Research in Transportation Economics 38 (2013) 139e148

Contents lists available at SciVerse ScienceDirect

Research in Transportation Economics

journal homepage: www.elsevier.com/locate/retrec

Effect of transportation infrastructure on economic growth in India: The VECM

approach
Rudra P. Pradhan a, *, Tapan P. Bagchi b
a
Vinod Gupta School of Management, Indian Institute of Technology Kharagpur, India
b
Narsee Monjee Institute of Management Studies (Mumbai), Shirpur Campus, Dhule 425405, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper examines the effect of transportation (road and rail) infrastructure on economic growth in
Available online 14 June 2012 India over the period 1970e2010. Using Vector Error Correction Model (VECM), the paper finds
bidirectional causality between road transportation and economic growth. It also finds bidirectional
Keywords: causality between road transportation and capital formation, bidirectional causality between gross
Transport infrastructure domestic capital formation and economic growth, unidirectional causality from rail transportation to
Economic growth
economic growth and unidirectional causality from rail transportation to gross capital formation. The
VECM
paper suggests that expansion of transport infrastructure (both road and rail) along with gross capital
formation will lead to substantial growth of the Indian economy. Therefore, within its stated scope, this
study suggests that a suitable transport policy should be retained to boost transportation infrastructure
and hence sustainable economic growth in India.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction
The transportation would be a key facilitator to sustainable
economic growth is rarely questioned. In India in particular,
transportation has been noted to be a critical infrastructure
required for economic growth (Raghuram & Babu, 2001). Indeed,
the benefits and importance of transportation infrastructure to
economic growth have been recognized for a long time (see Phang,
2003). A well-oiled transportation infrastructure expands the
productive capacity of a nation, both by increasing the mobilization
of available resources, and by enhancing the productivity of those
resources. The support for this assertion is straightforward and
there are many ways we can justify it. First, transportation infra-
structure can enter in the production process as direct input and in
many cases as an unpaid factor of production. Second, trans-
portation infrastructure may make other existing inputs more
productive. For instance, a well-designed road allows goods to be
transported to market in less time and hence, reducing the trans-
portation cost in the production process. Third, transport infra-
structure can act as magnet of regional economic growth by
attracting resources from other regions, which is called agglomer-
ation. In this vein one would recall that throughout the growth of
civilization, most centers of economic activities flourished along
* Corresponding author. Tel.: þ91 3222 282316; fax: þ91 3222 278027.
E-mail addresses: rudrap@vgsom.iitkgp.ernet.in (R.P. Pradhan), tapan.bagchi@
nmims.edu (T.P. Bagchi).
riverbanks and coast lines where water was the convenient prime
carrier of raw materials, goods and labor.
Transport infrastructure can also affect economic growth by
changing aggregate demand. For instance, transportation infra-
structure construction can create and increase demand for inter-
mediate inputs from other sectors and stimulate multiplier effects in
the economy. Similarly, the targets of “universal education and
health care for all” would be tough to reach without the provision of
transport infrastructure. In short, transport remains a crucial infra-
structure that boosts economic development (Esfahani & Ramirez,
2003; Phang, 2003; Sanchez-Robles, 1998; Shah, 1992; Short &
Kopp, 2005; Wang, 2002; WDR, 1994). But at the same time, we
must acknowledge the need for in depth analysis to establish any
“self-evident” causality implied here. Of necessity such analysis
would be complex and must recognize the multidimensional nature
of the links between transport, location, development and many
other new factors relevant to our understanding of these processes
as these too affect economic growth (see Fig. 1). The present
approach to study a part of this will be empirical. The demand of the
transport sector itself is likely to grow further with economic
growth, population growth, industrialization, urbanization and so
forth (Gramlich, 1994; Ramanathan & Parikh, 1999). Public trans-
portation infrastructure, in fact, can influence economic output by
crowding in or out private inputs such as labor and capital. An
increase in public transport infrastructure would attract more
private investments if there is a complementing relationship
between them; or it could reduce private investments when public

0739-8859/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.retrec.2012.05.008
140 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148
Infrastructure Investment
Investment
Multiplier
Travel Effects
Network Accessibility
Welfare Primary Benefits
Gains
Externalities
Externalities Allocative Externalities
Environmental Transport
Network
Economies
Relative Prices Economic Growth
and Land Rent
Fig. 1. Evaluation of economic growth benefits from the transport infrastructure. Source: Banister & Berechman, 2001
capital has a substitute effect on private inputs (Jiang, 2001). For
illustration one notes that transport infrastructure accounts now for
a major share of energy consumption in India, especially for petro-
leum products and the consequent development in that sector.
Historically, two possible methods may help one examine the
nexus between transport infrastructure and economic growth.
These are, respectively, cost benefit analysis and macro-
econometric modeling. In the former, the rates of return of infra-
structure projects are examined by calculating all the benefits and
costs of infrastructure projects. The latterdmacro-economic
modelsdhowever, provide three approaches. These include the
production function approach (Aschauer, 1989; Eisner, 1991;
Munnell, 1992), the cost function approach (Gillen, 1996; Khanam,
1999; Lynde & Richmond, 1992; Morrison & Schwartz, 1996; Nadiri
& Mamuneas, 1996) and the causality approach (Herranz-Loncan,
2007; Ramanathan, 2001). However, the first two approaches do
not give adequate attention to the direction of causalitydthe
beacon for effective policy formulation, while the third approach
provides high attention to the direction of causality. This paper
utilizes the third approach. The focus of this endeavor would be to
empirically investigate whether adding transport infrastructure
stimulates economic growth or economic growth itself acts as
a stimulus for any consequent growth in transport infrastructure.
As already noted, understanding such dependency between
transport infrastructure and economic growth would be vital in the
effective design and implementation of transport policies for an
Activity
Spatial
Labour Agglomeration Firm’s
Market Cost Reduction, Spatial
and Organizational
Changes
economy aspiring to grow. For instance, augmentation of transport
infrastructure will in general be expected to aid higher growth in
the economy. So transport infrastructure could be thought of as
a useful factor in even predicting the future of an economy both in
the short run and long run. In fact, a plausible “feedback hypoth-
esis” asserts that transport policies that improve efficiency in the
transport sector may not have a detrimental or draining impact on
economic growth; it may actually enhance the quality of the
transport infrastructure to help facilitate meeting the expanding
demand of the economy. Besides, transport infrastructure is also
a vital social asset; it structures space and determines mobility. It
influences trade flows as well as industrial and residence locations.
Its construction and maintenance absorb significant resources
though its highly visible and public nature sometimes raises
important policy concerns especially on its environmental effects
(more roads and parking space for cars vs. public transport) (Short
& Kopp, 2005).
However, the link between transport infrastructure’s growth
causing economic growth appears to be relatively weak in
a country like India. Still, since substantial economic expansion is
expected in India over the next two decades, building of transport
infrastructure to cause or facilitate it should at least be modest. But
there are other drivers of the transport sector. The recent high
growth in India is resulting not only from the country’s rapid
economic growth in a feed-back mode, but also due to population
expansion and changing economic lifestyles. The reverse is now
 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148 141

becoming increasingly visible; for instance, economic expansion in Table 1

India is causing notable growth of roads and railways. While the Rail and Road traffic volumes and modal shares.

former approach is called as demand driven (economic expansion Year X1 X2 X3 X4 X5 X6

causing the transport sector to grow), the latter is as supply driven 1950e51 44 67 12.09 44.80 56.09 111.8
(expanding transport facilities help the economy to grow). Such 1960e61 88 78 32.53 105.0 120.5 183.0
nexus suggests that economic growth and transport infrastructure 1970e71 127 118 82.36 263.1 209.4 381.1
1980e81 159 209 178.4 664.8 337.4 873.8
may perhaps be jointly determined to help in the formulation of
1990e91 243 296 566.7 1615 809.7 1911
appropriate policies, because economic growth is closely related to 2000e01 312 457 899.3 2128 1211 2585
transport as higher economic growth requires added transport
Note: X1: Rail freight traffic; X2: Rail passenger traffic; X3: Road freight traffic;
infrastructure. Likewise, augmentation of transport infrastructure X4: Road passenger traffic; X5: Total freight traffic; X6: Total passenger traffic.
needs support from a higher level of economic growth. So, two Source: Chaudhury, 2005.
related hypotheses now exist in the literature about the nexus
between transport infrastructure and economic growth. These are,
respectively, transport infrastructure led-growth hypothesis economic growth observed in India during 1970e2010. The rest of
(TILGH), and growth led-transport infrastructure hypothesis the paper is organized as follows: Section 2 presents a brief on the
(GLTIH). It must be noted that for policy formulation purposes, such Indian transport sector. Section 3 describes econometric method-
causality may not be known a priorida key motivation for the ology used in this study and data descriptions. Section 4 presents
present study. empirical results and the discussion thereof. The final section offers
Investigations of these two hypotheses (TILGH and GLTIH) are conclusion and policy implications.
readily available in the literature (see Kulshreshtha, Nag, &
Kulshreshtha, 2001; Ramanathan, 2001), yet their outcomes are 2. Transport infrastructure development in India
very inconsistent as well as controversial. This is mostly due to the
differences in country-specific studies, time periods, methodology, In discussing macroeconomic benefits of infrastructural invest-
and the different proxies for transport infrastructure and economic ment, Keynes pointed out that effective demand is the toll to be
growth used. The differences of these results are specified on the paid to stimulate employment and income. In order to increase the
basis of direction of causality and its long term impact on transport effective demand, the emphasis should be on infrastructure
policy. Such dependency may be represented in three possible investment. This is because such action can accelerate economic
ways: unidirectional causality between transport infrastructure developmentdby both backward and forward effects. India has
and economic growth, bidirectional causality between transport witnessed a marked increase in the transportation network in the
infrastructure and economic growth and no causality between three recent decades. However, this has been impressive only in the
transport infrastructure and economic growth. “post globalization” era (1991 onward) rather than in the “pre-
The link between infrastructure provision and economic growth liberalization” period for independent India. To be sure, in
has been the subject of extensive discussion during the last decade subsectors of transport, achievements are substantial. In short,
(see Banister & Berechman, 2001). Llanto (2007), using data on road growth of transport sector post globalization has well exceeded
quality for both the total road network and local road network, what was achieved over 40 prior years. Still, the level attained so far
finds that the quality of roads has a positive and significant effect on is quite low when compared with international norms. Rather
regional growth. This finding is in line with that of Mody and Wang importantly, in India one has noticed a gradual transition from rail-
(1997), who also concluded that transport has had a positive impact dominated transport to a road-dominated one (see Table 1).
on economic growth in China. Fernald (1999), using data from 29 Besides, the contribution of transport sector to GDP, as expected,
manufacturing industries form 1953e1989, finds causality from has been rising. It rose from 3.8% in 1980e81 to 4.6% in 1990e91
roads to productivity. Fernald’s findings endorse that the produc- and then to 5.5% in 2000e01. This is now visible in both road and
tivity decline in US manufacturing after 1973 might have been rail infrastructure sectors (see Table 2).
a result of lower public spending on road infrastructure. Canning
and Bennathan (2000), using cointegration methods for a panel 3. Data and econometric modeling
of 41 countries, find that the length of paved roads is highly
correlated with physical and human capital. But they also observed In this section we recall some crucial aspects of the effective use
that the marginal return to roads declines rapidly if the length of of the analytical tools employed in this study. To begin, the state of
roads is increased in isolation from other inputs. This implies that the transport infrastructure may be presented in many ways, such
infrastructure investments are not sufficient by themselves to yield as the availability of and quality of roads and railways (Canning,
large changes in output. This finding is in line with that of Gannon 1998; Kumar, 2001). The present paper, however, uses road and
and Zhi (1997), who also concluded that transport access is rail lengths to help capture the interface between transport infra-
complementary to other services such as health and education. structure and economic growth. This is because the primary modes
Studies by Fan, Jitsuchon, and Methakunnavut (2004) in rural India,
China and Thailand also estimate the effect of infrastructure
investments on economic growth and poverty. The results from Table 2
Contribution of the transport infrastructure to GDP in India.
these studies consistently show the importance of road invest-
ments in promoting economic growth and poverty reduction. In Year X1 X2 X3 X4 X5
India, public investment in rural roads was found to have had the 1980e81 0.28 2.42 0.79 0.31 3.8
largest positive impact on agricultural growth (Fan, Hazell, & 1985e86 0.61 2.59 0.66 0.34 4.2
1990e91 0.97 2.71 0.64 0.27 4.6
Thorat, 1999) while in China and Thailand, road investments were
1995e96 1.48 3.01 0.71 0.28 5.5
found to have contributed significantly to growth in non-farm 2000e01 1.48 3.15 0.69 0.22 5.5
sectors in particular and overall economic growth in general (Fan,
Note: X1: Rail contribution to GDP; X2: Road contribution to GDP; X3: Water
Zhang, & Zhang, 2002, 2004). contribution to GDP; X4: Air contribution to GDP; X5: Total transport infrastructure
The present study is aimed at delving a bit deeper into the contribution to GDP. Source: National Accounts Statistics of India, EPW Research
speculated nexus between the state of transport infrastructure and Foundation Mumbai.
142 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148

of transportation in India are drivable roads and serviced rail tracks. Table 4
To analyze the nexus between transport infrastructure and Unit root test results.

economic growth, the following function is used: Variables Level data Conclusion

Constant Constant with trend

EG ¼ fðTINF; GCFÞ (1)
GDP 3.61 0.20 Non-stationary
where EG is economic growth, TINF is the total transport infra- GDCF 2.07 0.39 Non-stationary
structure and GCF the gross domestic capital formation. For the ROAD 0.48 2.28 Non-stationary
RAIL 2.01 1.77 Non-stationary
present empirical analysis we use GDP as a proxy for economic ROAD & RAIL 0.34 2.85 Non-stationary
growth, road transport and railway transport as a proxy for trans- First difference
port infrastructure and gross domestic capital formation as a proxy GDP 5.09 6.69* Stationary
for infrastructure investment. The data employed for this research GDCF 5.96 6.76* Stationary
ROAD 6.47 6.38* Stationary
are annual and cover the period 1970 to 2010 obtained from World
RAIL 7.51 7.95* Stationary
Development Indicators reported by World Bank and Infrastructure ROAD & RAIL 6.99 6.91* Stationary
Statistics, Ministry of Statistics and Programme Implementation,
Note: GDP: Gross Domestic Product; GDCF: Gross Domestic Capital Formation;
Government of India, New Delhi. Note that we use the transport ROAD: Road Transportation Infrastructure; and RAIL: Railway Transportation
infrastructure in two different ways in this paperdfirst at the Infrastructure; and * indicates Statistically Significant at 1%.
individual leveldroad and rail separately, and then at group level,
which combines the effect of both road infrastructure and railway where ADF regression tests for existence of unit root of Zt, the
infrastructure. Principal component analysis was used to achieve logarithmic values of all model variables at time t. The null (and
the same (Pradhan, 2007). Both road and rail infrastructures were alternative) hypothesis that tests the existence of unit root in the
measured in kilometers, so combining these two would not be variable Zt is H0: a3 ¼ 0 against H1: a3 s 0.
technically defective even though they do not represent compa- The nature of stationarity indicates the presence of cointegra-
rable magnitudes of capacity. Furthermore, we combined these two tion and causality, based on VAR and VECM representations. Engel
by the process of standardization, which is statistically feasible and and Granger (1987) showed that, if two variables are individually
traceable to earlier literature in which investigators combined rail integrated of order one and cointegrated, then a causal relationship
and road (see Chaudhury, 2005 for instance). In the analysis all data may exist between them in at least one direction. Hence, it is
were expressed logarithmically in order to include the proliferative necessary to also test for cointegration among the time series
effect of time series. Table 3 provides the summary statistics for all variables (Yoo & Ku, 2009). In fact, Granger (1986) argued that ‘A
variables used in this study. test for cointegration can thus be thought of as a pre-test to avoid
In this work Granger causality test (Granger, 1988) based on the “spurious regression” situations’ and to know the common trend of
Vector autoregression (VAR) is applied to study the nexus between the time series variables (Engle & Yoo, 1987).
the transport infrastructure and economic growth. This test, Cointegration test is meant to reveal the existence of long run
however, is very conditional on the stationarity of the time series equilibrium relationship between the variables. A lack of cointe-
variables involved. If the underlying time series are non-stationary, gration suggests that such variable have no long run equilibrium
the stability condition required of VAR is not met, implying that the relationship and in principle, they can wander arbitrarily far away
test statistic of Granger causality is invalid. If variables are non- from each other (Dickey, Jansen, & Fuller, 1991).
stationary, cointegration and Vector Error Correction Model The hypothesis that tests this is the null of non-cointegration
(VECM) are recommended to investigate the relationship between against an alternative that cointegration exists. This uses the
such variables. So the first and prime condition for using VAR Johansen (1988) maximum likelihood ratio test and looks at two
would be to test stationarity of the variables. Many macroeconomic test statistics (Johansen & Juselius, 1990), namely, the trace statis-
time series contain unit roots dominated by stochastic trends tics and the maximum eigenvalue statistics,
detectable by the method given by Nelson and Plosser (1982). A
stochastic trend is determined by testing the presence of unit roots X
n
in time series data (Dritsaki & Dritsaki-Bargiota, 2005; Elliot, ltrace ¼ T Logð1  ^
li Þ (3)
Rothenberg, & Stock, 1996). The Augmented Dickey Fuller (ADF) i ¼ rþ1

test (Dickey & Fuller, 1981) is used to test for detecting unit roots.
lmax ¼ TLogð1  ^lrþ1 Þ (4)
X
p
DZt ¼ a1 þ a2 t þ a3 Zt1 þ bi DZti þ εt (2)
i¼1 Table 5
Cointegration test results.

Table 3 Models Null Trace 5% critical Max 5% critical

Summary statistics. hypothesis statistics value Eigenvalue value
Model 1 H0: r ¼ 0 43.93* 24.28 26.95* 17.79
Statistics GDP GDCF ROAD RAIL ROAD & RAIL H0: r  1 16.98* 12.32 11.27* 11.22
Mean 6.04 5.43 6.30 4.79 1.76 H0: r  2 5.810* 4.130 5.810* 4.130
Median 6.01 5.39 6.30 4.79 1.76 Model 2 H0: r ¼ 0 40.62* 24.28 30.73* 17.79
Maximum 6.59 6.21 6.63 4.80 1.79 H0: r  1 9.89 12.32 9.59 11.22
Minimum 5.68 4.95 5.96 4.78 1.74 H0: r  2 0.31 4.130 0.31 4.130
Std. Dev. 0.27 0.35 0.18 0.01 0.01 Model 3 H0: r ¼ 0 38.04* 24.28 28.51* 17.79
Skewness 0.42 0.59 0.26 0.45 0.13 H0: r  1 9.550 12.32 7.530 11.22
Kurtosis 2.10 2.44 2.26 1.90 2.04 H0: r  2 2.020 4.130 4.130 4.130
Jarque-Bera 2.44 2.74 1.32 3.24 1.61
Note: Model 1: The cointegration between GDP, GDCF and road transportation;
Probability 0.29 0.25 0.52 0.19 0.44
Model 2: The cointegration between GDP, GDCF and rail transportation; Model 3:
Note: GDP: Gross Domestic Product; GDCF: Gross Domestic Capital Formation; The cointegration between GDP, GDCF and transportation infrastructure (both road
ROAD: Road Transportation Infrastructure; and RAIL: Railway Transportation and rail); r denotes the number of cointegrating vectors; and * indicates statistically
Infrastructure. significant at 5%.
 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148 143

Table 6 hypothesis ‘GDP does not Granger cause TINF, given GCF’ is rejec-
VAR Lag order selection. ted, and so on. If the variables are cointegrated, then VECM models
DV AIC SIC HQ LR shown in (8)e(10) would be applied.
Model 1 GDP 111.29 108.83 13.69 14.04
GDC 73.29 70.02 13.20 13.81 X
p X
p

ROA 89.89 87.43 12.57 14.44 DGDPt ¼ f1 þ a11;l DGDPtl þ b12;l DTINFtl
Model 2 GDP 109.91 107.45 13.69 14.04 l¼1 l¼1
(8)
GDC 75.15 71.88 13.20 13.81 X
p
RAI 205.09 202.64 12.57 14.44 þ f13;l DGCFtl þ d1 ECt1 þ zt
Model 3 GDP 110.34 107.88 13.69 14.04 l¼1
GDC 75.25 71.98 13.20 13.81
R&R 187.37 184.91 12.57 14.44
X
p X
p
Note: GDP: Gross Domestic Product; GDC: Gross Domestic Capital Formation; ROA: DTRNFt ¼ f2 þ a21;l DTINFtl þ b22;l DGDPtl
Road Transportation Infrastructure; RAI: Railway Transportation Infrastructure; l¼1 l¼1
R&R: Road and Rail Transportation Infrastructure; AIC: Akaike Information Crite- (9)
X
p
rion; SBC; Schwarz Information Criterion; HQ: Hann-Quinn Information Criterion
and LR: Sequential Modified LR test statistic.
þ f23;l DGCFtl þ d2 ECt1 þ xt
l¼1

where ^lrþ1 , . ^ln are (n  r) smallest estimated eigenvalues. The X

p X
p
null hypothesis of r cointegrating vectors is tested here against the DGCFt ¼ f2 þ a31;l DGCFtl þ b32;l DGDPtl
alternative hypothesis of r þ 1 cointegrating vectors. As discussed l¼1 l¼1
(10)
above, cointegration would determine the types of model (VAR/ X
p

VECM) that we may then use to tell the direction of causality þ f33;l DTINFtl þ d3 ECt1 þ xt
between transport infrastructure and economic growth. For l¼1

instance, if the variables are stationary and not cointegrated, the where, EC is error correction termdthe estimated residual from the
following VAR model must be used: cointegration regression. A significant coefficient implies that past
equilibrium errors play a role in determining the current outcomes.
X
p X
p
The short run dynamics are captured through the individual coef-
GDPt ¼ f0 þ f1i GDPtl þ f2j TINFtj
i¼1 j¼1
ficients of the difference terms. One difficulty that a researcher may
(5) face in estimating a VAR model is the appropriate specification of
X
p
þ f3k GCFtk þ z1t the model. The Akaike Information Criterion (Akaike, 1974) and
k¼1 Schwarz Bayesian Criterion (Schwartz, 1978) are used for the same,
as they often perform better than other alternatives (Maddala,
X
p X
p 1992; Mills & Prasad, 1992). We also apply certain diagnostic tests
TINFt ¼ h0 þ h1i TINFti þ h2j GDPtj on the residuals of the model. In this study we used the Lagrange
i¼1 j¼1 test (LM) for the possible existence of autocorrelation and hetero-
(6)
X
p skedasticity and the Ramsey’s Reset test for the functional form of
þ h3k GCFtk þ z2t the model (Ramsey, 1969).
k¼1
4. Results and discussion
X
p X
p
GCFt ¼ l0 þ l1i GCFtl þ l2j TINFtj The present empirical analysis starts with checking stationarity
i¼1 j¼1
(7) of the time series variables as that is the prime requirement for
X
p
cointegration and causality testddetecting causality to facilitate
þ l3k GDPtk þ z3t policy formulation. The ADF unit root test has been applied to
k¼1
examine the stationarity of the time series data. The estimated
where GDP is used as a proxy for economic growth, TINF is used as results of unit root test are reported in Table 4. The results indicate
magnitude of transport infrastructure and GCF is used to represent that no time series variables appear to be stationary in variable
gross domestic capital formation. The null hypothesis ‘TINF does levels, since computed test statistics could not reject the null
not Granger cause GDP, given GCF’, is tested here via a standard F- hypothesis of non-stationarity. This means that means the variables
test, being rejected if the 42j in Equation (5) is statistically signifi- are non-stationary in their level data and suggests that stationarity
cant. If h2j in Equation (6) is statistically different from zero, the null be checked at a higher order of differencing. In the present case it is

Table 7
Results of causality tests.

Models Dependent variables DGDP DGDC DROA DRAI DR&R ECM Inferences
Model 1 DGDP e 5.01* 2.84* e e 2.05* GDC  GDP; ROA  GDP
DGDC 4.38* e 2.02** e e 3.84* GDP  GDC; ROA  GDC
DROA 2.85* 2.97* e e e 2.87* GDP  ROA; GDC  ROA
Model 2 DGDP e 3.91* e 2.24** e 2.84* GDC  GDP; RAI  GDP
DGDC 4.11* e e 2.89* e 2.84* GDP  GDC; RAI  GDC
DRAI 0.36 0.55 e e e 0.964 GDP s> RAI; GDC s> RAI
Model 3 DGDP e 4.13* e e 2.44** 3.48* GDC  GDP; R&R  GDP
DGDC 4.22* e e e 2.87* 2.88* GDP  GDC; R&R  GDC
DR&R 1.05 1.05 e e e 0.96 GDP s> R&R; GDC s> R&R

Note: GDP: Gross Domestic Product; GDC: Gross Domestic Capital Formation; ROA: Road Transportation Infrastructure; RAI: Railway Transportation Infrastructure; R&R: Road
and Rail Transportation Infrastructure; and * (**): statistically significant at 1% (5%) level.
144 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148

found that when the first differences of the variables are considered,
the null hypothesis of unit root is rejected at 1% significance level.
Hence, the differences become stationary and consequently the
related variables get characterized as integrated of order one, 1 (1).
Following that the three series are integrated of order one, the
cointegration relationship between them are established by using
Johansen’s maximum likelihood (ML) test. The results of cointe-
gration test (ltra and lmax) indicate that the time series variables
(transport infrastructure, economic growth and gross capital
formation) are cointegrated (see Table 5) and hence, long run
equilibrium relationship may exist among them. This gave us the
hope for striving toward the stated mission of this studydtesting
certain hypothesized causality. Model specification is also
checked here to validate the ML test results (see Table 6). On the
basis of cointegration results, VECM is deployed to know
the direction of causality. The estimated results of VECM
are reported in Table 7. The analysis finds the existence of bidi-
rectional causality between economic growth and gross capital
formation (GDP <¼> GCF). Bidirectional causality is also found
between road infrastructure and economic growth (ROA <¼> GDP)

Respons e of GDP to Choles ky

One S.D. Innovations
.025

.020

.015

.010

.005

.000
1 2 3 4 5 6 7 8 9 10

GDP GDCF ROA D

Res pons e of GDCF to Choles k y

One S.D. Innovations
.05

.04

.03

.02

.01

.00
1 2 3 4 5 6 7 8 9 10

GDP GDCF ROA D

Res pons e of ROAD to Choles k y

One S.D. Innovations
.03

.02

.01

.00

-.01
1 2 3 4 5 6 7 8 9 10

GDP GDCF ROA D

Fig. 2. Impulse response functions (between GDP, GCF and Road).

 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148 145

and road infrastructure and gross domestic capital formation
(ROA <¼> GCF).
A point needs to be made here regarding bidirectional causality.
If the rise in one variable causes the rise of another, and is also true
when the variables are switched, the variables would reinforce each
other and be self-reinforcing. This situation, therefore, would be
easy to handle by the policy maker (Chontanawat, 2010). However,
when this is not so, analysis would have to include other factors and
policy formulation then would not be trivial (Riman & Akpan, 2010).
In the present case, for the 1970e2010 period checked, the rein-
forcements were in the right direction and self-reinforcing.
This study also finds unidirectional causality from railway
infrastructure to economic growth (RAI  GDP) and railway infra-
structure to gross capital formation (RAI  GCF). Unidirectional
causality is also discovered to exist from total transport (both road
and railway) to economic growth (R&R  GDP) and total transport
to gross domestic capital formation (R&R  GCF). In summary, for
India, road transport infrastructure is observed to have a substan-
tial influence on both gross capital formation and economic
growth, since there is bidirectional causality among them. On the
contrary, railway transport infrastructure determines economic
growth and capital formation at the unidirectional level only.

Res pons e of GDP to Choles k y

One S .D. Innovations
.03

.02

.01

.00

-.01
1 2 3 4 5 6 7 8 9 10

GDP GDCF RA IL

Res pons e of GDCF to Choles k y

One S .D. Innovations
.06

.04

.02

.00

-.02
1 2 3 4 5 6 7 8 9 10

GDP GDCF RA IL

Res pons e of RA IL to Choles k y

One S .D. Innovations
.0012

.0008

.0004

.0000

-.0004

-.0008
1 2 3 4 5 6 7 8 9 10

GDP GDCF RA IL

Fig. 3. Impulse response functions (between GDP, GCF and Rail).

146 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148

The results were further verified through generalized impulse
response functions (GIRFs). GIRFs trace the effect of a one-time
shock to one of the innovations on current and future values of
endogenous variables. The generalized impulse responses provided
more insight into how shocks to transport infrastructure and gross
capital formation affected economic growth and vice versa. The
results of generalized impulse responses for economic growth,
transport infrastructure (road and railway) and gross capital
formation are represented in Figs. 2e4 respectively. The GIRFs
provided the support of causality status between these three time
variables in the multivariate VECM system. In all the cases, the
GIRFs were very responsive to the results of VECM. To complement
this study, we report also the diagnostic tests for serial correlation
(LM test), autoregressive conditional heteroskedasticity (ARCH
test), heteroskedasticity (White test) and stability test (Ramsey
test). These are provided in Table 8. The observations confirm the

Res ponse of GDP to Choles k y

One S.D. Innovations
.030

.025

.020

.015

.010

.005

.000
1 2 3 4 5 6 7 8 9 10

GDP GDCF CI

Res ponse of GDCF to Cholesk y

One S.D. Innovations
.05

.04

.03

.02

.01

.00
1 2 3 4 5 6 7 8 9 10

GDP GDCF CI

Response of CI to Choles ky
One S.D. Innovations
.0020

.0015

.0010

.0005

.0000

-.0005

-.0010
1 2 3 4 5 6 7 8 9 10

GDP GDCF CI

Fig. 4. Impulse response functions (between GDP, GCF and Road & Rail).
 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148
Table 8
Short run diagnostic tests.
Variables LM ARCH White Ramsey
GDP 13.03* 4.60* 2.296 0.908
GDCF 4.911* 4.56* 11.46* 24.76*
ROAD 4.717* 6.45* 12.12* 3.487*
Note: LM: Serial Correlation LM Test; ARCH: ARCH Test; Ramsey: Ramsey Test;
White: White Heteroskedasticity Test; * indicates that the test statistics is
significant.
stability of the model on the nexus or link between economic
growth, transport infrastructure and gross capital formation in the
Indian economy.
5. Conclusion and policy implications
This study has examined the presence of any nexus between
transport infrastructure and economic growth in India using the
data for the 1970e2010 period. This analysis is deemed to be vital in
the effective design and implementation of transport policies in
this rapidly growing economy. Using cointegration and Granger
causality test, this study has concluded the following:
1. Bidirectional causality exists between road transport infra-
structure and economic growth. That means that road transport
causes economic growth and vice versa. There are many reasons
for it. First, road transport is one of the basic inputs in the
production process; hence, an increase in road transport would
be expected to have a positive effect on economic growth. This is
consistent with the findings of Llanto (2007). Similarly, there are
at least two reasons why economic growth has boosted road
transport development in India. First, economic growth has
resulted in an expansion in the commercial and industrial
sectors, and in particular in the manufacturing sector, to facili-
tate which road transport would be a basic input. Second, higher
disposable income has increased demand for better road
infrastructure for household entertainment.
2. Bidirectional causality exists between road transport infra-
structure and gross domestic capital formation. This implies
that when road transport steps up, both in quantity and quality,
government can charge toll tax and that can contribute to
government’s revenue, a part of which can contribute to
increased gross domestic capital formation. On the contrary,
when there is higher gross domestic capital formation, it may
be used for expanding economic investment including that in
road transport; hence, building roads contributes to economic
growth and a part of this growth can expand road trans-
portation in India.
3. Bidirectional causality exists between economic growth and
gross domestic capital formation. This implies that when the
economy grows, people’s individual income also increases and
a part of the increased income can contribute to gross domestic
capital formation. Similarly, when gross domestic capital
formation increases, it can be used to raise socio-economic
investment and hence to contribute to economic growth.
4. A unidirectional causality exists from railway infrastructure to
both economic growth and gross domestic capital formation.
This result reveals that lowering the railway infrastructure in
the country’s transport mix would lead to a fall in economic
growth and gross domestic capital formation. So, in India,
railway indeed seems to play an important role in the transport
portfolio. This implies that boosting economic growth in India
would require an enormous railway infrastructure, even if
there are many other factors contributing to economic growth,
railway being only one of such factors.
147
To summarize, for India, transport infrastructure not only
influences economic growth but also gross capital formation. It is
therefore suggested that increasing transport facility (both road
and railway) along with gross capital formation will lead to more
pervasive economic growth in India. The achievement of higher
economic growth through transport infrastructure would be due to
its various direct and indirect benefits imparted to the economy.
But it is easy to observe that in India, the level of transport infra-
structure is not so good, both in quantity and quality, in contrast to
that in developed countries. The positive results implied in this
study would therefore be much better if substantial improvement
in transport infrastructure occurred. Thus, since transport infra-
structure is found to be a “big deal” for economic growth, a suitable
transport policy should be retained to maintain sustainable
economic growth in the country. Given the geographical spread of
population and industries in India, an appropriate modal mix of
transport is deemed to be necessary to promote economic devel-
opment (Raghuram & Babu, 2001). A piecemeal approach (with
seemingly unclear policies driving it) to such a vital issue, however,
may lead to serious consequences (e.g. wastage of resources) and
may slow down economic growth in the long run (Pradhan, 2007).
Hence, in order not to adversely affect economic growth, efforts
must be made to encourage government investment in transport
infrastructure and to overcome the constraints (e.g., land acquisi-
tion) facing the expansion of the nation’s transport infrastructure.
In short, this study, based on the country’s macroeconomic data,
suggests that the Indian government must upgrade and expand the
country’s transport infrastructure with top priority actions. Such
a policy would greatly help in sustaining the good momentum India
has built to raise its economic prosperity.
References
Akaike, H. (1974). A new look at the statistical model identification. IEEE Transaction
on Automatic Control, AC-19, 716e723.
Aschauer, D. A. (1989). Public investment and productivity growth in the group of
seven. Economic Perspectives, 13, 17e25.
Banister, D., & Berechman, Y. (2001). Transport investment and the promotion of
economic growth. Journal of Transport Geography, 9, 209e218.
Canning, D. (1998). A database of world stocks of infrastructure, 1950e95. World
Bank Economic Review, 12, 529e547.
Canning, D., & Bennathan, E. (2000). The social rate of return on infrastructure
investmentsIn Policy Research Working Paper Series 2390, . Washington, D.C.:
World Bank.
Chaudhury, P. D. (2005). Model split between rail and road modes of transport in
India. Vikalpa, 30, 17e33.
Chontanawat, J. (2010). Modelling the causal relationship between energy consumption
and economic growth in Asia. Working Paper. Bangkok, Thailand: KMU Technology.
Dickey, D. A., & Fuller, W. A. (1981). Likelihood ratio statistics for autoregressive
time series with a unit root. Econometrica, 49, 1057e1072.
Dickey, D. A., Jansen, D. W., & Fuller, W. A. (1991). A primer on cointegration with an
application to money and income. Review Federal Reserve Bank of St. Louis, 73, 58e78.
Dritsaki, C., & Dritsaki-Bargiota, M. (2005). The causal relationship between stock,
credit market and economic development: an empirical evidence of Greece.
Economic Change and Restructuring, 38, 113e127.
Eisner, R. (1991). Infrastructure and regional economic performance. New England
Economic Review. Federal Reserve Bank of Boston. pp. 47e58.
Elliot, G., Rothenberg, T. J., & Stock, J. H. (1996). Efficient tests for an autoregressive
time series with unit root. Econometrica, 64, 813e836.
Engle, R. F., & Granger, C. W. J. (1987). Cointegration and error correction: repre-
sentation, estimation and testing. Econometrica, 55, 251e276.
Engle, R. F., & Yoo, B. S. (1987). Forecasting and testing in cointegrated systems.
Journal of Econometrics, 35, 143e159.
Esfahani, H. S., & Ramirez, M. T. (2003). Institutions, infrastructure and economic
growth. Journal of Development Economics, 70, 443e477.
Fan, S., Hazell, P., & Thorat, S. (1999). Government spending, agricultural growth and
poverty: An analysis of inter-linkages in rural India. IFPRI Research Report No. 110.
Washington, D.C.: International Food Policy Research Institute.
Fan, S., Jitsuchon, S., & Methakunnavut, N. (2004). The importance of public invest-
ment for reducing rural poverty in middle-income countries: The case of Thailand.
DSGD Discussion Paper No. 7. Washington, D.C.: International Food Policy
Research Institute.
Fan, S., Zhang, L., & Zhang, X. (2002). Growth, inequality, and poverty in rural China:
The role of public investments. IFPRI Research Report No. 125. Washington, D.C.:
International Food Policy Research Institute.
148 R.P. Pradhan, T.P. Bagchi / Research in Transportation Economics 38 (2013) 139e148
Fernald, J. G. (1999). Roads to prosperity? Assessing the link between public capital
and productivity. The American Economic Review, 89, 619e638.
Gannon, C., & Zhi, L. (1997). Poverty and transport. Discussion Paper TWU-30.
Washington, D.C.: World Bank.
Gillen, D. W. (1996). Transportation infrastructure and economic development:
a review of recent literature. Logistics and Transportation Review, 32, 39e62.
Gramlich, E. M. (1994). Infrastructure investment: a review essay. Journal of
Economic Literature, 32, 1176e1196.
Granger, C. W. J. (1986). Developments in the study of cointegrated variables. Oxford
Bulletin of Economics and Statistics, 48, 213e228.
Granger, C. W. J. (1988). Some recent developments in a concept of causality. Journal
of Econometrics, 39(1e2), 199e211.
Herranz-Loncan, A. (2007). Infrastructure investment and Spanish economic
growth, 1850e1935. Explorations in Economic History, 44, 452e468.
Jiang, B. (2001). A review of studies on the relationship between transport infra-
structure investments and economic growth. Canada Transportation Act Review.
Vancouver: British Columbia.
Johansen, S. (1988). Statistical analysis for cointegration vectors. Journal of Economic
Dynamics and Control, 12, 231e254.
Johansen, S., & Juselius, K. (1990). Maximum likelihood estimation and inference on
cointegration with application to the demand for money. Oxford Bulletin of
Economics and Statistics, 52, 169e210.
Khanam, B. (1999). Contributions of highway capital to output, cost and productivity
growth: evidence from the Canadian goods producing sector. Ph. Dissertation. York
University, Canada.
Kulshreshtha, M., Nag, B., & Kulshreshtha, M. (2001). A multivariate cointegrating
vector auto regressive model of freight transport demand: evidence from Indian
railways. Transport Research Part A, 35, 29e45.
Kumar, N. (2001). Infrastructure availability, foreign direct investment Inflows and
their export-orientation: A cross country exploration. New Delhi: Research and
Information System for Developing Countries.
Llanto, G. M. (2007). In H. Hill, & A. Balisacan (Eds.), Infrastructure and regional
growth. Australia: Australian National University.
Lynde, C., & Richmond, J. (1992). The role of public capital in production. The Review
of Economics and Statistics, 74, 37e43.
Maddala, G. S. (1992). Introduction to econometrics. New Jersey: Prentice Hall.
Mills, J., & Prasad, K. (1992). A comparison of model selection criteria. Econometric
Reviews, 11, 201e233.
Mody, A., & Wang, F. (1997). Explaining industrial growth in Coastal China:
economic reforms and what else. World Bank Economic Review, 11, 293e325.
Morrison, C. J., & Schwartz, A. E. (1996). State infrastructure and productivity
performance. American Economic Review, 86, 1095e1111.
Munnell, A. H. (1992). Infrastructure investment and economic growth. Journal of
Economic Perspectives, 6, 189e198.
Nadiri, M. I., & Mamuneas, T. P. (1996). Highway capital and productivity growth.
Lansdown, VA US: Eno Transportation Foundations Inc.
Nelson, C. R., & Plosser, C. I. (1982). Trends and random walks in macroeconomic
time series: some evidence and implications. Journal of Monetary Economics, 10,
139e162.
Phang, S. (2003). Strategic development of airport and rail infrastructure: the case
of Singapore. Transport Policy, 10, 27e33.
Pradhan, R. P. (2007). Does infrastructure play a role in urbanization: evidence from
India? Journal of Economics and Business, 6, 81e92.
Raghuram, G., & Babu, R. (2001). Alternative means of financing railways. In
G. Raghuram, R. Jain, S. Sinha, P. Pangotra, & S. Morris (Eds.), Infrastructure
development and financing: Towards a publiceprivate partnership. New Delhi:
Macmillan.
Ramanathan, R. (2001). The long-run behavior of transport performance in India:
a cointegration approach. Transport Research Part A, 35, 309e320.
Ramanathan, R., & Parikh, J. K. (1999). Transport sector in India: an analysis in the
context of sustainable development. Transport Policy, 6, 35e45.
Ramsey, J. B. (1969). Test for specification errors in classical linear least squares
regression analysis. Journal of Royal Statistical Society B, 350e371.
Riman, H. B., & Akpan, E. S. (2010). Causality between poverty, health expenditure
and health status: evidence from Nigeria using VECM. European Journal of
Economics, Finance and Administrative Sciences, 27, 120e128.
Sanchez-Robles, B. (1998). Infrastructure investment and growth: some empirical
evidence. Contemporary Economic Policy, 16, 98e108.
Schwartz, R. (1978). Estimating the dimension of a model. Annals of Statistics, 6,
461e464.
Shah, A. (1992). Dynamics of public infrastructure, industrial productivity and
profitability. Review of Economics and Statistics, 74, 28e36.
Short, J., & Kopp, A. (2005). Transport Infrastructure: investment and planning
policy and research aspects. Transport Policy, 12, 360e367.
Wang, E. C. (2002). Public infrastructure and economic growth: a new approach
applied to East Asian Economies. Journal of Policy Modelling, 24, 411e435.
WDR. (1994). World Development Report, World Bank. Oxford: Oxford University
Press.
Yoo, S., & Ku, S. (2009). Causal relationship between nuclear energy consumption
and economic growth: a multi-country analysis. Energy Policy, 37, 1905e1913.