Impact of Public Debt on Economic Growth: Evidence from Indian

States

Asit Ranjan Mohanty​1 ​and Bibhuti Ranjan

Mishra​2

Abstrac
t

This study examines the impact of public debt on economic growth by taking other
control

variables like institutional credit and commercial electricity consumption. It uses

panel data of

14 major (non-special category) States in India for the period 1980-81 to 2013-14.
After

establishing long-run relationship among the variables, panel long-run estimates are
drawn

using both DOLS and FMOLS methods. Results from both the methods suggest
positive and

statistically significant impact of all the variables on economic growth. To test causal

relationships among the variables, Dumitrescu-Hurlin pairwise causality test is

employed. The

results indicate existence of bi-directional causality between public debt and

economic growth.

One way causality is revealed from economic growth to electricity consumption and
from

economic growth to credit. The policy implication is that, the sub-national

governments in

India should not think public debt as a burden but expand it for productive spending
to reap

higher economic
 growth.

Keywords: Public Debt, Economic Growth, Public Finance, Panel

Analysis

1​
Prof. Asit Ranjan Mohanty is Professor in Finance, Xavier University and Chair Professor of Centre of

Fiscal Policy and

​ Taxation, email: ​[email protected] ​2 ​Dr. Bibhuti Ranjan Mishra is Research

Associate in Centre of Fiscal Policy and Taxation, Xavier University, Bhubaneswar.,

​ email:
[email protected]
Impact of Public Debt on Economic Growth: Evidence from Indian
States

1.
Introduction

The impact of public debt on economic growth has remained a key issue in the
academia. Over the

past decade and especially after the financial crisis in 2008, the level of public debt is
expanding

in international, national and sub-national level. Heavy dependence on public debt could
retard

investment and economic growth. The ‘debt overhang’ hypothesis mentions that if the
anticipated

external debt of a country is more than it’s repayment ability, then the increased cost of
servicing

debt can impede investment (Krugman, 1988). If a major chunk of foreign capital is
used for

interest payments, then a meagre amount will remain to finance for investment that
could constrain

growth. This is regarded as the crowding-out effect of public debt (Diaz-Alejandro,

1981).

However, another school of thought states that, if public debt is used in productive
activities, then

the economy may expand without creating any macroeconomic instability (Burnside and
Dollar,

2000). As far as public debt is concerned, broadly it could be divided into types, one is
external

debt and the other is domestic debt. The two types of debt may have distinct impact on
economic

growth. The rationale behind dependence on domestic debt is that it saves the home
country from

the adverse external shocks and foreign exchange risk, and helps in the progress of
domestic

financial markets (Barajas and Salazar, 1999, 2000). But, Beaugrand et al. (2002) are
of the view

that the cost of domestic debt is more than the cost of

external debt.

Most of the recent literature on public debt and growth nexus centered on non-linear
(inverted U-

shape) relationship between the two and estimation of the threshold limit of public debt
share to

GDP (See Smyth and Hsing, 1995; Blavvy, 2006; Reinhart and Rogoff, 2010; Reinhart
et al., 2012;

Kumar and Woo, 2010; Cecchetti et al., 2011; Checherita-Westphal and Rother, 2012;
Furceri and

Zdzienicka, 2012; Herndon et al., 2013; Chen et al., 2016). However, for a developing
country like

India and its underdeveloped states, our hunch is that the optimum level of public debt
has not yet

been reached. The average public debt to gross domestic product ratio across the
major 14 states

for different periods from 1980-81 to 2013-14, varies from 19.1% to 35.3%. Therefore, a
positive

relationship between public debt and economic growth is

expected.

Earlier studies mainly focused on the impact of external debt in economic growth and
therefore

neglected the role of domestic debt. Further, the analysis is limited to cross-country
analysis or
time series analysis. Hence, this study will explore on the impact of domestic debt on
economic

growth along with other control variables by using a panel data set of 14 major states
in India.

The present study tries to assess the impact of public debt on economic growth using a
production

function approach where other relevant inputs like credit and electricity are also taken
as

explanatory variables. Inclusion of the other relevant variables also helps in removing
omitted

variable bias. Using Dynamic OLS (DOLS) and Fully Modified OLS (FMOLS) the study
gets

positive and statistically significant impact of public debt, institutional credit and
commercial

electricity consumption on economic growth. For causality analysis Dumitrescu-Hurlin

panel

causality test is also

employed.

The remainder of this study is as follows. Section 2 gives a brief overview of literature.
Section 3

outlines the theoretical framework and data used in the study. Econometric
methodology applied

in the analysis is presented in Section 4. Section 5 outlines the results and finally
section 6 provides

the conclusion and policy implication of the

study.

2. Brief overview of
Literature

2.1. Public Debt and Economic

Growth

Recent studies dealing with nexus between external debt and growth found that, the
relationship

between the two could be non-linear (inverted-U type shape). This means there could
be threshold

limit up to which debt can induce growth and thereafter higher debt can reduce growth.
The

threshold limit of external debt is estimated to be 38.4 percent of GDP by Smyth and
Hsing (1995);

21 percent by Blavvy (2006); 85 percent by Cecchetti et al. (2011); 90-100 percent by

Checherita-

Westphal and ​Rother (2012); ​and 90 percent by Chen et al. (2016). On the other hand
the study by

Reinhart and Rogoff (2010) concludes that in advanced and emerging market
economies (EMEs),
debt to GDP ratio of about 90 percent is growth reducing. If the ratio is below 60 percent
then it

can retard economic growth in only EMEs. Subsequently, Herndon et al. (2013) try to
replicate

the study by Reinhart and Rogoff (2010). By making some correction they find that the
relationship

between debt ratio and economic growth is similar in the two situations. So far as
causality between

public debt and economic growth is concerned, Panizza and Presbitero (2014) do not
find any

causality between the two, whereas Puente-Ajovín and Sanso-Navarro get bi-directional
causality

between public debt and economic growth. Lof and Malinen (2014) get support of one
way

causality from growth to

debt.
2.2.Credit and Economic
Growth

It is well known that financial deepening results in higher growth through different
channels like

more credit with financial liberalization promotes investment and innovation resulting in
more

efficient investment and thereby growth. In the literature various studies are done to
explore the

relationship between financial development and economic growth.​3 ​One strand of

studies

including Goldsmith (1969), focuses to measure the strength of the relationship

between the

financial development and economic performance. Others try to identify the channels
through
which the two are related. Pioneering work of McKinnon (1973) and Shaw (1973)
reflects that

financial liberalization positively affects saving and therefore more investment

culminating in

higher economic growth. Later on various papers using the endogenous growth models
take

financial development as a physical capital generating technological progress and

increasing the

efficiency in investment (See Bencivenga and Smith, 1991; Greenwood and Jovanovic,
1990). But

not all economists are convinced with the role of credit in generating growth. Robinson
(1952)

states that ‘where enterprise leads, finance follows’ which means economic growth
creates the

condition of financial arrangement and thereby financial development. Lucas (1988) is

of the view

that the role of financial development in growth is ‘over

stressed’.

Early papers find a positive relationship between financial development and economic
growth

using cross-country analysis (See King and Levine, 1993; Demirguc-Kunt and
Maksimovic, 1996;

Levine and Zervos, 1998; among others). But these studies do not deal with causality
analysis and

do not pays any attention to time series properties of the data. Ram (1999) gets
evidence of weak

negative relationship between financial development and economic growth and opines
that cross-

country studies have various limitations like heterogeneity issue in slope coefficients of
various
countries. Moreover, the results based on these type of analysis is sensitive to the
sample of

countries, computation method, frequency of data, functional form of the relationship

etc. So the

reliability of cross-country analysis is questioned by Khan and Senhadji (2003), Chua

and Thai

(2004), and Hassan et al. (2011) etc. Most of the time series studies get evidence of
causality

between credit and economic growth with no consensus on the direction of causality
(See

Demetriades and Hussein, 1996;; Luintel and Khan, 1999; Bell & Rousseau, 2001;
Calderon and

Liu, 2003; Bhattacharya and Sivasubramanian, 2003 and Liang and Teng, 2006 among
others).

3​
See Levine for a detailed survey on financial development-growth

nexus.
Few studies get no evidence of causality between the two (see Eng and Habibullah,
2011 and

Mukhopadhyay et al.,
2011).

To overcome the limitations of time series and cross-section analysis, various studies
use panel

data to assess the nexus between the financial development and economic growth (See
Levine et

al., 2000; Christopoulos and Tsionas, 2004; Beck and Levine, 2004; Hassan et al.,
2011, ​Gaffeo

and Garalova, 2014. ​etc.). Recent study by Kar et al. (2011) could not get any clear cut
relationship

between financial development and growth as the results are country specific in Middle
East and

North African (MENA) countries. However, two studies by Arestis et al. (2014) and
Valickova et

al. (2014) find evidence of positive relationship between financial development and
economic

growth by employing
meta-analysis.

2.3.Electricity Consumption and Economic

Growth

In the production process of an economy energy plays an important role. Among

various energy

variables electricity is most flexible form and a key infrastructural input for development.
There

are a number of studies exploring the relationship between consumption of electricity

and

economic growth. In studying the relationship between the two, the prime focus was on
the

causality issue since the publication of Kraft and Kraft (1978). The causality analysis is
important

in making policy decision because if unidirectional causality is found from growth to

electricity

consumption, then conservation policies can be implemented without hurting economic

growth.

In contrast to this if unidirectional causality runs from electricity consumption to growth,

then any

strategy to reduce electricity consumption may reduce growth. The causality from
growth to

electricity could be justified on the ground that demand for electricity will increase with
upsurge

in population, rapid urbanization and industrialization, and rise in standard of living.

Similarly, the

causality from consumption to electricity could be thought up as electricity is a major

infrastructure

and an input for production. Causality from economic growth to electricity is supported
by many

studies (see Kraft and Kraft, 1978; Ghosh, 2002; Narayan and Smyth, 2005; Mozumder
and

Marathe, 2007; and Jamil and Ahmad, 2010; Ciarreta and Zarraga, 2010; and Shahbaz
and Feridun,

2012 etc.). On the other hand, the reverse causality from electricity consumption to
growth is

revealed by other empirical papers (see Hsiao, 1981; Stern, 1993; Aqeel and Butt,
2001; Shiu and

Lam, 2004; Altinay and Karagol, 2005; Lee and Chang, 2005; Narayan and Singh,
2007; Yuan et

al., 2007; Tang, 2008, Odhiambo, 2009; and Chandran et al., 2010 etc.). Some other
find
bidirectional causality between electricity consumption and growth (See Yang, 2000;
Jumbe,

2004; Zachariadis and Pashourtidou, 2007; Lean and Smyth, 2010; Tang et al., 2013;
Osman et

al., 2016 among others). Some studies report no causality between the two (e.g. Ozturk
and

Acaravci, 2011; Yoo and Kwak, 2010; and Wolde-Rufael, 2006; etc.). Thus, the
relationship

between electricity consumption and economic growth is

ambiguous.

3. Theoretical Framework and

Data

In the growth literature, various schools of thoughts propose distinct relationships

between public
debt and economic growth. In the Classical sense, Ricardian Equivalence states that if
a

government borrows today, then, it has to repay this borrowing, in future by raising
taxes above

the normal level and the impact of debt on growth will be neutralized (Ricardo, 1817). In
the

Keynesian framework, foreign aid or foreign investment is required to fill the

saving-investment

gap (Todaro and Smith, 2003). Solow (1957) maintained that in the short-run fiscal
policy can

have some impact on level of per-capita income but in the long-run the impact is
neutral. In a neo-

classical set up Diamond (1965) formally brought the public debt as a variable
explaining growth.

He opines that internal debt reduces the available capital stock due to substitution of
public debt

for physical capital. As per the endogenous growth models, both fiscal and monetary
policies play

a crucial part in determining potential economic growth. Public debt can result in
technical

progress and thereby can influence growth (Villanueva, 1972). But ​Saint-Paul (1992)
using the

endogenous growth caveat states that higher debt is always associated with lower growth. ​The
recently

developed new growth theories explore the relationship of public debt and growth nexus
by

bringing utilization and governance aspects of public debt (See Zak and Knack, 2001
and

Acemoglu and Robinson,

2006).
As per our objective, the study relies on recently developed panel data analysis as it
has many

advantages over the pure cross-sectional or time-series analysis as noted by Osman et

al. (2016).

The study employs annual data for the period 1980-81 to 2013-14 for 14 major
non-special

category states in India. The key variables in the study include Gross State Domestic
Product

(GSDP) proxy used for real income, real public debt, real institutional credit to private
sector, and

commercial consumption of electricity (in Gigawatt). All the variables are transformed
into natural

logarithms prior to estimation and respectively denoted as LRY, LRD, LRC, and LCCE.
As State

wise data on private investment is not available for India, commercial bank credit to
private sector
is taken as a proxy for private investment. Nominal values of public debt and credit to
private

sector by commercial banks are deflated by the GSDP deflators of the respective states
to obtain

real values of the concerned variable. All variables except electricity are in Rs. crore at
2004-05

prices. Real GSDP data is collected from National Account Statistics published by
Central

Statistical Organization. Public debt and Credit variables are collected from State
Finance: A Study

of Budgets, published by the Reserve Bank of India and electricity data is taken from
EPWRF and

indiastat database. Institutional credit to private sector is a proxy for financial

development and
electricity consumption is a proxy for energy
use.

Table 1. Descriptive
Statistics
LRY LRD LRC LCCE Cross-section 14 14 14 14
Time Series 34 34 34 34 Observation 476 476 476 476 Mean 11.57 10.12 10.12 6.72
Median 11.53 10.05 9.97 6.70 Maximum 13.71 11.86 13.89 9.55 Minimum 10.04 8.22
7.58 3.86 Std. Dev. 0.74 0.80 1.18 1.15 Skewness 0.32 0.13 0.59 -0.02 Kurtosis 2.49
2.25 3.07 2.46 Note: Variables are in Natural Log.

Table 2. Correlation
Matrix
LYR LRD LRC LCCE LYR 1.000
----
-
----
-

LRD 0.895 1.000

43.680 ----- 0.000 -----

LRC 0.953 0.816 1.000

68.392 30.690 ----- 0.000 0.000 -----

LCCE 0.915 0.801 0.905 1.000

49.219 29.107 46.236 -----
0.000 0.000 0.000 ----- Note: Figures are correlation, t-statistics
and prob. respectively.

Table 1 depicts the descriptive statistics of all the logarithmic transformed variables. The
standard

deviation of all the variables are revealing that the data of all the series are dispersed
around the

mean. It permits us to move forward and use the data for further
estimation.
Table 2 presents the correlation matrix. The correlation coefficient between all variables
are very

high and they are intertwined which provides us the clue to estimate their
relationship.

4. Econometric
Methodology:

4.1.The
Model

Following, a neo-classical production function framework, the functional form of the

proposed

Model can be written as

follows,

LRY = f(LRD,LRC,LCCE) (1)

Where, LRY = log of real

GSDP,

LRD = log of real

debt,

LRC = log of real credit

and

LCCE = log of commercial consumption of

electricity.

The Model can be written

as:

LRY​it =
​ β​0 +
​ β​1​LRD​it +
​ β​2​LRC​it +
​ β​3​LCCE​it +
​ ε​it (2)
​

Where, ε​it =
​ μ​i +
​ θ​it​,

μ​i ∼​ (0,σ​μ​2​) and θ​it ∼​ (0,σ​θ​2​) are assumed to be independent of each other and among
themselves. μ​i and
​ θ​it​signify country specific fixed effects and time variants effects,
respectively. The subscripts ​i ​and ​t ​denote state (i = 1...14) and time period considered(t
= 1980/

81......2013/14), respectively.

The coefficients β​1​, β​2 ​and β​3 ​are the long-run elasticity estimates of gross income with
respect to public debt, credit, commercial electricity consumption, respectively. The
coefficient of all the

explanatory variables are expected to be

positive.
Any inference done on regression analysis by applying non-stationary series could lead
to spurious

regression. Therefore, unit-root testing and co-integration analysis are useful tools for
empirical

analysis. If a linear combination of two or more non-stationary series is stationary, then

long-run

cointegration relationship can be established among the variables. Testing for unit-roots
and

cointegration in panel data is helpful in establishing relationships among variables.

Maddala and

Wu (1999) mention that testing unit-root and cointegration in panel data increases the
power of

the test as compared to the respective tests done in Time series data. This paper uses
four steps in

its analysis: 1. Check for stationary properties of the data using panel unit root test; 2. if
the series

are stationary then test for the cointegration relationship, 3. in case the series are
cointegrated,

FMOLS and DOLS Methods are applied to measure the elasticity of income with
respect to public

debt, credit, and electricity consumption. 4. Causality analysis using Dumitrescu-Hurlin

procedur
e.

4.2.Panel Unit Root

Test:

Panel unit-root test can be done with two types of specification, one with a common or

homogeneous unit-root process and the other with an individual or heterogeneous

unit-root

process. The former specification can be implemented by Levin, Li and Chu (2002)
(LLC),

Breitung (2000), and Hadri (1998) panel unit root tests, while the latter specification can
be

fulfilled by Im, Pesaran and Shin (2003) (IPS), Fisher-ADF and Fisher-PP panel unit
root tests

(Maddala and Wu, 1999). This paper employs four methods of panel unit root tests that
are LLC,

Breitung, IPS and Fisher-PP Test i.e. two tests from each
categories.

The ADF specification of LLC panel unit root test is given

below.
+ δ​i​t + ∅​t +
​ u​it (3)
​
n​∆Y​i,t =
​ α​i +
​ ρY​i,t−1 ​+ ∑ φ​k​∆Y​i,t−k

k=1

This form allows two-way fixed effects i.e. the intercept varies over state as well as time
period,

captured by α​i and

​ ∅​t respectively.
​ These two coefficients are basically (i − 1) state
dummies and (t − 1) time dummies, respectively. The coefficient of lagged Y​i is
​
constrained to be homogeneous across all units of the panel. The null hypothesis H​0
and alternative hypothesis H​1 ​are H​0​: ρ = 0 and H​1​: ρ < 0.
In making the relevant standardization, Breitung (2000) method is different from LLC in
two

counts. Firstly, it removes the autoregressive portion but not the deterministic portion of
the ADF

equation. Secondly, the proxies for standardization are transformed and de-trended. He
considers

the following form to estimate the panel unit root

test.
k=1

n+1 ​Y​it ​= α​it ​+ ∑ β​ik​X​i,t−k + u​t (4)

​

With H​0​: ∑ ​n+1 ​k=1 β​ ​ 1 = 0 (non-stationary) and H​1​: ∑ ​n+1 ​k=1 β​

​ ik − ​ ik −
​ 1 < 0 (stationary)
for all i.

IPS (Im, Pesaran and Shin 2003) test allows heterogeneity on the coefficient of the
lagged Y​i variable
​ by extending the work of LLC. The IPS procedure of testing panel unit
root is based on

the following
model.
+ δ​i​t + ∅​t +
​ u​it (5)
​
n​∆Y​it =
​ α​i +
​ ρ​i​Y​i,t−1 +
​ ∑ φ​k​∆Y​i,t−k

k=1

With H​0​: ρ​i ​= 0 for all i and H​1​: ρ​i ​< 0 for at least one i. The t-statistic, applicable for a
balanced panel can be obtained by t = 1/N∑ ​N​t=1 ​t​ρi ​. Here, t​ρi ​denotes an individual ADF
t-

statistic to test H​0​: ρ​i =

​ 0 for all
i.

Another alternative method of panel unit root tests applies Fisher’s (1932) results to
develop tests

which combine the p-values from individual unit root tests. This Method is suggested by
Maddala
and Wu, and by
Choi.

If π​i ​is defined as the p-value from any individual unit root test for cross-section ​i​, then
under the null of unit root for all N cross section, the asymptotic result would be,

​
N​−2∑log(π​i​)
(6)

i=1

In the present paper we have reported the asymptotic χ​2 ​statistics using Phillips-Perron
individual

unit root tests. Choi’s result of standard normal statistics are not reported due to
similarity of

results. The null and alternative hypothesis for the Fisher-PP test are same as
the IPS test.
The tests are done with model with intercept for all tests except the Breitung test. AIC
criteria is

followed to select lag length. Bandwidth is selected by taking Newey-West method

using Barlett-

Kernel spectral
technique.

4.3.​Panel co-integration
Test:

Two varieties of co-integration test is done by following Pedroni (1999, 2004) and Kao
(1999).

Pedroni considers the following

regression:
+ ε​it (7)
​
M​Y​it =
​ α​i +
​ δ​i​t + ρY​i,t−1 +
​ ∑ β​mi​X​mi,t

m=1
In this model the X and Y are assumed to be stationary at first difference or I(1). Pedroni
test is a

residual based test where the residual obtained from equation (6) is tested for
stationarity by

estimating the following auxiliary

regression:

ε​it =
​ ρ​i​ε​i,t−1 +
​ u​it (8)
​

Or, ε​it = ​ ∑ ​ρ​j=1 ​iω​

​ ρ​i​ε​i,t−1 + ​ ij​∆ε​i,t−j +
​ v​it (9)
​

for each
cross-section.

Co-integration statistics given by Pedroni is divided into two categories. The first
category consists

of Panel v-statistic, Panel rho-statistic, Panel PP-statistics, Panel ADF-statistics which

are based

on pooling along the ‘with-in’ dimension. It is done by pooling the autoregressive

coefficients

across the different sections of the panel for the unit-root test on the residuals. The
second category

is formed by Group rho-statistic, Group PP-statistic, Group ADF-statistic which are

based on

pooling the ‘between’ dimension. It is done by averaging the autoregressive coefficients

for each

member of the panel for the unit-root test of the residuals. The Kao cointegration test is
also a

residual based test. It distinguishes from the Pedroni test in specifying cross-section
specific

intercepts and homogeneous coefficients on the panel

regressors.
 4.4.Estimation of Panel
Co-integration

Once a cointegrating relationship between the variables is established, it is useful to

estimate the

long-run parameters. In the presence of cointegration, OLS estimates give spurious

coefficients.

Therefore a number of alternative estimators are proposed such as DOLS and FMOLS
among
others. The major weakness of DOLS estimator is that it does not take care of the
cross-sectional
heterogeneity issue. Therefore, Pedroni (2000) suggested to use the FMOLS estimator
that deals
with the cross sectional heterogeneity, endogeneity and serial correlation problems. In
small
samples the FMOLS is believed to give consistent estimates. First we estimated the
DOLS
estimator and then to check the robustness of the results FMOLS estimator is applied.
Following, Mark and Sul (1999), the study applies a simple weighted DOLS estimator
that permits
for heterogeneity in the long-run variances. Similarly, feasible pooled (weighted)
FMOLS
estimator developed by Pedroni (2000) and Kao and Chiang (2000) is used which
considers
heterogeneous cointegrated panels with differences in long-run variances across
cross-sections.
4.5.Dumitrescu-Hurlin causality test:
After getting long-run coefficient of three explanatory variables determining the real
income, it is
useful to draw information on the causal link between them which can have greater
policy
implications. Therefore, this study also attempts to explore on the causal link between
the variables
by applying Dumitrescu-Hurlin test. This test has two advantages over the Granger
(1969)
causality test in that in addition to the fixed coefficient accounted in Granger causality
test, it
considers two dimensions of heterogeneity. One for the regression model used to test
Granger
causality and the other is heterogeneity in the causal relationship. The detailed
derivation of the
Dumitrescu-Hurlin test can be found in Dumitrescu and Hurlin (2012).
5. Results and Implications
5.1. Panel Unit Root Test:
The empirical analysis starts with the testing of panel unit root in log-levels of real
income, real
public debt, real credit and electricity consumption by following four varieties of panel
unit root
tests viz. LLC, Breitung, IPS and Fisher-PP panel unit root test.
Table 3. Panel Unit Root Test
Variable
LLC Breitung IPS Fisher-PP
(Chi-square
(t-statistics) (t-statistics) (W-Statistics) ​
Statistics)
Levels LRY 7.7235 2.9181 11.6800 0.0736
LRD -2.2760** 1.0919 1.7309 19.4651
LRC 4.4096 1.6611 9.0721 0.4573 LCCE 2.4438 3.6317 5.6921 20.6416 First Differences
LRY -14.0529* -3.3658* -14.2520* 344.643* LRD -6.8155* - 7.2473*
-7.5748* 137.297* LRC -6.0747* -5.4904* -5.4913* 186.949* LCCE
-16.1223* -5.2121* -17.1570* 337.172*

Note. 1. All tests are done on the model with an intercept except the Breitung test which is
done for a model of intercept with trend. 2. Lag Lengths are set based on the AIC criterion. 3.
For LLC and PP- Fisher test Barlet Spectral Method with Newey-West Bandwidth is applied.
4. * and ** depicts statistical significance at 1% and 5% level, respectively.

The results of panel unit root tests are reported in Table 3. It demonstrates that for the
log-level

series, the null of non-stationarity in variables could not be rejected for all the variables
except the

real debt variable. Real debt is found to be stationary in level by only the LLC test,
which is refuted

by other Panel Unit root tests. First difference of the log-levels variables are found to be
stationary
at 1 percent level of significance. Thus, the results indicate that the four variables
contain a panel

unit root in
level.

5.2. Panel
Co-integration:

As the panel unit root test results reveal that all the variables are difference stationary,
we proceed

to check the cointegration or long-run association of the variables by using Kao test and
Pedroni

Test. The model for Kao Test assumes no deterministic trend, while model for Pedroni
Test

assumes deterministic trend with intercept.​4 ​Lag length is set by AIC criteria and Barlett
spectral

estimation procedure is applied with Newey-West automatic bandwidth. Table 4 reveals

that both

the Kao and Pedroni Test of cointegration suggest long-run relationship among the four
variables

at 1% level of significance with the exception of group-rho (between dimension) statistic

in

Pedroni
test.

Table 4. Panel Co-integration Tests: Kao Test and Padroni Test ​Method Statistic Kao
Residual Cointegration Test ADF Stat. - 6.645*

Pedroni residual cointegration test Panel v-Statistic 8.151* Panel rho-Statistic -2.410*

4​
We could also reject the null of no cointegration by Panel ADF-test (with-in dimension) and Group ADF

statistics (between-dimension)
​ using Pedroni cointegration test with the assumption of no deterministic
trend.
 Panel PP-Statistic -6.315* Panel ADF-Statistic -6.625* Group rho-Statistic -0.223 Group
PP-Statistic -4.881* Group ADF-Statistic -6.676* Note: 1. Trend assumption [Kao: no
deterministic trend, Pedroni: Deterministic trend with intercept]; H​0​: No Co-integration; 2.
Newey–West automatic bandwidth selection and Bartlett kernel. 2. * depicts statistical
significance at 1% level.

5.3. Panel Long-run

Estimates

After establishing the cointegration relationship among the variables we proceed to

estimate the

long-run elasticity using DOLS and check the robustness by applying FMOLS method.
In the

DOLS method one lead and one lag is taken. Trend specification is set with no
deterministic trend

and the panel option is set as Pooled (weighted) in both the methods. The long-run
variance

weights are derived by Barlett-kernel method with Newey-west automatic bandwidth

and NW

automatic lag in both the

method.

Table 5 depicts the results from DOLS estimates. It reveals that public debt, institutional
credit

and electricity consumption are significantly affecting income. The coefficients of

explanatory

variables are elasticity. The coefficient for public debt in explaining income is found to
be 0.33

which can be interpreted as a 1% rise in public debt will increase the income by 0.33%.
The

positive relationship between the public debt and economic growth could be explained
by the fact

that, at State level, the threshold limit of debt to GSDP is not yet been reached. The
coefficient for

credit and electricity variables are found to be 0.34 and 0.07 respectively. It is
noticeable that the

impact of public debt and institutional credit on income is similar and the impact of
electricity

consumption is very low. This is due to the fact that the share of electricity consumption
to GSDP

in each State is quite

small.

Table 5. Results from DOLS Estimation (Dependent variable LRY) ​Variable Coefficient
Std. Error t-Statistic Prob. LRD 0.3306 0.0346 9.5419 0.0000 LRC 0.3390 0.0320 10.6031
0.0000 LCCE 0.0729 0.0198 3.6836 0.0003 R-squared 0.9893 Mean dependent var.
11.5894 Adjusted R-squared 0.9841 S.D. dependent var. 0.7076 S.E. of regression 0.0891
Sum squared resid. 2.3111
Long-run variance 0.0129

Table 6 gives the estimated results obtained from FMOLS estimate. The results fairly
resemble

with the DOLS Method though the coefficients of various explanatory variables slightly
differ. As

compared to the DOLS, FMOLS assigns a slightly low coefficient to public debt variable
and in

turn the coefficients of the credit variable in explaining income has increased. All the
explanatory

variables are affecting income positively and significantly. In this case also the elasticity
of income

on commercial electricity consumption is

low.

Table 6. Results from FMOLS Estimation (Dependent variable LRY) ​Variable Coefficient
Std. Error t-Statistic Prob. LRG 0.2626 0.0097 26.9531 0.0000 LRC 0.3615 0.0115 31.4851
0.0000 LCCE 0.0734 0.0070 10.5364 0.0000 R-squared 0.9805 Mean dependent var
 11.5950 Adjusted R-squared 0.9798 S.D. dependent var 0.7331 S.E. of regression 0.1041
Sum squared resid 4.8253 Long-run variance 0.0085

5.4. Dumitrescu-Hurlin pairwise causality

test:

The result of pairwise Dumitrescu-Hurlin panel causality test is reported in Table 7. The
result

suggests that bi-directional causality exists between public debt and economic growth
which is in

conformity with Puente-Ajovin and Sanso-Navarro (2015)​. ​Only, unidirectional causality

is found

from economic growth to credit at 10% level of significance. One way causality from
economic

growth to electricity consumption is also

established.

Table 7. Results from Pairwise Dumitrescu-Hurlin Panel

Causality Tests ​Sample: 1980 2013 Lags: 2
Null Hypothesis: W-Stat. Zbar-Stat. Prob. LRD does not homogeneously cause LRY 3.9778
2.9063 0.0037 LRY does not homogeneously cause LRD 3.8930 2.7708 0.0056 LRC does
not homogeneously cause LRY 2.7907 1.0084 0.3133 LRY does not homogeneously cause
LRC 3.2070 1.6739 0.0941 LCCE does not homogeneously cause LRY 2.3116 0.2424
0.8085 LRY does not homogeneously cause LCCE 14.4135 19.5908 0.0000 LRC does not
homogeneously cause LRD 3.4730 2.0992 0.0358
LRD does not homogeneously cause LRC 8.6022 10.2997 0.0000 LCCE does not
homogeneously cause LRD 2.4535 0.4692 0.6389 LRD does not homogeneously cause
LCCE 8.6784 10.4216 0.0000 LCCE does not homogeneously cause LRC 3.5864 2.2805
0.0226 LRC does not homogeneously cause LCCE 9.6580 11.9877 0.0000

6. Conclusion and Policy

Implication

This study uses panel data of 14 major (non-special category) States in India during the
period
1980-81 to 2013-14 to examine the influence of public debt on economic growth by
controlling

other relevant variables like institutional credit and commercial electricity consumption.
After

establishing long-run relationship among the variables, DOLS (pooled weighted)

Method is

applied to derive the elasticity of size of the economy (income) on public debt, credit
and

consumption of electricity. The study finds that economic growth is significantly and
positively

affected by public debt and credit. The impact of public debt and institutional credit are
high and

similar. The influence of electricity consumption is found to be low. To check for

robustness,

FMOLS (pooled weighted) estimates are also reported which give similar results.
Dumitrescu-

Hurlin pairwise causality test reveals existence of bi-directional causality between public
debt and

economic growth. One way causality is revealed from economic growth to electricity
consumption

and from economic growth to

credit.

The analysis reveals that at State level, expansionary debt policy will be helpful for the
economy

in generating higher economic growth. High economic growth will further increase the
provision

of institutional credit to private sector and increase the demand for energy consumption
for

commercial purpose. One can expand the paper by exploring on the composition of
public debt
and its impact on economic
growth.

Referenc
es

Acemoglu, D. and Robinson, J.A. (2006), ​Economic Origins of Dictatorship and Democracy,​
Cambridge University Press, Cambridge.

Altinay, G., Karagol, E. (2005), “Electricity consumption and economic growth: evidence from
Turkey”, ​Energy Economics,​ Vol. 27, pp. 849–856.

Aqeel, A., Butt, M.S. (2001), “The relationship between energy consumption and economic
growth in Pakistan”, ​Asia-Pacific Development Journal,​ Vol. 8, pp. 101–109.
Arestis, P., Chortareas, G. and Magkonis, G. (2015), “Financial Development and Economic
Growth: A Meta-Analysis”, ​Journal of Economic Surveys, V ​ ol. 29 No.3, pp. 549–565.

Barajas S. and Salazar, N (1999), “Interest spreads in banking in Colombia”, IMF Staff Papers,
46, pp. 196- 224.

Barajas, S. and Salazar, N. (2000), “The impact of liberalization and foreign investment in
Colombia’s financial sector”, ​Journal of Development Economics,​ Vol. 46 No. 2, pp. 157-196.

Beaugrand, P., Loko, B. and Mlachila, M. (2002), “The choice between external and domestic
debt in financing budget deficits: the case of central and West African countries”, IMF Working
Paper No. 02/79, IMF, Washington, DC.

Beck, T., & Levine, R. (2004), “Stock markets, banks, and growth: Panel evidence”, ​Journal of
Banking and Finance​, Vol. 28, pp. 423–442.

Bell, C., & Rousseau, P. L. (2001), Post-independence India: A case of finance-led

industrialization​?,​ ​Journal of Development Economics​, Vol. 65, pp. 153–175.

Bencivenga, Valerie and Bruce D. Smith (1991), “Financial intermediation and endogenous
growth”, ​The Review of Economic Studies, ​Vol. 58 No.2, pp. 195-209.

Bhattacharya, P.C and Sivasubramanian, M. N. (2003), “Financial development and economic

growth in India: 1970–1971 to 1998–1999”, ​Applied Financial Economics, ​Vol.13 No. 12, pp.
925-929.

Blavy, R. (2006), “Public debt and productivity: the difficult quest for growth in Jamaica”, IMF
Working Paper No. 06/235, IMF, Washington, DC.

Breitung, J. (2000), “The local power of some unit root test for panel data”, ​Advances in
Econometrics,​ Vol. 15, pp. 161–77.

Burnside, C. and Dollar, D. (2000), “Aid, policies, and growth”, ​American Economic Review​,
Vol. 90 No. 4, pp. 847-868.

Calderon, C., & Liu, L. (2003), “The direction of causality between financial development and
economic growth”, ​Journal of Development Economics,​ Vol. 72, pp. 321–334.

Cecchetti, S., Mohanty, M.S. and Zampolli, F. (2011), “The real effects of debt”, presented at
the Achieving Maximum Long-Run Growth Symposium sponsored by the Federal Reserve
Bank of Kansas City, Jackson Hole, Wyoming, 25-27 August.

Chandran, V.G.R., Sharma, S., Madhavan, K. (2010), “Electricity consumption-growth nexus:

​ ol. 38, pp. 606–612.
the case of Malaysia”, ​Energy Policy, V

Checherita-Westphal, C., Rother, P. (2012), “The impact of high government debt on economic
growth and its channels: an empirical investigation for the euro area”, ​European Economic
Review​, Vol. 56 No. 7, pp. 1392–1405.
Chen, C., et al. (2016), “Optimal government investment and public debt in an economic growth
model”, ​China Economic Review,​ http://dx.doi.org/10.1016/j.chieco.2016.08.005

Christopoulos, D.K. and Tsionas, E.G. (2004), “Financial development and economic growth:
evidence from panel unit root and cointegration tests”, ​Journal of Development Economics​,
Vol.73, pp. 55-74.

Chuah, H., & Thai, W. (2004), “Financial development and economic growth: Evidence from
causality tests for the GCC countries”, IMF Working Paper.

Ciarreta, A., Otaduy, J., Zarraga, A. (2009), “Causal relationship between electricity
consumption and GDP in Portugal: a multivariate approach”, ​Empirical Economics Letters​, Vol.
8, pp. 693–701.

Demetriades, P.O. and Hussein, K.A. (1996), “Does financial development cause economic
growth? Time- series evidence from 16 countries”, ​Journal of Development Economics​, Vol. 51,
pp. 387-411.

Demirguc-Kunt, A. and Maksimovic, V. (1996), “Financial constraints, use of funds and firm
growth: an international comparison”. Working Paper. World Bank.
Diamond, P.A. (1965), “National debt in a neoclassical growth model”, ​The American Economic
Review​, Vol. 55 No. 5, pp. 1126-1150.

Diaz-Alejandro, C. (1981), Southern Cone Stabilization Plans, Economic Stabilization in

Developing Countries, The Brookings Institution, Washington, DC.

DiPeitro, W.R. and Anoruo, E. (2012), “Government size, public debt and real economic growth:
a panel analysis”, ​Journal of Economic Studies,​ Vol. 39 No. 4, pp. 410 – 419.

Dumitrescu EI, Hurlin C. (2012), “Testing for granger non causality in heterogeneous panels”.
Economic Modeling,​ Vol. 29, pp. 1450–60.

Eng, Y., and Habibullah M. S. (2011), “Financial Development and Economic Growth Nexus:
Another look at the Panel Evidence from Different Geographical Regions”, ​Bank and Bank
Systems,​ Vol. 6 No. 1, pp. 62-71.

Furceri, D., Zdzienicka, A. (2012), “How costly are debt crises?”, ​Journal of International Money
and Finance,​ Vol. 31 No. 4, pp. 726–742.

Gaffeo, E. & Garalova, P. (2014), “On the finance-growth nexus: additional evidence from
Central and Eastern Europe countries”, ​Economic Change Restructuring,​ Vol. 47 No. 89. doi:
10.1007/s10644-013- 9143-x.

​ ol.
Ghosh, S. (2002), “Electricity consumption and economic growth in Taiwan”, ​Energy Policy, V
30, pp. 125–129.

Goldsmith, Raymond W. (1969), ​Financial Structure and Development​. New Haven, CN: Yale
University Press.
Granger CWJ. (1969), “Investigating causal relations by econometric models and cross spectral
methods”. ​Econometrica​, Vol. 37, pp. 424–38.

Greenwood, J. and Boyan J. (1998), “Financial development, growth, and the distribution of
income”, ​Journal of Political Economy,​ Vol. 98 No. 5, pp. 1076-l 108.

Hadri, K. (1998), ​Testing for stationarity in heterogeneous panel data​.

Mimeo.

Hassan, M.K., Sanchez, B., and Yu Jung-Suk (2011), “Financial development and economic
growth: New evidence from panel data”, ​The Quarterly Review of Economics and Finance ​Vol.
51, pp. 88–104.
Herndon, T., Ash, M. and Pollin, R. (2013), “Does high public debt consistently stifle economic
growth? A critique of Reinhart and Rogoff”, PERI Working Paper No. 322, Political Economy
Research Institute, Amherst.

Im, K.S., M.H. Pesaran, and Y. Shin. (2003), “Testing for unit roots in heterogeneous panels”,
Journal of Econometrics, ​Vol. 115, pp. 53–74.

Jamil, F., Ahmad, E. (2010), “The relationship between electricity consumption, electricity prices
and GDP in Pakistan”, ​Energy Policy,​ Vol. 38, pp. 6016–6025.

Jumbe, C.B.L., (2004), Cointegration and causality between electricity consumption and GDP:
empirical evidence from Malawi, ​Energy Economics​, Vol. 26, pp. 61–68.

Kao, C. (1999), “Spurious regression and residual-based tests for co-integration in panel data”,
J​ournal of Econometrics​, Vol. 90, pp. 1–44.

Kao, Chihwa and Min-Hsien Chiang (2000), “On the Estimation and Inference of a Cointegrated
Regression in Panel Data”, in Baltagi, B. H. et al. eds., ​Nonstationary Panels, Panel
Cointegration and Dynamic Panels​, 15, Amsterdam: Elsevier, pp. 179–222.

Kar, M., S. Nazilioglu, and H. Agir (2011), “Financial Development and Economic Growth Nexus
in the MENA Countries: Bootstrap Panel Granger Causality Analysis”, ​Economic Modelling,​ Vol.
28, pp. 685- 693.

Khan, M., & Senhadji, A. (2003), “Financial development and economic growth: A review and
new evidence”, ​Journal of African Economies,​ Vol. 12, pp. ii89–ii110.

King, R. G. and Levine, R. (1993), “Finance and growth: Schumpeter might be right”, ​Quarterly
Journal of Economics​, Vol. 108, pp. 717–37.

Kraft, J., Kraft, A., (1978), “On the relationship between energy and GDP”, ​Journal of Energy
Development​, Vol. 3, pp. 401–403.

Krugman, P. (1988), “Financing vs. Forgiving a debt overhang”, ​Journal of Development

Economics,​ Vol. 29 No. 29, pp. 253-268.
Kumar, M.S., Woo, J. (2010), “Public Debt and Growth”. IMF Working Papers 10/174
International Monetary Fund.

Lean, H.H., Smyth, R. (2010), “On the dynamics of aggregate output, electricity consumption
and exports in Malaysia: evidence from multivariate Granger causality tests”, ​Applied Energy,​
Vol. 87, pp. 1963–1971.
Lee, C.C., Chang, C.P. (2005), “Structural breaks, energy consumption, and economic growth
revisited: evidence from Taiwan”, ​Energy Economics​, Vol. 27, pp. 857–872.

Levin, A., C.F. Lin, and J. Chu. (2002), “Unit root tests in panel data: Asymptotic and finite
sample properties”, ​Journal of Econometrics​, Vol. 98, pp. 1–24.

Levine, R. (1997), “Financial development and economic growth: views and agenda”, ​Journal of
Economic Literature​, Vol. 35, pp. 688–726.

Levine, R. and Zervos, S. (1998), “Stock markets, banks and economic growth”, ​American
Economic Review​, Vol. 88, pp. 537–58.

Levine, R., Loyaza, N., Beck, T. (2000), “Financial intermediation and growth: causality and
causes”, ​Journal of Monetary Economics,​ Vol. 46, pp. 31–71.

Liang, Qi and Teng, J.Z. (2006), “Financial development and economic growth: Evidence from
China”. ​China Economic Review,​ Vol. 17, pp. 395–411.

Lucas, R. E. (1988), “On the mechanics of economic development”, ​Journal of Monetary

Economics,​ Vol. 22, pp. 3–42.

Luintel, K. B., & Khan, M. (1999), “A quantitative reassessment of the finance–growth nexus:
Evidence from a multivariate VAR”, ​Journal of Development Economics,​ Vol. 60, pp. 381–405.

Maddala, G.S., and S. Wu. (1999), “A comparative study of unit root tests with panel data and a
new simple test”, ​Oxford Bulletin of Economics and Statistics​, Vol. 61, pp. 631–52.

Mark, Nelson C. and Donggyu Sul. (2003), “Cointegration Vector Estimation by Panel DOLS
and Long- run Money Demand”, ​Oxford Bulletin of Economics and Statistics​, Vol. 65, pp.
655–680.

McKinnon, Ronald I. (1973), ​Money and Capital in Economic Development (​ Washington, DC:
Brookings Institution.

Mozumder, P., Marathe, A. (2007), “Causality relationship between electricity consumption and
GDP in Bangladesh”, ​Energy Policy​, Vol. 35, pp. 395–402.

Mukhopadhyay, B., Pradhan, R. P. and Feridun, M. (2011), “Finance-growth Nexus Revisited

for Some Asian countries”, ​Applied Economics Letters​, Vol. 18 No. 16, pp. 1527- 1530.

Narayan, P.K., Singh, B. (2007), “The electricity consumption and GDP nexus for the Fiji
Islands”, ​Energy Economics,​ Vol. 29, pp. 1141–1150.
Narayan, P.K., Smyth, R. (2005), “Electricity consumption, employment and real income in
Australia evidence from multivariate Granger causality tests”, ​Energy Policy​, Vol. 33, pp.
1109–1116.

Odhiambo, N.M. (2009), “Energy consumption and economic growth nexus in Tanzania: an
ARDL bounds testing approach”, ​Energy Policy,​ Vol. 37, pp. 617–622.

Osman, M., Gachino, G. and Hoque, A. (2016), “Electricity consumption and economic growth
in the GCC countries: Panel data analysis”, ​Energy Policy​, Vol. 98. Article in Press.

Ozturk, I., Acaravci, A. (2011), “Electricity consumption and real GDP causality nexus: evidence
from ARDL bounds testing approach for 11 MENA countries”, ​Applied Energy​, Vol. 88, pp.
2885–2892.

Pedroni, P. (1999), “Critical values for co integration test in heterogeneous panels with multiple
regressors”, ​Oxford Bulletin of Economics and Statistics,​ Vol. 61, pp. 653–78.

Pedroni, P. (2000), “Fully Modified OLS for Heterogeneous Cointegrated Panels”, in Baltagi, B.
H. ed. ​Nonstationary Panels, Panel Cointegration and Dynamic Panels​, 15, Amsterdam:
Elsevier, pp. 93–130.

Pedroni, P. (2004), “Panel co integration: Asymptotic and finite sample properties of pooled
time series tests with an application to the PPP hypothesis”, ​Econometric Theory,​ Vol. 20, pp.
597–625.

Puente-Ajovín, M. and Sanso-Navarro, M. (2015), “Granger causality between debt and growth:
Evidence from OECD countries “, ​International Review of Economics and Finance,​ Vol. 35, pp.
66–77.

Ram, R. (1999), “Financial development and economic growth: additional evidence”, ​Journal of
Development Studies, V​ ol. 35, pp. 164–74.

Reinhart, C.M. and Rogoff, K.S. (2010), “Growth in a time of debt”, ​American Economic Review,​
Vol. 100 No. 2, pp. 573-578.

Reinhart, C.M., Reinhart, V.R., Rogoff, K.S. (2012), “Public debt overhangs:
advanced-economy episodes since 1800”, ​Journal of Economic Perspectives,​ Vol. 26 No. 3, pp.
69–86.

Ricardo, D. (1817), ​On the principles of political economy and taxation,​ Cambridge University
Press, Cambridge.
Robinson, J. (1952), ​The generalization of the General Theory, in The Rate of Interest and
Other Essays​, Macmillan, London.

Saint-Paul, G., (1992), “Fiscal policy in an endogenous growth model”, ​Quarterly Journal of
Economics,​ Vol. 107 No. 4, 1243–1259.

Salahuddin, M., Gow, J., Ozturk, I. (2015), “Is the long-run relationship between economic
growth, electricity consumption, carbon dioxide emissions and financial development in Gulf
Cooperation Council countries robust?”, ​Renewable and Sustainable Energy Review​, Vol. 51,
pp. 317–326.
Shahbaz, M., Feridun, M. (2012), “Electricity consumption and economic growth empirical
evidence from Pakistan”, ​Quality & Quantity,​ Vol. 46 No. 5, pp. 1–17.

Shaw, Edward S. (1973), ​Financial Deepening in Economic Development,​ New York: Oxford
University Press.

Shiu, A., Lam, P.L., (2004), “Electricity consumption and economic growth in China”, ​Energy
Policy​, Vol. 32, pp. 47–54.

Smyth, D. and Hsing, Y. (1995), “In search of an optimal debt ratio for economic growth”,
Contemporary Economic Policy,​ Vol. 13 No. 4, pp. 51-59.

Solow, R. (1957), “A contribution to the theory of economic growth”, ​Quarterly Journal of

Economics,​ Vol. 70 No. 1, pp. 65-94.

Stern, D.I. (1993), “Energy and economic growth in the USA”, ​Energy Economics​, Vol. 2, pp.
137–150.

Tang, C.F., Shahbaz, M., Arouri, M. (2013), “Re-investigating the electricity consumption and
economic growth nexus in Portugal”, ​Energy Policy,​ Vol. 62, pp. 1515–1524.

Todaro, M.P. and Smith, S.C. (2003), ​Economic Development​, Addison Wesley Series, New
York, NY.

Valickova, P., Havranek, T. and Horvath, R. (2015), “Financial Development and Economic
Growth: A Meta-Analysis”, ​Journal of Economic ​Surveys, Vol. 29 No.3, pp. 506–526.

Villanueva, D.P. (1972), “Public debt, efficient labour, and growth: comment”, Journal of Money,
Credit and Banking, Vol. 4 No. 4, pp. 1016-1018.

Wolde-Rufael, Y. (2014), “Electricity consumption and economic growth in transition countries:

A revisit using bootstrap panel Granger causality analysis”. ​Energy Economics,​ Vol. 44, pp.
325–330.

Yang, H.Y. (2000), “A note of the causal relationship between energy and GDP in Taiwan”,
Energy Economics,​ Vol. 22, pp. 309–317.

Yoo, S.H., Kwak, S.Y. (2010), “Electricity consumption and economic growth in seven South
American countries”, ​Energy Policy​, Vol. 38, pp. 181–188.

Yuan, J., Zhao, C., Yu, S., Hu, Z. (2007), “Electricity consumption and economic growth in
China: cointegration and co-feature analysis”, ​Energy Economics,​ Vol. 6, pp. 1179–1191.

Zachariadis, T., Pashourtidou, N. (2007), “An empirical analysis of electricity consumption in

Cyprus”, ​Energy Economics​, Vol. 29, pp. 183–198.

Zak, P.J. and Knack, S. (2001), “Trust and growth”, ​Economic Journal​, Vol. 111 No. 470, pp.
295-321.