A Study on electronic toll collection systems in expressways in Sri Lanka

Transcription

1 A Study on electronic toll collection systems in expressways in Sri Lanka A Rodrigo and L Edirisinghe # Faculty of Humanities and Social Sciences, Colombo International Nautical and Engineering College (CINEC) Campus, Malabe, Sri Lanka # edirisinghe@cinec.edu Abstract This paper focuses on the determinants that affects, when introducing prepaid and postpaid Electronic Toll Collection systems to Sri Lankan expressways. Purpose of this research is to identify the user preference between prepaid and postpaid Electronic Toll Collection methods and to identify the determinants that affect such a choice. The study is focusing on analyzing the new implementation of Electronic Toll Collection in Sri Lanka, hence the drawbacks of the current cash based tolling system is identified. At the moment cash is the only available option for the expressway users. With the rapid growth of the traffic density, even the expressways are eventually getting congested especially near the toll gates. As a result, Sri Lankan government is on the process of introducing electronic payment mechanisms in very near future but only one prepaid payment option been considered under their new implementation plans where the users do not have a liberty to make a choice, which is of little benefit. Therefore to check the user preference on electronic toll collection methods, five widely implemented methods been considered which were subsequently divided into two, prepaid and postpaid. Based on the literature, a structured questionnaire survey and interviews were carried out in order to collect primary data. 230 responds were able to collect within the months of November and December 2014 which were used to test the significance of the identified variables using Person s chi squared statistics and binary logistic regression analysis. Secondary data has been collected using literature and previously published reports from RDA, National Transport Commission, Ministry of Transport and The Central Bank. Based on the results it was concluded that only the convenience and vehicle category have a significant association with the dependent variable of Prepaid or postpaid ETC method while combined effect of convenience and vehicle category too have a significant relationship to the dependent variable. Keywords electronic toll collection; srilankan expressways; prepaid and postpaid payments; user preference. I. INTRODUCTION For a country, transportation is the web that links people, cultures, cities, towns and villages in to a one common 283 network of relationships. People use transport services to commute from one place to another and fulfil their day to day necessities. Transportation distributes varied resources to every nook and corner of the country which are originally concentrated in to a one place. In early ages, people used non motorized transportation modes for their activities and eventually it was rapidly developed subsequent to the industrial revolution utilizing the new technologies. At present, technology has become an integral part of the transportation and people tend to demand much more complex facilities for their transport related activities. Out of the various means of local transportation options which are in existence, road transportation is one of the highlighted areas of modern development plans in the country. Road transportation counts for more than 90% of the land transport in the island and covers more than 12,000 km of A and B class roads within the country. After the end of the long prevailed civil war in the country, opportunities has enlarged to the public and people tend to enjoy the benefits derived out of those opportunities. Hence the commuting patterns of the population have changed drastically and there was a rapid increase in traffic volumes during the past few years but, there had been a shortage of capacity to meet this demand. As per the sources of Departments of Motor Traffic, during last few years overall vehicle population shows a growth rate of about 56% while in year 2010 total number of vehicle registrations increased by 76%. From the preliminary studies carried out to meet this demand, it has been realized that rehabilitation and widening of existing roads to cope up with future traffic needs at a meaningful level of service is a difficult task without large scale acquisition and demolition of buildings and relocating service utilities. Therefore with the intention of developing and upgrading the road transportation systems, Sri Lankan government has expedited the major proposal to construct expressways for its road network. Expressway is a novel concept to Sri Lanka which was only initiated in few years back. For the moment, over 150 kms of designated expressways are serving the southern part of the country. It travels from Kottawa to

2 Matara (126.1 km) and was initially opened to public from Kottawa - Pinnaduwa (95.3 km) on 27th November 2011 and the extended section from Pinnaduwa - Godagama (30.8 km) has been opened to public on 15th March As the second access control expressway, the Colombo - Katunayake Expressway (E03) was opened to the public on 27th October 2013 which runs from New Kelani Bridge to Katunayake (25.8 km). Simultaneously, on the very same day of opening the Pinnaduwa - Godagama section, the Kottawa to Kothalawala section of the Outer Circular Highway (E02) also opened for the public which has a total length of 10 km. As for the plans, northern end of the Outer Circular Highway is located at Kerawalapitiya while the southern end is located at Kottawa. Design speed of all the above 03 expressways are 100 kmph. Since the E02 is still under construction, the Kottawa to Kothalawala section of this Outer Circular Highway is currently operating as a part of the Southern Expressway (E01). All the expressway users are charged a fee which named as a user fee or toll, according to the distance they travel and the type of the vehicle they are using. According to the most recent statistics, ratio of vehicles against the population is 1:4, with a staggering 4.4 million vehicles which are in operation. Due to that, commuters spend more time on the roads, burning extra money for fuel where the opportunity cost is very high. At the same time, the number of vehicles plying in the expressways has increased to a massive number just within 03 years from the inception of its operation. From the day of opening up to date there were millions of vehicles travelled in E01 and E03 respectively. And also as per the Road Development Authority, in average more than 15,000 Vehicles travel in E03 generating a daily income of nearly 5 million rupees while according to the Annual Report of the Central Bank of Sri Lanka the Southern Expressway earns a daily income of Rs. 3.4 million in 2013 from around 8,700 vehicles per day. Therefore without any doubt the trend would be continue to the future, leading to the fact that the expressways would also get congested like other urban roads with more number of vehicles using the expressways. Hence in future, current conventional cash based tolling mechanism would not sustain and more efficient traffic management systems would be required at the interchanges to handle the increased vehicle traffic flow. affects the convenience. Due to the current system there are major environmental concerns as well. Expressways issue 02 paper tickets per each vehicle for a single journey. If we sum up the numbers i.e more than 17 million paper tickets per year which is a considerable amount of garbage dump in to the environment. In the current toll collection method expressway users have to stop the vehicle to pay the toll which is adversely effect to the environment so as to the fuel economy of the vehicle. Acceleration, deceleration and stop incidents at the entrance and exit would lead to an extra fuel consumption and increased air pollution due to emission of hydrocarbon, carbon monoxide, and nitrogen oxide around the toll booth areas. In order to overcome the drawbacks in the current system, there is a clear need to introduce electronically enabled payment options which are more convenient, secure, flexible and fast compared to the traditional cash based tolling system. The daily growth of the technology has changed the way of selling the products and services all over the world and electronic payments is one of the crucial areas of the modern technology which had influenced almost every industry so as the expressway tolling mechanisms in the world. In 1959, Nobel Economic prize winner William Vickrey was the first to propose a system of electronic tolling for the Washington Metropolitan Area. The technology has evolved over time and Norway has been the world's pioneer in the widespread implementation of this technology where Electronic Toll Collection (ETC) was first introduced in Bergen, in 1986, operating together with traditional cash tollbooths. ETC is an element of Information and Communication Technology (ICT) that allows for nonstop toll collection. At present this technology has widely spread all over the world. For Sri Lanka, access controlled expressways is a novel concept so as the electronic toll collection systems. In doing this research paper, five major ETC methods have been identified which are in practice around the world. Identified payments options were further divided into two categories under prepaid and postpaid methods. In this study, attention was paid to analyse the user preference between prepaid and postpaid ETC methods and to determine the customer expectation level using the identified independent variables. In Sri Lanka at the moment, user fee can only be paid using conventional cash. Due to this, users suffer from the poor service quality particularly at the exit interchanges. Drivers need to have cash with them and sometimes they have to face the bad experience of toll superintendents asking for cash change which negatively 284 II. RESEARCH METHODOLOGY In order to arrive at the aforesaid aims casual research methodology has been used to analyse the relationship between independent variables and dependant variables of prepaid or postpaid ETC method. For this study, the purpose was to generalise the results to the whole

3 population of expressway users on selecting ETC method, hence the simple random probability sampling method was chosen to acquire the necessary information. Based on resources and time frame and considering the confidence internal is 5 % at 95% significance level, required sample size was limited to responses were collected in return where the questionnaire survey able to cover 71% of the planned samples. Primary data study were obtained from two sources; structured questionnaire survey and interviews. Google document as an electronic version and printed questionnaire in Sinhala and English mediums were distributed among the sampling unit of in a way to cover all the vehicle categories and both expressways. Using the Google form 139 responses were collected and 91 responses were collected using printed questionnaire. Sinhala questionnaire was used to capture data from freight vehicles and other passenger transports such as expressway bus services. Apart from the questionnaire survey, telephone and personal interviewing has been carried out especially with the private bus owners and transport executives of distribution companies. Secondary data were collected from the literature and the sources available from National Transport Commission, Ministry of Transport, Road Development Authority and Central Bank Report, Sivapragasam (2013) et al on their research of the factors affecting on internet banking adoption, they have used Technology Acceptance Model (TAM) introduced by Davis (1985). Carrying out this case study under the TAM, the researchers have tested four basic demographic attributes influence the internet banking adoption using the 2-way chi square statistic and the factors are age, gender, education and monthly income. These variables were given scores depending on its nature. Nominal scaling is gender and given scales 1 and 2 respectively. Each variable of age level, education and monthly income were categorized into four which was measured using the ordinal scaling from 1 to 4. Type of the vehicle and vehicle categories were also used as variables. Both passenger and freight vehicles were considered under the type of the vehicle where measurement is done according to the nominal scaling of 1 and 2. Currently, the user fee is levied under four categories of vehicles which have been measured using nominal scale from 1 to 4. Ownership of the vehicle was analysed under 04 ownership models and nominal scale from 1 to 4 is used. Frequency of travelling on expressways divided into four categories and ordinal measurement from 1 to 4 is used. 285 Total distance of km from Kothalawala to Godagama in E01 has divided in to four sections of 30 km intervals where no. 1 allocated to less than 30km while no. 4 allocated to more than 90km. Total length of 25.8km from New Kelani Bridge to Katunayake has divided in to two categories where no. 1 and 2 assigned up to 16 km and more than 16km respectively. And also 05 purposes of travelling have been identified which were given the nominal scores from 1 to 5 respectively and awareness of the ETC has measured based on nominal scaling of 1 and 2. Through a review of past studies on electronic payment methods (Quibria, 2008), five way evaluation criteria has been identified initially to analyse the Impact of the ETC. After the pilot survey, minor changes were done to the wording of the question used to represent attitudinal factors of the drivers. Each criterion was given a score of 1 to 5 (1= high negative impact, 2= low negative impact, 3= neutral, 4=low positive impact, and 5= high positive impact) to test five attitudinal factors; convenience, flexibility, security and risk, speed of the payment and positive externalities. Dependant variable of the study is the selection of either prepaid or postpaid ETC method. Identified five ETC methods were divided in to two categories according to the instance the payment being made, i.e prior to the journey or after the journey on the expressway. This variable is measured using nominal scale of 1 and 2 respectively which is as follows. III. EXPERIMENTAL DESIGN To check the reliability of the identified variables, a reliability analysis has been carried out. (Green et al., 1977) By this analysis cronbach s alpha coefficient of has been derived which verifies that there is good internal consistency among the variables. Suppose that we measure a quantity which is a sum of components,, therefore the Cronbach's is defined as Where scores, current is the variance of the observed total test is the variance of component i for the sample of persons When analyzing the customer preference on introducing ETC systems, a descriptive analysis has been carried out. The influence of each independent variable to the dependant variable was tested using chi square statistics

4 by the tool Statistical Package for Social Sciences (SPSS). This is a statistical hypothesis testing which checks whether the sampling of the test statistic is a chi squared distribution when the null hypothesis is true. (Sivapragasam et al, 2013) The test will apply to the categorical variables and it is used to determine whether there is a significant association between two variables. The chi square statistic used to conduct this test is the same as in the Goodness of fit test as shown in (1)....(1) Where Oi is the observed number of cases and Ei is the expected number of cases After identifying the factors which has a significant association, to see the impact of those factors on selecting either prepaid or postpaid ETC option, a binary logistic has carried out using the SPSS. By interpreting the test results, model of best fit is identified and implications of the exp(b) is described. (Bigne et al, 2005) IV. RESULTS AND DISCUSSION A. Pearson s Chi Squared Test Interdependence between identified variables and user preference on prepaid or postpaid ETC methods have been checked using this test. For each variable, null hypotheses and the alternative hypotheses is stated where the significance of the relationship will be determined using the significance level of 5%. If the respective p-value of the derived test statistic is less than 0.05, null hypotheses have rejected and alternative hypothesis have accepted. (Sivapragasam et al, 2013) In the below table, three independent variables including education level, vehicle category and ownership of the vehicle are significant at 5% level (p-value >0.05) with 1 degree of freedom. Therefore null hypothesis should be rejected at 5% significance level. And also it can be observed that, Frequency of travelling in E01, awareness about of ETC, convenience and Positive Externalities only has marginal difference between the p-value and significance level, those four variables has also counted in the. Though the variable alone is insignificant, there may be a dependency exist in combination with another factor hence the variable is model fitting. Table 2. Summary of p-values Independent Variable P- valu e Significant/ Insignificant Gender.863 Insignificant Age.211 Insignificant Education Level.001 Significant Income.196 Insignificant Vehicle Type.190 Insignificant Vehicle Category.000 Significant Ownership of the Vehicle Frequency of travelling in E Significant.078 Insignificant but.644 Insignificant Frequency of travelling in E03 Distance of travelling in.565 Insignificant E01 Distance of travelling in.585 Insignificant E03 Purpose of travelling.144 Insignificant Awareness about of ETC.063 Insignificant but Flexibility.395 Insignificant Convenience.084 Insignificant but Security.606 Insignificant Speed.454 Insignificant Positive Externalities.082 Insignificant but Binary Logistic Regression Analysis Using the above selected variables, binary logistic has been carried out. In this aspect, above covariates were used with a variety of combinations in both single and coupled combinations. Finally one model is selected based on the goodness of fit, p-value and Nagelkerke R Square values and used for the model fitting and analysis which is as follows. 1) Block 0 - Beginning Block: As per the below table initial -2 Log Likelihood is which measures the degree of discrepancy between the observed values and predicted values from the model. 286

5 Iteration Table 3. Iteration History -2 Log likelihood Coefficients Constant Step a. Constant is included in the model. b. Initial -2 Log Likelihood: c. Estimation terminated at iteration number 3 because parameter estimates changed by less than.001. Under the beginning block, SPSS initially produced an output which corresponds to the constant model. Results are shown in the below table. Table 4. Variables in the Equation Step 0 Constant B S.E Wal d Df Sig Exp(B ).783 As for the above table, fitted constan t model is, logit (P /1-P) = where P is the probability of using ETC methods and also; (P /1-P) = Exp (B) = is called as the odd ratio difference in the value of the -2LL statistic between Model1 and Model 2 is called the likelihood ratio (LR) test statistic and it was used to test the stated null hypothesis. SPSS provides these values in the following table of Omnibus Tests of Model Coefficients Table 6. Omnibus Tests of Model Coefficients Chi-square df Sig. Step 1 Step Block Model Since the method of analysis is ENTER, at the step 1 it can be seen that the LR test statistic has a chi-square distribution of which is equivalent to the difference between -2LL statistic of Model1 and Model2. Test statistic is significant at 5% level (p-value <0.05) with 7 degree of freedom. Therefore the null hypothesis can be rejected at 5% significance level. So it can be concluded that the additional explanatory variables in the Model 2 are required. Fitted model is a technically sound model for the randomly selected data. SPSS provides two statistical values which are namely Cox & Snell R Square and Nagelkerke R Square. Those statistics can be used to measure the proportion of variation in the output being explained by the model. The results are shown in the following below table. 2) Block 1: Method = Enter (Likelihood Ratio: The values of -2Log Likelihood for model 1 and model 2 are illustrated in the below table. Model 1 is nested within the model 2 therefore it needs to be checked whether the additional explanatory variables in the model 2 is required to decide on the final mode. Table 5. -2Log likelihood for model 1 and model 2 Model -2Log Likelihood Model1(Constant model) Model Hypothesis Testing: To check the aforesaid requirement, below stated hypothesis is used. H0: Additional explanatory variables in the Model 2 not required. H1: Additional explanatory variables in the Model 2 are required. The -2Log Likelihood (-2LL) value of the model1 is larger than the corresponding value of the model 2. The 287 Step -2 Log likelihood Table 7. Model Summary Cox & Snell R Square Nagelkerke R Square a a. Estimation terminated at iteration number 5 because parameter estimates changed by less than.001 The Cox & Snell R Square is and the derived Nagelkerke R Square is Out of the two statistics, Nagelkerke R Square is selected for the explanation which can be interpreted as the model explains about 13.9% of the variation of the selected data set. Hosmer and Lemeshow Goodness-of-Fit Test test is used to examine the overall fitness of the model to the observed data. Below hypothesis is used in this aspect and the results are sown in the below table. H0: There is no difference between observed and predicted values. H1: There is a difference between observed and predicted values.

6 Table 8. Hosmer and Lemeshow Test Step Chi-square df Sig Chi-square value is and it is insignificant at 5% significance level (p-value>0.05). Therefore null hypothesis should not be rejected at 5% significance level. It can be concluded that there is no difference between observed and predicted values at 5% significance level. St ep 1 a According to the above table 9, (P-Value<0.05) the best fit model can be written as; logit = 2.985* Vehicle Category 0.926* Convenience * (Convenience by Vehicle Category) Where: P is the probability of using prepaid electronic toll collection methods Here Table 9. Variables in the Equation B S.E. Wal d = Exp (B) is called as the odd ratio df Sig. Education_Level Vehicle_Category Type_of_ownershi p_of_the_vehicle Frequency_E01_N ew Convenience Convenience by Vehicle_Category Convenience by Positive_externalit ies Constant Exp(B ) a. Variable(s) entered on step 1: Education_Level, Vehicle_Category, Type_of_ownership_of_the_vehicle, Frequency_E01_New, Convenience, Convenience * Vehicle_Category, Convenience * Positive_externalities. 288 Reference category used here is last category. Therefore based on the above details, following associations among the variables can be interpreted. Exp (B) for Vehicle Category is This means that the odds of ETC method being postpaid are times the odds of payment method being prepaid as a result of the vehicle category. Exp (B) for Convenience is This implies that the odds of ETC method being postpaid are times the odds of payment method being prepaid due to the convenience of the payment method. Exp (B) for Convenience by Vehicle Category is This means that the odds of ETC method being postpaid are times the odds of payment method being prepaid as a combined effect of convenience and vehicle category. While the above determinants are included in the model; Gender, Age, Education Level, Income, Vehicle Type, Ownership of the Vehicle, Frequency of travelling in E01, Frequency of travelling in E03, Distance of travelling in E01, Distance of travelling in E03, Purpose of travelling, Awareness about of ETC, Flexibility, Security, Speed and Positive Externalities are not significantly associated with the preference of selecting ETC method. V. CONCLUSION Based on the above findings, it can be concluded that out of the used independent variables only the vehicle category and convenience are closely associated with the preference for ETC methods. Out of the two payment methods, both vehicle category and convenience are closely associated with the prepaid ETC methods. Also it can be concluded that the vehicle category and convenience has a negative relationship with postpaid ETC methods. When it comes to the combine effect of convenience and vehicle category, it can be concluded that the combined effect is associated with more likelihood of using the postpaid ETC methods. According to the Nagelkerke R Square, that model explains about 13.9% of the variation of the total data set. Further using this model, preference for selecting prepaid ETC methods can be predicted and overall correct percentage is 56.1%. Freight vehicle users more preferred towards using postpaid electronic payment methods. Through the personal interviewing carried out regarding the freight vehicle operations, most preferred method of payment is the postpaid payments from the freight vehicle controller s perspective. Category 03 comprises of Motor Lorries as mentioned earlier under fright vehicles and more specifically this category consists of the Motor Coaches (expressway public buses). With the data collected from those bus drivers, majority of them are not interested on the ETCs. They were highly

7 entertained to use whatever the payment method assigned to them by their employer/owner and if the ETCs to be assigned, majority of them are preferred to use postpaid payment options. Further, drivers and the helpers are love to use current cash based toll collection as they have ample cash on hand where the expressway toll ticket is produced to the owner at the end of the day. As per the comments of bus operators, majority of the busses are on leasing terms and it is convenient for them to pay the cash then and there on each journey rather than paying millions of rupees for their entire fleet at once on either prepaid or postpaid terms. Only 67.8% of the expressway users have heard about the ETCs before while rests of the respondents were not aware about the ETCs. People with higher educational qualifications are more aware of the ETCs compared to the less educated drivers. Therefore when implementing ETCs, it would be needed to carry out awareness programes to provide knowledge on the ETCs especially to the less educated drivers. The study was limited to test identified variables relegated to the ETCs but who are interested could be carried out more profound analysis as further research by applying all the variables of Technology Acceptance Model (TAM) once the ETCs are implemented. This would be great if an enlarged sample size can be taken to this aspect covering the entire country. REFERENCES Bigne, E., Ruiz, C., & Sanz, S. (2005). The Impact Of Internet User Shopping Patterns And Demographics On Consumer Mobile Buying Behaviour. Journal of Electronic Commerce Research, VOL. 6, NO.3, 2005, Central Bank of Sri Lanka. (2013). Annual Report - Economic and Social Infrastructure - Road Development. Colombo: Central Bank of Sri Lanka. Davis, F. (1985). A Technology Acceptance Model for Empirically Testing New End - User Information Systems; Theory and Results. Massachusetts Institute of Technology: unpublished Doctoral Dissertation. DIVISION)_RDA, I. (2014). Expressways. Retrieved November 2014, from Expressway Operation Maintenance And Management Division : Green, S. B; Lissitz, R. W; Mulaik, S. A. (1977). Limitations of coefficient alpha as an index of test unidimensionality. Educational and Psychological Measurement, 37, Quibria Nasreen (2008). The Contactless Wave: A Case Study in Transit Payments. Emerging Payments Industry Briefing. Federal Reserve Bank of Boston. Sivapragasam, Y., & Peiris, T. (2013). The factors affecting on internet banking adoption - A case study BIOGRAPHY OF AUTHORS Lalith Edirisinghe is a PhD candidate in Transportation Planning and Logistics Management He obtained his MBA in International Trade & Logistics - Awarded by Sri Jayawardanapura University and Canberra University. He holds a Postgraduate Diploma in Business Management with a Distinction pass and Executive Diploma in Marketing with Distinction pass both Awarded by University of Colombo. He is a Chartered Marketer and a Chartered Logistician. Having started his carrier as a Cadet in Ceylon Shipping Corporation now counts 25 years work experience in three mega shipping lines and Board of Investment of Sri Lanka. 289