R4TLI Conference Proceedings 2022

ISSN : XXXXX Paper Reference: R4TLI-XXX

EVALUATION OF COST EFFICIENCY OF BUS ROUTES IN KANDY,

SRI LANKA
P.M.G.D.M. Heratha, W.R.S.S. Dharmarathnab , Vasantha Wickramasingheb, , A.G.H.J. Edirisingheb

a,b
Faculty of Engineering, University of Peradeniya, Peradeniya, 20400, Sri Lanka
a
E-mail: e14135@[Link]

ABSTRACT: Bus transportation plays a significant role in Kandy, Sri Lanka as it is a reasonably priced
form of transportation. Route efficiency is a critical performance indicator for both bus service providers
and passengers. So, understanding and optimizing the cost efficiency of bus routes is crucial for
improving urban mobility and ensuring accessibility for all. Therefore, this study addresses the cost
efficiency of bus routes within the district of Kandy, aiming to identify the possible adjustments to fill the
supply-demand gap in bus routes in Kandy, considering the route cost-efficiency. In the city of Kandy,
four major types of areas were identified depending on the service level and population density (Lowe et
al., 2022). According to that, the areas which need an adjustment in supply and demand have been
identified. A range of individual route inputs are used to determine the route efficiency. The data
collection was done through collecting relevant documents from relevant depots, the Department of
Census and Statistics and through past research papers etc. Stochastic Frontier Model (SFM) is used to
analyze the input factors influencing the bus routes’ efficiency. SFM assumes that there is a parametric
function between inputs and outputs (K. Cullinane et al, 2006). To apply logarithmic frontier model
revenue per kilometer is taken as the dependent variable (output) and some bus route attributes like
number of buses allocated for a specific route, distance covered by the route, fare of the route, bus
capacity etc. are taken as independent variables. As a key goal of this study, by identifying the factors
affecting the cost efficiency of bus routes, the SFM is used construct a comprehensive model of cost
efficiency for bus routes. This statistical approach enables the identification of areas where operational
framework needs to be improved and allows for a more comprehensive evaluation of the efficiency levels.
Each route's efficiency is computed, and the results are used to recommend measures for improvement.

Keywords: Bus Transportation, Cost Efficiency, Stochastic Frontier Model, SFM, Route Characteristics

1. INTRODUCTION

Bus transportation is a significant part of daily life in Kandy because it's a reasonably priced mode of
transportation. While bus transportation plays a significant role, the quality of the services provided is
critical. In city of Kandy, four major types of areas were identified (Lowe et al., 2022) depending on the
service level and population density as,
[Link] Scored – High Density areas
[Link] Scored – Low Density areas
[Link] Scored – High Density areas
[Link] Scored – Low Density areas
 R4TLI Conference Proceedings 2022
ISSN : XXXXX Paper Reference: R4TLI-XXX

Figure 1: Areas depending on the service level and population density.

According to above mentioned four areas, low scored-high density areas and high scored-low density
areas are the areas which need an upgrade considering the supply and demand. As this study is to upgrade
unbalanced routes to a balanced system, routes in these two areas have considered. The design of bus
routes to maximize revenue generation is one of the many factors influencing the overall performance of
bus transportation. The characteristics of each bus route have an impact on the financial efficiency of the
bus service. In order to apply the logarithmic frontier model, the revenue per kilometer is considered as
the dependent variable (output), and the number of buses assigned to a particular route, the distance
traveled, the route fare, the bus capacity, etc. are considered as the independent variables.
For the data analysis purpose Stochastic Frontier Model (SFM) is chosen over Data Envelopment
Analysis (DEA) as SFM accounts for random shocks and noise, offering a more realistic picture of
efficiency and SFM estimates technical inefficiency, pinpointing areas for improvement. The purpose of
this study is to use the SFM to analyze the input factors impacting the efficiency of bus routes and to
evaluate the features of particular routes. Each route's efficiency is computed, and the results are used to
recommend measures for improvement. In order to enhance the performance of the bus route, strategic
recommendations would be made in light of the outcome.
.
2. RESULTS AND DISCUSSION

The maximum likelihood estimates of stochastic frontier model are made using Stata , and the result is
presented in table . The ordinary least square (OLS) coefficients provide a starting point for the maximum
likelihood estimate process. The goodness to fit of the estimated regression equation is evaluated by R 2 of
the least square method, which looks reasonably high at 0.6102. This implies that the inputs to the model
do satisfactorily explain the model output. The R-squared value of 0.61 indicates that the model explains
approximately 61% of the variation in revenue per kilometer across the bus routes. The remaining 39% of
the variation could be attributed to unobserved factors or limitations of the model. The OLS can also be
used for measuring the residual term ε, where as the inefficiency parameters can be calculated using the
half normal SFM results.
Rather than using the raw revenue data, the standardized revenue generated per kilometer (rev/km) is
taken as the dependent variable, and the logarithm of all variables is used for the analysis. A positive
coefficient for trip length (trip length) indicates that longer routes tend to generate higher revenue per
kilometer. This is likely because longer routes have the opportunity to serve more passengers, offsetting
the increased operational costs associated with longer distances. The negative coefficient for the number
of buses (no of buses) suggests that routes with more buses assigned might have a lower revenue/km.
 R4TLI Conference Proceedings 2022
ISSN : XXXXX Paper Reference: R4TLI-XXX

This could be due to factors like increased operational costs (more buses to maintain) or passenger
dispersion across a larger number of buses, reducing overall passenger load per bus.

Table 1: Model Estimation Results

Dependent variable: Revenue
Least squares Half-normal SFM
per km
Independent Variables β t-ratio β t-ratio
constant 1710.002 3.16 1710.064 0.11
Route length -7.91013 -0.35 -7.91013 -0.36
Fare -2.530328 -0.46 -2.530328 -0.47
Capacity -16.75666 -1.84 -16.75666 -1.89
No of buses -645.2417 -2.71 -645.2417 -2.78
Trip length 0.0224194 4.62 0.0224194 4.73
λ - - 0.0001067 -

- - 0.0766869 -

- - 718.834 -

Log likelihood , R2 - , 0.6102 - ,-1015.5643

3. CONCLUSION

The study reveals significant variations in ridership, revenue, and resource allocation (buses) across
bus routes in [Link] Frontier Analysis (SFA) can be a valuable tool to account for these
variations and assess the efficiency of each route by considering factors beyond the bus depot's control
(passenger fluctuations, external factors).By applying SFM, the research can identify routes that are
underperforming relative to their resources and characteristics. These routes might benefit from:
 Increased or optimized bus allocation
 Schedule adjustments to improve efficiency and ridership
 Exploring network connectivity improvements to enhance passenger access to certain areas
The findings can inform strategic decision-making for the bus depot to optimize resource
allocation, improve route efficiency, and potentially increase ridership and revenue.
Limitations
 The result doesn't reveal the specific SFM results (technical inefficiency scores).
 Further analysis is needed to identify specific underperforming routes and potential areas for
improvement.
Future Research Directions
 Investigate the reasons behind ridership variations across routes (socioeconomic factors, route
connectivity).
 R4TLI Conference Proceedings 2022
ISSN : XXXXX Paper Reference: R4TLI-XXX

 Explore the impact of service quality (waiting time, route reliability) on ridership and
revenue.
 Analyze the cost structure of the bus company to understand the profitability of different
routes.

REFERENCES

1. Gebeyehu, A., & Takano, M. (2006). Stochastic frontier analysis of bus route efficiency in Addis
Ababa, Ethiopia. Transportation Research Part A: Policy and Practice, 40(7), 553-570.
2. Asmael, M., & Waheed, A. (2020). Assessing the operational efficiency of urban bus routes using
data envelopment analysis. Transportation Research Part E: Logistics and Transportation Review,
137, 101911.
3. Farsi, M., Filippi, C., & Nijkamp, P. (2006). Efficiency analysis of public transport systems:
Accounting for unobserved heterogeneity. Transportation Research Part A: Policy and Practice,
40(1), 3-19.
4. Deng, W., & Yan, R. (2016). Evaluating bus route efficiency from a passenger perspective: A
network DEA approach with epsilon-based measures. Transportation Research Part E: Logistics and
Transportation Review, 94, 22-38.
5. Lowe, W. et al. (2022) “Analysis of Network Features and Travel Characteristics for Enhancing Bus-
Based Public Transport System-Case Study in Kandy, Sri Lanka,” Journal of the Eastern Asia Society
for Transportation Studies, 14, pp. 1146–1157.