Decision-Making Support of Sustainable and Efficiency of Railway Project: Case Study China-Pakistan

The China-Pakistan Economic Corridor (CPEC) is a strategic economic project aiming at increasing regional connectivity for economic development. The economic corridor will connect Pakistan’s Gwadar port with Kashgar in Western China between 2014 and 2030 by developing a transport infrastructure network consisting of road and rail. It is not only expected to be beneficial for Pakistan and China but is also expected to have positive spillover effects on other countries by enhancing geographical connectivity.


INTRoDUCTIoN
The Belt and Road Initiative (BRI) is a strategic vision proposed by China, that focused on connectivity and economic support on a transcontinental scale.The range of activities that will be part of Belt and Road initiative is very wide, including infrastructure, policy coordination, financial and people to people exchange.Transportation infrastructure contributes to trade and economic integration, which further leads towards sustainable development.Moreover, excellent infrastructure provides a suitable environment for foreign investors (Ramachandran,2011).In this initiative, Chinese government aims to connect China with its OBOR partners through the Silk Road Economic Belt (SREB), Maritime Silk Road (MSR), and Digital Silk Road (DSR) (Nurmyrat,2017).The importance of One Belt One Road (OBOR) is increasing day by day, and aaccording to official information, by January 2021, 133 to 140 countries had signed cooperation agreements for the OBOR Initiative since 2013 (Ghanem, Li,2021).For countries and organizations to join the OBOR, China and the respective country or organization sign a Memorandum of Understanding (MoU).More recently, China now imports more goods from countries that have signed up to the OBOR Initiative than it exports (Islam,2009).Chinese

LITERATURE REVIEw
A well-developed and reliable transport system plays an important role in economic development and attracting investment.The strategic location of Pakistan is exceptional and is well suitable to serve as the center of commercial activities.The contribution of transportation is around 10% to GDP and provides above 6% of employment opportunities in Pakistan.The consumption of transport sector is about 35% of the total energy yearly and accounts for about 15% of the government development projects (Esfahani,2003).The rail transport system in Pakistan is diverse and, helps a population of more than 191 million individuals.Pakistan Railways is a state-owned organization under Ministry of Railways, is also experiencing development in recent years under CPEC.
The significance of evaluating transportation modes concurrently stems from the fact that transport networks are the pillar of sustainable metropolitan improvement and in the public transport systems, consist of suburban railway, subway and light-rail lines (Jain,2000).Data Envelopment Analysis (DEA) has been applied in the field of transportation to evaluate efficiency.Numerous studies have used the DEA approach to investigate the efficiency of transport services (Barros,2010) and(LaoY.2009)airports and airlines (Barros,2011).In order to compete with other firms in the market, businesses such as railways, airways, banks, and private companies need to reach their optimal performance.Mohamad Said (2010) propose that one of the major objectives of business organizations is to improve efficiency.Efficiency can be measured with respect to maximization of output, minimization of cost or maximization of profits.An organization is considered as efficient if it is able to produce maximum outputs from given inputs or minimum inputs are used for the production of given outputs.Vincova(2005) suggested possible ways for productive units to measure their efficiency.DEA method has been used as a tool to measure efficiency and to estimate the relative efficiency of a chosen unit in a given group.Beriha et al. (2011) applied DEA, Constant Return to Scale (CRS) model to estimate technical efficiency and to identify the benchmarking units for safety performance in Indian industries.The number of accidents in units was taken as outputs, and percentage of annual budget for various safety activities was taken as inputs.Seven units out of thirty were found to be efficient under the CRS model.Based on this model, benchmarking was done for inefficient units to become efficient units.Ozbek et al. (2009) addressed the application of the DEA method as innovative and powerful method to help the decision-making process, to derive meaningful conclusions from obtained results to acknowledge the limitations of the DEA.This effort will help the researchers in improved utilization of DEA especially while addressing vain transportation-related problems (Osman, Xue,2020).Jiang (2009) used DEA approach to evaluate the transportation system efficiency for thirty-one major regions in China.According to his findings, the evaluation framework, which used six output variables and nine input variables, the results show that thirty-one major regions in China were efficient in transportation except Hubei and Chongqing.Furthermore, Savolainen (2007) and Osman (2020) used DEA to evaluate the relative technical efficiency of three European transportation systems: rail, maritime and air.The results show that privately-owned airline companies are significantly more efficient in operating their passenger services.Railways show huge variations between different countries and between different years within the same company in relative technical efficiency.The mode choice with rail substitute in an urban freight has been rarely modeled.Consequently, a choice modeling structure has been offered, that contains the choice of the kind of service and the mode transport on the base of earlier analysis and model at urban freight context (Savolainen,2007).Oum and Yu (1999) measured the productive efficiency of the railway's industry in nineteen OECD countries.They used the DEA method to measure the gross efficiency index whereas Tobit regression was used to identify the effects of the public subsidies.Similarly, Ghanem (2020) measured alternative methodologies for comparing and measuring the efficiency of railways and published a complete overview of productivity and efficiency in rail transport suggesting that the estimates are very sensitive to output specifications.Hilmola(2007) used DEA to analyze the efficiency and productivity of railway freight transportation in different European countries from 1980 to 2003.In a similar way, partial productivity analysis is also used to support DEA evaluation.The efficiency analysis shows that the most efficient railway freight transportation is located in the Baltic States.Based on partial productivity analysis, the productivity of locomotives and railway tracks should be in primary target of productivity improvement in highly efficient years.Caulfield et al. (2013) employed the DEA method to identify the most efficient solution for the airport route and to establish the reasons for its inefficiency.The findings and research method could be used in other public transport investment cases to ascertain the best case for investment and could be used for a cost-benefit analysis.Yu and Lin (2008) proposed the multi-activity network DEA represents both efficiency and effectiveness.This model is applied to twenty selected railways for the year 2002 for estimation of efficiency of passenger and freight simultaneously.Lan and Lin (2005) proposed a four-stage DEA approach to estimate the technical efficiency and service effectiveness of railway transport.The results present strong evidence that efficiency and effectiveness scores are overvalued.Jitsuzumi and Nakamura (2010) suggested a method based on DEA to examine the reasons for inefficiency in Japanese railway operations and to calculate the further optimal subsidy levels.Yu (2008) applied the traditional DEA and network DEA to examine the efficiency and effectiveness of 40 world railways in the year 2002.The results obtained from network DEA are compared with traditional DEA indicate that performance measures are quite different in terms of magnitude and using different DEA models to assess railway performance.Arrigo (2014) and Maudos (2001) examined the State aid for rail transport as a factor for railway development.The results of these studies can be summarized as follows: The classifications of railway market aim to deal with the following problems: transport problems related to improving the competitive position of the railway transport with regard to other modes of transportation; infrastructural problems rehabilitation and development of the railway lines and stations; technical problems, improving the technical level of rolling stocks; integrated problems concerned with the evolution of the productivity of the railways.Movahedi et al. (2007) did not differentiate efficiency and effectiveness and evaluate efficiency by taking passenger-km and tone-km as outputs.It is necessary to differentiate efficiency and effectiveness by clarifying the non-storable nature of railway transport to identify the reasons for poor performance in order to suggest more practical development policies.From the review of previous literature, most of the earlier studies focused on evaluating the performance of railway industry nationally or provincially, however, very few studies focused on analyzing the performance regionally and not a single study is found to examine the performance of the railway industry in OBOR countries.The aim of this study is to fill the gap in the literature by application of DEA model in the railway industry of OBOR countries to examine the performance by differentiating efficiency and effectiveness.Secondly, to identify the reasons for poor performance during different time-periods.

Data Envelopment Analysis (DEA)
Data Envelopment Analysis (DEA) is a linear programming technique based on the pioneering work of Farrell (1957) to measure the relative efficiency of each unit called Decision-Making Units (DMUs).DEA enables to assess the efficiency of a firm, agency or such other unit that uses the inputs to generate a set of outputs relative to other units.In this research, the DMUs of One Belt One Road countries are the different years from 1978 to 2018.These DMUs use a variety of inputs to produce outputs, and DEA analysis attempts to identify the most efficient DMUs and points out specific inefficiencies in the other DMUs.DEA has been applied in the field of transportation to examine efficiency.Several studies have used the DEA approach to examine the efficiency of transport services airports and airlines, (Eurostat,2014(Eurostat, ,2016) ) and (National Bureau of Statistics of China,2018).
There are two types of DEA models, constant returns to scale (CRS) and variable returns to scale (VRS) model.Charnes and Rhode (1978) proposed the CRS model and called it CCR.Banker, Charnes, and Cooper (1984) suggested the VRS model and called it BCC.Input-oriented constant returns to scale is used in this study.Constant returns to scale (CRS) model is one of the most basic DEA models, and CRS model uses linear programming to extend single input/output efficiency measures to the multi-input/multi-output.Constant returns to scale and variable returns to scale models can be applied to the input or output orientation.The input-oriented measure keeps output fixed, explore the proportion of the possible reduction in inputs while output-oriented measure keeps the inputs fixed, and explore the possible proportional expansion in output.With the most fundamental Data Envelopment Analysis method is DEACCR (Banker, Charnes Cooper, 1984).The mathematical function of DEACCR method is offered with number of Decision-Making Units (DMUs) to be measured.Each DMU has inputs, and different outputs (Banker, Seiford,2004) and (Charnes, Lawrence M. Tone,1994).For a set of technical efficiency of n DMUs donated by DMUj= (j=1, 2,... n) each DMU has m inputs donated by: xi (i = 1, . .., m) .The input weight is vi (i= 1, 2, . .., n).Each DMU has q outputs donated by: yr(r = 1,2, . .., q).The output weight is ur(r=1, 2, . .., q).The linear function used to present the input-oriented CCR analysis is presented as: 1 2 1 2 , ... ; , ..., ; , ..., u r , v i : weights of input and output.y r , x i : the experiential variables of inputs and outputs amount of (DMUs).

SySTEM INPUTS AND oUTPUTS
In this study, input and output variables used in DEA are considered as internal factors.An important feature of DEA is that it does not assume a functional form.DEA does not need any production function equation of a parametric form for the solution of the specified model.Therefore, any variable can be included in the model without the need to specify functional or parametric relationships (Markovits,2015) and (Charnes,1994).DEA does not make a priori distinction between the relative importance of any two inputs or two outputs.In other words, all variables that are included in the model have an equal opportunity to influence the calculated efficiency.To select the input and output variables for the DEA model used in this study, we take into account, factors with cost and benefit in nature.Factors having cost in nature are considered as input variables and factors with benefit in nature are considered as output variables.Based on the review of past studies (Nuzzolo,2015), (Beriha,2011), (Jitsuzumi,2010) and (Caulfield,2013) availability of data, five input variables and two output variables were chosen for Data Envelopment Analysis.Figure1 presents the framework for evaluating railway efficiency.Data used in this study were collected from dependable sources (Eurostat, 2016), (National Bureau of Statistics of China,2018) and Eurostat.Publications Office of the European Union Eurostat Regional,2014) Table 1 and 2. To process the data MAXDEA ULTRA Software was used, the DMU that ensures maximum railways goods transported and railways passengers carried performance with their existing inputs were determined, if the efficiency score is 1, and if the inputs or outputs of this DMU do not have mixed inefficiency, this country is deemed efficient.

RESULTS AND DISCUSSIoN
To process the data "MAXDEA ULTRA" Software was used, the DMU that ensures maximum railways goods transported and railways passengers carried performance with their existing inputs     4 and 5 show the technical efficiency scores in terms of DEACCR and DEABCC, countries were not found efficient in this analysis either needed to decrease their inputs to achieve the efficiency under this condition without reducing outputs, and that depend on the weight of the input values of their reference countries.

Clarification the Results
Based on the result above, and for deep analysis to investigate whether railway transport of OBOR countries utilizes its railway performance, and identify the current situation of rail connection of China -Pakistan Economic Corridor, the following techniques data analysis involved the data compilation that retrieved from the MAXDEA ULTRA software which includes efficiency scores, slacks, listing peer, benchmark projection, and strong efficiency projection.

Slack Analysis
The analysis results of an efficiency of OROB countries between 1977-2012 and 2013-2018 railway (Input-Oriented -CRS and VRS analysis) show inefficient of some OBOR countries because there are slacks values and the inputs values are not equal to 0, assuming that there is capability for comparatively enhancement and the result obtained is not the best, for example China as an efficient country to the inefficient.China is reference efficient country for inefficient countries such as Belgium, Austria, Spain, Belarus, Bosnia and Israel in term of CRS (1977CRS ( -2012)), by using VRS  China is a reference efficient country for inefficient countries, such as Belgium, Austria, Spain and Belarus.Table 6.

Benchmark and projection of the inefficient countries.
Listing the references benchmarks and target values for benchmark and projection (Input-Oriented CRS and VRS analysis), China is used as reference country for some OBOR countries, which means the projection point of inefficient countries on the frontier is made up of a linear combination of input and output of China, and the projection values of inputs and outputs as presented in

Strong efficiency projection (China is reference efficient country)
If the proportionate movement is finished for the inefficient DMUs, the country after enhancement may be a weak efficiency because there is a slack value needs to change decreasing or increasing to be a strong efficiency country.Solution point of inefficiency country on the frontier country to be referenced on to inefficiency country and represents its objective value.To convert inefficient country to become strong efficiency, a proportionate movement of slacks values movement is essential and needed.In the present study to project the inefficient country into strong efficiency is by finishing proportionate movement and slack values movement Table 8.

Strong efficiency projection (Pakistan is inefficient)
The results from Tables 9, 10 using CRS and VRS (1977-2012), (2013-2018) analysis indicate that many DMUs are inefficient country but their efficiency scores are near to 1 such as Pakistan.Pakistan railway operates at 89.8 percent level of overall technical efficiency, inputs could be decreased by 10.2 percent, and its four reference frontier DMUs.Pakistan is inefficient in terms of CRS and VRS (1977-2012and 2013-2018) and its score are higher than 0.9 but not equal to 1, the results of input and output targets indicate the quantity of the inputs and outputs required for achieving efficiency    competitive and efficient.This investment will be used to upgrade existing lines, modernizing and improving rail connections and extending the rail network.In the OBOR region, the existence of a rail connection between trading partners especially China and Pakistan according to CPEC is associated with a large impact on improving trade.This dissertation attempts to estimate the overall technical, pure technical and scale efficiency of rail network CPEC linked with OBOR and develop a manner of assessing the comparative efficiency of railway between 1977-2012 and 2013-2018, the study employed Non-radial methods of Data Envelopment Analysis (DEA), efficiency scores and technical efficiency were ranked and compared, while techniques data analysis MAXDEA ULTRA is applied to examine slacks, listing peer, benchmark projection, and strong efficiency projection.The slack analysis results indicate that all slacks of efficient DMUs are zero, and considered as fully efficient DMUs, and the target value for inputs and outputs are same as original value.For inefficient DMUs, slack values are non-zero and show how much the outputs need to increased and inputs need to be decreased to improve efficiencies.The results of the efficiency and performance of the railway in which DMUs were selected in order to calculate the efficient DMUs of OBOR as follow.European Union countries, the efficient DMUs were Sweden, France, Italy Latvia and Netherlands in terms of CRS andVRS (2013-2018), and Luxembourg, Lithuania, Estonia and Ireland are efficient in terms of VRS (2013VRS ( -2018)).According to the East Asia and Pacific countries (EAP), the rail performance in Thailand is efficient in terms CRS and VRS (1977-2012and 2013-2018).The rail performance efficiency of DMUs in Middle East and North Africa (MENA) was Egypt in terms CRS and VRS (1977-2012and 2013-2018).The rail performance efficiency of South Asia (SAR) countries were only India is efficient in terms of CRS and VRS.The rail performance efficiency of Europe and Central Asia countries (ECA) were Russia, Ukraine and Kazakhstan in terms of CRS and VRS (1977-2012and 2013-2018).
To attain overall technical efficiency, Pakistan Railway is required to increase output O2 Freight, ton-kilometers by 257, and decrease input I 1 (Number of passenger transport vehicles) by 3293 and I5 (Number of employed persons) by 4600 in terms of CRS and required to increase output O2 (Freight-Ton-kilometers) by 188, and decrease input I 1 (Number of passenger transport vehicles) by

I 1 :
Total track-kilometre I 2 : Number of passenger transport vehicles I 3 : Number of locomotives owned I 4 : Number of freight wagons owned I 5 : Total number of employed persons(1000) Output variables O 1 : Passenger-(passenger-km) O 2 : Freight (Ton-kilometres)

Figure 1 .
Figure 1.Framework of Evaluating Railway Efficiency of CPEC countries such as China is a reference efficient country for inefficient countries, such as Belgium, Austria, Spain, Belarus, Bosnia and Israel.UsingVRS (1977-2012) and VRS (2013-2018)  analysis indicate that 25 countries in 1977-2012 and 26 countries in 2013-2018 having an efficiency score of 1 have efficiently used their resources better than other inefficient countries, most of efficiency scores of 2013-2018 have increased according to them scores in 1977-2012, and few countries in 1977-2012 have decreased efficiency scores, the most efficient countries are ten countries from EU, one country in each of EAP and SAR, two countries from MENA, eight countries from ECA, China have an efficiency score of 1.Many countries are inefficient country but their efficiency scores are near to 1,such as Pakistan and Indonesia .Inefficient countries need to reduce their inputs values, as much as the weights of the input values of their reference frontier countries such as China is a reference efficient country for inefficient countries, such as Belgium, Austria, Spain and Belarus.
and need to reduce its inputs values.To attain overall technical efficiency, Pakistan Railway is required to increase output O 2 (Freight -Ton-kilometers) by 257, and decrease input I 1 (Number of passenger transport vehicles) by 3293 and I 5 (Number of employed persons) by 4600 in terms of CRS and required to increase output O 2 (Freight-Ton-kilometers) by 188, and decrease input I 1 (Number of passenger transport vehicles) by 3378 and I5 (number of employed persons) by 3670 in terms of CRS and VRS.Tables 9,10.Pakistan is a central part of OBOR Initiative and needs to increases the technical efficiency of Pakistan railways evidently.Likewise, the development of highways in Pakistan would decrease the technical efficiency of railways because road transport is used as substitute for rail transportation.Pakistan has a railway connection with neighboring countries of India, Iran, and Turkey while future projects are underway to construct railway connection with China and Afghanistan.Pakistan

Table 1 Inputs and Outputs of (OBOR)countries railways, average 1977-2012
were determined, if the efficiency score is 1, and if the inputs or outputs of this DMU do not have mixed inefficiency, this country is deemed efficient.Table3shows the efficiency scores of the OBOR countries railway from years 1977 to 2012 (before OBOR initiative) and from 2013 to 2018(after OBOR initiative) are presented in terms of CCR and VRS.The results from Table3 using CRSandCRS (2013CRS ( -2018) )analysis indicate that 17 countries in 1977-2012 and 18 countries in 2013-2018 having an efficiency score of 1 have efficiently used their resources better than other inefficient countries, most of efficiency scores of 2013-2018 have increased according to them scores in 1977-2012, and few countries in 1977-2012 have decreased efficiency scores, the most efficient countries are five countries from EU, one country in each of EAP, MENA, six countries from ECA .China have an efficiency score of 1.Many countries are inefficient country but their efficiency scores are near to 1, such as Pakistan and Indonesia .Inefficient countries need to reduce their inputs values, as much as the weights of the input values of their reference frontier

Table 2 . Continued Table 3 Efficiency score based on CRS and VRS
With Input-Oriented model and two different DEA approaches used, the technical efficiency of efficient countries is 1 in both models.The efficient countries have very good ability to convert the railway indicators into useful outputs.Average efficiency of railway performance in OBOR countries between1977-2012 and 2013-2018, according to models of DEACCR and DEABCC, technical efficiency score (CRS) 0.68 and pure technical efficiency score (VRS) is 0.79.Average scale efficiency is 0.85, but in 2013-2018, technical efficiency score(CRS) is 0.69 and pure technical efficiency score (VRS) is 0.82 and average scale efficiency is0.83,Tables

Table 7
, similar results can be achieved for the other inefficient DMUs.

Table 6 Efficiency score and slacks (1977-2012) Railway performance CRS (Belgium)
Three main locomotive workshops and thirty-five smaller workshops are responsible form maintenance.Pakistan railways have fallen behind in routine and periodic maintenance due to resource limitations, leading to weakening of the whole infrastructure.Tracks are in poor condition with speed limitations that result in unnecessary interruptions and high transport costs.Ministry of railways launched a special project to update and enhance the carrying capacity of the rail network.This project will fulfill the numerous gaps by replacing the locomotives that are no longer fit for use, modernizing or buying new rolling stocks, renovating severely deprecated sections of tracks and updating plant and machinery at different workshops.These measurements will help to increase the carrying capacity, increase speed and safety standards.

Table 8 Strong efficiency projection (1977-2012) Railway performance CRS (Belgium)
CoNCLUSIoNRail transportation plays an important role in the movement of goods.Efficient rail transport infrastructure facilitates the faster, safe and low-cost transfer of goods and has a positive impact on trade.CPEC will provide an enormous investment in enhancing the rail network to make it

Table 10 Strong efficiency projection (2013-2018) Railway performance CRS (Pakistan)
(number of employed persons) by 3670 in terms of CRS and VRS.Pakistan is a central part of OBOR Initiative and needs to increases the technical efficiency of Pakistan railways evidently.Likewise, the development of highways in Pakistan would decrease the technical efficiency of railways because road transport is used as substitute for rail transportation.The CPEC project emphases major upgrades to Pakistan's aging railway system, including rebuilding of the entire Main Line 1 railway.Pakistan is a central part of OBOR Initiative and needs to increases the technical efficiency of Pakistan railways evidently.Likewise, the development of highways in Pakistan would decrease the technical efficiency of railways because road transport is used as substitute for rail transportation