Challenge D: A world of services for passengers

An analysis of traffic demand sensitivity corresponding private rail transit transferring discounts

Jun LEE Myung-Joo HAN Senior Researcher Doctor’s Course Railway Transport&Logistics Research Department of Environmental Planning Department Graduate School of Environmental Studies Korea Railroad Research Institute National University 437-757, 360-1 Woram-Dong, -si 151-742, 599 Gwanak-ro, Gwanak-gu Gyeonggi-Do, Korea Seoul, Korea Fax: +82-31-460-5021 Fax: +82-2-871-8847 E-mail: [email protected] E-mail: [email protected]

Yong-Taek LIM Young-Ihn LEE Professor Professor Division of Transportation and Department of Environmental Planning Logistics System Engineering Graduate School of Environmental Studies Chunnam National University Seoul National University San 96-1, Dundeok-dong, Yosu city 151-742, 599 Gwanak-ro, Gwanak-gu Chunnam, 550-749, KOREA Seoul, Korea Fax:+82-061-659-3340 Fax: +82-2-871-8847 E-mail:[email protected] E-mail: [email protected]

ABSTRACT : This study estimated demand for transfer discount of the private rail transit in order to prepare the systematic standard such as a method to apply transfer discount, and a method to deal with the loss of the private rail transit due to applying transfer discount with regard to calculating fares of the private rail transit. As a result of the analysis, it is analyzed that travel demand of most private lines is increased when discounting transfer. Some features could be found here, first, it is analyzed that total demand sensitivity is influenced depending on the number of transfer stations for each private line, and there is a difference of demand increase also according to the line's role(branch or main line's function).

Key Words: Private Rail Transit, Transfer Discount

1. INTRODUCTION

According to the SOC Project, there has been a continuing expansion of railways and motorways being constructed with private capital funds due to the insufficient national funds. In particular, the private rail transit is in discussion about linking up with the railways which are run by public institutions. The private rail transit has generally chosen independent fare system not to consider the transferring discounts. Planning the construction and operation of the Seoul metropolitan transport system, the private rail transit is in need to give the transferring discounts to meet the demands of the public.

For this to happen, to solve problems, the standard system for transfer discount application plan, avoidance system for private transit rail loss and an demand analysis of rail transportation with transferring discounts are needed.

Based on such setting, the aim of the study is to analyze the demand sensitivity of private transit rail’s transfer discount system of each rail network.  Analysis on the current status of fares and transfer for the metropolitan private rail transit.  Construction of a network reflecting each fare and transfer system for analyzing demand  Derivation of demand sensitivity due to transfer discount

2. Analysis on the current status of metropolitan private rail transit Private rail transit projects are divided into BTO (Build-Transfer-Operate), BTL (Build-Transfer- Lease), BOT (Build-Operate-Transfer), BOO (Build-Own-Operate) etc. to carry out depending on the property of private investment project. For the fare, the private rail transits constructed by a private investment project have selected an independent fare system in general. The independent fare system has a negative utility because the basic fare is paid twice over in user's position. On the contrary, in operator's position, there is an advantage that could collect fares simply and avoid difficulties distributing fares between several operators arisen from transfer discount. Since operators of the private rail transit projects are different from ones of the existing subway lines so that arrangement is required when distributing their revenue, most private investment projects have proposed the independent fare system.

The site of this study is on the main seven rail lines which the transfer discount systems applies – AREX( Airport Railroad), SinBundang Line, Sosa-Wonsi Line, Sosa-Daegok Line, LRT, LRT, LRT. There are different fares available for these rail lines, such as independent fare system, distance ratio fare system and section fare system, as well as analyzing the current system for fare and connection transfer situation.

Table 1 Metropolitan Private rail transit Beginning Opening transfer Line section Distance Year Year station · ∼Gangnam· 2010 18.5km SinBundang Line 2005 7 Gangnam∼Yongsan 2011 7.49km

AREX(Korail · Airport∼ Airport· 2007 40.3km 2001 6 Airport Railroad) Gimpo Airport∼Seoul KTX station 2010 20.7km Sosa-Wonsi Line - · Sosa∼Wonsi 2014 23.4km 2 Sosa-Daegok · Gimpo Airport∼Sosa· Daegok∼Gimpo - 2015 17.8km 6 Line Airport Yongin LRT 2005 Yongin 2009 18.404km 1 Uijeongbu LRT 2007 Uijeongbu 2011 11.085km 1 Gwangmyeong - Gwangmyeong - 10.305km 2 LRT Figure 1 Private rail transit

Examination of elasticity variation due to transfer discount

In general, the fare for public transportations is a variable having the largest effect on travelers when selecting transportation mode and routes, which is a major factor with a negative sign that traffic volume is decreased as the fare is increased. An analysis from domestic and foreign existing literatures shows that price elasticity is generally inelastic for the public transportation demand. Most of the preceding studies used the multiple regression model to estimate elasticity, and it is analyzed that estimated value of elasticity is within -0.21~-0.79.

Table 2 price elasticity for public transportation literatures elasticity The demand for public transport : The effects of ㆍ Bus short-term: -0.4, long-term: -1.0 fares, quality of service, income and car ownership (Pualley, 2005) ㆍ Subway short-term: -0.3, long-term: -0.6

ㆍ Subway short-term: -0.2, long-term: -0.4 Transport Research Lab. UK(TRL) ㆍ Rail short-term: -0.7, long-term: -1.1

Demand and Revenue Implications of an Integrated ㆍ Bus -0.21 Public Transport Policy (Matas, 2004) ㆍ Subway -0.37

ㆍ Bus short-term: -0.4, long-term: -0.9 The Demand for Local Bus Services in England ㆍ UK short-term: -0.53 (Dargay, 2002) ㆍ France short-term: -0.39

A Review of New Demand Elasticities with Spacial ㆍ Bus -0.41 Reference to Short and Long Run Effect of Price -0.79 Changes(Goodwin, 1992) ㆍ Subway

Transit Price Elasticities and Cross-Elasticities ㆍ Subway, UK short-term: -0.54, long-term: - (Litman, 2004) 0.75

ㆍ Subway, France short-term: -0.3, long-term: -0.59

Causes of Bus Transit Ridership Change in Seoul ㆍ Bus -0.27 ~ -0.32 (Ha, 1997)

3. Establishment of a methodology to analyze traffic demand when discounting transfer

General methods to estimate traffic demand use 4-step model estimating traffic demand, which is composed of trip generation, trip distribution, mode choice, and assignment.

Table 3 Fundamental data Fundamental Contents of fundamental data data Traffic zone Metropolitan area (Seoul, Incheon, Gyeonggi-do) Network: 1142 zones and network Urban and County area are constructed by Dong and Eup·Myeon unit, respectively.

Transportation mode O/D of SDI(Seoul Development Institute) O/D - Passenger: 4 modes O/D of car, taxi, bus, and subway are utilized. - Freight: It was used the freight O/D quantity provided by KTDB(Korea Transport Database).

The basic and goal year of this study is set to 2007 and 2021, respectively. This is based on the time when all the 7 private rail transit lines, which are analyzed in this study, are to be opened. Since this study is aimed at analyzing demand for transfer discount of the metropolitan private rail transit, spatial scope is set to the metropolitan area including Seoul.

Table 4 analysis scope Division Scope Basic year: 2007 Temporal scope Goal year: 2021 (the time when all the 7 lines are to be opened)

Spatial scope Metropolitan area including Seoul

Limits and basic premises of analysis

It is necessary to explain limits and basic premises of the analysis first before carrying out analysis of this study. Because errors could be avoid for the analysis only if the basic limit of this analysis is recognized and the derived analysis result is interpreted under the basic established premise. The limits of the analysis in this study are as follows.

First, there is a difficulty in considering transfer between modes. In the current metropolitan transportation DB, the OD and transit line is separately constructed for each 4 modes (car, taxi, bus, subway) and it is separated to be assigned for each means. Therefore, since there is a limit that could not accurately estimate transfer or connected travel between buses and subways, this study restrictively reflected transfer between buses and subways by a method discounting bus fares in the time to access subway.

Second, there is a limit when applying demand of the existing report. Since the driving details is different for each private investment project and there are various reports even in a project, the time analyzing demand is not identical. Therefore, there is a difference in the reflection of the future development plan and the network, and the fare level is also different.

Since there are such limits, this study established the following premises for the analysis.  This study analyzes only demand sensitivity for transfer discount between subways (including LRT) and it reflects as a method discounting bus fares in the time to access subway for transfer with buses.  The variation rate is applied for transfer discount based on demand of the existing report.  Finally, the purpose of this study is to decide demand sensitivity due to transfer discount, which it is assumed that demand in the existing report is reasonable.

Mode Choice Model

In this study, the mode choice model of KTDB was applied.

eUK PK()  n eUi i

Uk = Utility of Mode K

U i = Utility of Mode i n= number of modes

Uijm1( Ttime ) ijm   2 ( T cos t ) ijm  ( DUM ) m

U ijm =Utility Fn. Mode m

()Ttime ijm = total travel time of Mode m

(T cos t )ijm = total travel cost of Mode m

()DUM m =random part

Table 5 Utility Fn. of mode choice model Division Ttime Tcost Dum auto - taxi 2.55838 -0.39896 -0.01704 bus 0.0776925 subway -0.0579425

Travel Time and Travel Cost

In order to apply the utility function to a mode choice model, it is required a process to calculate total travel time and total travel cost.

A standard to calculate travel time for each mode used values assumed appropriately in this study, and the standard was set to describe it close to reality. Travel time for each mode is divided into in-vehicle and out-vehicle time, the former could be found as actual in-vehicle travel time, and the latter as (access time + waiting time + boarding/alighting time) in general. This study used in-vehicle time of the shortest travel path from assignment results for in-vehicle time, and access time, waiting time, boarding/alighting time are calculated by assuming appropriately as follows.

Table 6 Standard to calculate travel time for each transportation mode Division In-vehicle time Access time (boarding/alighting time) Waiting time Travel time of the Car shortest travel path from No No assignment results Travel time of the Taxi shortest travel path from No 5 minutes assignment results Bus In-vehicle time from Access distance within 500m: 8.12 minutes 1/2 of headway assignment results Access distance above 500m: 500Won+(Access distance- 0.5km)/20km/h*60(Assuming accessed by feeder bus) Access distance within 500m: 8.12 minutes In-vehicle time from Access distance above 500m: Subway 1/2 of headway assignment results 500Won+(Access distance- 0.5km)/20km/h*60(Assuming accessed by feeder bus)

The standard on travel cost was also set to describe it close to reality for each mode.

For buses and subways, the in-vehicle travel distance is calculated to apply their travel cost following the current fare system, and for taxies, travel cost is calculated by adding 100 won of additional fare every travel distance (144m) to 1,900 won of basic fare.

Table 7 Standard to calculate travel cost for each transportation mode Division Travel cost Access ㆍ parking cost - Average parking cost 2,400won/hour - Use the regression equation for car's travel cost in won Car - Average parking cost for 1~5 grade unit place Travel distance within 10km: 900Won Bus No Travel distance above 10km: every 5km 100won - basic fare : 1,900won (within 2 ㎞) Taxi No - ×100won Travel distance within 10km: 900Won - 10 ㎞

Reflecting network cost

In order to reflect fares of the private rail transit, additional generalized cost is applied to some links and nodes of railway networks to implement it, and the MRS(marginal rate of substitution) of mode choice model is applied to the generalized cost. Equivalent travel time for travel cost becomes (- 0.01704)/(-0.039896) = 0.43(minute/100won). Therefore, this study used this equivalent travel time to convert travel cost into travel time to reflect it. The network is constructed in order to divide non- transfer and transfer nodes to enforce the fare discount selectively.

4. Estimation of demand sensitivity following transfer discount

4.1 Demand applied in this study (the existing reports)

This study was to analyze demand sensitivity following transfer discount of the private rail transit, which analyzed demand sensitivity following transfer discount for the case applying a transfer discount system for each individual private line. Therefore, the demand sensitivity was analyzed when discounting transfer between the private rail transit and the subway (including LRT) and the transfer with buses was reflected by a method discounting bus fares in the time to access subway through the select link analysis.

Demand prior to transfer discount of each private rail transit line was based on the basic planning report of each private rail transit. In other words, this study is to estimate sensitivity for transfer discount based on demand of the basic planning report.

4.2 Establishment of a scenario This study established the demand sensitivity, which follows several discount rates by applying a method to discount a fixed rate for each transfer between the private rail transit and the metropolitan public transportation network, as an analysis scenario. Even though the transfer-basis discount system could be introduced through a relatively simple method, it is difficult to find a reasonable transfer discount rate if the fare system and level are different for each line, and it also has a problem to rearrange the discount rate whenever fare is changed.

 Transfer discount rate: discount 10%, 20%, 30%, 50% of the travel fare

Table 8 Scenario Division non enforced Transfer discount rate discount 10% discount 30% discount 20% of the discount 50% of Scenario proposed fare of the travel of the travel travel fare the travel fare fare fare

4.3 Estimation of demand sensitivity following transfer discount

[Fig. 2] and [Table 9] show the result analyzing the demand sensitivity due to transfer discount rates of 10%, 20%, 30%, 50% for 7 private rail transit lines. As shown in the table, if the transfer discount is enforced in 2021, it indicates that travel demand for most lines is increased comparing to the case not enforced. If travel fare is discounted by 10% for AREX, it was analyzed that non-transfer and transfer demand is increased by 4.31% and 3.36%, respectively, and total demand is increased by 10.24%. It is considered, for AREX, that there is much transfer demand because scheduled speed is high and the travel fare is relatively expensive so that fare discount is larger than other lines.

In addition, since SinBundang Line is transferred to various existing lines such as Line No. 1~4, 7, 9, and Bundang Line etc. so that possibility of changing a path is high, transfer demand is significantly increased due to fair discount. It was analyzed that non-transfer and transfer demand is increased by 1.48% and 3.29%, respectively, when 10% of travel fare was discounted for transfer.

When 30% of transfer discount was applied, it is shown that AREX (+25.26%), SinBundang Line (+19.08%), Sosa-Wonsi Line (+19.92%), and Sosa-Daegok Line (+10.07%) had a large variation of demand, and for Sosa-Daegok Line, it is decided that variation of demand is considerably increased because 6 stations of all 8 stations in the line are transfer stations.

On the contrary, since Yongin LRT and Uijeongbu LRT include only one transfer station, and they play a role of branch line responsible for traveling within the area, it is considered that variation of demand due to transfer discount is low. Figure 2 variaion of demand following transfer discount

Table 9 variaion of demand following transfer discount (trip/day) 10% of transfer 20% of transfer 30% of transfer 50% of transfer non enforced discount discount discount discount Line non- non- non- non- non- transfer transfer transfer transfer transfer transfer transfer transfer transfer transfer Gwangmyeong 69,771 23,187 69,878 23,423 70,991 23,453 72,260 25,035 68,658 23,123 LRT 1.62% 0.28% 1.78% 1.30% 3.40% 1.43% 5.25% 8.27% Sosa-Wonsi 109,356 25,915 109,640 26,548 111,126 27,490 112,220 29,071 109,141 24,578 Line 0.20% 5.44% 0.46% 8.01% 1.82% 11.85% 2.82% 18.28% Sosa-Daegok 42,493 107,818 42,734 109,024 43,103 114,110 43,290 125,395 42,320 104,381 Line 0.41% 3.29% 0.98% 4.45% 1.85% 9.32% 2.29% 20.13% AREX(Korail 219,726 136,461 223,807 140,277 228,866 147,145 236,003 165,481 Airport 210,642 132,024 Railroad) 4.31% 3.36% 6.25% 6.25% 8.65% 11.45% 12.04% 25.34% 142,858 61,287 142,946 61,337 143,075 61,384 143,337 61,484 Yongin LRT 142,762 61,237 0.07% 0.08% 0.13% 0.16% 0.22% 0.24% 0.40% 0.40% SinBundang 703,903 268,266 711,741 270,542 717,309 285,589 723,712 312,627 693,636 259,712 Line 1.48% 3.29% 2.61% 4.17% 3.41% 9.96% 4.34% 20.37% 95,100 41,222 95,377 41,384 95,675 41,558 97,052 41,902 Uijeongbu LRT 94,809 41,053 0.31% 0.41% 0.60% 0.81% 0.91% 1.23% 2.37% 2.07%

5. Conclusion and the future study

This study estimated demand for transfer discount of the private rail transit in order to prepare the systematic standard such as a method to apply transfer discount, and a method to deal with the loss of the private rail transit due to applying transfer discount with regard to calculating fares of the private rail transit.

The private rail transit lines, which are subjects of this study, are total 7 lines such as AREX, SinBundang Line, Sosa-Wonsi Line, Sosa-Daegok Line, Yongin LRT, Uijeongbu LRT, Gwangmyeong LRT, and as a result of analyzing current status of the corresponding line's fares and the connected transfer lines. The status of transfer is that, for the light rail transit, the number of transfer stations is small as 1~2 because it plays a role of the branch line, and the other lines are variously transferred to the existing metropolitan subway lines. In particular, for Sosa-Daegok Line, 6 of total 8 stations are able to be transferred.

In order to analyze demand sensitivity for transfer discount, the cost according to transfer fare is converted into the time to reflect it. Travel demand, which is basis of the analysis, is analyzed on the basis of demand in the existing reports for the time when not enforcing after dividing into cases discounting 10%, 20%, 30%, and 50% of travel cost, respectively, and it is presented by dividing into non-transfer and transfer demand.

As a result of the analysis, it is analyzed that travel demand of most private lines is increased when discounting transfer. Some features could be found here, first, it is analyzed that total demand sensitivity is influenced depending on the number of transfer stations for each private line, and there is a difference of demand increase also according to the line's role(branch or main line's function). In other words, for the case including many transfer stations in the line such as SinBundang Line, Sosa- Wonsi Line, or the case taking charge of the main line's function such as AREX, it could be found that their demand is quite sensitively varied for transfer discount.

On the contrary, for the branch lines like Gwangmyeong LRT, Uijeongbu LRT, Yongin LRT etc., it is shown that their transfer demand is slightly varied because of the property that there are few transfer stations and it is transferred at the terminal station. In addition, it is analyzed that the longer the line and the larger the number of transfer stations is, the more sensitive the demand sensitivity for transfer is.

In addition, even though this study analyzed the demand of transfer discount based on the demand in the existing report, there could be a difference from actual measured demand in the opening year because of neighboring development project and the future road/railway construction, therefore, it is considered that it is also needed the future study for it.

REFERENCES

Dargay, J. M. and M. Hanly, "The Demand for Local Bus Services in England", JTEP, Vol. 36, No. 1(Jan 2002), 99. 73-91. Goodwin, P. B., "A Review of New Demand Elasticities with Spacial Reference to Short and Long Run Effect of Price Changes", Journal of Transport Economics and Policy, Vol. 26(May 1992), pp. 155-169. Litman, T., "Transit Price Elasticities and Cross-Elasticities", JPT, Vol. 7, No. 2(2004), pp. 37-58. Matas, A., "Demand and Revenue Implications of an Integrated Public Transport Policy: The Case of Madrid", Transport Reviews, Vol. 24, No. 2(March 2004), pp. 195-217. Neil Paulley, "The demand for public transport : The effects of fares, quality of service, income and car ownership", Transport Policy, Vol. 13(2006), pp. 295-306.