Energy Policy 46 (2012) 7887

Contents lists available at SciVerse ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Economic evaluation of current conditions of competition and

efciency of automotive and rail systems in Colombia
Carlos Andres Vasco Correa n
Applied Microeconomics Group, Universidad de Antioquia, Calle 67 Nro 53-108, Medelln, Colombia

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 10 May 2011
Received in revised form
12 March 2012
Accepted 13 March 2012
Available online 13 April 2012

This study uses microeconomic data from the transportation systems of land cargo in Colombia: rail
and trucking, to determine their degree of allocative efciency through the non-parametric method
Data Envelopment Analysis DEA. The average overall efciency found was 74.4% for trains and 20.56%
for trucks. These gures indicate that rail is more efcient in the allocation of resources. This means
that trains in Colombia use their input better than trucks to maximize their production, given the costs
and technological characteristics of each system. This is a signal for the design of a public policy for
investment in transportation infrastructure that seeks to raise the competitiveness of Colombian
exports, investing not only in roads but in complementary systems such as railways too.
& 2012 Elsevier Ltd. All rights reserved.

Keywords:
Allocative efciency
Data envelopment analysis
Colombia freight infrastructure

1. Introduction
Colombia is currently undergoing a delay in transportation
infrastructure. First, the large cities and industrial areas are far
from seaports, its process of geographical conguration is due to
the as occurred colonization in the 19th century and the strategy
of import-substitution model in the 20th century. Second, the
characteristics of the Colombian landscape composed of rugged
mountains, ood valleys, geological fault lines, among others,
make the topography an obstacle for the competitiveness and
productivity of the country against the rest of the world. Accompanied by a design of development and investment policies for
transport infrastructure with a regional vision, in lieu of a
national vision, aligned then are a set of factors that led Colombia
to todays backlog in its transportation infrastructure, thus harming a cross-sectional productive apparatus (Banco Mundial, 2006).
On the other hand, Colombia is immersed in a process of
internationalization with the goal of transforming its productive
structure toward an economy that produces goods and services
with an increased added value, a process that allows the country
to engage in international markets. The signing of free-trade
treaties invites the State to identify those bottlenecks that prevent Colombian products from arriving with competitive prices to
these markets. The strategy addressed has been granting large
transport infrastructure projects, the great majority roads, leaving
aside complementary systems such as railways (Bamrud, 2010).

The transportation infrastructure is a key factor in economic

development, more than a consequential relationship. Throughout history the Colombian State has developed its transport policy
in a planning framework of short and medium term, which has
not allowed for construction of infrastructure that boosts trade
ow. Roadways receives a little more than 80% of the private and
public investment, the market is regulated by means of a table of
minimum freight price, such as the presence of barriers to entry.1
Atypical working conditions that in recent years led to friction
between the cargo generators and the trucking sector, manifesting itself in national strikes, that in the absence of alternative
transport systems that could put pressure on their early resolution, instead ended by immobilizing the ow of cargo by weeks,
seriously affecting the entire logistics chain with lack of inventories, loss of the product, and price increases (Restrepo, 2008).
All the factors mentioned above: infrastructure in poor condition, absence of multi-modal systems, public policies for investment planning in the short and medium term, thinking regional,
have Colombia with very high export costs. In Fig. 1 are compared
the costs of export a container by country. For Colombia more
than half correspond to transportation costs, which together with
tax, customs and trade transaction costs, weaken the competitiveness of exports.
The need to carry out a plan to develop infrastructure was
confronted by the Mexican government, when it signed the
NAFTA trade treaty with its neighboring Canada and the United
States. Then, proceeded to carry out an analysis of competition
and efciency of systems in their current state, to determine what

Tel.: 57 3137921189; fax: 57 4 2195843.

E-mail addresses: carlosvasco@gmail.com,
carlos.vasco@economicas.udea.edu.co
0301-4215/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.enpol.2012.03.029

Import quotas for cargo vehicles.

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

79

Fig. 1. Comparative export costs and times.

Source: (World Bank, 2012).

should be the focus of public policies for investment in infrastructure. As a result of this study were found indicators of a
steady progress in railway system in front of a slow and erratic
growth roadway system, which provided clear signs of what
should be the priority (Lopez et al., 2004).
The transport sector in Colombia requires changes, it is necessary
to resolve their structural problems and to promote their adaptation
toward a market that is increasingly competitive and globalized. A
study like this is needed to determine if the current policy of
investment in transport infrastructure, which was focused on the
vast majority roads in detriment of rail, matches the ndings of level
of competition and efciency of both systems. It is necessary to
determine which are the characteristics of trucks and rail systems,
every one of its peculiarities, its advantages and disadvantages,
which are translated into an analysis of competition where price and
an analysis of allocative efciency in both transport systems is built
using the methodology non-parametric analysis of data envelopment DEA as the most appropriate given the data acquired.
The document is divided as follows: Section 2, an analysis
of competition and characteristics of the trucking and railroad;
Section 3, methodology used, where the following are some essential theoretical aspects about Data Envelopment Analysis used in
this study. Section 4, where we show a review of studies associated
with the transport in the Colombian economy, through a review of
literature on the topic, at national and international level. Section 5
discriminates the process of data collection, statistical analysis,
correlation and empirical results under DEA and its interpretation.
Finally, a section with the conclusions and some suggestions arising
from the work carried out.

2. Analysis of competition
Colombia counts with 164,000 km of roads, of which only 10%
corresponds to primary network. From the main grid only 9.82% are
in very good conditions, 31.09% in good conditions, and a large
percentage of the other 59.12% are in fair, poor and very bad
conditions according to the National Highway Institute INVIAS.
The road network is concentrated mostly in the Andean region,
where the three biggest cities are located, Bogota, Medelln and Cali,
without a backbone or transverse to dual carriageway to interconnect
the cities with the ports or between them. Nearly two thirds of the
network is in a bad status, in places geographically unstable, which
makes its maintenance and operation extremely hard on standard
conditions. More than 80% of cargo ow is carried away by these
means, in detriment of taking advantage of benets offered by the
other modes. The State allocates almost entirely all transport investment resources on roads (Ministerio de Transporte, 2009).

Recently, they are making an aggressive project award of dual

carriageway, such as Ruta del Sol which will communicate
Bogota with the Atlantic ports; Tunel de la lnea on Buenaventura Bogota road, which shall communicate the capital with
the port on the Pacic Ocean and the city of Cali; Autopistas de la
a, which will connect Medellin with the two highways
Montan
mentioned above and the port of Uraba.
On the other hand, the countrys rail network is deployed over
3458 Km, both cross-sectional shape, such as longitudinal, in
particular following the valley of Magdalena river. Of these
3458 km, 1991 are left, 1322 are inactive and 145 are a private
network. This network is currently in an advanced state of
deterioration as well as lagging technologically. Some sections
are incomplete or missing, which was formerly the main mode of
transport, today by several variables and determinants, ended
relegated and dismantled, in general because of problems in the
design of concession agreements, where the State was who bore
the majority of risk investment, the contractor was limited to
make prots of the system until it went bankrupt.
From a technical point of view, the railway has certain advantages and disadvantages if it is likened with roadway trucking
systems. The railroad has a greater capacity; their equipments
had a greater useful life, and allows systematization in the operation
and control in comparison with trucks. A technical aspect is key to
understanding the difference between both, and that is that there is
a much lower friction among wheel of the train and the railroad,
versus that of the rim of the truck and the road; while a train has a
resistance of progress from 4 to 5 pounds per ton, a truck faces a
resistance of between 25 and 30 pounds per ton.2 As a result, a
railroad consumes much less energy, and therefore less fuel than a
truck; while the train consumes on average 0.00102 gallons per ton,
a truck requires 0.0048, i.e., 4.7 times more. On the subject of load
capacity, each train car can be loaded with 60 t, versus 35 on an
articulated lorry type CS.
But the conguration of the Colombian landscape-type mountainous country in the main cities, introduces a disadvantage for
the rail, and perhaps an explanation to why the migration from
one mode to another. The rail does not allow slopes of more than
2.5%, the curves radius may not be pronounced, and the distance
between the rails on which the train operates, must be equal to
avoid derailments; in contrast, the trucks are more exible in this
aspect in circumventing holes, taking alternative ways, in summary provide more exibility in the operation (A. de G. Ingeniera,
2008).

Both faced in a at terrain and rectum.

80

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

Fig. 2. Transported tonnes by road and rail (excluding coal). (Railways right axis.)
Source: (Ministerio de Transporte, 2010).

Fig. 3. Public transportation investment on roads and rails.

Source: (Ministerio de Transporte, 2010).

However, even with the limitations and strengths of each of

the modes of transport, the statistics of cargo moved reect that
the advantages of rail is not being employed. According to records
the use of the rail is very small, almost zero in the last decade
in comparison with roads as can be seen in Fig. 2. In 2009
the railway transported 300 thousand tons, while road 177.05
million t, a proportion of 0.1% versus 81.6% with respect to the
national total.3
But the foregoing is not another thing that the consequence of
government investment policies, which in the past 20 years
invested in roads much of the resources allocated to transport
infrastructure, as can be seen in Fig. 3.
In the area of rates, the prices of the load per ton are dependant
on the route. By way of example, Ferrocarril del Oeste S.A., in its path
Buenaventura to Cali, a ton in rail has a price of USD$20 while a
cargo truck USD$24. In terms of price there is an advantage of 16%,
although in terms of time, it takes 10 h to perform the journey,
unlike the 5 h of the truck.

Air modes or waterborne were not included into this analysis.

3. Methodology used
The concepts of productivity and efciency are commonly
associated as synonyms, when in fact they differ. Determine the
productivity of a rm includes the aggregation of products and
inputs, and perform a ratio between both. In the meantime the
efciency assess whether the level of inputs and production of the
rm is associated with its limit of potential output.
Efciency can be approached from different points of view, either
technical or cost. Technical efciency is obtained by using physical
data of the inputs and technical relationships, allocative efciency
for its part, is obtained from prices and costs, both allow you to
obtain the conditions of economic efciency of a company.
There are different methodologies that allow you to obtain the
degree of efciency of a rm, some of them are:






Ordinary least squares method applied to production models

Indices of total factor productivity
Analysis of Data Envelopment Analysis DEA
Stochastic frontier.
The rst two methodologies are mostly used in aggregated
time series and provide indicators of technical change and

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

81

productivity rates. The DEA method and stochastic frontier on the

other hand, are applied to cross sectional data and reect the
relative level of efciency among multiple rms.
The four methods are also characterized by their use of different
methodologies for estimating, while the rst and the last used
parametric methods for the econometric estimation, the DEA and
the rates of total factor productivity use non-parametric methods for
linear programming (Coelli et al., 1998).
3.1. Data Envelopment Analysis
Implicitly uses methodologies such as least squares and total
factor productivity, assume that all the rms are efcient, that are
located on the edge of production possibilities.4 By means of the
methodology DEA it is possible to estimate borderlines and
determine efciency degree of each signature, in contrast with
the borderline built with the available data.
In the previous paragraph it was noticed that there are two
components of economic efciency, technical efciency, which
reects rms capability to obtain the maximum given production
inputs, and allocative efciency, which expresses the skill of rms
to use inputs in optimal proportions, given prices and production
technology.
A rm that manufactures a single product on the basis of two
inputs is located at point P. The Firm is inefcient, since it is
located far from the black isoquant from Fig. 4. This inefciency is
represented by the distance between points Q and P. Magnitude
that is nothing but what should be done to reduce inputs without
reducing level of production.
Technical efciency is usually expressed in percentage terms
as QP=0P.
That would be the percentage in which all the inputs should be
reduced to achieve the technical efciency.
ET

0Q
0P

Now, economic efciency is composed additionally of the

allocative efciency, which can be found if one takes the isocost
blue line, and the rm is located in the point Q0 . There, the
economic unit would be technically efcient, since it is located,
like the point Q on the isoquant, but it would be locative efcient
since rms takes into account the costs of inputs, standing at the
tangent point between isocost and isoquant.
EL

0R
0Q

0Q 0R
0R

n
0P 0Q
0P

In Fig. 4 the isoquant used is assumed known, in practice this

is difcult, and therefore an entire efciency analysis cannot be
performed. Then we must estimate the isoquant through the
available data, either using methodologies of parametric type,
estimating the parameters of a function Constant Elasticity of
Substitution CES or a CobbDouglas, or using non-parametric
methodologies, estimating a linear function convex to collect all
the data as can be seen in Fig. 5.
This methodology answers the question how much inputs
should be reduced proportionately so as not to alter the production, in other words seeking to minimize costs. This approach is
known as the approach of input, but it has its dual problem, and is
the maximization of the production and with it the income.
4

Fig. 5.

Then we have that economic efciency is dened as the merge

between Eq. (1) and (2), as seen in Eq. (3).
EE ET nEL

Fig. 4.

So-called DMU in the literature (Decision Making Units).

Unlike in the case focused on inputs, explained by means of

Fig. 5, we now have a signature with two products instead of one,
and only one input instead of two. You can see in Fig. 6 how a rm
that is located at point A is inefcient. The magnitude of this
inefciency is given by the distance between points A and B, called
technical efciency, where the signature is located on the borderline of possibilities of production, and the distance between points
B and C, known as the allocative efciency, distances that coupled
with provided economic efciency and that would lead to the
signature to be placed in the point B0 .
There are then two approaches to determine the degree of
efciency, the rst from the point of view of inputs, where costs
are minimized, and another point of view where it ends up by
maximizing income. It is not then an approach where prots are
maximized, a key aspect that is of interest to rms, but it will not
be addressed in this research (Coelli et al., 1998).
The DEA methodology has certain advantages and disadvantages when used to estimate the degree of technical and allocative
efciency. The available data for this study, in authors judgment,
maximizes their benets and minimizes their disadvantages that
are listed in Table 1.

82

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

3.2. DEA model with constant returns to scale

constraints to the model.

There are K inputs and M outputs for the ith signature, these
inputs and outputs are represented in the column vectors xi and
yi, respectively. The matrix of inputs X KN and the array of
products Y MN that represent the data for all the signatures.
The problem to be resolved would be:

miny, l y,
s:a: yi Y l Z0
yxi X l Z 0
N10 l 1
lZ0

miny, l y,
s:a: yi Y l Z0

yxi X l Z0
lZ0
Being y a scalar that represents the efciency degree of the ith
rm. y r1, where y 1 indicates that the rm reaches the
technical efciency and is on the isoquant in the approach of
the input, or on the border of potential production in the product
approach.

where N1 is a vector of ones. This restriction allows the size of the

rms to be further restrictions within the model; small rms are
compared with similar size rms.
In general, for different types of returns to scale, in Fig. 7 you
can appreciate the solutions for the case of an input (x) and a
production (q). The dotted line represents the constant returns of
scale CRS while the continuous line of variable returns of scale
VRS. When you assume CRS, the distance from all points to the
dotted line, allows you to calculate the degree of efciency. In the
case of VRS distance is from each point, the line formed by those
Decision Making Units DMUs to be seen as more efcient.

3.3. DEA model with variable returns of scale

4. Studies in transport efciency by applying DEA
Assume consistent returns of scale is to consider that all the
rms operate in perfect competition, without restrictions on
credit among other assumptions, that when those ease, it takes
the model to introduce a multitude of variables that would
prevent operation in an optimal way.
The DEA model of returns with variables of scale is similar to
the DEA with consistent returns, the difference lies in adding

The Data Envelopment Analysis as non-parametric for the

measurement of the relative efciency of organizational units,
in situations where there are multiple inputs and outputs, was
developed by A. Charnes, W. W. Cooper and E. Rhodes in the 1970s.
Their use in different topics is broad with more than 5000
works, compiled in www.deabib.org. With regard to transportation,

Fig. 6.

Fig. 7.

Table 1
Advantages and disadvantages DEA methodology.
Source: (Perdomo & Mendieta, 2007), (Coelliet al., 2003).
Advantages

 Does not involve a functional form for the border or a particular







distribution for inefciency parameter

Flexibility in the use of information, you can use variables discrete and
continuous whatever the unit of measure.
Being a non-parametric analysis, is less susceptible to errors in specifying
Easy-to-understand
Evaluates the presence of economies of scale.
For each company, identies a set of comparators (efcient rms with
combinations of inputs and similar products)

Disadvantages








Is sensitive to atypical data in the sample.

Requires complete information
Obtains localized results of efciency, without being able to analyze their accuracy
When comparing units is not possible to know if the difference is solely due to errors
in the data, creating doubts about the reliability.
The traditional tests or assumptions are not feasible.
Requires large samples to obtain robust estimates.

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

associated with studies of logistics, selection of industrial equipment, utilities regulation, and cost-effectiveness studies, there are
more than 300 works (Gattou et al., 2010).
Nowadays also efciency measurement of transportation using
DEA has gained popularity among academics; some examples of
applications are British bus industry (Cowie and Asenova, 1999),
Spanish airports (Martin and Roman, 2001), European railways
(Cantos et al., 1999; Hilmola, 2007) and North American container
ports (Turner et al., 2003; Cullinane et al., 2005). Recently, Savolainen
(2007) analyzed the European railways and airlines transportation
system using DEA, but each transportation mode separately.
Castello Taliani and Giralt Escobar (2008) showed a study
using DEA with the objective of describing the characteristics of a
group of companies for the delivery of goods by road, in the
Spanish provinces of Madrid and Guadalajara, in order to classify
them as a function of its technical efciency and the constraints of
this efciency between the years 20042006. As output variables
they used the sales gures, as input costs in materials, personnel,
depreciation and other expenses. These data Taliani and Giralt
applied DEA assuming variable returns of scale in such a way to
reduce the effect on the selection of companies that determine
the borderline.
At the national level, the Ministry of Mines and Energy (2008)
conducted a thorough analysis, oriented toward the identication
of strengths and weaknesses of the national transportation
system in each of the mining districts or with mineral potential
in Colombia. In a rst part, they made an inventory of the infrastructure of the country, state of operation, to then build indicators of
efciency.5 For each modes of land transport: inland waterway, sea,
rail and road, prepared a count of the regulations associated with it,
i.e., what are the policies of the State, as reected in the National
Development Plan and in the documents of the Council CONPES
(CONPES, 2007; Ministerio de Minas y Energa, 2008).
The Ministry of Transport (2008) in the report Diagnosis of
the transport sector performed a similar and complementary
analysis, with the one developed by Ministry of Mines and Energy.
The rst part takes over the macroeconomic indicators of the
transport sector, highlighting its importance as a fundamental
determinant in foreign trade and as a recipient of large investments6 by the State. Then it describes the situation of various
modes of transport: road, rail, river, sea, air and urban massive.
Transaction costs, regulatory systems, representative companies,
and tonnage statistics were considered. Just some of the elements
that allow for a clear and precise vision of what is the current
status of the system (Ministerio de Transporte, 2008).
In the Strategic Transport Plan, using available data, and on
the basis of a similar analysis used in 2008, researchers from the
Ministry of Transport used STAN software to analyze the feasibility and viability of various transportation infrastructure projects, in different modes. With this software, generating a net
present value NPV, simulating the generation of employment, the
promotion of trade (increase in ow), reducing poverty indicators,
among other factors and variables, both endogenous and exogenous, which would affect the expected result of the project, it was
determined given scarce resources and limited investment, which
were the priority projects that were needed to be built (Ministerio
de Transporte, 2002).
According to the literature researched, a study has not been
carried out where the DEA methodology for gauging the degree of
efciency of the cargo transportation systems in Colombia, it has
only been addressed by Mendieta and Perdomo in a study in
5
Indicators that are due to observation of the level of resources planned to
invest versus actual resources invested in the culmination of infrastructure
projects.
6
Planned but not implemented.

83

Table 2
Inputs and outputs covered by the study.
Inputs

Outputs

Fuel
Lubricants
Filters
Grease
Repairing and maintenance
Salaries

Load capacity

Bogota, to passengers transport companies, where the parametric

methodology is used and by estimating the production boundary
stochastic screening, as well as the methodology non-parametric
analysis of data envelopment to determine the degree of efciency, with similar results with both methodologies, but observe
the advantages and disadvantages explained in Table 1 (Mendieta
and Perdomo, 2008).
As an annotation, it is known that the new government of
Colombia at the head of the Ministry of Transport and the
National Planning Department are working on the development
of a transport master plan, where the institutional environment of
the sector would be adjusted, as an incentive to private pension
funds, foreign investors, as well as the government itself, with the
proceeds from the sale of part of its shareholding in the oil
company ECOPETROL, these resources are to be invested in the
transportation infrastructure, including rails.
Data collection process the microeconomic data from this
study were obtained by means of personal survey of rails in the
department of Valle del Cauca, interviewing the nance director,
responsible for control and trafc and sales staff. For the rail data
in the Guajira, the consulting rm A de G ingeniera.7 Trucking
data was obtained via secondary information from the Ministry of
Transportt, National Planning Department DNP, National Bureau
of Statistics (DANE) and the Colombian Federation of conveyors of
freight by road COLFECAR. This information describes the main
components of the variable cost, for the different types of trucks
and locomotives, 2010 data (Ministerio de Transporte, 2010a,b).8
The input variables and product referred to in this study were
(Table 2)9 :
All inputs are found in terms of cost per kilometer traveled.
The output, tons capacity to carry. The restriction on the time to
gather all data did not allow an analysis of technical efciency,
since only data was obtained for the railways; it was not possible
to make a comparison between systems. The weights of inputs
and their total share of the cost of operating trucks can be seen in
Fig. 8. The inputs referred to in this study are variable, xed costs
such as taxes or insurance, were excluded from this research. C2,
C3 and CS are truck categories by axles, check Table 3.
The railways were chosen taking into account that in Colombia
there are currently three lines operating, two of them as concessions
in, the Pacic and Atlantic coasts, and the private line of Cerrejon.
The typology of trucks, were chosen according to the percentage

7
I am grateful for the valuable collaboration of Carolina Cheng Perez, Ferrocarril
del Oeste, as well as Gustavo Arias de Greiff, A de G Ingeniera (Greiff, 2010).
8
The trucks were divided according to number of axles and load capacity. The
railways were discriminated according to the model of the locomotive and
capacity of drag. Both systems discriminate according to the topography of the
terrain.
9
DEA is a methodology which attempts to measure the output (input) of a
group of homogeneous products (factors of production). Both trucks and trains
share the same output (tons per Km) and the same inputs (fuel, salaries, lters),
but the capital factors differ. While trains have a lifespan of 30 years, trucks are
depreciated over 5 or 10 years. Thats why capital factors were not included into
the analysis.

84

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

distribution of the typology of vehicles (not stock) on the roads of

the country by year 2010, presented in Table 4, and the availability
of information found for the inputs across all systems.

The fuel consumption varies signicantly according to the slopes

of eld facing either the truck or the locomotive. In addition, as
mentioned in the introduction, the topography of the country is
not uniform. The countrys roads are distributed in background
(8040 km), wave (4879 km), and mountainous (3852 km) elds, it is
therefore precedent discriminate performance by slope.
Once the gures were compiled, we proceeded to construct the
matrix data, which was in a multidimensional fashion to onedimensional format (cost/km), then introduced them into the
software with results presented below.
5. Statistical analysis, correlation and empirical results under
DEA and interpretation
For the analysis of data the DEAP software 2.1 and the web
page www.deaos.com were employed. In previous section the CRS
and VRS models were presented, whose results applied to the
data obtained are presented below, performing various simulations or case studies, including all DMUs, then excluding the
Cerrejon outlier, and then a sectorial analysis (Coelli, 1996).
Table 4
Percentage distribution trucks thousands of vehicles km/Day.
Source: (Ministerio de Transporte, 2010).

Fig. 8. Costs weights to carriers by vehicle type.

Source: (DANE, 2010).

C2 (%)

C3-C4 (%)

CS (C5 and C6) (%)

Others (%)

50.85

11.01

9.18

13

16.45

Table 3
Decision making units.

Ferrocarril del Oeste

Ferrocarril (Costa Rica)c

Ferrocarril Cerrejon
Autotransporte
(Ministerio de Transporte)f

DMU

Description

U12up
U12down
U10up
U10down
Apolo
Conve
B36cerre
C2

Diesel-electric locomotive General Electrica, reference U12, working on steep terrain.

Diesel-electric locomotive General Electric, reference U12, working in hilly terrain or at
Diesel-electric locomotive General Electricb, reference U10, working on steep terrain.
Diesel-electric locomotive General electric, reference U10, working in hilly terrain orat
Apollo ferrobus tractive unit type (UT-300)
Tractive unit diesel series 80d
General electric diesel locomotive, B36-7 referencee
C2up

C3

C2wavy
C2hilly
C3up

CS

C3wavy
C3hilly
CS(up)

2-axle truck, working on at, rolling or mountainous

3-axle truck, working on at, rolling or mountainous

Articulated truck, working on at, rolling or mountainous

Colfecar (2010)
a

CS(wavy)
CS(hilly)
CS

Articulated truck, undifferentiated average data according to terrain type

For more info visit: http://www.thedieselshop.us/Data%20U12C.HTML.

For more info visit: http://www.thedieselshop.us/Data%20GE%20U10B.HTML.
Data found were included Costa Rica Railway, looking at strengthening the data, and given the similarity in the conguration of land and machinery available,
as a substitutein the absence of data from the Colombian Northern Railroad (Autoridad Reguladora de Servicios Publicos Costa Rica, 2008).
d
(Modelo de regulacion economica del servicio publico de transporte remunerado de personas, modalidad ferrocarril. y ajuste tarifario de ocio para el servicio de
transporte de pasajeros modalidad ferrocarril urbano San Pedro Pavas, 2008).
e
For more info visit: http://www.thedieselshop.us/DataB36.HTML.
f
Truck tipology are regulated by the Ministry of Transport via resolution 4100 (Ministerio de Transporte, 2004).
b
c

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

This investigation was a study of analysis of technical efciency of the railway and the trucks, not conducted due to the
reasons outlined on page ve. The rail is more efcient technically
than the trucks by denition. Following are the results of the
degree of allocative efciency found, with one being the degree of
maximum efciency (Tables 5 and 6).
The choice between CRS or VRS is an entire area of research.
The use of VRS should be given only when you have shown that
there is such a behavior in the industry, since when it is adopted,
implies that will always be an enveloping closer to the points that
an CRS approximation. When using VRS, very small units or very
large tend to be overvalued, as is the case of the compiled data,
where the rail by their economies of scale, it has the capacity to
carry many more tons that trucks (Dyson et al., 2001).
Using the results obtained for the case of consistent returns of
scale, with all available units, it was found that the unit of reference
is the Cerrejon railway. There is a big difference in levels of efciency

85

for the railways with respect to the trucks that does not exceed 10%,
similar to the locomotive U12 in steep terrain. Excluding the outlier
of Cerrejon, the results are not altered to a large extent, now as units
of reference are the railways of Costa Rica, and the units U12 and
U10 on at terrain, the trucks on average reach an efciency of 22%
while the rail of 74%.
If the results are analyzed by sector, the trucks are all very
close to the efciency, taking as reference points to articulated
lorries and three axles on at terrain. For their part, the locomotives have as a reference point Cerrejon, their results are not as
homogeneous as they are in trucks.
In DEA analysis, inputs should not be correlated with one
another, since otherwise, they must be removed from the analysis. Although there is strong correlation between fuel and wages,
and fat, and lters, fuel and lubricants, as can be seen in Table 1,
what the methodology requires is to abolish those who have
correlation one by one, avoid building a singular matrix (Table 7).

Table 5
Degree of efciency under CRS.

Ferrocarril del Oeste

Ferrocarril (Costa Rica)

Ferrocarril Cerrejon
Mean values (Railroad)
Autotransporte (Ministerio de Transporte)d

Colfecar (2010)
Mean values (Autotransport)

DMU

ALL

Without Cerrejona

U12up
U12down
U10up
U10down
Apolo
Conve
B36cerre

0.099
0.474
0.135
0.525
0.756
0.501
1
0.5651
0.069
0.069
0.069
0.098
0.073
0.073
0.079
0.079
0.079
0.075
0.075

0.209
1
0.257
1
1
1

C2up
C2wavy
C2hilly
C3up
C3wavy
C3hilly
CS(up)
CS(wavy)
CS(hilly)
CS

0.744
0.153
0.15
0.15
0.216
0.216
0.216
0.261
0.261
0.261
0.172
0.2056

Trucks onlyb

NA
0.912
0.912
0.912
1
0.979
0.973
1
1
1
1
0.9688

Rail onlyc
0.099
0.474
0.083
0.324
0.756
0.501
1
0.4624

NA

One of the disadvantages of this method is the sensitivity by outliers, and the train of Cerrejon is one of those cases. Simulation seeks to isolate this effect.
In this simulation does not take into account the inputs of fat or lters, to meet thetheoretical standards DMU at least as many as twice the number of input and
outputs.You can understand this as the degrees of freedom model, drawing a parallel to suchparametric models.
c
Were exempted from the simulation inputs: grease and lters and maintenance categories.
d
The types of trucks is regulated by (Resolucion por medio de la cual se adoptan los lmites de pesos y dimensiones en los vehculos de transporte terrestre automotor
de carga por carretera, 2004).
b

Table 6
Degree of efciency under VRS.

Ferrocarril del Oeste

Ferrocarril (Costa Rica)

Ferrocarril Cerrejon
Mean values
Autotransporte (Ministerio de Transporte)

Colfecar (2010)
Mean values

DMU

ALL

Without Cerrejon

U12up
U12down
U10up
U10down
Apolo
Conve
B36cerre
0.99
C2up
C2wavy
C2hilly
C3up
C3wavy
C3hilly
CS(up)
CS(wavy)
CS(hilly)
CS
0.962

0.9988
0.999
1
1
1
1
1
0.99
1
1
1
1
1
1
1
1
1
0.6214
0.962

0.999
1
1
1
1
1
NA
1
1
1
1
1
1
1
1
1
0.626
1

Trucks only

Rail only
0.999
0.999
1
1
1
1
1

0.99
1
1
1
1
1
1
1
1
1
1
NA

86

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

Table 7
Partial correlation between inputs.
Source: Author calcs.
Correlation

Fuel

Filters

Grease

Lubricants

Repairingmaintenance

Salaries

Fuel
Filters
Grease
Lubricants
RepairingMaintenance
Salaries

1.000000
0.183263
 0.234633
0.786884
 0.562683
0.917879

1.000000
0.808860
 0.006415
 0.277969
0.066329

1.000000
 0.212193
 0.151167
 0.215404

1.000000
 0.466761
0.907468

1.000000
 0.575631

1.000000

The DEA analysis dispersed for each of the simulations, both

under CRS as VRS, which were the DMU reference for each case,
i.e., the lambdas referred to in the approach of Section 3.2 and 3.3
above. Annex 2 presents the results of the simulation for the CRS
case, when considering all DMUs with the exception of Cerrejon.
All the trucking units have as a reference point U12 locomotive on
at terrain, as well as unity apollo. The railway units apollo,
conventional, U10 and U12 on at terrain have no reference point
other than themselves. Units U12 and U10 in steep terrain are as a
reference point to themselves but in conditions of at terrain.
Also, each of the simulations sheds which are the slacks and
improvements for each DMU. These results would be subject to
interpretation if it were a study where the DMU units were more
homogeneous and had a larger sample size, so they are not the
subject of analysis in this study.

6. Conclusions and policy suggestions

Colombia is currently undergoing a backlog of transport
infrastructure, in part because of the difculty of building and
operating under the conditions of its topography, and the design
of development policies and investment with regional vision, in
detriment of a national vision. As a result, the export costs are
very high if compared with countries in the region and the main
trade partners with which Colombia has or is on its way to
establish trade treaties. The strategy pursued by the government
has been granting large transportation infrastructure projects, of
which the great majority to trucking, leaving aside an alternative
such as rail, which offers several advantages with respect to the
trucks, that is not being exploited.
In Section 1, it was found that more than 80% of loading ow in
the country moves via roads, which today more than half are
found in regular, bad or very bad conditions. The State is taking
the lead, thinking up large projects of double lanes to meet the
need of communication and reduce costs and times. The railway,
for its part, is currently in a state of neglect, since two-thirds are
inactive, and it is a narrow-gage line, that is to say an obsolete
technology, which makes getting machinery from outside the
country more.
Taking into account the statistics of investment in transportation infrastructure over the past twenty years, the majority of
resources are invested in roads, so it is no wonder of the reasons
for the low amount of tons that one means carries with regard to
the other and the differences in trends looking to the future. This
indicates that they are not taking advantage of the strengths
offered by the railway, for example on parallel routes to the
Magdalena river, which could reduce the costs of imports and
exports to the capital, Bogota.
In section three it was veried that in Colombia a study has
not been carried out where the methodology DEA for gauging the
degree of efciency of the transport systems of Colombia. Studies
were carried out for diagnosis of the current conditions, and a
methodology for calculating net present value to analyze the

feasibility and practicability of the different projects. The magnitude of the sunk costs of the rail road and rolling stock equipment
made it seem like an unattractive alternative in comparison with
the other road projects, without taking into account strategic and
technical elements or other things like efciency in the operation.
As a result of the implementation of the methodology presented in section two, section ve found that the rail is much
more efcient than the automotive system. That when a sectorial
analysis which only takes into account trucking, it notes that all
are efcient, which could indicate that for the government in
determining where to invest its resources, if it ignores the rail, it
will indicate in the data that it is efcient. While the topography
of the pathways is an important element, the fact of including
them in the analysis does not yield results that indicate that the
rail, which faces the same challenges as the trucks of steep climbs
to overcome the western mountain range between the port of
Buenaventura and Cali, is less efcient than cargo trucks, a
situation reected in the level of rates between one and another.
As policy recommendations on the transportation of goods, the
government should consider the investment in the railways that,
despite its technological limitations and budget, offer lower prices
and more efciency. Aimed as a multimodal system that serves
the needs of the country to increase its competitiveness. Not
thinking of replacing a system, but in the contrary, to establish
conditions for cargo transport in Colombia to be competitive. The
choice of investment projects in infrastructure should not be
guided only by returns of investment in terms of a classical
approach, it is necessary to take into account the technical and
strategic value that investment in railways would bring.

Appendix A. Supporting information

Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.enpol.2012.03.029.

References
A. de G. Ingeniera, 2008. El ferrocarril: ventajas, desventajas, otras consideraciones y su importancia para el pas. Paipa.
Autoridad Reguladora de Servicios Publicos Costa Rica, 2008. Modelo de
regulacion economica del servicio publico de transporte remunerado de
personas, modalidad ferrocarril. y ajuste tarifario de ocio para el servicio de
transporte de pasajeros modalidad ferrocarril urbano San Pedro Pavas,
RESOLUCION RRG-8873-2008.
Bamrud, J. (2 de Febrero de 2010). Colombia: Positive Business Outlook. Obtenido
de http://www.latinbusinesschronicle.com/app/article.aspx?id 3962.
Banco Mundial, 2006. Infraestructura logstica y de calidad para la competitividad
de Colombia. Finance, Private Sector and Infrastructure Management Unit,
Washington.
Cantos, Pedro, Pastor, Jose M., Serrano, Lorenzo, 1999. Productivity, efciency and
tehnical change in European railways: a non-parametric approach. Transportation 26 (4), 231248.
Castello Taliani, E., Giralt Escobar, S., 2008. Analisis de la eciencia en costes de las
empresas de transporte de mercanca por carretera: una aproximacion
emprica del DEA. Revista iberoamericana de contabilidad de gestion, 93120.

C.A. Vasco Correa / Energy Policy 46 (2012) 7887

Coelli, T., 1996. A Guide to DEAP v. 2.1: A Data Envelopment Analysis (Computer)
Program. University of New England, Armidale.
Coelli, T., Estache, A., Perelman, S., Trujillo, L., 2003. Una introduccion a las
medidas de eciencia para reguladores de servicios publicos y de transporte.
Alfaomega, Mexico D.F.
Coelli, T., Rao, P., ODonnell, C., Battese, G., 1998. An Introduction to Efciency and
Productivity Analysis. Kluwer Academic, New York.
Colfecar, 2010. lndice de Precios al Transporte de Carga. Transporte y Coyuntura, 32.
CONPES, 2007. Poltica nacional de transporte publico automotor de carga.
Ministerio de Transporte Mincomercio DNP, Bogota.
Cowie, Jonathan, Asenova, Darinka, 1999. Organisation form, scale effects and
efciency in the British Bus Industry. Transportation 26 (3), 231248.
Cullinane, Kevin, Ji, Ping, Wang, Teng-fei, 2005. The relationship between privatization and DEA estimates in the Container Port Industry. Journal of
Economics and Business 57, 433462.
DANE, 2010. Ponderaciones clases de costos e ndices total nacional. Indice de
Costos del Transporte de Carga por Carretera ICTC. Departamento Nacional de
Estadstica, Bogota.
Dyson, R.G., Allen, R., Camanho, A.S., Podinovski, V.V., Sarrico, C.S., Shale, E.A.,
2001. Pitfalls and protocols in DEA. European Journal of Operational Research,
245259.
Gattou, S., Becker, D., Chandel, J., Sander, M., 2010. deabib.org a bibliographic
database about data envelopment analysis. Version 0.70.
Greiff, G. A., (20 de Marzo de 2010). Ferrocarril Asesora. (C. A. Correa,
Entrevistador).
Hilmola, Olli-Pekka, 2007. European railway freight transportation and adaptation
to demand decline efciency and partial productivity analysis from period
of 19802003. International Journal of Productivity and Performance Management 56 (3), 205225.
Lopez Castro, M. A., Hernandez Garca, S., Torres Vargas, G., 2004. Evaluacion
Economica de las Actuales Condiciones de Competencia y Complementariedad
entre el Ferrocarril y el Autotransporte. Recuperado el Febrero de 2010, de
http://www.imt.mx/archivos/Publicaciones/PublicacionTecnica/pt261.pdf.

87

Martin, Juan Carlos, Roman, Concepcion, 2001. An application of DEA to measure

the efciency of Spanish airports prior to privatisation. Journal of Air Transport
Management 7, 149157.
Mendieta, J.C., Perdomo, J.A., 2008. Fundamentos de Economa del Transporte:
teora, metodologa y analisis de poltica. CEDE, Bogota (Centro de Estudios
para el Desarrollo Economico).
Ministerio de Minas y Energa, 2008. Analisis de la Infraestructura de Transporte y
Energetica de los Distritos Mineros de Colombia. Fundacion para el Desarrollo
del Quindo.
Ministerio de Transporte, 2002. Plan Estrategico de Transporte. Ocina de Planeacion, Bogota.
Ministerio de Transporte, 2008. Diagnostico del Sector Transporte 2008: cifras
o 2007. Grupo de Planicacion Sectorial, Bogota.
correspondientes al an
Ministerio de Transporte, 2004. Resolucion por medio de la cual se adoptan los
lmites de pesos y dimensiones en los vehculos de transporte terrestre
automotor de carga por carretera, 4100.
Ministerio de Transporte. (2009). Diagnostico del transporte. Bogota: Grupo de
Planicacion Sectorial.
Ministerio de Transporte, 2010a. Polticas costos de transporte de carga por
carretera. Colfecar.
Ministerio de Transporte, 2010b. Transporte en Cifras. Grupo de Planicacion
Sectorial, Bogota.
Perdomo, J.A., Mendieta, J.C., 2007. Factores que afectan la eciencia tecnica y
asignativa en el sector cafetero colombiano: una aplicacion con analisis
envolvente de datos. Desarrollo y Sociedad, 145.
Restrepo, J. C. (8 de Agosto de 2008). Reexion sobre etes. Portafolio, pag. 1.
Savolainen, V,2007. Relative technical efciency of European transportation
systems. Master Thesis. Department of Industrial Management. Lappeenranta
University of Techonology.
Turner, Hugh, Windle, Robert, Dresner, Martin, 2003. North American container
port productivity. Transportation Research E 40, 339356.
World Bank, 2012. Doing Business Report. Washington DC.