Analysis of Passenger Transportation Demand: Case Study of the Customs House, the Thai-Laos Friendship Bridge, Nong Khai Province

Thailand and the Laos PDR have a very long border of 1,810 kilometers which has been involved with international economic transactions in terms of trade, investment, tourism and passenger transportation over a long period. There are 36 custom check points allowing passenger transportation between the two countries by international bus, train, car, or private hire van. The trend of passenger transportation has recently increased because of the rise of tourists and investors as well as the supporting promotion policy of both countries particularly for the passenger transportation business. The growth of this business can be attributed to the capacity of Nong Khai province with respect to tourist attractions and destinations, food, culture and the natural environment. Therefore, it is very interesting to examine and estimate the passenger transportation demand model in order to investigate the impact of the variables that determine the model. This study also proposes policies that could enhance the effectiveness of entrepreneurs’ strategies. Moreover, the results of this study can be used as guidelines for both entrepreneurs and related government organisations. This study aims to explore the impact of factors influencing passenger transportation demand via estimating the passenger transportation demand model. The appropriate combination estimation techniques are used. The study period was between July 2007and June 2010 and monthly data was used to estimate the model. The data used to develop in this study has been mainly obtained from the Bank of Thailand data base, the annual report of the entrepreneurs, the interview and International Financial Statistics from the IMF. The sample and data covered only the two countries in the Mekong Sub-region according to the objectives of the study. Moreover, the data is both time series and non-time series data. Therefore, it can be affirmed that the data used to construct and estimate the econometric model is suitable. The following linear model was constructed based on the general passenger transportation demand. The variables were developed to suit the passenger transportation between Thailand and Laos. ..................................(1) Where: Qa is the total number of passengers who use the service provided (people/month) Pa is the fee paid by customers who use international bus service to travel from Nong Khai to Vientiane deflated by public transportation price (2007 base year, baht/person) Pb is the fee paid by customers who use the regular bus service to travel from Nong Khai to Vientiane deflated by public transportation price (2007 base year, baht/person) GDP is the gross domestic product of Thailand (as 1998 constant price, million baht) EX is the exchange rate (Baht/Kip) HO is the number of holidays in each month consisting of weekends and special occasion holidays TR is the frequency of service obtained from the service schedule From equation (1) above, it can be stated that all the independent variables determine and influence the passenger transportation demand that is represented by total number of passenger who use the service provided (Qa). Additionally, the price of substitute goods is combined in the equation (1) that is expressed by Pb that is consistent with general demand theory. According to equation (1), the linear function can be written in the linear econometric modelling form as follows: …………… (2) Where: 0 is the constant term 1 - 6 is the coefficient of each independent variable t is the error term 5.2) Justification of the Model Estimation Methods The model estimation methods undertaken have to be appropriate in order to propose the results as well as policy implications and recommendations are plausible. As a consequence, it is needed to conduct multi steps and estimation techniques by following. 5.2.1) The Unit Root Test 5.2.2) The Co-integration Test 5.2.3) The Coefficient Test 5.2.4) The Error Correction Mechanism (ECM) Test 6.1) Unit Root Test It can be explained that for all independent variables, the stationary level of time series data used in the estimated model needed to be tested first via the Augmented Dickey Fuller (ADF) test. This technique is an appropriate method of checking the mean and variance of data. The first order stationary level or I(0) has three different model testing formats: 1) No intercept and no trend equation 2) Only intercept equation 3) Both intercept and trend equation The comparison result between ADF t-Statistic and McKinnon critical value has to be considered to accept or reject the null hypothesis that each independent variable has no unit root or stationary. The empirical comparison result above found that the null hypothesis can be accepted and that the stationary of almost independent time series variable is the same level because of the ADF t-Statistic is lower than the McKinnon critical value at different level of significance of 0.01, 0.05 and 0.10 respectively. Nevertheless, there are some independent variables which rejected the null hypothesis but accepted the alternative hypothesis consisting of HO and TR. The alternative hypothesis showed that there are both intercept and trend variables in the estimated model at 99% of the significance level. The above mentioned independent variables had to be tested to make all independent variables at the same stationary level. The higher stationary level was tested by using the first different order estimation technique to obtain the most accurate and reliable empirical results possible. This procedure revealed that all independent variables are stationary at the same level of the first different order or I(1). The different levels of significance of 0.01, 0.05 and 0.10 are used to support this claim. 6.2) Co-integration Test For the next step, the long-term relationship of the time series data was tested by using a co-integration process of the Engle and Granger test consisting of the following procedures: Firstly, the residuals of error term ( t ) obtained from an estimated equation using OLS had to be taken to test the stationary level of the total number of customers who used the service (Qa) as well as other independent variables. The integration of zero order or unit root test by using ADF test was then calculated. The final step was to take the residual from the estimated equation by OLS to test the stationary level via the same process. Regarding the empirical results, it can be stated that the ADF test is less than the McKinnon critical value at a significant level of 0.01. As a consequence, it can be concluded that the error term has the stationary characteristic or integration of order zero or I(0). Therefore, the total number of passengers who used the international transportation service between Thailand and Laos (Qa) has a long-term equilibrium relationship with each independent variable. However, the multicollinearity can be detected by considering the correlation coefficient of two independent variables of Pa and Pb that is equal to 0.9998. Consequently, the independent variable of substitute good (Pb) can be deleted from the estimated equation. Furthermore, the first conclusion that can be drawn is that the independent variables Pa and Pb are nearly perfect substitute goods, which is consistent with demand theory. 6.3) Coefficient Test The next step after the co-integration process was coefficient testing in order to detect any heteroskedasticity and autocorrelation problems. Heteroskedasticity were tested by co-integration estimation results and autocorrelation problems were tested by using the Durbin-Watson (DW) statistic. The outcome was that no problems were detected. It can be implied that the total number of passengers (Qa) has a long-term equilibrium relationship with all independent variables. This relationship can be written as follows: (1.73)* (3.14)*** (-2.66)** (1.10) (2.07)* = 0.53 = 0.40 = 17,551.96 = 2.30 = 0.011** Note: 1. *** is statistic level of significance at 1% (0.01) 2. ** is statistic level of significance at 5% (0.05) 3. The number in blanket is t-statistic From the equation above, it can be explained that the total number of passengers from Thailand to Laos (Qa) has a positive relationship with the fee (Pa), the frequency of service provided (TR) and the GDP. On the other hand, the total number of passengers from Thailand to Laos (Qa) has a negative relationship with the exchange rate (EX). 6.4) Error Correction Mechanism (ECM) Test The ECM is a mechanism that enables an examination of the short-term equilibrium adjustment process. Therefore, the error equilibrium adjustment term is the linkage between short-term and long-term adjustment mechanism. This mechanism and linkage can be expressed as the following equation. A change of the total number of passengers using the transportation service between Thailand and Laos influences the change of fee (Pa), the frequency of service provided (TR) and the GDP in the same positive direction. However, a change in the total number of passengers influences the exchange rate (EX) and the number of holidays (HO) in the opposite direction. The negative coefficient of error correction term is consistent with the equilibrium adjustment theory. According to this theory, the value of error will be much lower leading to an adjustment for long-term equilibrium. It can be stated that the speed of adjustment to the long-term equilibrium is -46.27% if there are situations that influence the total number of passengers that deviate from equilibrium in each period. The Breusch-Godfrey serial correlation LM test is used to test and accept the hypothesis that there no autocorrelation occurred.