Revisited the Relationship Between Economic Growth and Transport Infrastructure in India: An Empirical Study

Abstract

This study empirically re-examines the relationship between transport infrastructure and economic growth in India for the period 1990–2017. Multivariate dynamic models are applied to estimate the relationship between economic growth and different modes of transport infrastructure namely road, rail and air transports in the vector error correction model framework. The results reveal that road and air transports have significant positive contribution to economic growth in the long-run while rail transport is insignificant. This study further examines the said issue using unit free index variables and has constructed a composite index of transport infrastructure using principal component analysis to analyse the nexus between aggregate transport infrastructure and economic growth in India in the post globalisation era. The results of the study indicate the bidirectional causality between aggregate transport infrastructure and economic growth. Results of this study suggest incorporating feedback issue in policy formulations.

JEL Codes: C22, O18, R4

Keywords

Multivariate model composite index transport infrastructure globalisation VECM PCA causality

I. Introduction

The economists and policy makers recognise transport infrastructure as a crucial factor for sustainable economic growth. Many studies (Pradhan & Bagchi, 2013; Sanchez-Robles, 1998; Shah, 1992; Short & Kopp, 2005; Tripathi & Goutam, 2010) show that integrated and well-functioning transport facilities are necessary for achieving sustainable economic growth. Recently, according to the World Bank, India emerges as the fastest growing economy in the World.¹ One of the possible reasons behind this success might be the outcome of several factors particularly, massive investment in infrastructure which was initiated in road infrastructure² in India since early in the twenty-first century. In this context, the relevant question arises whether transport infrastructure is the cause of economic growth in the post globalisation era in India. This study revisits the ongoing debate over this issue in the context of India.

Demand for infrastructure especially, transport infrastructure for economic growth is age old. Various economic theories have been put forwarded to justify the role of transport infrastructure in a country’s economic progress. Amongst them some supported the view that enlarged availability of transport facilities is the essential pre-condition for economic development (Barro, 1990; Barro & Sala-i-Martin, 1992, 1995; Hirschman, 1958; Wagner, 1958, etc.). However, Wagner (1958) opposed the view of Hirschman (1958) and argued that demand for transport facilities would grow if and only if development takes place. So, there is a debate over the causal relationship between transport infrastructure and economic growth. In other words, whether transport infrastructure causes economic growth or economic growth itself is the cause of demand for transport infrastructure. Empirical literature provides controversial role of transport infrastructure in the process of economic growth. Aschauer (1990), Khadaroo and Seetanah (2008), Tripathi and Goutam (2010), Pradhan and Bagchi (2013) and Mohmand et al. (2016) supported the view that the availability of well-developed transport facilities is the cause of economic growth. Opposing the above said view, Gramlich (1994), Ramanathan and Parikh (1999) and Maparu and Mazumder (2017) suggested that economic growth causes transport sector to grow further. Thus, the outcomes of the empirical studies (Bosede et al., 2013; Khadaroo & Seetanah, 2008; Nwakwze & Mulikat, 2010; Stephan, 2000; Tong et al., 2014; Wessel, 2019) are inconclusive and still unsettled.

In the context of India, few studies have been conducted to estimate the nexus between transport infrastructure and economic growth (Ghani et al., 2014; Maparu & Mazumder, 2017; Pradhan & Bagchi, 2013; Sahoo & Dash, 2011; Tripathi & Goutam, 2010). Overall, the results of these studies revealed that transport infrastructure plays a crucial role in economic growth. In most of the cases either road transport or both road and rail transports have been considered as proxy for transport infrastructure. However, the multidimensional facet of transport infrastructure will not be reflected adequately if we consider transport infrastructure as one-dimensional phenomenon, that is, use them separately in a bivariate framework. This is because the use of bivariate model may mislead to the biased causality inferences due to the omission of relevant variables (Lutkepohl, 1982). This study attempts to provide a better estimation technique by taking different sub-sectors of transport infrastructure as explanatory variables to estimate the nexus between transport infrastructure and economic growth unlike previous studies that have taken only one sub-sector of transport infrastructure as explanatory variable under ceteris paribus assumption. The use of different sub-sectors of transport infrastructure in a multivariate structure³ helps us to capture the multidimensional aspects of transport infrastructure. However, this may create problems like multicollinearity if all the indicator variables are highly correlated to each other. Therefore, this study develops a composite time series index of transport infrastructure using principal component analysis (PCA)⁴ to avoid such problem and revisits the relationship between transport infrastructure and economic growth in India in the post globalisation period.

The rest of the study is organised as follows: Section 2 summarises the existing literature. Section 3 presents a brief overview of transport infrastructure development in the post globalisation period in India. Section 4 describes the sources of data and econometric methodologies used in the study. Section 5 analyses the empirical results. Finally, conclusions and policy suggestions are presented in Section 6.

II. Review of Literature

The nexus between transport infrastructure and economic growth had been started with Antle (1983) when he measured the effects of transportation and communication infrastructure on aggregate agriculture productivity using a Cobb–Douglas production function for 47 less developing countries and 19 developed countries. He found a strong and positive association between the level of infrastructure and aggregate agriculture productivity. This finding is in line with Aschauer (1989), who found that the elasticity of gross domestic product (GDP) with respect to core (such as street lights, highways, airports, mass transit services, sewerage and electricity and gas)infrastructure was 0.24 and concluded that in the U.S. core infrastructure contributed more to productivity than other forms of infrastructure. Eberts (1990), Munnell (1990), Garcia-Mila and McGuire (1992), etc. have also found a high output elasticity of some public capital infrastructure. The positive contribution of transport infrastructure has also been addressed by Fernald (1999) where he reported that output elasticity of highway capital in U.S. economy for the period 1953–1989 was 0.35. He concluded that industries who used road transport intensively, have a faster growth of factor productivity than others. Stephan (2000) measured the effects of public infrastructure (consisting transport and human capital infrastructure) to local private production using a panel data set of 327 German counties and found that transport and human capital positively contribute to the productivity and output of local private sector. Fan and Zhang (2004) used 1996 Agricultural Census dataset of China and estimated the effects of rural infrastructure (road density) on both farm and nonfarm productions. Using a simultaneous equation system they concluded that the role of rural infrastructure and education is much higher to the productivity of nonfarm sector than agriculture productivity. Khadaroo and Seetanah (2008) examined the association between transport capital and economic growth for Mauritius over the period 1950–2000 using a dynamic time series analysis in a vector error correction model (VECM) framework and found a positive contribution of transport infrastructure to the economic performance of Mauritius. Tripathi and Goutam (2010) examined the long-run equilibrium relationship between road transport, employment, output and gross capital formation in India from 1970–1971 to 2007–2008. They used vector autoregressive approach (VAR) to analyse the impact of road transport on such macroeconomic variables. The results of their study revealed that road transportation has a significant and positive long-run relationship with economic growth and gross public capital formation. This result is in line with Pradhan and Bagchi (2013), who have also showed a positive contribution of transport infrastructure (consisting road and rail) to economic growth in India during 1970–2010. Using VECM, they found bidirectional causality between road infrastructure and economic growth and road infrastructure and gross domestic capital formation, unidirectional causality from railway infrastructure to economic growth, and gross domestic capital formation, and finally, unidirectional causality from total transport to economic growth and gross capital formation in India. Similar results have been found by Mohmand et al. (2016). They have measured the impact of transportation infrastructure on economic growth in Pakistan using a panel data of developed and less developed provinces. The results of their study found bidirectional causality between transport infrastructure and economic growth in case of rich and much developed provinces and unidirectional causality between economic growth and transportation infrastructure in underdeveloped provinces. Maparu and Mazumder (2017) examined the causal relationships between transport infrastructure (road, rail, air and port infrastructure), economic development and urbanisation in India from the period 1990 to 2011. They used several time series estimation techniques such as, Engle–Granger cointegration test, Johansen cointegration test, VECM and Granger causality test to conduct the analysis of their study. Their results showed that in the long-run, transport infrastructure is cointegrated with economic development and the directional of causality is from economic development to different sub-sectors of transport infrastructure in most of the cases and drawing support in favour Wagner’s law. However, no causation has been found from urbanisation to transport infrastructure but the reverse is not true as unidirectional causation runs from highway and port transport to urbanisation. Wessel (2019) analysed the effects of specific mode of transport infrastructure on trade using a gravity equation model with European trade flows. The results of the study showed that improvement of certain types of transport infrastructure have different trade effects, among them rail and air infrastructures are more responsive to quality improvements in the corresponding infrastructure while road density rather than road quality has a positive trade effect.

III. State of Indian Transport Infrastructure in the Post Globalisation Period

This section comprises the state of different sub-sectors of transport infrastructure along with their contributions to GDP in India. Table 1 represents the state of different modes of road transport in India over the last three decades. Table 1 shows that share of national highways has increased from 1.67% in 1990–1991 to 2.15% in 2015–2016. The share of state highways on the other hand has decreased from 6.31% in 1990–1991 to 3.75% in 2015–2016. The share of Public Works Department roads, rural roads and urban roads has also increased during 1990–1991 to 2015–2016.

Table 1.

Category-wise Share of Road to Total Road Length.

Year	National Highways	State Highways	Other Public Works Department Roads	Rural Roads	Urban Roads	Project Roads
1990–1991	1.67	6.31	25.26	47.09	9.26	10.40
2000–2001	2.33	5.34	29.76	43.34	10.19	9.04
2010–2011	1.87	4.32	26.52	48.80	10.87	7.61
2015–2016	2.15	3.75	NA	64.54	10.84	6.78

Source: CMIE Database of India.

Note: NA: Not available.

Table 2 represents the state of railway transport infrastructure in India over the last three decades. It shows that double line railway track has increased from 22.98% in 1990–1991 to 31.85% in 2015–2016 while single line railway track has decreased from 77.02% in 1990–1991 to 68.15% in 2015–2016. Total number of railway stations and railway bridges has also increased from 1990–1991 to 2015–2016. Total number of level crossing has decreased from 1990–1991 to 2015–2016.

Table 2.

State of Railway Transport in India.

Year	Single Line (% Share of Total Railway Line)	Double/Multiple Line (% Share of Total Railway Line)	No. of Stations	No. of Bridges	Level Crossings
1990–1991	77.02	22.98	7100	116000	37117
2000–2001	74.60	25.40	6843	119984	38561
2010–2011	70.18	29.82	7133	133160	32735
2015–2016	68.15	31.85	7216	140919	28607

Source: CMIE Database of India.

Table 3 represents the state of airport traffic statistics in India in the period of post economic reforms. During 1995–1996 to 2017–2018, domestic aircraft movements have increased from 3.15 to 18.87 lakhs. Total number of domestic passenger has also increased from 255.6 lakhs in 1995–1996 to 2432.78 lakhs in 2017–2018. Total number of domestic cargo handled has also increased from 2.126 lakh tonnes in 1995–1996 to 12.13 lakh tonnes in 2017–2018.

Table 3.

Airport Traffic Statistics in India.

Year	Domestic Aircraft Movements (Lakhs)	Number of Domestic Passenger (Lakhs)	Domestic Cargo Handled (Lakh Tonnes)
1995–1996	3.147	255.64	2.126
2000–2001	3.866	280.176	2.884
2010–2011	10.936	1055.227	8.527
2017–2018	18.866	2432.779	12.131

Source: CMIE Database of India.

Table 4 provides the composition of various sub-sectors of the transport sector in terms of GDP. Table 1 shows the contribution of transport sector in India’s GDP for the period of 1999–2000 to 2010–2011. The share of transport sector in India’s GDP has increased from 6.0% in 1999–2000 to 6.5% in 2010–2011. In particular, the share of road transport infrastructure in GDP has increased while that of railways has fallen during 1999–2000 to 2010–2011. Table 4 suggests that road transport is the most dominant segment in India’s transport sector with a share of 6.5% in GDP in 2010–2011 in comparison to 1.0% GDP share of railways in the same year. However, the share of air and water transport infrastructures in GDP has remained either constant or grown marginally during this period. Therefore, the entire increase in percentage share of transport sector in GDP since 1999–2000 has come from road transport sector only.

Table 4.

Share of GDP of Different Transport Modes in India During 1999–2000 to 2010–2011.

Year	Transport	Road	Rail	Air	Water
1999–2000	6.0	3.8	1.3	0.2	0.2
2000–2001	6.0	3.9	1.3	0.2	0.2
2001–2002	6.0	3.9	1.2	0.2	0.2
2002–2003	6.2	4.1	1.2	0.2	0.2
2003–2004	6.3	4.3	1.2	0.2	0.2
2004–2005	6.7	4.8	1.0	0.2	0.2
2005–2006	6.7	4.8	1.0	0.2	0.2
2006–2007	6.7	4.8	1.0	0.2	0.2
2007–2008	6.7	4.7	1.0	0.2	0.2
2008–2009	6.6	4.8	1.0	0.2	0.2
2009–2010	6.5	4.7	1.0	0.2	0.2
2010–2011	6.5	4.7	1.0	0.3	0.2

Source: Central Statistical Organization.

Notes: Data up to 2003–2004 are at 1999–2000 prices. Data from 2004–2005 onwards are at 2004–2005 prices.

All shares in GDP are inclusive of Financial Intermediation Services Indirectly Measured.

IV. Data and Methodology

For conducting the analysis, this study takes three components of transport infrastructure namely, road density (total road length per 1,000 sq. km), railway density (total railway route length per 1,000 sq. km) and air density (domestic aircraft flown per 1,000 sq. km), and GDP per-capita (constant 2011–2012 in rupees) as a proxy for economic growth in India for the period 1990–2017. All the indicators of transport infrastructure are taken from the Economic Outlook of India—the CMIE database while GDP per-capita is obtained from the Handbook of Statistics, Reserve Bank of India (RBI). All the variables are used in real terms and then transformed to natural logarithms, that is, road density as lnRODN, rail density as lnRLDN, air density as lnARDN and economic growth as lnGDPPC.

This study applies VECM to assess the long-run equilibrium relationship between road, rail and air transports and economic growth as well as their short-run dynamics. Since, the use of the VECM (Equation (1a) in Appendix) requires the series to be cointegrated with the integration of order one, that is, I (1), therefore, Engle–Granger cointegration test (1987) and Johansen cointegration test (1988) are applied to check the cointegrating nature of the variables. Engle and Granger test of cointegration method applies augmented Dickey–Fuller test (ADF) test on the residuals (Equation (1b) in Appendix) estimated from the cointegrating regression between the variables. On the other hand, Johansen test of cointegration (1988) proposes two different likelihood ratio tests namely: the Trace test (Equation (1c) in Appendix) and Maximum Eigenvalue test (Equation (1d) in Appendix) to analyse the long-run associations among the variables.

V. Results and Discussion

Basic Results

The underlying assumption of time series empirical work is based on stationary data that might avoid spurious regression.⁵ Therefore, it is necessary to check the nature of the data generating process (DGP)⁶ of a time series variable that is, whether this process is stationary or not before using it in time series estimation. A DGP is said to be stationary (weekly) if its mean and variance are time-invariant and its covariance depends on time difference only. We use two different ways to detect the stationary nature of a time series, namely visual inspection of the data plots (Figures 1 and 2) and numerical judgement with unit root tests (see Table 5). Figure 1 displays lnGDPPC, lnRODN, lnRLDN, lnARDN, income index (GDPPC) and index of transport infrastructure (TRNINF) for the period of 1989–90 to 2016–2017. All the series show a tendency to drift upwards over time (see Figure 1) at their levels and therefore, they are non-stationary in mean. Otherwise, they are stationary in mean at their first difference (see Figure 2).

Figure 1.

Source: The authors.

Figure 2.

Source: The authors.

The ADF test (Dickey & Fuller, 1981) and Phillips–Perron (PP) test (Phillips & Perron, 1988) are performed to examine the stationary properties of the variables and their order of integration. We use both ADF and PP tests mainly because of their methodological differences. In ADF test, the serial correlation and heteroscedasticity in the errors are corrected parametrically by incorporating augmented terms in the model whereas, the PP test non-parametrically corrects serial correlation and heteroscedasticity in the residuals of error terms by modifying the ADF test statistics (Das, 2019, pp. 326–327). Results of ADF and PP tests are reported in Table 5. The results of Table 5 show that lnGDPPC, lnRODN, lnRLDN, lnARDN, normalised GDPPC and index of TRNINF are non-stationary at level as the null hypothesis of unit roots for both the test cannot be rejected for all the variables. However, first differencing makes them stationary. Thus, all the variables are non-stationary with integration of order one, that is, I (1).

Table 5.

Results of Unit Root Tests of GDPPC and TRNINF Index.

Variables Name	Augmented Dickey–Fuller Test		Phillips–Perron Test		Concluding Remarks
Variables Name	At Level	First Difference	At Level	First Difference	Concluding Remarks
lnRODN	1.2501	–4.5976***	1.5378	–4.5785***	I(1)
lnRLDN	1.4705	–6.1392***	1.9046	–6.1392***	I(1)
lnARDN	0.3879	–4.3504***	0.3218	–4.3504***	I(1)
lnGDPPC	–2.4362	–5.3375***	–2.4459	–8.0168***	I(1)
TRNINF	2.2517	–4.2445**	2.8068	–7.3024***	I(1)
GDPPC	0.6075	–4.1460**	1.1112	–4.6031***	I(1)

Source: Authors’ calculation.

Notes: *** and ** denote the level of significance at 1% and 5%, respectively.

I(1) indicates non-stationary nature of the variable with the integration of order one.

Results

Results of ADF and PP test indicate that lnGDPPC, lnRODN, lnRLDN, lnARDN and index of TRNINF and normalised GDPPC are integrated of order one, that is, all the variables are first difference stationary. In this context, this study applies two types of cointegration techniques—multivariate cointegration and bivariate cointegration. Engle and Granger (1987) showed that two I(1) time series may cointegrate if their linear combination is I(0). Existence of cointegration may exhibit a long-run equilibrium relationship and in such situation error correction model can be applied to estimate the long-run relationship along with short-run dynamics among the variables. The next section discusses the nexus between transport infrastructure and economic growth in a multivariate framework.

Multivariate Model

The multivariate analysis of this study examines the cointegrating relationship⁷ among lnGDPPC, lnRODN, lnRLDN and lnARDN. In this context, Johansen cointegration test (1988)⁸ is applied to find out whether any long-run relationship exists among them. Table 6 reports the results of Johansen cointegration test. Both Trace statistics and Max Eigenvalue statistics indicate that there is one cointegrating equation among the variables, implies that all the variables are cointegrated.

Table 6.

Results of Johansen Cointegration Test.

No. of Co-integrating Equations	Trace Statistics	Critical Value	Max. Eigenvalue Statistics	Critical Value
None*	50.5429	47.8561	27.9102	27.5843
At most 1	22.6327	29.7971	16.0476	21.1216

Source: Authors’ calculation.

Notes: Both the Trace and Max. Eigenvalue tests indicate 1 cointegrating equation at the 0.05 level.

* denotes rejection of the hypothesis at the 0.05 level.

Cointegration provides long-run equilibrium relationship between two or more non-stationary variables. However, it does not say anything about short-run forces that keep the long-run equilibrium relationship intact (Bhaumik, 2015, p. 273). In this context, the study applies VECM to assess the long-run relationship among road transport, rail transport, air transport and economic growth in India along with short-run dynamics. The result of estimated long-run relationship is presented in Table 7. The estimated long-run relationship between GDP per-capita and different modes of transport infrastructure is:

lngdppc = 5.2337 + 0.9085 \ln r d d n + 0.1308 \ln a r d n

(1)

Equation (1) indicates that an increase of one unit of lnrddn (road density) leads to an increase of 0.9085 unit of lngdppc (GDP per-capita) in the long-run. Similarly, an increase of 1 unit of lnardn (air density) raises 0.1308 unit of lngdppc in the long-run. The results thus reveal a positive and significant long-run relationship of road and air transports with economic growth in India. The coefficient of railway transport is found insignificant in the long-run.

Table 7.

Estimated Long-Run Relationship.

Variables	Coefficient	Std Errors	t-Statistics
lnGDPPC	1.0000
lnRODN	–0.9085***	0.1172	–7.7502
lnRLDN	–0.8380	1.0942	0.7658
lnARDN	–0.1308***	0.0289	–4.5267
Constant	5.2337

Source: Authors’ calculation.

Notes: *** denotes statistical significance at 1% level.

Optimum lag length is 2 as per Akaike information criterion and Schwarz information criterion.

The coefficient of the error correction term (ECT₋₁) is reported in Table 8. The coefficient of ECT₋₁ is negative and statistically significant at 5% level. It indicates that any short-run deviation from the long-run equilibrium relationship will automatically be corrected in the next year, and estimated speed of error correction is around 56%. It implies that the process will converge towards its long-run equilibrium value.

Table 8.

Short-run Dynamics.

	Coefficient	SE	t-Statistics
ECT (–1)	–0.5613**	0.2776	–2.0214

Source: Authors’ calculation.

Note: ** denotes statistical significance at 5% level.

The study then performs some diagnostic tests to check the model’s validity.⁹ Results of the Lagrange multiplier (LM) test, Jarque–Bera test and CUSUM of squares test (Tables A1 and A2, and Figure A1) suggest that the error correction equations are not subject to residually auto-correlated up to lag 2, normally distributed and stable.

The study further investigates the direction of causality among the variables. The direction of causation on the nexus between growth and transport infrastructure development is debatable and still an unsettled issue. However, this issue needs to be settled in the effective design and implementation of better transport policies for an emerging country like India. Therefore, the study applies VECM based causality test to assess the direction of causality, whether it is from transport infrastructure (or more specifically, different modes of transport infrastructure) to economic growth and vice versa.

Table 9 presents the results of Wald test. The test does not reject the null hypothesis of non-causality in case of any of the mode of transport infrastructure to economic growth and vice versa. This implies that in the short-run neither any sub-sector of transport infrastructure causes economic growth nor economic growth causes them. However, the study finds a short-run causation from road transport to air transport, implying that growth of road transport infrastructure is a pre-condition for the growth of air transport infrastructure.

Table 9.

Results of VEC Granger Causality/Block Exogeneity Wald Test.

Direction of Causality	No. of Lags	Chi-square	Concluding Remarks
Road transport ↓ Economic growth	₁	0.7273	Cannot reject H₀
Railway transport ↓ Economic growth	1	0.1421	Cannot reject H₀
Air transport ↓ Economic growth	1	0.4371	Cannot reject H₀
Economic growth ↓ Road transport	1	0.0917	Cannot reject H₀
Rail transport ↓ Road transport	1	1.1489	Cannot reject H₀
Air transport ↓ Road transport	1	0.3970	Cannot reject H₀
Economic growth ↓ Railway transport	1	1.2387	Cannot reject H₀
Road transport ↓ Railway transport	1	0.4099	Cannot reject H₀
Air transport ↓ Railway transport	1	0.1998	Cannot reject H₀
Economic growth ↓ Air transport	1	1.7002	Cannot reject H₀
Road transport ↓ Air transport	1	14.4952***	Reject H₀
Rail transport ↓ Air transport	1	3.9536	Cannot reject H₀

Source: Authors’ calculation.

Note: *** denotes statistical significance at 1% level.

Overall, from the multivariate analysis, it is observed that road, rail and air transport infrastructures are cointegrated with economic growth in India. Road and air transports have a positive and significant long-run relationship with economic growth. However, no short-run causality is found from any of the sub-sector of transport infrastructure to economic growth and vice versa.

The analysis of above said multivariate level variables may have some limitations, if any. As per literature, there is also certain problem in bivariate levels. To overcome these problems and to re-examine the nexus between growth and infrastructure development focusing transport infrastructure, this study designs a time series composite index of transport infrastructure using PCA. The corresponding Eigenvalues and the different principal components for each sub-sectors of transport infrastructure are reported in Tables A3 and A4. Design of composite index of transport infrastructure will reduce the dimension of the data set and make them unit free. Despite this, it also provides relevant coefficients of all the three indicators rather incorporating them into a single equation. In the section below, we discuss the empirical results between transport infrastructure index and normalised GDP per capita or income index.

Nexus Between Transport Infrastructure Index and Income Index

ADF and PP unit roots tests (Table 5) suggest that TRNINF and GDPPC are non-stationary and having integrated in order one, that is, I(1). Therefore, cointegration can be tested between GDPPC and TRNINF. Engle and Granger test of cointegration applies ADF test on residuals (Table 10) of the estimated cointegrating equation:

G D P P C = - 0.027 + 0.500 T R N I N F + 0.007 t

(2)

Table 10 indicates that estimated residual series generated from Equation (2) is stationary at level as the ADF test rejects the null hypothesis of having unit roots at 1% level.

Table 10.

Results of Unit Root Test of the Estimated Residuals of Equation 1.

Augmented Dickey–Fuller Test at Level
Residual	t-Statistics	p-Value	Variable’s Type
Residual	–4.8720***	.0030	I(0)

Source: Authors’ calculation.

Note: *** denotes statistical significance at 1% level.

The result indicates that transport infrastructure and economic growth are cointegrated and a long-run equilibrium relationship may exist between them. In such situation, Granger causality test can be applied to assess the direction of causality. Table 11 presents the estimated results of long-run relationships along with short-run dynamics.

Table 11.

Estimated Long-run and Short-run Dynamics Between GDPPC and TRNINF Index.

Variables	Coefficients	Std Errors	t-Statistics	Prob.
TRNINF	0.5000***	0.0242	20.6300	.0000
Trend	0.0071***	0.0015	4.8595	.0001
Constant	–0.0279**	0.0107	–2.5962	.0156
ECT (–1)	–0.5668**	0.2302	–2.4620	.0230
R ²	0.9956
Adj. R²	0.9956
DW	1.9104
Sum sq. Resid.	0.0103
Log likelihood	70.9503
F-statistics	3073.98
AIC	-4.8536
SIC	-4.7109

Source: Authors’ calculation.

Notes: *** and ** denote the level of significance at 1% and 5%, respectively.

AIC: Akaike information criterion; SIC: Schwarz information criterion.

This study applies ordinary least square method to estimate the long-run relationship between transport infrastructure and economic growth. Equation 1 shows the estimated long-run relationship between transport infrastructure and economic growth. Equation 1 indicates that normalised per-capita income increases by 0.50 points for every incremental point of transport infrastructure index to maintain the long-run equilibrium, Ceteris peribus. Thus, in the long-run, transport infrastructure has a positive relationship with economic growth in India in the period of post liberalisation. Time trend is also positive and significant and bears a positive relationship with economic growth. The negative and significant error correction term indicate that transport infrastructure would make a short-run adjustments at an approximate speed of 57% towards its long-run equilibrium path whenever there induce any shock to the system, respectively. The study then performs some diagnostic tests to check whether the model is free from serial correlation, whether the residuals of the error are normally distributed and finally, whether the model is stable or not. The results of LM test, Jarque–Bera test and CUSUM of squares test are reported in Tables A5, A6 and Figure A2.

In such situation, Granger causality test may be applied to assess the direction of causality. So, the study further performs vector autoregressive (VAR) base Granger causality (Granger, 1969) test to assess the direction of causality between overall transport infrastructure and economic growth. In this context, we examine optimum lag length for VAR model. The optimal lag length is one as per Akaike information criterion and Schwarz information criterion (Table A7). The results of Granger causality test are presented in Table 12.

Table 12.

VAR Granger Causality/Block Exogeneity Wald Tests.

Causal Relation	Chi-Square	Prob.	Remarks
∇(TRNINF) → ∇(GDPPC)	11.8879***	.0026	Causality exists
∇(GDPPC) → ∇(TRNINF)	23.6353***	.0000	Causality exists

Source: Authors’ calculation.

Notes: *** denotes statistical significance at 1% level. →

Lag length for the model is 1 as per Schwarz information criterion and Akaike information criterion. Here, x y means x is a cause of y.

The result of the causality test indicates that transport infrastructure causes economic growth and economic growth is also a cause of transport infrastructure in the short-run. This indicates the existence of bidirectional causality between transport infrastructure and economic growth.

There has been a long debate between endogenous growth theories and Wagner’s law on the issue of direction of causality between infrastructure and economic development. Empirical literature was unable to confirm whether the causality is from infrastructure to economic development or vice versa. The findings of this study may settle the above said issue and definitely help to form appropriate transport policies for an economy which aspires to grow.

Conclusion

This study has reinvestigated the relationship between transport infrastructure and economic growth in the post liberalisation era in India. For the said purpose, we have collected data from the RBI and the CMIE for the period of 1990–2017. Applying techniques of unit root tests, we have examined the nature of data and observed that concerned variables are non-stationary, that is, I(1). Next, the study investigates the cointegrating long-run equilibrium relations. This investigative study has done in two ways using (a) multivariate level variables and (b) unit free bivariate index variables. To make unit free, this study constructs a composite index for transport infrastructure using the PCA. This composite index represents overall transport infrastructure.

Initially, this study applies a multivariate dynamic framework to examine the nexus between economic growth and transport infrastructure in general, and different modes of transport namely road, rail and air transport infrastructure with economic growth in particular. Applying econometric techniques, this study confirms the cointegrating relationship between economic growth and transport infrastructure. Using cointegration and Granger causality test, the results of this study may be summarised as follows:

Economic growth is directly related with road and air transport. Road transport infrastructure has a significant positive contribution to economic growth in the long-run. This indicates that an increase in road transport would have a positive effect on economic growth in the long-run. In the long-run, air transport has also played an important role in economic growth in India. No significant long-run relationship is found between railway infrastructure and economic growth in India. This could be due to the fact that in the post globalisation era, the extension of railway network (total railway length) has not been increased much and remained constant over the years (Figure A3). The results of the Granger causality test from the multivariate dynamic frame work suggest no short-run causality from any of the sub-sector of transport infrastructure to economic growth and vice versa.

Unit free bivariate model is used to replace the multivariate model. Results of unit free bivariate model provide long-run relation between transport infrastructure and economic growth with short-run dynamics. Overall transport infrastructure is reflected in transport infrastructure index, which is significant in the long-run and has a positive contribution to the economic growth in India since 1990. However, bidirectional causality exists between overall transport infrastructure and economic growth in the short-run, indicating that expansion of transportation system will cause economic growth and on the contrary, high growth in per-capita income will also facilitate the demand for transport infrastructure.

From the policy perspective, the results of this study suggest that increasing transport facilities (namely road, rail and air) might be an effective way through which sustainable economic growth can be achieved. An extra care should be taken to increase the total railway length as it is more environmental friendly than other sub-sectors of transport infrastructure.

Appendix A

[\begin{matrix} Δ \ln g d p p c_{t} \\ Δ \ln r o d n_{t} \\ Δ \ln r l d n_{t} \\ Δ \ln a r d n_{t} \end{matrix}] = [\begin{matrix} α_{10} \\ α_{20} \\ α_{30} \\ α_{40} \end{matrix}] + \sum_{i = 1}^{p} [\begin{matrix} α_{11 i} & α_{12 i} & α_{13 i} & α_{14} \\ α_{21 i} & α_{22 i} & α_{23 i} & α_{24 i} \\ α_{31 i} & α_{32 i} & α_{33 i} & α_{34 i} \\ α_{41 i} & α_{42 i} & α_{43 i} & α_{44 i} \end{matrix}] [\begin{matrix} Δ \ln g d p p c_{t - i} \\ Δ \ln r o d n_{t - i} \\ Δ \ln r l d n_{t - i} \\ Δ \ln a r d n_{t - i} \end{matrix}] + [\begin{matrix} λ_{1} \\ λ_{2} \\ λ_{3} \\ λ_{4} \end{matrix}] {E C T}_{t - 1} + [\begin{matrix} ε_{1 t} \\ ε_{2 t} \\ ε_{3 t} \\ ε_{4 t} \end{matrix}]

(1a)

Δ ε_{t}^{\land} = α ε_{t - 1}^{\land} + \sum_{i = 1}^{ρ} β_{i} Δ ε_{t - i}^{\hat{n}} + μ_{t}

(1b)

J_{trace} = - T \sum_{i = r + 1}^{n} \ln (1 - {\hat{ρ}}_{i}

(1c)

J_{\max} = - T \ln (1 - {\hat{ρ}}_{r + 1}

(1d)

Figure A1.

Source: The authors.

Figure A2.

Source: The authors.

Figure A3.

Source: The authors.

Table A1.

Results of Breusch–Godfrey Correlation Lagrange Multiplier Test.

F-statistic	p-Value	Remark
1.5466	.2526	Cannot reject

Source: The authors.

Table A2.

Results of Jarque–Bera Normality Test.

F-statistic	p-Value	Remark
0.6931	.7070	Cannot reject

Source: The authors.

Table A3.

Eigenvalues and Variance Explained by Principal Components.

Principal Components	Eigenvalues	Percentage of Variance	Cumulative Variance
1	2.9327	0.9776	0.9776
2	0.0526	0.0175	0.9951
3	0.0146	0.0049	1.0000

Source: The authors.

Table A4.

Components Loadings for Different Sub-sectors of Transport Infrastructure.

Variables	PC 1	PC 2	PC 3
Road density	0.580503	–0.230756	–0.780877
Rail density	0.573947	0.796214	0.191384
Air density	0.577582	–0.559281	0.594646

Source: The authors.

Table A5.

Results of Breusch–Godfrey Correlation Lagrange Multiplier Test.

F-statistic	p-Value	Remark
0.0011	.9732	Cannot reject

Source: The authors.

Table A6.

Results of Jarque–Bera Normality Test.

F-statistic	p-Value	Remark
0.6709	.7150	Cannot reject

Source: The authors.

Table A7.

Lag Order Selection Criteria for the VAR Model.

Lag	LR	FPE	AIC	SIC	HQIC
Optimum Lag Length for Per Capital GDP and Road Transport
0	NA	0.000206	–2.813946	–2.717169	–2.786078
1	160.7962*	2.58e-07*	–9.497395*	–9.207065*	–9.413790*
2	2.616652	3.12e-07	–9.314305	–8.830422	–9.174964

Source: The authors.

Notes: * denotes lag order selected by the criterion.

AIC: Akaike information criterion; FPE: Final prediction error; GDP: Gross domestic product; HQIC: Hannan–Quinn information criterion; LR: Sequential modified LR test statistic (each test at 5% level); SIC: Schwarz information criterion.

Footnotes

Acknowledgements

We are grateful to the referees for their valuable comments and suggestions. We are also thankful to the Chairman and participants of XXIXth Annual Conference at Jadavpur University during 16–17 December 2019 and Summer School at Presidency University during 29 July–2 August 2019 for their constructive comments.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Notes

References

Antle

J. M.

(1983). Infrastructure and aggregate productivity: International evidence. Economic Development and Cultural Change , 31(2), 609–620.

Aschauer

D. A.

(1989). Is public expenditure productive? Journal of Monetary Economics , 23, 177–200.

Aschauer

D. A.

(1990). Why is infrastructure important? Is there a shortfall in public capital investment? Conference Series of Federal Reserve Bank of Boston.

Barro

R. J.

(1990). Government spending in a simple model of endogenous growth. Journal of Political Economy , 98, 103–125.

Barro

R. J.

, & Sala-i-Martin

(1992). Convergence. Journal of Political Economy , 100, 223–251.

Barro

R. J.

, & Sala-i-Martin

(1995). Economic growth (2nd ed.). McGraw-Hill.

Bhaumik

S. K.

(2015). Principle of econometrics: A modern approach using EViews . Oxford University Press.

Bosede

, Abalaba

, & Afolabi

(2013). Transport infrastructure and economic growth in Nigeria. International Journal of Humanities and Social Science Invention , 2, 26–31.

Das

(2019). Econometrics in theory and practice . Springer.

10.

Dickey

D. A.

, & Fuller

W. A.

(1981). Likelihood ratio statistics for Autoregressive time series with a unit root. Econometrica , 49, 1057–1072.

11.

Dinda

(2016). Land acquisition and compensation policy for development Activity. Journal of Land and Rural Studies , 4(1), 111–118.

12.

Eberts

R. W.

(1990). Public infrastructure and regional economic development. Economic Review , Q I, 15–27.

13.

Engle

R. F.

, & Granger

C. W. J.

(1987). Cointegration and error correction: Representation, estimation and testing. Econometrica , 55(1987), 251–271.

14.

Fan

, & Zhang

(2004). Infrastructure and regional economic development in rural China. China Economic Review , 15(2004), 203–214.

15.

Fernald

J. G.

(1999). Roads to prosperity? Assessing the link between public capital and productivity. The American Economic Review , 89(3), 619–637.

16.

Garcia-Mila

, & McGuire

T. J.

(1992). The contribution of public provided inputs to States’ economies. Regional Science and Urban Economics , 22(2), 229–241.

17.

Ghani

, Grover Goswami

, & Kerr

W. R.

(2014, March). Highway to success: The impact of golden quadrilateral project for the location and performance of Indian manufacturing. The Economic Journal , 126, 317–357.

18.

Gramlich

E. M.

(1994). Infrastructure investment: A review essay. Journal of Economic Literature , 32, 1176–1196.

19.

Granger

C. W. J.

(1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica , 37(3), 423–438.

20.

Hirschman

(1958). The strategy of economic development . Yale University Press.

21.

Johansen

(1988). Statistical analysis of cointegrating vectors. Journal of Economic Dynamics and Control , 12, 231–254.

22.

Johansen

(1995). A statistical analysis of cointegration for I (2) variables. Econometric Theory , 11, 25–59.

23.

Khadaroo

, & Seetanah

(2008). Transport and economic performance: The case of Mauritius. Journal of Transport Economics and Policy , 42(2), 255–267.

24.

Lutkepohl

(1982). Non causality due to omitted variables, Journal of Econometrics , 19, 367–378.

25.

Maparu

T. S.

, & Mazumder

T. N.

(2017). Transport infrastructure, economic development and urbanization in India (1990-2011): Is there any causal relationship? Transportation Research Part A , 100, 319–336.

26.

Mohmand

Y. T.

, Wang

, & Saeed

(2016). The impact of transportation infrastructure on economic growth: Empirical evidence from Pakistan. The International Journal of Transportation Research . https://doi.org/10.1080/19427867.2016.1165463

27.

Munnell

A. H.

(1990). Why has productivity declined? Productivity and public investment. New England Economic Review , 3–22.

28.

Nwakwze

M. N.

, & Mulikat

A. Y.

(2010). Transportation and economic growth in Nigeria. Journal of Research in National Development , 18(2).

29.

Phillips

P. C. B.

, & Perron

(1988). Testing for a unit root in time series regression. Biometrika , 75(2), 335–346.

30.

Pradhan

R. B

, & Bagchi

P. T

(2013). Effect of transportation infrastructure on economic growth in India: The VECM approach. Research in Transport Economics , 38, 139–148.

31.

Ramanathan

, & Parikh

J. K.

(1999). Transport sector in India: An analysis in the context of sustainable development. Transport Policy , 6, 35–45.

32.

Sahoo

, & Dash

R. K.

(2011). Economic growth in South Asia: Role of infrastructure. The Journal of International Trade and Economic Development , 21(2), 217–252.

33.

Sanchez-Robles

(1998). Infrastructure investment and growth: Some empirical evidence, Contemporary Economic Policy , 16, 98–108.

34.

Shah

(1992). Dynamics of public infrastructure, industrial productivity and profitability. Review of Economics and Statistics , 74, 28–36.

35.

Short

, & Kopp

(2005). Transport Infrastructure: Investment and planning policy and research aspects. Transport Policy , 12, 360–367.

36.

Stephan

(2000). The contribution of transport and human capital infrastructure to local private production: A partial adjustment approach . www.ssoar.info

37.

The Economic Times . (2015, June 11). India’s growth rate set to surpass China this year: World Bank. https://economictimes.indiatimes.com/news/economy/indicators/indias-growth-rate-set-to-surpass-china-this-year-world-bank/articleshow/47621953.cms?from=mdr

38.

Tong

, Yu

T. E.

, & Roberts

R. K.

(2014). Dynamics of transport infrastructure, exports and economic growth in the United States. Transportation Research Forum , 53(1), 65–81.

39.

Tripathi

, & Goutam

(2010). Road transport infrastructure and economic growth in India. Journal of Infrastructure Development , 2, 135–151.

40.

Wagner

(1958). Three extracts on Public Finance. In R. A. Musgrave & A. T. Peacock (Eds.), Classics in the theory of public finance . Macmillan.

41.

Wessel

(2019). Evaluating the transport mode-specific trade effects of different transport infrastructure types. Transport Policy , 78, 42–57.