See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254120910

Road Transport Infrastructure and Economic Growth in India

Article  in  Journal of Infrastructure Development · December 2010

DOI: 10.1177/097493061100200204

CITATIONS READS

5 6,808

2 authors, including:

Shruti Tripathi
National Institute of Public Finance and Policy
7 PUBLICATIONS   36 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Indian macroeconomic structure View project

All content following this page was uploaded by Shruti Tripathi on 28 October 2015.

The user has requested enhancement of the downloaded file.

Article

Road Transport Infrastructure and Journal of Infrastructure Development

2(2) 135–151

Economic Growth in India © 2010 India Development Foundation

SAGE Publications
Los Angeles, London,
New Delhi, Singapore,
Washington DC
10.1177/097493061100200204
Shruti Tripathi http://joi.sagepub.com
Research Scholar, Indira Gandhi Institute of
Development Research, Mumbai, [email protected]

Vikash Gautam

SE
Research Scholar, Indira Gandhi Institute of
Development Research, Mumbai, [email protected]

U
L
IA
Abstract
Creation of road transport infrastructure, through its direct and indirect effects, has a bearing on sus-
C
tainability of growth and overall development of a country. It provides knowledge spillovers resulting
ER
from the whole agglomerated area via network dynamic externalities. The models based on cost or
production function that incorporate infrastructure but simply assume a positive effect are no longer
M

satisfactory to take to the data, because they ignore any feedback effect. In this article, therefore, we
M

have twin objectives: first, we examine whether road transport infrastructure has a long-run equilib-
O

rium relationship with the macroeconomic variables such as output, employment and gross private
C

capital formation or not. Second, we use vector autoregression (VAR) approach to analyse the impact
of road transport infrastructure on macroeconomic variables. The elasticities that are estimated in
R

the VAR model differ from the production function elasticities, as they incorporate feedback effect
FO

between the variables in the model, as opposed to the ‘ceteris paribus’ elasticities which are estimated
in production function studies.
T

JEL Classification: B41, C01, C22, R42

O
N

Keywords: Network dynamic externalities, cost function, production function, feedback effect, VAR,
output, employment, gross private capital formation, elasticity

1. Introduction

Creation of road transport infrastructure, through its direct and indirect effects, has a bearing on
sustainability of growth and overall development of a country. Apart from improving connectivity, the
development of roads can open up hitherto unconnected regions to trade and investment and step up
access to goods, services and employment opportunities. This is reinforced by the fact that spending on
infrastructure has large multiplier effects. In the long run this is achieved by getting higher returns from
136 Shruti Tripathi and Vikash Gautam

private investments, but only if this effect is greater than the negative impact of the increased tax rates
needed to pay for it.
The conventional theoretical models on infrastructure use ‘black box’ models of exogenous
and endogenous growth which simply assume a positive effect of infrastructure on macroeconomic
aggregates. However, these models do not specify sufficient form to allow an interpretation of the
controversial empirical results. There are at least three important reasons why economic benefits from
transport infrastructure may exceed benefits as conventionally measured (Crafts 2009): first, transport
projects may lead to economies of agglomeration, that is, positive externalities in terms of the productivity
benefits that firms gain from being located close to other firms through knowledge spillover, access to
more suppliers and larger labour markets. Second, reduced transport costs would save resources and may
lead to increased output. This effect would be larger, the bigger the price–cost margins. Third, transport

SE
projects may lead to improved labour supply, for example, cheaper commuting may give people better
access to jobs and also it can help in specialisation of labour.1
Therefore, a satisfactory model, from the public policy point of view, must be able to go behind

U
the link between growth and public capital, understanding and explaining the mechanisms that allow

L
infrastructure to lead to significant economic effects. A major challenge for such an approach is to

IA
analyse the dynamic externalities associated with transport infrastructure.

C
Empirically, most of the works on dynamic externalities are classified into three heads: Marshall-
Arrow-Romer (MAR) externalities, Jacobs externalities and Porter externalities. MAR externalities
ER
concern knowledge spillover among firms in a single industry; Jacobs externalities are associated with
knowledge spillover resulting from the diversity of industries; and Porter externalities refer to the
M

knowledge spillovers caused by the competition within industries (Glaeser et al. 1992). However, there
M

is another type of externality which is very less talked about in the literature and which concerns the
O

complete geographical scope of the subject under study. This is called as network dynamic externality
(Zheng 2010). Network dynamic externalities represent knowledge spillovers resulting from the whole
C

agglomerated area. Therefore, by definition, it is the network dynamic externality, which is relevant for
R

understanding the dynamics of transport infrastructure.

FO

In this article we attempt to address two basic issues. First, we present empirical evidence on the
dynamic impacts of road transport on macroeconomic variables in the Indian context based on vector
autoregression (VAR) model. Our choice of VAR model stems from the fact that it imposes less
T

theoretical a priori restrictions on the variables. Second, we focus on a cointegration analysis of dynamic
O

externalities to investigate whether road transport infrastructure has a long-run equilibrium relationship
N

with the macroeconomic variables or not. Our specification for the cointegration analysis is based on the
grounds of network dynamic externalities.
The organisation of the article is as follows: Section 2 sketches the state of road transport infrastructure
in India. Section 3 presents a review of literature dealing with models on infrastructure. Section 4 gives data
description and preliminary analysis. Section 5 presents the model for VAR analysis and its results. Section
6 presents the model for cointegration analysis and its results, and finally, Section 7 concludes the article.

2. State of Indian Road Transport Infrastructure

Transport sector accounted for 6.4 per cent in India’s gross domestic product (GDP) in 2007–08. The
composition of various sub-sectors of the transport sector as a percentage of GDP is given in Table 1.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 137

Table 1. Share of Different Modes of Transport in GDP

Railways Road Water Air Services* All
1999–2000 1.2 3.8 0.2 0.2 0.5 5.8
2000–01 1.2 3.9 0.2 0.2 0.5 6
2001–02 1.2 3.8 0.2 0.2 0.5 5.9
2002–03 1.2 4.1 0.2 0.2 0.5 6.2
2003–04 1.2 4.3 0.2 0.2 0.5 6.3
2004–05 1.2 4.5 0.2 0.2 0.5 6.6
2005–06 1.2 4.5 0.2 0.2 0.5 6.5

SE
2006–07 1.2 4.5 0.2 0.2 0.5 6.4
Source: Central Statistical Organisation.

U
Notes: *Services incidental to transport.
The share of railways in GDP is exclusive of Financial Intermediation Services Indirectly Measured (FISM); for other modes it

L
includes FISM.

IA
C
The table suggests that road transport is the most dominant segment in India’s transportation sector.
ER
It has a share of 4.5 per cent in India’s GDP in comparison to mere 1.2 per cent share of railways
in 2006–07. It may be noted that the entire increase in percentage share of transport in GDP since
M

1999–2000 has come from road transport sector only, with share of other modes remaining nearly
M

constant.
Table 2 shows that over seven years from 1999–2000 to 2006–07 the average annual growth in road
O

transport sector at 9.4 per cent is much higher than the overall annual average GDP growth of 6.9 per
C

cent in the same period. The growth rates across various modes have varied, with road transport growing
R

at a much higher rate compared to other competing modes.

FO

Table 2. Average Annual Growth Rates (%) in Transport Sector and GDP*
T

Sector 2006–07/1999–2000
O

Overall GDP 6.9

N

Transport 8.6
Railways 6.8
Road 9.4
Water 7.9
Air 7.7
Services ** 6.4
Source: Central Statistical Organisation.
Notes: *GDP at factor cost in real terms (1999–2000 prices).
**Services incidental to transport.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

138 Shruti Tripathi and Vikash Gautam

Table 3. Outlay and Expenditure for Road Transport Sector during IX and X Plan (in Rs millions)
Outlay Expenditure Outlay and Expenditure
Year Centre States Total Centre States Total Centre States Total
IX Plan 60 73,552.6 74,152.6 427.8 59,253.7 5,968.15 17.22 1,429.89 1,447.11
(1997–2002)
X Plan 210 92,069 94,169 1,793.7 55,400.7 5,719.44 30.63 3,666.83 3,697.46
(2002–07)
2002–03 30 22,094.9 22,394.9 297.5 12,010.3 1,230.78 0.25 1,008.46 1,008.71
2003–04 40 13,613.2 14,013.2 354.8 8,475.5 883.03 4.52 513.77 518.29
2004–05 44 14,985 15,425 368.3 7,663.7 803.2 7.17 732.13 739.3

SE
2005–06 46 16,587.2 17,047.2 31.81 11,705.1 1,202.32 14.19 488.21 502.4
2006–07 50 2,174.06 2,224.06 45.5 1,554.61 1,600.11 4.5 619.45 623.95

U
Source: Planning Commission, Government of India.

L
IA
Table 3 outlines the outlay and expenditure for road transport sector from 1997–98 to 2006–07

C
which covers the ninth and the tenth Five-year Plan periods. Larger chunk of the allocation comes
from the state governments as most of the roads fall in state list (all except national highways which
ER
is about 2 per cent of the total length of roads). There was an increase of nearly 30 per cent in total
outlays from ninth plan to the tenth plan period. But besides there being a need for further capacity
M

creation, nearly 19.5 and 39.3 per cent of the allocations remained unutilised in the ninth plan and the
M

tenth plan period, respectively. Road transport demand is expected to grow by around 10 per cent per
O

annum in the backdrop of a targeted annual GDP growth of 9 per cent during the eleventh Five-year
Plan.
C
R

3. Literature Review
FO

There have been various classes of models that have tried to model the impact of transport infrastructure,
T

as a form of public capital, on macroeconomic variables. There are two streams of work which have
O

attempted to introduce public capital into the production function: one, where public capital enters as a
N

standard input to the production function along with labour and physical capital (Barro 1990), and the
other where public capital affects the productivity of the standard inputs by affecting the technology
variable (Duggal et al. 1999; Shioji 2001). In the Barro (1990) model there are two effects of the size
of government: An increase in tax reduces growth, while an increase in expenditure increases growth.
When government is small, the second effect is likely to dominate. This approach has been criticised on
the grounds that it assumes that firms know the marginal cost of infrastructure and can, therefore, use it
in the optimisation problem.2 In models which incorporate public capital in the technology constraints,
an increase in public capital improves productivity and therefore, output via an external effect. These
models yield the result that the elasticity of output with respect to public capital is greater in the long
run because public capital influences output not just because of its own productivity impact but also by
attracting more private investment to the region.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 139

Following this, the relation between public infrastructure and economic growth has been mainly
captured by using following approaches: the production function approach, the cost and profit function
approach and the VAR approach. We discuss each of these below.

3.1 The Production Function Approach

The production function approach to the study of the effects of public infrastructure on economic growth
and productivity starts from a neoclassical production function

Yt = F(At , KPt , Lt , KGt ) ............ (1)

SE
Y represents output, KP represents the stock of private capital, L the number of employed workers (or the
number of hours), KG, generally measured in monetary terms, represents either the overall stock of public

U
capital, or narrower aggregates like core infrastructure and transport infrastructure and A represents the
total factor productivity level of the country. Most studies assume a Cobb-Douglas functional form for

L
equation (1) commonly estimated in logarithm form as

IA
C
yt = a + αkpt + βlt + γkgt + εt ............(2)
ER
M

Here εt is the error term and α, β and γ are the parameters.3 The γ coefficient, which tells how much output
would increase if the public infrastructure stock were increased by 1 per cent, represents the elasticity of
M

output with respect to public infrastructure capital.

O

There are a series of remarks on such an approach: first relates to the assumption of a Cobb-Douglas
C

functional form to represent the underlying production technology. It is based on some simplifying
assumptions that often do not hold true in the data. For instance, the Cobb-Douglas function imposes a
R

unitary elasticity of substitution between inputs which implies that increases in public capital raise the
FO

marginal and average product of both labour and private inputs. Second, returns to scale are independent
of the scale of output, thereby forcing the same value for returns to scale to hold for every observation
T

unit. A third remark acknowledges that the public infrastructure stock is treated symmetrically to the
O

private inputs, which would be strictly valid only if it could be safely assumed that there exists a market-
determined unit price of public infrastructure that is known to individual firms (Duggal et al. 1999); but
N

as this condition is unlikely to hold, the marginal productivity theory according to which input factors are
paid according to their marginal product is likely to be violated. Fourth, production functions like those
represented in equation (2) have been usually estimated assuming that capital and labour are exogenous,
and therefore, it is likely to suffer from simultaneity bias which might arise for a variety of reasons
(Griliches and Mairesse 1995; Nickell 1981).

3.2 The Cost and Profit Function Approach

The cost function approach, in contrast, to production function approach is built on the assumption
that, given market determined input prices and conditional on output and on other technical variables,

Journal of Infrastructure Development, 2, 2 (2010): 135–151

140 Shruti Tripathi and Vikash Gautam

producers combine inputs to minimise the costs of producing output. A cost function can, therefore, be
represented as follows:

C = C(W, Y, T, KG) ............(3)

Here, W is a vector of factor prices (usually labour, capital and intermediate inputs), Y is the output level,
T is time varying level of technology and KG is the stock of public infrastructure.4 In this framework,
input quantities and costs are endogenous, while factor prices and output levels are assumed exogenous.5
This approach uses flexible functional forms like the translog or the generalised Leontief which, unlike
Cobb-Douglas form, do not impose any a priori restriction on technology (Afraz et al. 2006).
An additional advantage of the cost function over the production function approach is that the

SE
estimation of a cost function allows to test whether the stock of public infrastructure is a substitute or a
complement for each private input. In other words, it is possible to compute the effect that an extra rupee
of public infrastructure has on the private sector in terms of cost savings in a given year which is also

U
known as the shadow price of public infrastructure.6 A positive shadow value is necessary condition for

L
the public capital stock to be productive.

IA
This functional framework has been criticised widely for assuming that output is exogenous to the system.

C
This has led some authors (for instance, Demetriades and Mamuneas 2000) to estimate a profit rather than
a cost function. The profit function is the result of producers who choose input (output) quantities, given
ER
output (input) prices, in order to maximise firm’s profit. This yields a function analogous to equation (3):
M

Π = Π (P, W, T, KG) ............(4)

M
O

The profit function has, as arguments, the input and factor prices, assumed exogenous, and a trend
variable to proxy for technical progress and the stock of public infrastructure.7
C
R

3.3 The VAR Approach

FO

Most of the empirical studies that have assessed the impact of public infrastructure capital on economy have
adopted a ‘structural’ approach, based on either the production or the cost function. Vector autoregressive
T

(VAR) models, instead, impose less theoretical a priori restrictions between the variables. They rather
O

facilitate the ascertaining of whether there are substantial feedback effects among variables. In addition,
N

the VAR approach estimates of the long-run effect of public infrastructure on output incorporates both
the direct effect that public infrastructure has on output and also the indirect effects through the impact
that it has on private inputs.
Pereira and Andraz (2001) implement a vector error correction model (VECM) for 12 US industries.
They observe a positive effect of public investment on private capital investment, with an estimated
public–private capital elasticity of 0.397. But, at industry level, there is evidence of crowding-in effects
only in 5 out of 12. The effects of public investment on employment appear to be very small: In the
long run, 51 long-term jobs are created per million dollars in public investment. Kamps (2004) uses a
VECM for 22 OECD countries over the period 1960–2001 to test the relationship which exists between
macroeconomic variables like output, private capital, employment and public capital. He found that, in
the long run, the elasticity of output with respect to public capital was positive and significant for 12

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 141

countries, and not significantly different from zero for the remaining 9. Furthermore, his results show
that public capital reacts to GDP shocks, which suggests that public capital is endogenous.
At a country level, Pereira and Roca (2003) working on Spain found positive elasticity of output with
respect to the public capital stock. According to the estimates presented, long-run rates of return of public
capital are about 5 to 8 per cent. Higher rates of return for public investments in transport infrastructure
are found in Pereira and Andraz (2005) for Portugal (15 per cent). They also examined the impact on
output of narrower categories of transport investments, and found that investments in ports had the
highest return, followed by national roads, municipal roads, airports and railways. Mamatzakis (1999)
for Greece and Flores de Frutos, Garcia-Diez and Perez (1998) for Spain also concluded that the effect
of public capital on output was positive. However, Sturm, Kuper and Groote (1999) for the Netherlands
and Litgarth (2000) for Portugal were not able to detect any positive impact of public capital on output.

SE
4. Data Description and Preliminary Analysis

U
L
We use annual data for the period 1970–71 to 2007–08. The macroeconomic variables we consider are

IA
output (GDP), employment, gross private capital formation (GPTCF) and gross public capital formation
(GPBCF). We consider three measures of road transport—length of all roads, length of national highways
C
and length of state highways. We also use data on registered number of vehicles to calculate a proxy
ER
for dynamic externality defined as ratio of length of all roads to number of registered vehicles. The data
series on GDP, employment, GPTCF and GPBCF are taken from economic surveys and data on length
M

of all roads, length of national highways, length of state highways and number of vehicles are taken from
M

Centre for Monitoring Indian Economy (CMIE) report on infrastructure. GDP, GPTCF and GPBCF are
measured in crore rupees (1999–2000 series), employment is measured in millions of workers and length
O

of roads is measured in kilometres. Number of vehicles is measured in thousands. We normalise all the
C

variables by converting them in natural log.

R

As the first step of the preliminary analysis we do unit root tests (or alternatively, stationarity tests) for
the variables noted earlier. We use the augmented Dickey-Fuller (ADF) method to test the null hypothesis
FO

of unit root (Table 4). The optimal lag structure is chosen using Schwarz’s Bayesian information criterion
T

Table 4. ADF Unit Root Test

O

Variables Test Statistic (p-value) Result

N

GDP –2.07 (0.54) Non-stationary

Employment –3.81 (0.04) Stationary
GPTCF –2.69 (0.24) Non-stationary
GPBCF –1.89 (0.54) Non-stationary
Length of national highway –0.69 (0.97) Non-stationary
Length of state highway –4.63 (0.01) Stationary
Length of total road –2.31 (0.42) Non-stationary
Dynamic externality –1.54 (0.95) Non-stationary
Critical value –3.5 (0.05)

Journal of Infrastructure Development, 2, 2 (2010): 135–151

142 Shruti Tripathi and Vikash Gautam

Table 5. Zivot-Andrews Test

Variable Test Statistic Result
GDP –2.859 No structural break; presence of unit root
GPTCF –3.377 No structural break; presence of unit root
GPBCF –2.681 No structural break; presence of unit root
Length of national highway –3.266 No structural break; presence of unit root
Length of state highway –4.012 No structural break; presence of unit root
Length of total road –2.502 No structural break; presence of unit root
Dynamic externality –4.488 No structural break; presence of unit root
Critical values –4.80 (at 5% significance)

SE
(SBIC). In all the cases except for employment and length of state highways, the p-value of the ADF test

U
is greater than the 0.05—the critical value. Therefore, we cannot reject the null hypothesis of unit root
in these variables. For these variables we conduct the same test by doing the first difference and find that

L
the p-values are less than the critical value of 0.05. Therefore, we rejected the null hypothesis of unit root

IA
at first difference of data series.
C
A problem common with the unit root tests is that they do not allow for the possibility of a structural
ER
break. We follow Zivot and Andrews (1992) to test the null of unit root against the alternative of
stationarity caused by endogenous breaks in the data series. Test results, given by Table 5, show that
M

there is no structural break in the data; therefore, all the variables are I(1) variables.
M
O

5. VAR Analysis
C

Since employment and length of state highways are stationary variables and we wanted to analyse the impact
R

of infrastructure variables on macroeconomic variables, we thought of going ahead with the VAR analysis.
FO

This section presents empirical evidence of the dynamic impacts of road transport and gross public capital
formation on macroeconomic variables in India based on VAR models.8 We determine the specifications
of the VAR models using growth rates of the variables mentioned earlier. We estimate four VAR models
T

having one variable from GPBCF, length of all roads, length of national highways and length of state
O

highways in each model. All the models, however, include aggregated output, employment and GPTCF.
N

5.1 The Unrestricted VAR Model

A p-th order VAR model, denoted VAR (p), can be expressed as follows:

Xt = A1 Xt–1+A2 Xt–2 + ... + Ap Xt–p + et ............(5)

where Xt = [x1t,…, xkt] is a set of variables collected in a (k × 1) vector, Aj denotes a k× k matrix of

autoregressive coefficients for j = 1, 2, …, p. The vector εt = [ε1t, ε2t,…, εkt] is a k dimensional white noise
process, that is, E[εt] = 0, E[εtεt'] = Ω, and E[εtεs'] = 0 for s ≠ t, with Ω a (k × k) symmetric positive
definite matrix.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 143

5.2 Structural Identification

The next step is to identify VAR models in order to give the impulse response functions a structural
interpretation. In this section, we identify the VAR models by assuming that the relation between the
reduced-form disturbances, εt, and the structural disturbances, et, takes the following form:

1 0 0 0 ε gt  1 0 0 0  egt 
a     
 21 1 0 0  ε pt  = 0 1 0 0  e pt 
 a31 a32 1 0  ε nt  0 0 1 0  ent 
       
 a41 a42 a43 1 ε yt  0 0 0 1  eyt 

SE
There are six unknown parameters in the representation (left lower triangle) as well as four unknown

U
parameters in the diagonal covariance matrix in the structural disturbances, Σ. Since there are 10 distinct
elements in the covariance matrix of the reduced-form residuals, , the model is just identified. This

L
set of identifying assumptions is an example for the recursive approach originally proposed by Sims

IA
(1980). Note that after the initial period the variables in the system are allowed to interact freely, that is,

C
for example, shocks to real GDP can affect public capital in all periods after the one in which the shock
occurs.
ER
The assumptions on the contemporaneous relations between the variables can be justified as follows:
Movements in government spending, unlike movements in taxes, are largely unrelated to the business
M

cycle. In particular, government spending on capital items involves large decision and implementation
M

lags. Therefore, it seems sensible to assume that public capital is not affected contemporaneously by
O

shocks originating in the private sector. In a similar vein, private capital is largely unrelated to the
business cycle (King and Rebelo 1999). Employment, while being strongly pro-cyclical, in general lags
C

the business cycle (Stock and Watson 1999). Thus, it seems appropriate to assume that employment is
R

unaffected contemporaneously by output shocks. Ordering output last can be justified by, for example, a
FO

production function which shows that the three inputs affect output contemporaneously. Because of our
choice of method of identification, for each variable, there is a single estimated elasticity for government
capital.
T
O

5.3 Identifying and Measuring the Effect of Innovations in Public Investment

N

5.3.1 Impulse Responses

To plot the response of all the series to a shock on a single scale, they are normalised using their
innovation variances. We consider the effects of one standard deviation shock in the rates of growth
of GPBCF and infrastructure variables. We expect these shocks to have at least temporary effect on
the growth rates of other variables which get converted into permanent effects on the levels of these
variables.9 Figures 1 to 4 give the impulse responses for all the series for one standard deviation shock to
GPBCF and infrastructure variables. The specification for the underlying model is done on the basis of
Bayesian information criterion. In all the cases, first-order specification was selected; a higher order was
not considered due to the relative small size of the sample.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

144 Shruti Tripathi and Vikash Gautam

1.0
0.8
0.6
0.4
0.2
0.0
–0.2
–0.4
0 1 2 3 4 5 6 7 8 9

Government Capital Formation Employment

SE
Private Captial Formation GDP

U
Figure 1. Plot of Response to Government Capital Formation

L
IA
The plots show that the output effect of a shock to GPBCF is positive at all plotted horizons up to
C
the endpoint of 10 years, but for a shock to national highway, the output response is negative or close
ER
to zero at all plotted horizons. Since the own impulse response of GPBCF to a shock is positive for
all the cases, the negative responses of output for national highway cannot be easily reconciled with
M

GPBCF responses unless GPBCF is conceived to have a negative marginal productivity in that sector.
M

The impulse responses of GPTCF to a shock to GPBCF show that both are complements in the medium
run. However, in the case of length of all roads, public investment in infrastructure seems to substitute
O

GPTCF in the short run. The effect of shock on GPBCF, length of state highway and length of total roads
C

on employment is positive. This supports the traditional Keynesian hypothesis according to which em-
ployment will rise in response to an increase in government spending. However, in the case of shock to
R

length of national highway the result differs from the rest.10

FO
T

1.0
O

0.8
N

0.6

0.4

0.2

0.0

–0.2
0 1 2 3 4 5 6 7 8 9
Length of the State Highways Employment
Private Captial Formation GDP

Figure 2. Plot of Responses to Length of the State Highways

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 145

1.0
0.8
0.6
0.4
0.2
0.0
–0.2
–0.4
–0.6
0 1 2 3 4 5 6 7 8 9
Length of the National Highways Employment

SE
Private Captial Formation GDP

U
Figure 3. Plot of Responses to Length of the National Highways

L
IA
There are two opposing forces determining the response of GPTCF to a shock to GPBCF: first, the

C
resources for financing an additional unit of GPBCF compete with the resources available to the GPTCF.
ER
And second, an increase in public capital has positive effect on the marginal productivity of private
capital; all other things being equal this induces a rise in private investment. If public capital accumulates
gradually, then the first force will dominate the second in the short run, whereas in the medium to long
M

run the second force will dominate (Baxter and King 1993).
M
O

5.3.2 Accumulated Elasticity

C

In this section, we estimate the long-run elasticities of macroeconomic variables, with respect to GPBCF and
infrastructure variables.11 The long-term elasticity is obtained by allowing all the variables to respond over
R

time to the shock to GPBCF and infrastructure variables growth in the first step. The long-run elasticities
FO
T

1.0
O

0.8
N

0.6

0.4

0.2

0.0

–0.2
0 1 2 3 4 5 6 7 8 9
Length of the Roads Employment
Private Captial Formation GDP

Figure 4. Plot of Responses to Length of the Roads

Journal of Infrastructure Development, 2, 2 (2010): 135–151

146 Shruti Tripathi and Vikash Gautam

Table 6. Long-term Accumulated Elasticities

Public Investment Variable Output Employment GPTCF
GPBCF 0.012 0.051 0.008
Length of national highways 0.000 –0.014 –0.008
Length of state highways 0.014 0.013 0.050
Length of total roads 0.018 0.033 0.049

with respect to public capital reported in Table 6 account for the dynamic interaction between the variables
in the system.12 The results show that for most of the cases the long-run elasticity of output and employment
with respect to GPBCF and infrastructure variables (excluding national highways) is positive, giving support

SE
to the hypothesis that public capital is productive. It also suggests that in the long term, public investment
crowds in private capital. However, increases in the length of national highways have the opposite effect, not

U
only it crowds out GPTCF but it also has no impact on output and negative impact on employment.

L
5.3.3 Marginal Effect

IA
Table 7 presents the contemporaneous structural parameters from the VAR analysis. As can be seen from

C
the table while GPBCF, length of state highways and length of total roads have positive effect on output
ER
and employment, they have negative effect on GPTCF. However, increase in length of national highway
has negative effect on output and employment, confirming to our earlier results.
M
M

5.4 An Explanation for Contrasting Findings on National Highways

O

National highways account for nearly 2 per cent of total road length in India. Since the early 1990s
C

there has been a sharp slowdown in transport sector in general and national highways in particular.
Between 1991–92 and 2002–03 capital formation in infrastructure as a proportion of GDP was nearly
R

halved, from 6.34 to 3.5 per cent. The main reason behind this fall was pursuit of a macroeconomic
FO

adjustment programme under which the government tried to rein in fiscal deficit through a cutback in
public investment in general, and infrastructural investment in particular. In this period while the ratio of
public investment to GDP came down from 8.8 to 6.1 per cent, for public investment in infrastructure the
T

fall was steeper and investment in national highways was largely ignored. It was presumed that private
O

entrepreneurs would more than make up for the fall in public investment in this sector. The presumption
N

proved unfounded and the country had to pay a heavy price in terms of lost production and employment
with increasing infrastructural bottlenecks (Rakshit 2006).

Table 7. Marginal Effects of Public Investment on Private Sector Variables

Public Investment Variable Output Employment GPTCF
GPBCF 0.0403 0.0088 –0.4038
Length of national highways –0.0589 –0.0139 0.0969
Length of state highways 0.1004 0.0244 –1.0535
Length of total roads 0.1214 0.1175 –0.8534

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 147

6. Network Dynamic Externality

We use a model similar to the one used in Zheng (2010) where we suppose that a macroeconomic
variable (Mt ) is dependent on network dynamic externality (Xt ) at time t as follows:

M kt = b(Xt )g ............(6)

Here superscript k = {output, government capital formation, private capital formation} (employment is
a stationary variable hence we keep it out of the analysis). b and g are parameters to be estimated. To
represent network dynamic externality, we use the following measure:

Lt

SE
Xt , ............ (7)
Rr

U
where Lt is the total length of roads and Rt is the number of registered vehicles, both at time t. Xt in

L
expression (7) can be interpreted as the potential velocity of movement of people and firms who use

IA
roads as their mode of transportation. This is likely to result into knowledge spillovers and bring about
dynamic externalities.13
C
ER
In order to conduct the estimation exercise, we transform the model in expression (6) into logarithmic
form:
M

ln M kt = ln b + g ln Xt + et ............(8)
M

where et is the error term which is assumed to be white noise. We estimate this equation using the
O

data explained earlier. The Johansen (1988) maximum likelihood (ML) test is applied to examine the
C

cointegration between Xt and M kt. The null hypothesis of r cointegrating vectors is tested against the
R

alternative hypothesis of r + 1 cointegrating vectors.

The results in Table 8 show that there exists one cointegrating relationship for each possible linear
FO

relationship between macro variable and dynamic. However, there exists no cointegrating relationship
T

Table 8. Cointegration Rank Test

O

Hypothesised Number of Cointegrating Vectors Trace Statistic p-Value

N

Between GDP and Xt

None 30.238 0.000
At most 1 3.467 0.063
Between GPTCF and Xt
None 16.137 0.040
At most 1 0.646 0.422
Between GPBCF and Xt
None 4.367 0.872
At most 1 0.213 0.645
Critical value 0.05

Journal of Infrastructure Development, 2, 2 (2010): 135–151

148 Shruti Tripathi and Vikash Gautam

between GPBCF and Xt (Dickey et al. 1991).14 Therefore, we can say that network dynamic externality
contribute to the long-run growth of macro variable implying that concentration of industries would
bring about an increase in output and capital formation.
We also test for cointegrating relationship between GPBCF, length of all roads, length of national
highways with the macroeconomic variables—output, GPTCF. Following the Engle–Granger procedure,
in each case we perform three tests, each one with different infrastructure pertaining variable. The ADF
tests cannot reject the null hypothesis of no cointegration. Johansen’s trace and maximum Eigenvalue
tests also failed to accept cointegration in any of the cases.

7. Conclusion

SE
In this article the attempt is to analyse the anatomy of road transport infrastructure in the Indian context.
The major points made in this article are as follows: first, the entire increase in percentage share of

U
transport in GDP since 1999–2000 has come from road transport sector only, with share of other modes

L
remaining nearly constant. Second, the growth rates across various modes of transport have varied with

IA
road transport growing at a much higher rate compared to other competing modes. However, the growth
of vehicular traffic on roads has been far greater than the growth in road network; as a result the main
arteries face capacity saturation. C
ER
We also present empirical evidence of the dynamic impacts of road transport on macroeconomic
variables in the Indian context. First, we examine if there is any long-term relationship between network
M

dynamic externalities resulting from road transport networks and macroeconomic variables. We notice
M

that there exists a long-term relationship between network dynamic externalities from road transport
networks and GDP, and also with GPBCF. However, there exists no such long-term relationship between
O

network dynamic externalities from road transport networks and GPTCF. The VAR approach estimates
C

of the long-run effect of public infrastructure on output, employment and GPTCF incorporate both the
R

direct effect that public infrastructure has on output and also the indirect effects through other channels.
The results suggest that the long-run elasticity of output with respect to public capital is positive,
FO

giving support to the hypothesis that public capital is productive. It also suggests that in the long term,
public investment crowds in both private capital and employment. However, increases in the length of
T

national highways seem to have just the opposite effect, not only it crowds out private investment and
O

employment but also have negative long-term impact on output and employment. The results imply that
N

government should be proactive in the provisions of infrastructure facilities (road) as we have a case for
‘development via excess’ where the government takes the initiative and the private sector follows.

Notes
  1. These would be recognised in traditional cost–benefit analysis by the commuter’s willingness to pay for the
time savings.
  2. The Barro (1990) model has since been extended in several different ways. For example, Fernald (1999)
incorporates transport services provided by the government into each sector’s production function. In ad-
dition, sectoral output depends on the transport services produced within the sector. Fernald models both
the network nature of public transport infrastructure (i.e., that the productivity of any particular node in the

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 149

transport network is inherently linked with the capacities of the other nodes and the set-up of the transport
system) and congestion effects (where the effect of an additional user on the congestion experienced by the
remaining users is not constant).
  3. If a version of equation (2) is estimated with panel data, with subscript i denoting the cross-sectional unit, it
could be re-written as follows: yit = ai + αkpit + βlit + γkgit + ct + εit . Here ai denotes an individual effect which
accounts for any unobservable time-invariant factor that might affect each unit of observation, and ct is a vector
of time dummies which accounts for any shock which is common to all units of observation.
  4. It is important to note that the information directly provided by the cost function is not directly comparable to
that derived from the estimation of a production function. While the latter provides a value for the elasticity
of output with respect to public infrastructure, the former provides information on the elasticity of costs with
respect to public infrastructure: If that is negative, then private costs fall when the stock of public infrastructure
is raised.

SE
  5. The assumption of exogenous factor prices is usually defendable—especially in panels where the units of ob-
servation are (relatively small) individual economic sectors or (probably to a lesser extent) regions. The endo-
geneity assumption of output might be perhaps more problematic, even if most of the literature seems to have

U
implicitly assumed away this problem.

L
  6. The shadow price of public infrastructure can be regarded as the average benefits accruing each year to the

IA
private sector firms when an additional rupee is spent on public infrastructure.
  7. Although the theoretical advantages of the profit function over the cost-function approach are fairly clear, the

C
latter is still a much more common approach among applied researchers who have, however, not yet addressed,
ER
to the best of our knowledge, the issues of output endogeneity in a cost function framework.
  8. We are considering GPBCF besides road transport because it is a comprehensive measure in the sense that it in-
cludes all publicly funded capital formation. It includes all the research and development undertaken by central
M

government laboratories, colleges and universities, and non-profit firms which may be more likely to generate
M

positive externalities and hence supporting growth.

  9. Such impulse responses have the attractive feature that they can be viewed as representative for typical shocks
O

that occurred during the sample period in the individual countries and hence results can be compared across
C

studies.
10. An explanation for contrasting findings on national highways is presented in Section 5.4.
R

11. Long term is defined as the time horizon over which the growth effects of a shock disappear, that is, the accu-
FO

mulated response functions converge. In our analysis, this is a time horizon of 5 to 10 years.
12. The long-run elasticities in Table 6 are different from the elasticities of a production function. In a production
function the elasticity of output with respect to public capital gives the percentage change in output for 1 per
T

cent change in exogenous public capital, holding fixed the private inputs and excluding feedback effects, for
O

example, from output to public capital.

N

13. Zheng (2010) use total area of densely inhibited districts as denominator in expression (6). We use number of
registered vehicles in place of this because of data availability issues. However, looking at some of the cities
(state capitals) we figure out that the correlation between population density and number of registered vehicles
is very high.
14. Eigenvalue test, not reported here, also gives the similar results.

References
Afraz, N., M. Aquilina, M. Conti and A. Lilico (2006), ‘Impact of Transport Infrastructure on Economic Growth’,
Annex 6 to Final Report of COMPETE: Analysis of the Contribution of Transport Policies to the Competitive-
ness of the EU Economy and Comparison with the United States, Karlsruhe, ISI.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

150 Shruti Tripathi and Vikash Gautam

Barro, R.J. (1990), ‘Government Spending in a Simple Model of Endogenous Growth’, The Journal of Political
Economy, 98(5, Pt. 2): 103–25.
Baxter, M. and R.G. King (1993), ‘Fiscal Policy in General Equilibrium’, American Economic Review, 83(3): 315–
33.
Crafts, Nicholas (2009), ‘Transport Infrastructure Investment: Implications for Growth and Productivity’, Oxford
Review of Economic Policy, 25(3): 327–43.
Demetriades, P.O. and T.P. Mamuneas (2000), ‘Intertemporal Output and Employment Effects of Public Infrastruc-
ture Capital: Evidence from 12 OECD Economies’, Economic Journal, 110(5): 687–12.
Dickey, David A., Dennis W. Jansen and Daniel L. Thornton (1991), ‘A Primer on Cointegration with an Application
to Money and Income’, Review, Federal Reserve Bank of St. Louis, March: 58–78.
Duggal, V., C. Saltzman and L. Klein (1999), ‘Infrastructure and Productivity: A Nonlinear Approach’, Journal of
Econometrics, 92(1): 47–74.

SE
Economic Survey (1996–97 to 2008–09), Chapters on Infrastructure, Ministry of Finance, Government of India.
Flores de Frutos, R., M. Garcia-Diez and Amal T. Perez (1998), ‘Public Capital and Economic Growth: An Analysis
of the Spanish Economy’, Applied Economics, 30(8): 985–94.

U
Glaeser, Edward L., Hedi D. Kallal, Jose A. Scheinkman and Andrei Shleifer (1992), ‘Growth in Cities’, Journal of

L
Political Economy, University of Chicago Press, 100(6): 1126–52.

IA
Griliches, Z. and J. Mairesse (1995), ‘Production Functions: The Search for Identification’, NBER Working Paper
No. 5067.

C
Fernald, John G. (1999), ‘Roads to Prosperity? Assessing the Link between Public Capital and Productivity’, Ameri-
ER
can Economic Review, 89(3): 619–38.
Johansen, S. (1988), ‘Statistical Analysis of Cointegration Vectors’, Journal of Economic Dynamics and Control,
12(2–3): 231–54.
M

Kamps, C. (2004), ‘New Estimates of Government Net Capital Stocks for 22 OECD Countries 1960–2001’, IMF
M

Working Papers, 67.

King, R.G. and S.T. Rebelo. (1999), ‘Resuscitating Real Business Cycles’, in J.B. Taylor and M. Woodford (eds),
O

Handbook of Macroeconomics, Volume 1, Part 2, pp. 927–1007. Elsevier.

C

Litgarth, J.E. (2000), ‘Public Capital and Output Growth in Portugal: An Empirical Analysis’, IMF Working Paper,
11.
R

Mamatzakis, E.C. (1999), ‘Testing for Long Run Relationship between Infrastructure and Private Capital Productiv-
FO

ity: A Time Series Analysis for the Greek Industry’, Applied Economics Letters, 6(4): 243–46.
Nickell, S. (1981), ‘Biases in Dynamic Models with Fixed Effects’, Econometrica, 49(6): 1417–26.
Pereira, A. and J. Andraz (2001), ‘On the Impact of Public Investment on the Performance of U.S. Industries’, Public
T

Finance Review, 31(1): 66–90.

O

Pereira, A. and Sagales O. Roca (2003), ‘Spillover Effects of Public Capital Formation: Evidence from the Spanish
N

Regions’, Journal of Urban Economics, 53(2): 238–56.

Pereira, A. and J.M. Andraz (2005), ‘Public Investment in Transportation Infrastructure and Economic Performance
in Portugal’, Review of Development Economics, 9(2): 177–96.
Rakshit, M. (2006), ‘Issues in Infrastructural Investment: National Highway and Development Programme’, ICRA
Bulletin, Money and Finance, January–June: 49–80.
Shioji, E. (2001), ‘Public Capital and Economic Growth: A Convergence Approach’, Journal of Economic Growth,
6(3): 205–27.
Sims, C.A. (1980), ‘Macroeconomics and Reality’, Econometrica, 48(1): 1–48.
Stock, J.H. and M.W. Watson (1999), ‘Business Cycle Fluctuations in US Macroeconomic Time Series’, in J.B.
Taylor and M. Woodford (eds), Handbook of Macroeconomics, Volume 1, Part A, pp. 3–64. Elsevier.
Sturm, J.E., G.H. Kuper and P. Groote (1999), ‘Output Effects of Infrastructure Investment in the Netherlands,
1853–1913’, Journal of Macroeconomics, 21(2): 355–80.

Journal of Infrastructure Development, 2, 2 (2010): 135–151

Road Transport Infrastructure and Economic Growth in India 151

Zheng, Xiao-Ping (2010), ‘A Cointegration Analysis of Dynamic Externalities’, Japan and the World Economy,
22(2): 130–40.
Zivot, Eric and Donald W.K. Andrews (1992), ‘Further Evidence on the Great Crash, the Oil-Price Shock, and the
Unit-Root Hypothesis’, Journal of Business & Economic Statistics, American Statistical Association, 10(3):
251–70.

SE
U
L
IA
C
ER
M
M
O
C
R
FO
T
O
N

Journal of Infrastructure Development, 2, 2 (2010): 135–151

 SE
U
L
IA
C
ER
M
M
O
C
R
FO
T
O
N

View publication stats