of the households under prvate housng n the three ctes. The second data set, the small and md-sze property, covers the observatons of prvate propertes less than 1,000 sq. ft. Ths data set s used to measure the stuaton of small and medum-szed propertes n the three ctes. In partcular, there are 943, 869, and 390 observatons n the small and md-sze property data sets n Hong Kong, Shangha, and Tape, respectvely. The thrd data set, luxury property, conssts of prvate propertes equal to or exceedng 1,000 sq. ft. The luxury property set s used to measure the stuaton of the luxury property 8 For the property rental market n Hong Kong, the data sources are Centalne Property (http://web.centanet.com/fndproperty/) and Mdland Realty (http://www.mdland.com.hk/ch/). For the Shangha property rental market, the data source s Koofang (http://shangha.koofang.com/), whereas for the Tape property rental market, the data sources are Happyrent (http://happyrent.rakuya.com.tw/) and Twhouses (http://www.twhouses.com.tw/). 9 The Metro Systems n Hong Kong, Shangha, and Tape are Mass Transt Ralway, Shangha Metro, and Tape Rapd Transt System, respectvely. 10 In Shangha and Tape, snce the property nformaton shown on the real estate webstes s not well organzed, the nformaton on the property estate s always mssng, whle only the street name and number of the property can be found. Therefore, t s not feasble to dentfy the property estate nformaton on the observatons n these two ctes. However, the property estate nformaton can be found on the real estate webstes n Hong Kong, and the observatons n Hong Kong are obtaned from 76 property estates. 11 In the Shangha and Tape property rental markets, only the bd nformaton s avalable so the bd prce on the property rental s used to estmate the actual property rental prce. 5
market. 12 In the luxury property, there are 143, 872, and 503 observatons for Hong Kong, Shangha, and Tape, respectvely. 3: Models and Methodology Two models are employed n the current research: a specfc model and a comparatve model. Snce more nformaton on the Hong Kong property rental market s avalable, a specfc model s constructed for Hong Kong, whle a comparatve model wth fewer parameters specfed s used for the markets n Shangha and Tape nstead of comparng results across the three markets, as less nformaton s publshed for the Shangha and Tape property rental market. For both models, the lnear model s adopted; as accordng to the Handbook on Resdental Property Prces Indces (Eurostat, 2013), the lnear model has much to recommend t when the property sze s ncluded as an explanatory varable. 3.1 Specfc Model The specfc model s constructed as follows: prce = β + β age + β age 1 + β FSD + β DMI 8 + β hopewell + β hendersonland 13 17 2 9 3 + e 2 + β sze + β hgh + β medum + β tme + β hutchson + β hanglung + β swre + β sunhungka + β newworld 10 14 4 5 11 15 6 12 + β cheungkong 16 7 (2) where prce denotes the property rental prce age denotes the property age sze denotes the property sze hgh s the dummy varable for hgh floor medum s the dummy varable for medum floor tme denotes the travel tme from the property to the CBD va the ralway systems 12 Hgh ncome people lvng n luxury property use subway nstead of drvng as ther basc commutng mode because parkng lots are lmted n CBD, whle traffc jams durng peak hours also nduce most of the hgh ncome group people to take subway to work. 6
FSD s the dummy varable for famous school dstrct DMI denotes the dstrct medan ncome swre s the dummy varable for the Swre Group sunhungka s the dummy varable for Sun Hung Ka Propertes Ltd. newworld s the dummy varable for New World Development Co. Ltd. hendersonland s the dummy varable for Henderson Land Development Co. Ltd. hutchson s the dummy varable for Hutchson Whampoa Ltd. hanglung s the dummy varable for Hang Lung Holdngs Ltd. cheungkong s the dummy varable for Cheung Kong Holdngs Ltd. hopewell s the dummy varable for Hopewell Holdngs Ltd. The varable age 2 s added to capture the nonlnear age effect on the property rental prce. The floor dummy varables, hgh and medum, are used to demarcate the general heght of the floor nstead of usng the exact floor number. The famous school effect matters because the resdental locaton of students s an mportant factor n school admssons. Lvng n dstrcts wth famous schools n the vcnty mples a greater chance for the chldren theren to be accepted nto a famous school, thus affectng the prce of the property. As a result, the famous school rato s appled to dfferentate dstrcts wth more famous schools. The rato s defned as follows: Number of band one schools n that dstrct Famous school rato = Total number of schools n that dstrct 13 (3) Table 1 presents the famous school rato of all 18 dstrcts n Hong Kong. The top four dstrcts wth the hghest famous school ratos are Central and Western, Wan Cha, Yau Tsm Mong, and Kowloon Cty, wth the ratos hgher than 1/3. The famous school dstrct (FSD) equals one f the property s located n one of these four 13 In Hong Kong, there are three bandngs that represent the rankng of secondary schools, wth band one schools defned as famous schools. 7
dstrcts and zero otherwse. The dstrct medan ncome 14 (DMI) denotes the medan monthly domestc household ncome n each dstrct, whch covers the households n prvate propertes only. The DMI s appled to measure the wealth effect n dfferent dstrcts, as the household ncome and purchasng power vary among dstrcts. In ths paper, eght dummy varables are used to capture the developer effect 15. Table 2 presents the summary statstcs of the three data sets. 3.2 Comparatve Model The major regressors reman n the comparatve model to compare the property rental markets n Hong Kong, Shangha, and Tape; however, the mnor regressors n Shangha and Tape markets are removed because of nsuffcent nformaton. The comparatve model s constructed as follows: prce β β + β + + + + 2 = 1 + 2age 3age β4sze β5 floor β6tme e (4) Informaton on the exact property floor number n the Shangha and Tape property rental markets s avalable. The floor varable n the above model denotes the correspondng floor number of the rental property. For the property rental market n Hong Kong, floor dummy varables are used nstead to measure the effect of the floor that the property s located. Tables 3 to 5 show the summary statstcs of the three data sets n the comparatve models n Hong Kong, Shangha, and Tape, respectvely. 14 The 2010 data on the medan monthly domestc household ncome are obtaned from the Hong Kong Census and Statstcs Department by request. 15 The eght major property companes refer to the Swre Group, Sun Hung Ka Propertes Ltd., New World Development Co. Ltd., Henderson Land Development Co. Ltd., Hutchson Whampoa Ltd., Hang Lung Holdngs Ltd., Cheung Kong Holdngs Ltd., and Hopewell Holdngs Ltd. 8
3.3 Estmaton of the Household Tme Value and the Household Shadow Wage The regresson coeffcent for travel tme (β 7 n the specfc model for Hong Kong or β 6 n the comparatve models for Shangha and Tape) s estmated. Ths tme coeffcent measures the addtonal property rental prce that households are wllng to pay n order to lve closer to the CBD, and thus for every mnute saved on travel per month. For Hong Kong, β = transportaton cost average workng days per month 2 7 + household tmevalue (5) whereas for Shangha and Tape, β average workng days per month 2 6 = transportaton cost + household tmevalue (6) In Equatons (5) and (6), the tme coeffcent s dvded by the average workng days per month to measure the tme coeffcent per day nstead of per month. Specfcally, the average workng days per month of the people n Hong Kong, Shangha, and Tape are 25, 21.75, and 22.4 days, respectvely 16. The coeffcent s also dvded by two as the travel tme accounts for a round trp. The left-hand sde of Equatons (5) and (6) ndcates how much a household s wllng to pay, whch comprses the transportaton cost and household tme value, n order to save a mnute of travel tme. In Equaton (7), transportaton cost 17 s calculated as the mean of the travelng cost dvded by the travel tme from each staton to the CBD. 16 The average workng days per month of the people n Hong Kong, Shangha, and Tape are obtaned from the Hong Kong Census and Statstcs Department, the General Offce of the State Councl of the People's Republc of Chna, and the Councl of Labor Affars n Tawan, respectvely. 17 Informaton on the transportaton cost s obtaned from the Hong Kong Mass Transt Ralway, Shangha Metro, and Tape Rapd Transt System, respectvely. 9
travelng cost Transporta ton cost = mean( ) (7) travel tme Snce the nformaton on the number of famly members per household s unavalable, the tme value, shadow wage and estmated monthly ncome calculated from the tme coeffcents are on a household bass. The household tme value can be obtaned by deductng the transportaton cost from the left-hand sde of Equatons (5) and (6). The obtaned household tme value s calculated on a mnute bass. In order to measure the household shadow wage, whch s then calculated on an hourly bass, the household tme value s multpled by 60 as shown n Equaton (8): Household shadow wage = 60 household tme value (8) In Equaton (9), the monthly household ncome can be estmated by multplyng the household shadow wage by the average workng days per month and the average workng hours per day. Estmated monthly household ncome = household shadow wage average workng days per month average workng hours per day (9) The average workng hours per day of the people n Hong Kong, Shangha, and Tape are 8.6, 8, and 8.2 hours, respectvely 18. The data on the transportaton cost, average workng days per month, and average workng hours per day n the three ctes are presented n Table 6. 4: Results The results of the specfc model and the comparatve model are shown n Sectons 4.1 and 4.2 respectvely. 18 Informaton on the average workng hours per day n Hong Kong, Shangha, and Tape s obtaned from the Hong Kong Census and Statstcs Department, the General Offce of the State Councl of the People's Republc of Chna, and the Councl of Labor Affars n Tawan, respectvely. 10
4.1 Results of the Specfc Model Table 7a llustrates the estmaton results for all three data sets of the specfc model n Hong Kong. For robustness check, we also estmate a model wthout the developer dummes n Table 7b. The estmated tme coeffcents are close to those n Table 7a. The coeffcents for the tme varables n all three data sets are statstcally sgnfcant at the 1% level. A negatve sgn ndcates a negatve relaton between the property rental prce and the travel tme between the property and the CBD,.e., the shorter the travel tme, the hgher the property rental prce. Ths fndng confrms the hypothess made n ths paper and the results obtaned by Tang (1975), and Tse and Chan (2003). The wllngness of households to save travel tme to and from the CBD by payng a hgher rent to lve n a property closer to the CBD s shown n the coeffcents. The coeffcents of the travel tme requred to arrve at the CBD ndcate how much, on average, households resdng n prvate propertes are wllng to pay n order to save a mnute of ther travel tme from ther home to the CBD per month. From Table 7a, the tme coeffcents n the all property data set, small and md-sze property data set, and luxury property data set are -145.6, -147, and -392.4, respectvely. Ths shows that for the all property data set, households n prvate propertes are wllng to pay HK$145.6 each month, on average, n order to save a mnute of ther travel tme, n other words, around HK$2,912 (HK$145.6x20) more to stay n a property whch reduces 20 mnutes travel to the CBD. Smlarly, by usng the tme coeffcents for the small and md-sze property data set and the luxury property data set, t s shown that households are wllng to pay an addtonal sum of HK$2,940 (HK$147x20) and HK$7,848 (HK$392.4x20) respectvely for a property that takes 20 mnutes less travel from the CBD. Usng the nformaton that the average workng days per month and the transportaton cost per mnute n Hong Kong are 25 days and HK$0.5556 respectvely, Equaton (5) can be used to calculate the tme value and the shadow wage of Hong Kong households n the three data sets. Based on the all property data set n Hong Kong, the household tme value s HK$2.36 per mnute, so the household shadow wage s HK$141 per hour on average. By multplyng the average workng 11
days per month and the average workng hours per day, the estmated monthly ncome of a household lvng n prvate property s HK$30,398. The regressors, age and age 2, are added to obtan the quadratc shape of the property age effect on the property rental prce. As shown n Table 7a, the values of β 3, whch are all sgnfcant at the 1% level, are negatve n all three data sets. β 3 also shows an nverse U-shape relaton between the property rental prce and property age. In other words, propertes bult n a later perod affect the rental prce postvely, whereas propertes bult earler have a negatve mpact on the rental prce. One explanaton for the nverse U shape s that the usable area of newly constructed propertes has declned sgnfcantly n recent years, and they are less preferred by households, whle older propertes wth larger usable areas are favored. As a result, the property rental prce ncreases n relaton to ncreasng property age at an early stage. However, when the property s too old, the qualty of the property s perceved to have declned, and the safety of the property s questoned. The turnng pont s when the perceved value of property starts to declne, whch trggers a rental prce downfall. The turnng pont can be obtaned by settng β 2β 2 age =. As shown n Table 7 and Equaton (11), the turnng ponts of property age n the all property data set, small and md-sze property data set, and luxury property data set are 20.9 years, 20.6 years, and 15.4 years, respectvely. 3 Table 8 dsplays the travel tme coeffcent, household tme value, household shadow wage, and the estmated monthly household ncome of all three data sets n Hong Kong. For example, smlar results can be obtaned n the luxury property market n whch the household tme value, household shadow wage, and the estmated monthly household ncome are HK$7.29 per mnute, HK$437.54 per hour, and HK$94,072 per month, respectvely. The household shadow wage and monthly household ncome n dfferent data sets can thus be estmated. 12
4.2 Results of the Comparatve Model Tables 9 to 11 show the estmaton results of the comparatve models of all three data sets n Hong Kong, Shangha, and Tape, respectvely. Smlar to the specfc model, the tme varables of all three data sets n the comparatve models of the three ctes are 99% statstcally sgnfcant. The relevant coeffcents shown n Tables 9 to 11 llustrate the property age effects on the property rental markets n Hong Kong, Shangha, and Tape. Only the cases of the Hong Kong and Tape markets are presented, as that of Shangha s not sgnfcant at the 10% level. For Hong Kong, the relaton between the property rental prce and the property age remans an nverse U shape. However, the turnng pont appears at 19.36 years rather than 20.9 years n the specfc model n Equaton (11). 19 The travel tme coeffcent, household tme value, household shadow wage, and estmated monthly household ncome of all three data sets n the comparatve model n the three ctes are calculated and lsted n Table 12. Note that the coeffcent of the tme varable for Hong Kong n Table 7a s -145.6 n the specfc model and -256.1 n the comparatve model. The estmated coeffcent of TIME of the specfc model should be more precse compared to the one from the comparatve model because more varables are used n the former model. For robustness check, we also estmate a model wthout the developer dummes n Table 7b. The estmated tme coeffcents are close to those n Table 7a, suggestng that there s no multcollnearty among the developer dummes and other varables. Snce fewer regressors are ncluded n the comparatve model, we wll only use the result of the comparatve model for comparson among the three ctes. 19 The property rental market n Tape exhbts characterstcs exactly opposte to those of ts Hong Kong counterpart: the former shows a U-shape relaton between the property rental prce and property age. In other words, the property age affects the property rental prce negatvely durng the early stage but postvely at a later stage. The negatve relaton before the turnng pont n ths U-shape pattern can be explaned by households who prefer newer flats than other smlar-qualty propertes. The postve relaton after the turnng pont n the U-shape pattern of the Tape market can be explaned by the proxmty of the propertes constructed durng earler decades to the CBD. Households prefer these earler constructed propertes due to ther locaton. Consequently, property rental prces start to ncrease when the property age reaches the turnng pont. Takng the all property data set as an example, the turnng ponts of property age n Hong Kong and Tape markets are 19.36 and 21.61 years, respectvely. 13
The ratos, as opposed to the absolute numbers, wll be the focus n the comparatve model. In partcular, the estmated coeffcent of the comparatve model wll manly be used to dentfy the rato of the tme values of ctzens n the three places. Takng the whole property data set as an example, the tme values are presented n Table 12. The tme value rato of the households n Hong Kong, Shangha, and Tape s approxmately 2.25: 1: 1.61. Ths fndng demonstrates that among these three ctes, travel tme s valued most n Hong Kong, as more has to be pad for property closer to the CBD; whereas t s valued least n Shangha. The extent of dfference between the households of these two ctes s more than double. Ths rato also ndcates that approxmately 1.61 and 2.25 tmes more rent have to be pad n Tape and Hong Kong than n Shangha, respectvely, for lvng closer to the CBD. The household shadow wage rato among the three ctes remans at 2.25: 1: 1.61, as ths rato s obtaned by smply multplyng the correspondng household tme value by a fxed number, 60, for all three ctes. Nevertheless, the estmated monthly household ncome rato s 2.78: 1: 1.70, as the two key components, the average workng days per month and the average workng hours per day, vary among the three ctes. Therefore, the estmated monthly household ncome of Hong Kong and Tape s 2.78 and 1.70 tmes that of Shangha respectvely. In order to verfy the valdty of ths rato n these three ctes, the rato s then used to compare wth dfferent salary ndexes of these ctes. Table 13 llustrates the average salary levels of varous occupatonal sectors. 20 The salary comparson ratos of the selected occupatonal sectors n the three ctes are then calculated based on the salary nformaton n Table 13 and the resultng ratos are presented n Table 14. Tables 13 and 14 show that among the three ctes, gven the same sector, Hong Kong has the hghest salary on average, whle Shangha has the lowest. The salary comparson ratos are consstent wth the estmated monthly household ncome rato and reflect the general stuatons n Hong Kong, Shangha, and Tape. 20 Occupatonal sectors consst of unversty graduate, polce, teacher, nformaton technology, logstcs and shppng, desgn, manufacturng, engneerng, real estate and property, and food and beverage. The salary nformaton on cvl servants such as polce s from the correspondng government salary ndex tables of the three ctes. The salary nformaton on the remanng occupatonal sectors s from Centalne Human Resources Consultants Lmted and Classfed Post n Hong Kong, Baca Recrutment Agent n Shangha and the 1111 Job Bank n Tape. 14
5: Concluson The central busness dstrct (CBD), where most of the commercal offces are located, plays a crtcal role n the economc development of a cty. Most people prefer to lve close to the CBD to save the commutng cost. Snce the rent that an ndvdual s wllng to pay to lve near the CBD depends on the value of the tme saved from shortenng ther commute, the dstance between property and the CBD s an mportant factor that determnes property rental value. The nearer to the CBD, the hgher the rental value of a property. Ths paper examnes how the rental dfferental between two locatons n a metropoltan s determned by the tme value of a household. The rental nformaton on propertes atop or adjacent to the ralway statons n the CBDs of Hong Kong, Shangha and Tape s analyzed. Compared wth alternatve publc transportaton modes, the ralway schedule provdes better nformaton on commutng tme for research purposes. Our emprcal result ndcates that the commutng tme to the CBD s an mportant factor n determnng the rent of a resdental property. All the tme coeffcents under the specfc model for Hong Kong and the comparatve model for Hong Kong, Shangha, and Tape are found to be statstcally sgnfcant at the 1% level. The household tme value, household shadow wage, and monthly ncome of the households n these three ctes can be recovered from our models. It s found that the tme value rato of the households n Hong Kong, Shangha, and Tape s about 2.25: 1: 1.61. The estmated level of the shadow wage n the three ctes and ther respectve ratos are consstent wth the emprcal data, whch provde evdence that the rental prce dfferental between two locatons n a metropoltan s a reflecton of the total value of the commutng tme dfferental. For future research along ths lne, one may nclude other Asan major ctes where publc transportaton s the man mode of commutng, and ctes wth multple CBDs. Fnally, as dfferent ncome groups perceve tme value dfferently, the threshold model of Hansen (2000) may be used to analyze f there s a threshold effect above whch the mpact of travel tme to the CBD on the value of a property has a substantal ncrease. 15
Table 1: Famous school ratos of 18 dstrcts n Hong Kong Dstrcts Famous school rato Central and Western 0.6364 Wan Cha 0.5385 Yau Tsm Mong 0.4 Kowloon Cty 0.3871 Sha Tn 0.3158 North 0.3 Eastern 0.2903 Sham Shu Po 0.2727 Tsuen Wan 0.2308 Ta Po 0.2273 Kwa Tsng 0.2258 Tuen Mun 0.2162 Yuen Long 0.2 Kwun Tong 0.1935 Sa Kung 0.1905 Wong Ta Sn 0.1818 Islands 0.125 Southern 0.0714 16
Table 2: Summary statstcs of the three data sets n the specfc model n Hong Kong between January 2011 and March 2011 Varable Mean Std. Dev. Mn Max All S & M Luxury All S & M Luxury All S & M Luxury All S & M Luxury prce 13727.83 12448.86 22161.85 6303.44 4057.42 10577.80 1500 1500 8000 82000 40000 82000 age 14.0460 14.9533 8.0629 10.6078 10.3860 10.1338 1 1 1 36 36 36 age 2 309.7127 331.3563 166.9860 353.6497 348.1050 357.9611 1 1 1 1296 1296 1296 sze 741.930 667.674 1231.601 258.847 169.980 205.407 292 292 1007 2416 999 2416 hgh 0.4742 0.4634 0.5455 0.4996 0.4989 0.4997 0 0 0 1 1 1 medum 0.2459 0.2460 0.2448 0.4308 0.4309 0.4315 0 0 0 1 1 1 tme 29.2947 29.4199 28.4685 9.9654 9.9150 10.2887 3 3 3 54 54 50 FSD 0.0783 0.0636 0.1748 0.2687 0.2442 0.3812 0 0 0 1 1 1 DMI 27771.92 27712.62 28162.94 5660.20 5536.05 6427.09 18000 18000 18000 33600 33600 33600 swre 0.0654 0.0594 0.1049 0.2473 0.2365 0.3075 0 0 0 1 1 1 sunhungka 0.1446 0.1400 0.1748 0.3518 0.3471 0.3812 0 0 0 1 1 1 newworld 0.0359 0.0286 0.0839 0.1862 0.1669 0.2782 0 0 0 1 1 1 hendersonland 0.0046 0.0042 0.0070 0.0677 0.0650 0.0836 0 0 0 1 1 1 hutchson 0.0276 0.0318 0.0000 0.1640 0.1756 0.0000 0 0 0 1 1 0 hanglung 0.0746 0.0742 0.0769 0.2628 0.2623 0.2674 0 0 0 1 1 1 cheungkong 0.1667 0.1919 0.0000 0.3729 0.3940 0.0000 0 0 0 1 1 0 hopewell 0.0414 0.0477 0.0000 0.1994 0.2133 0.0000 0 0 0 1 1 0 Note: All refers to the all property data set, S & M refers to the small and md-sze property data set, and Luxury refers to the luxury property data set. 17
Table 3: Summary statstcs of the three data sets n the comparatve model n Hong Kong between January 2011 and March 2011 Varable Mean Std. Dev. Mn Max All S & M Luxury All S & M Luxury All S & M Luxury All S & M Luxury prce 13727.83 12448.86 22161.85 6303.44 4057.42 10577.80 1500 1500 8000 82000 40000 82000 age 14.0460 14.9533 8.0629 10.6078 10.3860 10.1338 1 1 1 36 36 36 age 2 309.713 331.356 166.986 353.650 348.105 357.961 1 1 1 1296 1296 1296 sze 741.930 667.674 1231.601 258.847 169.980 205.407 292 292 1007 2416 999 2416 hgh 0.4733 0.4624 0.5455 0.4995 0.4988 0.4997 0 0 0 1 1 1 medum 0.2477 0.2481 0.2448 0.4319 0.4322 0.4315 0 0 0 1 1 1 tme 29.2947 29.4199 28.4685 9.9654 9.9150 10.2887 3 3 3 54 54 50 Note: All refers to the all property data set, S & M refers to the small and md-sze property data set, and Luxury refers to the luxury property data set. 18
Table 4: Summary statstcs of the three data sets n the comparatve model n Shangha between January 2011 and March 2011 Varable Mean Std. Dev. Mn Max All S & M Luxury All S & M Luxury All S & M Luxury All S & M Luxury prce 6152.29 3465.06 8830.27 6453.54 2181.01 8004.74 534.825 534.825 534.825 85572 27335.5 85572 age 8.7220 10.6306 6.8200 5.7137 6.8503 3.3462 0 0 0 83 83 23 age 2 108.701 159.883 57.696 206.879 277.922 57.507 0 0 0 6889 6889 529 sze 1044.199 657.624 1429.444 515.342 220.979 429.184 107.64 107.64 1001.052 4929.912 990.288 4929.912 floor 9.8093 7.2670 12.3429 8.0998 6.2522 8.9012 1 1 1 53 45 53 tme 29.3510 32.5167 26.1961 18.2652 17.3524 18.6128 0 0 0 81 81 81 Note: All refers to the all property data set, S & M refers to the small and md-sze property data set, and Luxury refers to the luxury property data set. For easy comparson, the property rental prces n the Shangha market are converted to Hong Kong dollars by multplyng ther average exchange rate from January 2011 to March 2011, whereas the sze unt of Shangha propertes s converted from square meters to square feet. 19
Table 5: Summary statstcs of the three data sets n the comparatve model n Tape between January 2011 and March 2011 Varable Mean Std. Dev. Mn Max All S & M Luxury All S & M Luxury All S & M Luxury All S & M Luxury prce 9251.25 7785.32 10387.86 3564.88 2456.10 3865.60 2147.2 2147.2 3489.2 25498 17446 25498 age 14.2231 12.4167 15.6237 9.2206 9.5759 8.6899 1 1 1 50 39 50 age 2 287.220 245.636 319.463 306.089 312.195 297.602 1 1 1 2500 1521 2500 sze 1137.512 836.482 1370.915 340.548 116.590 265.6154 286.654 286.654 1002.222 2489.55 999.377 2489.55 floor 6.2128 7.0462 5.5467 3.8668 4.3288 3.3336 1 1 0 26 26 22 tme 12.7503 13.4923 12.1750 7.8851 8.6196 7.2211 0 0 0 45 45 45 Note: All refers to the all property data set, S & M refers to the small and md-sze property data set, and Luxury refers to the luxury property data set. For easy comparson, the property rental prces n the Tape market are converted to Hong Kong dollars by multplyng ther average exchange rate from January 2011 to March 2011, whereas the sze unt of Tape propertes s converted from pyeong to square feet. 20
Table 6: Summary of transportaton cost, average workng days per month, and average workng hours per day Hong Kong Shangha Tape Transportaton cost 0.5556 0.1875 0.4383 Average workng days per month 25 21.75 22.4 Average workng hours per day 8.6 8 8.2 Note: The transportaton costs n Shangha and Tape are converted to Hong Kong dollars by multplyng ther average exchange rates from January 2011 to March 2011. 21
Table 7a: Estmaton results of the three data sets n the specfc model n Hong Kong between January 2011 and March 2011 VARIABLES All property data set Hong Kong Small and md-sze property data set Luxury property data set Age 509.6*** 353.0*** 1,106*** (40.53) (30.77) (218.8) age 2-12.22*** -8.566*** -35.80*** (1.18) (0.869) (7.479) Sze 19.15*** 16.24*** 30.93*** (0.468) (0.556) (2.296) Hgh 483.8** 404.7** 2,627** (226.6) (160.5) (1,241) Medum 577.4** 299.8 3,321** (261.2) (185.2) (1,452) Tme -145.6*** -147.0*** -392.4*** (12.32) (8.798) (122.7) FSD 6,985*** 5,296*** -2,511 (474.4) (350.2) (5,999) DMI -0.035-0.0159-0.690** (0.0226) (0.0156) (0.312) Swre 910.0** 791.5** 6,823*** (458.4) (340.9) (2,247) sunhungka 738.9** 124.5-710.5 (326.7) (250.4) (1,753) Newworld -1,056* -46.85-13,722*** (579.5) (437.6) (3,429) 22
VARIABLES All property data set Hong Kong Small and md-sze property data set Luxury property data set hendersonland 3,477** 2,462** -1,361 (1,423) (1,069) (6,408) Hutchson -506.8-35.95 0 (665.1) (448.5) (0) Hanglung 1,916*** -99.36 3,492 (569.7) (455) (2,413) cheungkong 1,076*** 659.8*** 0 (306) (208.8) (0) Hopewell -1,426** 392.6 0 (722.5) (534.5) (0) Constant 78.06 3,125*** 10,241 (973.5) (820.2) (13,804) Observatons 1,086 943 143 R-squared 0.767 0.75 0.787 Standard errors n parentheses *** p<0.01, ** p<0.05, * p<0.1 23
Table 7b: Estmaton results of the three data sets n the specfc model (property company dummy varables are excluded) n Hong Kong between January 2011 and March 2011 VARIABLES All property data set Hong Kong Small and md-sze property data set Luxury property data set Age 557.1*** 361.0*** 1,210*** (38.26) (29.12) (167.6) age 2-13.62*** -8.908*** -30.73*** (1.092) (0.818) (4.904) Sze 18.86*** 15.78*** 30.98*** (0.454) (0.518) (2.421) Hgh 479.4** 428.6*** 1,442 (228.9) (161.1) (1,253) Medum 564.3** 294.5 2,541* (263.9) (185.8) (1,450) Tme -162.3*** -149.5*** -382.1*** (11.20) (7.863) (112.9) FSD 6,680*** 5,039*** 7,718* (452.3) (329.6) (4,509) DMI -0.0367* -0.0216 0.0184 (0.0216) (0.0148) (0.202) Constant 1,014 3,872*** -13,011 (901.2) (726.1) (9,434) Observatons 1,086 943 143 R-squared 0.758 0.744 0.745 Standard errors n parentheses *** p<0.01, ** p<0.05, * p<0.1 24
Table 8: Summary of the travel tme coeffcent, household tme value, household shadow wage, and the estmated monthly household ncome n all three data sets n Hong Kong All property data set Hong Kong Small and md-sze property data set Luxury property data set Coeffcent of travel tme -145.6-147 -392.4 Household tme value 2.3564 2.3844 7.2924 Household shadow wage 141.384 143.064 437.544 Estmated monthly household ncome 30397.56 30758.76 94071.96 25
Table 9: Estmaton results of the all property data set n the comparatve model between January 2011 and March 2011 VARIABLES Hong Kong Shangha Tape age 648.3*** -46.76-312.6*** (42.53) (31.83) (31.65) age 2-16.74*** 1.147 7.231*** (1.208) (0.808) (0.946) sze 19.40*** 7.736*** 5.700*** (0.509) (0.218) (0.267) floor hgh: 532.0** 62.38*** 58.50** medum: 430.3 (257.5) (14.53) (23.68) (296.3) tme -256.1*** -96.31*** -166.0*** (10.81) (6.455) (11.53) Constant 2,559*** 571.9 6,889*** (711.2) (520.3) (441.8) Observatons 1,086 1,741 893 R-squared 0.693 0.595 0.458 Standard errors n parentheses ***p<0.01, **p<0.05,*p<0.1 Note: For easy comparson, the property rental prces n the Shangha and Tape markets are converted to Hong Kong dollars by multplyng ther average exchange rates from January 2011 to March 2011, whereas the sze unt of Shangha propertes s converted from square meters to square feet, and that of Tape propertes s converted from pyeong to square feet. 26
Table 10: Estmaton results of the small and md-sze property data set n the comparatve model between January 2011 and March 2011 VARIABLES Hong Kong Shangha Tape age 398.7*** -73.36*** -275.0*** (33.14) (15.33) (36.37) age 2-10.41*** 0.683** 6.303*** (0.928) (0.337) (1.114) sze 15.38*** 3.074*** 4.056*** (0.591) (0.264) (0.871) floor hgh: 454.0** 66.82*** 30.72 medum: 183.1 (184.2) (10.09) (23.94) (211.7) tme -205.9*** -63.11*** -132.6*** (8.018) (3.723) (11.86) Constant 5,469*** 3,681*** 7,831*** (660.2) (321) (756.8) Observatons 943 869 390 R-squared 0.665 0.451 0.361 Standard errors n parentheses ***p<0.01, **p<0.05,*p<0.1 Note: For easy comparson, the property rental prces n the Shangha and Tape markets are converted to Hong Kong dollars by multplyng ther average exchange rates from January 2011 to March 2011, whereas the sze unt of Shangha propertes s converted from square meters to square feet, and that of Tape propertes s converted from pyeong to square feet. 27
Table 11: Estmaton results of the luxury property data set n the comparatve model between January 2011 and March 2011 VARIABLES Hong Kong Shangha Tape age 1,262*** -149.2-340.8*** (160.1) (163.3) (48.74) age 2-35.37*** -4.99 7.844*** (4.507) (9.344) (1.417) sze 31.75*** 11.84*** 6.239*** (2.448) (0.41) (0.513) floor hgh: 1313 63.55*** 99.02** medum: 2,451* (1271) (21.42) (41.38) (1460) tme -633.7*** -120.1*** -202.0*** (48.7) (10.92) (19.50) Constant -4,486-4,431*** 6,564*** (3675) (1096) (935.0) Observatons 143 872 503 R-squared 0.731 0.619 0.396 Standard errors n parentheses ***p<0.01, **p<0.05,*p<0.1 Note: For easy comparson, the property rental prces n the Shangha and Tape markets are converted to Hong Kong dollars by multplyng ther average exchange rates from January 2011 to March 2011, whereas the sze unt of Shangha propertes s converted from square meters to square feet, and that of Tape propertes s converted from pyeong to square feet. 28
Table 12: Summary of the travel tme coeffcent, household tme value, household shadow wage, and the estmated monthly household ncome n all three data sets n Hong Kong, Shangha, and Tape All property data set Small and md-sze property data set Luxury property data set Hong Kong Shangha Tape Coeffcent of travel -256.10 tme -96.31-166.00 Household tme value 4.57 2.03 3.27 Household shadow wage 273.98 121.59 196.02 Estmated monthly household ncome 58906.56 21156.90 36005.58 Coeffcent of travel -205.90-63.11 tme -132.60 Household tme value 3.56 1.26 2.52 Household shadow wage 213.74 75.80 151.29 Estmated monthly household ncome 45954.96 13188.90 27789.18 Coeffcent of travel -633.70-120.10 tme -202.00 Household tme value 12.12 2.57 4.07 Household shadow wage 727.10 154.41 244.24 Estmated monthly household ncome 156327.36 26866.50 44861.58 29
Table 13: Comparson of the salary ndexes of the 10 selected occupatonal sectors n Hong Kong, Shangha and Tape n 2010 Hong Kong Shangha Tape Unversty Graduate 14300 3861 5736 Polce 17250 4975 13301 Teacher 19945 5437 12092 Informaton Technology 13000 6708 7630 Logstcs and Shppng 15000 4856 7563 Desgn 17000 7069 9141 Manufacturng 12000 6393 7671 Engneerng 18000 7357 9818 Real Estate and Property 18000 8685 9507 Food and Beverage 10000 4982 6348 Note: The salary ndexes of Shangha and Tape are converted to Hong Kong dollar for easy comparson. 30
Table 14: Salary comparson ratos of the 10 selected occupatonal sectors n Hong Kong, Shangha and Tape n 2010 Hong Kong Shangha Tape Unversty Graduate 3.70 1 1.49 Polce 3.47 1 2.67 Teacher 3.67 1 2.22 Informaton Technology 1.94 1 1.14 Logstcs and Shppng 3.09 1 1.56 Desgn 2.40 1 1.29 Manufacturng 1.88 1 1.20 Engneerng 2.45 1 1.33 Real Estate and Property 2.07 1 1.09 Food and Beverage 2.01 1 1.27 31
References: Alonso, W. (1964), Locaton and Land Use: Toward a General Theory of Land Rent, Cambrdge: Harvard Unversty Press. Bajar, P. and M. E. Kahn (2008), Estmatng hedonc models of consumer demand wth an applcaton to urban sprawl, n Baranzn, A., J. Ramrez, C. Schaerer and P. Thalmann (Eds.), Hedonc Methods n Housng Markets: Prcng Envronmental Amentes and Segregaton, New York: Sprnger, pp. 129-155. Eurostat (2013), Handbook on Resdental Property Prces Indces, Luxembourg: Eurostat. Hansen, B. E. (2000), Sample splttng and threshold estmaton, Econometrca, 68(3), pp. 575-603. Heckman, J. (1974), Shadow prces, market wages, and labor supply, Econometrca, 42(4), pp. 679-694. Henderson, J. V. (1977), Economc Theory and The Ctes, New York: Academc Press. Kan, J. F. and J. M. Qugley (1970), Measurng the value of housng qualty, Journal of the Amercan Statstcal Assocaton, 65(330), pp. 532-548. Mahan, B. L., S. Polasky and R. M. Adams (2000), Valung urban wetlands: a property prce approach, Land Economcs, 76(1), pp. 100-113. McMllen, D. P. (2006), Testng for Monocentrcty, n Arnott, R. J., D. P. McMllen (Eds.), A Companon to Urban Economcs, Oxford: Blackwell, pp. 128-140. 32
McMllen, D. P. and L. D. Sngell, Jr. (1992), Work locaton, resdence locaton, and the ntraurban wage gradent, Journal of Urban Economcs, 32(2), pp. 195-213. Mlls, E. S. (1969), The value of urban land, n Perloff, H. S. (Ed.), The Qualty of the Urban Envronment: Essays on New Resources n an Urban Age, Washngton, D.C.: Resources for the Future, pp. 231-253. Muth, R. F. (1969), Ctes and Housng: The Spatal Pattern of Urban Resdental Land Use, Chcago: Unversty of Chcago Press. Nelson, J. P. (1978), Resdental choce, hedonc prces, and the demand for urban ar qualty, Journal of Urban Economcs, 5(3), pp. 357-369. So, H. M., R. Y. C. Tse and S. Ganesan (1997), Estmatng the nfluence of transport on house prces: evdence from Hong Kong, Journal of Property Valuaton & Investment, 15(1), pp. 40-47. Stegman, M. A. (1969), Accessblty models and resdental locaton, Journal of the Amercan Insttute of Planners, 35(1), pp. 22-29. Tang, F. T. (1975), Detecton and Estmaton of Transportaton Impact wth Models of Suburban Resdental Property Sales Prces, Tape: Insttute of Economcs, Academa Snca. Tse, C. Y. and A. W. H. Chan (2003), Estmatng the commutng cost and commutng tme property prce gradents, Regonal Scence and Urban Economcs, 33(6), pp. 745-767. 33