FINANCIAL SUSTAINABILITY OF RAIL TRANSIT SERVICES CASE STUDY IN HONG KONG

Transcription

1 FINANCIAL SUSTAINABILITY OF RAIL TRANSIT SERVICES CASE STUDY IN HONG KONG Z. W. WANG PhD Student Depatment of Civil Engineeing The Hong Kong Univesity of Science and Technology Clea Wate Bay Hong Kong, P.R.C Fax: Hong K. LO Pofesso Depatment of Civil Engineeing The Hong Kong Univesity of Science and Technology Clea Wate Bay Hong Kong, P.R.C Fax: Abstact: In many moden cities, ail tansit sevices ae not financially sustainable without subsidy. Howeve, the situation in Hong Kong is diffeent. It is believed that the high population density povides a suitable envionment fo the ail tansit system to be financially sustainable. In this study, a peliminay analysis on the demand and supply patten of the ail tansit sevice in Hong Kong was undetaken to shed light on the theshold density below which the ail sevice cannot be sustainable without subsidy, consideing its opeations vis-àvis the competing modes faes and sevices. We investigated the Mass Tansit Railway (MTR) line in the new town Tseung Kwan O (TKO) in Hong Kong as a case study. Key Wods: Rail tansit sevice, Uban density, Financial sustainability 1. INTRODUCTION Public tansit sevices (PTS) ae impotant assets to any majo city. It is ideal if the PTS ae financially sustainable, with affodable faes and expedient quality. The key lies in how the uban fom and density ae planned to ensue thei sustainability. Indeed, the elationship between uban density and tanspot chaacteistics has been studied. It has been shown that uban density is positively coelated with PTS usage. Pevious studies, howeve, analyzed the poblem desciptively based on statistical appoaches. Newman and Kenwothy (1999) found that highe uban density is positively coelated with lowe levels of ca owneship and ca use, and highe levels of tansit use. Sinha (2003) pointed to a simila conclusion, that the fundamental vaiable to inceased tansit use and sustainability is uban population density o activity density. Fom the liteatue, thee is no doubt that high population density is positively coelated with high tansit usage. These desciptive esults ae impotant and stimulating. Yet they ae not sufficiently efined to be adopted as policy guidelines. Basically, how high should the uban density be fo the povision of sustainable PTS? A pesciptive answe povides impotant guidelines fo integating the planning of uban development with sustainable PTS. Ganted high density developments ae not fo evey egion, the answe at least allows plannes to undestand the implications of PTS quality and sustainability in elation to a cetain density development. The answe to this pesciptive question is specific to the type of PTS unde consideation. In this study, we use the ail sevice in Hong Kong as the case study. Rail systems ae impotant to moden cities; they ae efficient, high speed and capacity, and have little impact on congestion. Peviously, some studies investigated how tansit sevices can be optimized in tems of coveage, coodination with bus sevices, o sevice quality 1719

2 impovement (examples, Wiasinghe and Hudle, 1977; Chang and Schonfeld, 1991; Spasovic and Schonfeld, 1993). An impotant aspect of planning ail sevices is thei financial sustainability. In many cities, ail tansit sevices ae often not financially sustainable and must ely on govenment subsidies. In Hong Kong, howeve, ail sevices ae pofitable. It is believed that the special uban fom and high population density of Hong Kong contibute to this positive outcome. It is valuable and inteesting to develop a deepe undestanding of the easons behind the success of ail sevices in Hong Kong. Fo this pupose, we study the case of the Mass Tansit Rail (MTR) line in Tseung Kwan O (TKO), which is Hong Kong s 7 th new town lying within the Sai Kung Distict in the southeasten pat of the New Teitoies. The total development aea of TKO is about km², with a population of aound 350,000. TKO is pimaily seved by the MTR TKO line; meanwhile, public bus and minibus sevices offe the taveles altenative modes. The MTR TKO line now has fou stations: Tiu Keng Leng, Tseung Kwan O, Hang Hau and Po Lam. All stations ae eithe connected to o in the poximity of shopping centes and lage pivate housing developments. In essence, the uban developments aound the ail stations in TKO aea by and lage fall within the concept of tansit-oiented development poposed by Calthope (1993), which povide a use-fiendly envionment fo the ail tansit sevices. In this study, we pesent an analytical method to examine the supply and demand pattens fo sustainable ail sevices, and investigate how the population density and the sevice s opeation stategy affect its pofitability in the pesence of competing modes. 2. FORMULATION In this fomulation, only two tavel modes ae consideed, ail and diect bus sevices. The ail sevice is complemented by feede bus sevices to take passenges fom moe distant locations to the ail stations. Individual commutes choose one of the two modes accoding to thei tavel costs. If one chooses the diect bus, he o she will have to walk to the neaest bus stop to access the bus sevice; on the othe hand, if one decides to use tansit sevice, he o she will take a feede bus o walk to the ail stations. Competition does exist between the two modes. The ail tansit optimizes its sevice quality to attact passenge demand so as to be pofitable o at least financially sustainable. Nevetheless, with a low uban density, it can be the case that no matte how the ail opeatos optimize thei sevices, the demand is so low that the sevice simply cannot be financially sustainable. In othe wods, thee exists a theshold uban density below which the ail tansit sevice cannot be sustainable. The focus of this study is to investigate this elationship between uban density and financial sustainability of ail sevices. 2.1 Assumptions To simplify the analysis, we make seveal assumptions as descibed below. (i) Homogeneous population: All taveles inside the study aea ae assumed to behave similaly in thei mode choices, which may be easonable fo Hong Kong as 90% of all tips in Hong Kong ae seved by public tansit sevices. Auto usage is not common. In any case, this assumption can be elaxed by segegating taveles into diffeent classes. (ii) Stochastic tavel choice behavio: Taveles do not have pefect infomation about the tavel modes. The logit model is used to descibe passenges modal choices, in which thei total tavel costs, including the total tavel time and monetay cost, ae consideed. 1720

3 (iii) Highway congestion: The tavel time of taffic on the highway is detemined by the commonly used BPR function to take into account the delay due to taffic congestion. 2.2 Model Fomulation The vaiable and paamete definitions ae as follows: b : Public bus d( x, y ) : population density at location ( x, y ) : Rail sevice P : Pobability of choosing ail sevice C : Tavel cost on the ail sevice C b : Tavel cost on the diect bus sevice H : Headway of the ail sevice H b : Headway of the bus sevice l t : Tavel distance of the tain l b : Tavel distance of diect public bus l f : Tavel distance of feede bus to ail stations l w : Walking distance V t : Aveage tain speed V b : Aveage bus speed at fee flow condition V f : Aveage feede bus speed V w : Speed of walking VoT i : Value of time while inside the tain o vehicle VoT walk : Value of time fo walking VoT wait : Value of time fo waiting c : Road capacity F : Fae of the ail sevice F b : Fae of the diect bus sevice κ : Aveage constuction cost pe day ψ : Rail opeation costs : Passenge demands fo ail tansit D Tavel Cost The tavel cost includes both fae and tavel time, with the latte conveted into monetay tems via the value of time. Walking is consideed as the access mode to the ail stations o bus stops, with the walking distance simply taken as the Euclidean distance between the oigin location and the intended station (stop). The tavel time includes access time, waiting time at stations, and in-vehicle time. Fo ail uses, the tavel cost is expessed in (1). Taveles choose the station and the neaest-by feede bus with the minimum tavel cost. l l w f l t C = VoTwalk + + VoTi H VoTWait + F V w Vf V t (1) 1721

4 Fo diect bus uses, the access time includes only the walking time. The tavel time on the highway is detemined by the BPR function to take into account the delay due to taffic congestion. The total tavel cost is witten as below, whee f descibes the taffic flow on the highway. β lw l b f C = VoT + 1+ α VoT H VoT + F Vw V b c b walk i b Wait b (2) Patonage Estimation The logit demand split model is applied to captue tip makes choices between the two x, y in the modes in accodance to thei tavel costs. Fo taveles oiginating fom point ( ) study egion, the tavel costs associated with the two diffeent tavel modes ae calculated as in equations (1) and (2), espectively. Thus, the pecentage of passenge demand at point x, y choosing the ail tansit sevice can be expessed as: ( ) P ( x, y) = e C ( x, y) (, ) (, ) C x y Cb x y e + e (3) Let the aveage total geneation ate fo public tansit tips be ρ, which includes both the bus and ail sevice passenges, and the study egion is epesented by Ω, then the patonage fo the ail sevice can be estimated as: (, ) (, ) ρ (4) Ω D = d x y P x y dxdy By solving (1)-(4), we obtain the patonage fo the ail sevice. Note that this system of equations involves a fixed-point elationship. The modal split detemines the taffic volume on the highway, which in tun detemines the congested tavel time fo the bus sevice and hence its modal split. In this way, the modal split o taffic volume and tavel time involve a fixed-point elationship, which will be detemined by an iteation method Total Costs of The Rail Sevice In the model, the total costs of the ail tansit sevice include both the fixed, depeciation cost on the tansit capital and the vaiable opeation cost that mainly depends on its headway. As the calculation is based on the houly passenge flow, the costs of constuction and opeation ae also expessed po-ata into houly cost units. The evenue is expessed as a negative tem to offset the cost. ( ) TC = κ + ψ H F D (5) The ail tansit capital cost and depeciation cost data ae available fom the account books of the MTR Copoation. The details of these cost analysis ae discussed in Tang and Lo (2007). The opeation cost is pesented as a function of the headway, which is the main opeation paamete captued in this study. If we assume γ p as exogenous paamete epesenting the aveage ail sevice opeation cost pe caying capacity-km, then the tem γ ClK / H can be used to estimate the opeation cost ψ ( H ), whee the capacity pe tain vehicle C, ail line p 1722

5 length l, and sevice time K ae given paametes. The evenue is subject to the fae and passenge demand fo the ail sevice. Meanwhile, the passenge demand depends on the total tip geneation, which in tun depends on the uban density, and passenges mode choices as shown in (3). 3. SOLUTION PROCESS In this fomulation, the uban density and the opeation paamete of the ail sevice, i.e., headway, ae two independent vaiables unde investigation. Fo a paticula uban density, the ail sevice changes thei sevice quality by vaying the headway to optimize fo its financial pefomance (i.e., evenue minus opeation costs). If the optimal headway and the coesponding total cost fo diffeent levels of uban density can be obtained, we can then study the sensitivity of the total cost to the uban density and subsequently obtain the theshold uban density below which financial sustainability cannot be guaanteed. 3.1 Detemination of The Catchment Aea The catchment aea is a useful concept in tansit sevice planning, especially fo evaluating the patonage of tansit lines. In this study, modal competition occus between the ail and bus sevices. The catchment aea fo the ail sevice is obtained by passenges choice between the two modes accoding to thei espective peceived disutilities. Fo illustation puposes, Figue 1 shows schematically the catchment zones associated with a station that is located at the cente of the concentic cicles. The aea aound the station is divided into diffeent zones o concentic ings. The value within each ing shows the pecentage of potential tip makes living in that zone would likely use the ailway tansit sevice as detemined by the demand split model. The exact geometic shape of the boundaies between zones depends on the disutility function, and the demand split model used. With the catchment aeas and total potential tip makes detemined, we can estimate the patonage fo a specific station. 70% 30% Figue 1 Station with catchment zones aound it The headway is an impotant paamete eflecting the ail sevice quality. A shot headway means shot waiting time and hence deceases the total cost fo using the sevice. Theefoe, educing the headway will enlage the catchment aea. In this study, we also conside taffic congestion on the highway, which eflects the advantage of the ail sevice ove the bus sevice. The catchment aea fo the ail sevice will be inceased afte consideing the congestion effect on the highway. Theefoe, the consideation of taffic congestion endes the catchment aea changed with the uban density. Imagine two scenaios fo the same aea with the same infastuctue and tansit sevices but diffeent uban densities. Even if the qualities of ail sevices ae equal, the catchment aea fo the scenaio with highe density will be lage, because highe density (theefoe moe tips geneated) would lead to highe 1723

6 congestion on the highway, thus inceasing the competitive edge of the congestion-fee ail sevice. This patially explains why ail sevices benefit moe fom highe uban density. 3.2 Solution Pocess The solution pocess involves solving the system (1)-(4) and putting the esults into (5). By fist fixing the uban density and headway, we compute the taffic flow on the highway iteatively so that the system (1)-(4) is satisfied. Then the flow esults ae substituted into (5) to calculate the total cost fo the ail sevice. Subsequently, we vay the headway and obtain the total cost fo diffeent headway values. In this way, we detemine the optimal headway fo a given uban density. Afte that, we vay the uban density and epeat the above pocess. Eventually, we obtain the sensitivity of uban density to the total cost of the ail sevice. 4. CASE STUDY IN HONG KONG 4.1 Population Density We detemine the population density distibution fo the entie study aea. Fistly, we set up 65 mete by 65 mete gids in the new town TKO, as shown in figue 2. To simplify the poblem, the population densities ae classified into fou categoies: high, medium, low, and zeo. Each gid is classified manually as pe these fou density categoies, depending on the actual situation and chaacteistics of the buildings lying within each gid. To illustate the esult, diffeent density categoies ae highlighted in Figue 2: gids with high esidential densities ae labeled with dake colos. Subsequently, we use the data fom the 2001 Census to calibate the exact density (in population pe sq. km) fo these fou population densities. With efeence to the final epot of 2001 Census, Hong Kong, TKO aea is divided into 13 zones. The population in each zone is known pecisely. With the population known fom the census and the numbe of gids classified as high, medium o low density in each zone, we obtain fom this esult: low density is equivalent to 5210 pesons/km 2 ; and the density atios between the high, medium and low categoies ae 68:22:1. Note that the high density goes up to 354,380 pe km 2. This numbe is seemingly high, but one must conside that new pivate estates ae typically 50-stoey buildings and the gid size used fo this population density calculation is small, at km x km. In any case, the population density of any gid in this study aea is detemined with the oveall total pefectly consistent with the census data. Figue 2 Tseung Kwan O gid layout with diffeent density categoies 1724

7 4.2 Ca Owneship Based on the Tavel Chaacteistics Suvey 2002, the ca owneship in TKO is 8.5 %. Howeve, not 100% of ca ownes dive to wok. It is then assumed that 80% of the ca ownes would dive to wok, constituting about 6.8% of the population; the est of the population use public tansit sevices, eithe the diect bus sevices o the ail sevice. This estimate is consistent with the fact that 90% of all tips in Hong Kong ae seved by public tansit sevices. 4.3 Tanspotation Netwok The destinations of the tips geneated affect the tavel times and costs of tanspotation substantially. In this study, we assume that all tips end at an extenal location which is estimated to be the weighted centoid of the main business disticts. Based on the Tavel Chaacteistics Suvey 2002, the location of this centoid is calculated by weighting the attaction of the each zone. Fo this pupose, the estimated weighted centoid of the main business disticts is taken to be in the Kwun Tong Distict, which is used as the extenal destination of all tips leaving TKO Bus Netwok Buses ae assumed to go diectly fom each bus stop to the outlet point of the highway netwok, which is set to be the main oundabout at the TKO Tunnel. The bus lines ae schematically shown in Figue 3. To simplify the analysis, only Euclidean distance is used fo the calculation, as shown in Figue 3. The stop (station) locations ae displayed in the same geogaphical gid as shown in Figue 2, wheein the study aea is divided into 65m by 65m gids. The location of the each bus stop is collected fom the Centamap ( Rail and Feede Bus Netwok The ail netwok is set up in accodance with the actual TKO Line. Feede buses pick up passenges fom the whole TKO to the neaest MTR Stations. The feede bus lines ae depicted in Figue 3 with diffeent colos. To calculate the tavel time on each feede bus, we use the exact distance along each oute. Bus netwok Feede bus netwok X Bus stop location MTR Station Bus taffic sink/outlet Colo lines feede Euclidean distance of bus outes bus outes Figue 3 Bus and feede bus netwoks 1725

8 4.3.3 Tips Pe Peson And Houly Flow Pecentage Refeing to the Tavel Chaacteistics Suvey 2002, the aveage daily tip pe peson is So, thee ae daily tips geneated fom TKO. Additionally, one day is divided into 24 time peiods and the houly flow pecentage is pesented in Figue 4. Figue 4 Houly flow pecentages 4.4 Othe Paametes All the othe paametes ae calibated and detemined accoding to the actual situation of TKO, as shown in Table 1. The speed of the MTR is calibated by egession between tavel time and distance of diffeent OD pais. The calculated speed of the MTR tain has aleady taken into account dwelling time and deceleation and acceleation of the tain at stations. The same pocess is implemented to calibate the bus and feede bus speeds. The value of time fo in-vehicle time, waiting time and walking time efe to the Tavel Chaacteistics Suvey Table 1 Othe paametes V t V b V f V w VoT i VoT walk VoT wait c HK$0.5 HK$1.25 HK$0.75 km/min km/min km/min pe min pe min pe min 0.68 km/min passps/h 5. RESULTS 5.1 Headway Fistly, we analyze how the headway of the ail sevice affects pofitability. The headway is one of the most impotant opeation paametes fo ail sevices. It not only detemines the level of sevice and hence passenge demand, but also the opeation costs. Figue 5 shows a typical elationship between pofit and headway. Shote headways mean bette quality sevice, which attacts highe ail patonage, but induces high opeation costs. The optimal point lies somewhee in the middle whee the passenge demand is high and the opeation cost can be moe than ecoveed by the evenue. It should be noted that the headway is also limited by the capacity of the tain; it cannot be too shot fo safety easons. Moeove, in the off-peak hous, the headway cannot be too long in ode to maintain a cetain acceptable level of sevice. In the calculation, the passenge demand is based on the total geneated tips and the houly demand split detemined by the logit model. The pofit of the ail sevice is calculated at the headways fom 1 to 9 minutes. The calculated optimal headway is compaable with the MTR headway opeation schedule. Unde the pesent MTR headway schedule, the catchments 1726

9 aea of the MTR (defined as ove 50% of the esidents would like to choose MTR) coves the entie newly developed aeas in TKO. Pofit (HK$) Headway (min) Figue 5 The elationship between pofitability and headway The daily pofit of the ail sevice can be calculated by summing up the houly pofits unde the cuent MTR headway schedule. In the off-peak hous, the pofit is negative; howeve, peak hou pofits ae high enough to cove the loss in the off-peak peiod. It can be concluded that the ail sevice in TKO is self-sustainable unde the cuent situation. Counting the evenue fom the fae alone, the pofit is about one hunded thousand HK dollas pe day. This is not a big numbe as compaed with the investment and opeation costs of the MTR TKO Line. We emphasize that this numbe is vey peliminay given all the assumptions made and the lack of full set of infomation essential fo this calculation. 5.2 Population Density It is clea that population density is among the most influencing facto fo the financial sustainability of tansit sevices. In this section, using the MTR TKO line as the base case, we conduct a sensitivity analysis of how the ail line s pofit vaies with diffeent levels of uban density. It is inteesting and impotant to study how high the population density needs to be so as to ensue financial sustainability of the ail sevice. The sensitivity analysis applies a unifom scale facto to the cuent density of each gid in the study aea and plots how the pofit changes accodingly. The headway is detemined based on the headway analysis in the pevious section. Figue 6 shows how the houly pofit vaies when the population density scaling facto vaies fom 0.5 to 1.3. The x-axis depicts the hou of the day, wheeas the y-axis the houly pofit. The total daily pofits fo diffeent population densities ae obtained by summing the houly pofit fo each hou in Figue 6. The total daily pofits as a function of the density scaling facto ae shown in Figue 7. It shows that the ail sevice could be baely financially sustainable when the density is dopped to 90% of the cuent level. Even in a high population density aea as in TKO, Hong Kong, thee is only a small magin fo achieving financial sustainability. This explains patly why it difficult fo ail systems in othe big cites (New Yok o L.A.) o counties, which geneally have lowe population density, to be financially sustainable. 1727

10 Scaling Facto 0.5 Scaling Facto 0.7 Scaling Facto 0.9 Scaling Facto 1 Scaling Facto 1.1 Scaling Facto 1.2 Scaling Facto 1.3 Pofit (HK$) :30 7:30 8:30 9:30 10:30 11:30 12:30 13:30 14:30 15:30 16:30 17:30 18:30 19:30 20:30 21:30 22:30 23:30 0: Time Peoid Figue 6 Houly pofits fo diffeent population density Daily Pofit (HK$) Population Scaling Facto of RDR Figue 7 Total daily pofits fo diffeent population densities 5.3 Modal Competition This study also conducts a sensitivity of the bus fae, the ail sevice s main competito, to the pofitability of the ail sevice. This infomation is impotant fo the govenment to appeciate the effect of modal competition on the financial sustainability of the ail sevice, which may have impotant implications fo defining policies to egulate this competition. Fo this case study, based on the existing fae, the cuent bus fae is set to be HK$4.5. We vay the bus fae fom HK$3 to HK$6 and examine the esultant impact on the ail pofitability. The elationship between the bus fae and ail pofit is shown in the Figue 8. This esult shows that the MTR TKO line pofitability is sensitive to the bus fae: if the bus fae inceases by 10%, the ail pofit would incease by moe than 50%. On the othe hand, if the bus fae wee to dop by HK$1, the ail pofit would vanish. Theefoe, if the objective is to ensue the financial sustainability of the ail sevice, it is impotant to undestand the competition and define appopiate measues to egulate the faes of all modes. In Hong Kong, in the 1980 s, when the MTR system was fist stated, the govenment implemented the line potection policy, essentially eliminating bus lines that ovelapped with the ail sevice, which was found to be one citical policy fo the success of the ail sevice back then. Hee in this study, we 1728

11 find that some sot of potection policy is also needed. Even with Hong Kong s high density, it is just difficult to achieve financial sustainability of the ail sevice, if it wee left entiely to the maket pocess o the pivate secto y = 28479x R 2 = Pofit of Railway (HK$) HK$3.00 HK$3.50 HK$4.00 HK$4.50 HK$5.00 HK$5.50 HK$6.00 Bus Fae (HK$) Figue 8 Rail sevice pofitability against the bus fae Finally, this study also looks into the elationship between the pofitability of the bus company and its own bus fae. The esult is quite inteesting, indicating that the evenue of the bus company would incease with the bus fae. That is to say, the bus company would also be in favo of inceasing its fae. If the govenment wee to deegulate the contol on tansit faes totally, thee would be a win-win situation fo both the bus and ail sevices; both stand to make a bigge pofit. Of couse, this would be bad news to taveles. This analysis indicates the impotance of egulating the competition and the faes, as they would have diffeent implications fo diffeent paties. 6. CONCLUDING REMARK This pape developed a model to examine how the uban density affects the financial sustainability of ail sevice. The model was applied to the case study of the new town TKO in Hong Kong. We estimated how high the uban density level needs to be in ode to achieve financial sustainability of the ail sevice while in the pesence of competing modes. The bottom line is that Hong Kong s high uban density can baely achieve financially sustainability. This explains why it is difficult to achieve the same fo many othe majo metopolitan aeas that typically have lowe uban density than Hong Kong. Finally, we caution that many assumptions wee used to simplify the poblem and theefoe the numbes obtained as such should be used caefully. Ou main objective is to illustate the model and obtain some peliminay esults that do not violate ou fist ode obsevation. In ou futue studies, we will elax some of the assumptions and make the model moe accuate. In paticula, how the uban fom, in addition to the uban density, would affect the esults ae impotant extensions. Othe extensions to be consideed may include how the long-tem population and land-use changes would affect the povision and financial sustainability of diffeent foms of public tansit sevices. 1729

12 ACKNOWLEDGEMENT The authos thank C.W. Tsang and K.H. Man fo pepaing the data. The study is suppoted by the Competitive Eamaked Reseach Gants HKUST6161/02E and # fom the Reseach Gants Council of the HKSAR. REFERENCES Calthope, P. (1993). The Next Ameican Metopolis. Pinceton Achitectual Pess, New Yok. Chang, S.K. and Schonfeld, P. M. (1991). Multiple peiod optimization of bus tansit systems, Tanspotation Reseach Pat B, 25, Kenwothy, J. and Laube. F. (1999). Pattens of automobile dependence in cities: an intenational oveview of key physical and economic dimensions with some implications fo uban policy, Tanspotation Reseach Pat A, 22, Newman, P. and Kenwothy. J. (1999). Costs of automobile dependence: Global suvey of cities. Tanspotation Reseach Recod, Sinha, K. (2003). Sustainability and uban public tanspotation. ASCE Jounal of Tanspotation Engineeing, 129, Spasovic, L. N. and Schonfeld, P. M. (1993). Method fo optimizing tansit sevice coveage, Tanspotation Reseach Recod, 1402, Tang, S. and H. Lo The Impact of Public Tanspot Policy on the Viability and Sustainability of Mass Railway Tansit The Hong Kong Expeience. Submitted fo jounal publication. Wiasinghe, S. C. and Hudle, V. F. (1977). Optimal paametes fo a coodinated ail and bus tansit system, Tanspotation Science, 11, Hong Kong Govenment, Census and Statistics Depatment, Census, 2001 Hong Kong Govenment, Tanspotation Depatment, Tavel Chaacteistic Suvey,