Dynamics between Power Consumption and Economic Growth at Dynamics between Power Consumption and Economic Growth at Aggregated and Disaggregated (Sectoral) Level Using the Aggregated and Disaggregated (Sectoral) Level Using the Frequency Domain Causality Frequency Domain Causality

: We investigated the Granger causal relationship between the consumption of power both at the aggregate and sectoral level and economic growth in India using the frequency domain approach, which would help policy makers seek the efﬁcient allocation of electricity via proper policy initiatives at different frequencies. We ﬁnd that at the aggregate level, unidirectional causality runs from the total power consumption to economic growth, starting from the second up to the seventh quarter. In the sectoral context, the results are different. Since there is no causality between industrial power consumption and economic growth; therefore, an energy conservation policy can thus be implemented for the industrial sector. Moreover, since a bidirectional causality exists after 15 quarters for the commercial sector, a short-term policy but not an energy conservation policy could also be initiated for this sector. In the industrial and agricultural sectors, a promotional policy should be initiated because a unidirectional causality exists from sectoral power consumption to economic growth. Therefore, different and sector-speciﬁc policies would be more appropriate than a single policy for all power sectors in India in order to orient the efﬁcient utilisation of power towards better economic development.


Introduction
The road to better economic growth is paved with an efficient investment and better utilisation of infrastructure.Being one of the key infrastructural components, power (electricity) requires an efficient allocation of capital and other factor utilised in production.Otherwise, this sector would lead to mounting costs due to a lack of economic competitiveness.Further, as a factor of production, electricity directly or indirectly complements other factors of production such as labour and capital in the production process.Thus, any shortages of electricity will disrupt the manufacturing sector of an economy, which consequently may lead to the destabilisation of economic growth (Costantini and Martini 2010).Since India is an energy-led-growth economy with a perennial problem of power deficit, it is imperative on the part of policy makers to set out measures for the efficient consumption of electricity across the consuming sectors in order to boost economic development 1 .For instance, Chontanawat et al. (2008); Payne (2010); and Ozturk (2010) argue that inefficient use of energy may negatively impact economic growth.For this reason, it is essential to know how each consuming sector of electricity contributes to economic growth.
The study contributes to the existing literature in four key areas.(i) It is the first study that investigates the degree of short-run and long-run causality across different time scales for each consuming sector of electricity, in addition to the total power consumption in India.
(ii) It employs the novel frequency domain causality methodology developed by Breitung and Candelon (2006) to sectoral and total power consumption.At different frequencies, the causality between power consumption and economic growth is estimated, which is missing in the studies conducted in the time domain.(iii) The other studies have so far had an electricity supply-side focus (i.e., causality between different sources of supply of electricity and economic growth), but our focus in this study was on the demand-side of electricity (i.e., consumption of electricity sector-wise, including domestic, industrial, commercial, and agricultural sectors).(iv) This study has policy implications for the power sector in India 2 , as country-specific studies would allow researchers to consider institutional, structural, and policy reforms undertaken in the economy (Chandran et al. 2010).In addition to this, it could help in prioritising the consumption of power across the consuming sectors.
Additionally, cross-country studies have some limitations which could be overcome through country-specific studies (Soytas and Sari 2009;Chang et al. 2001;Stern 2000).Our study finds a unidirectional causality from electricity consumption to economic growth at the aggregate level, starting after two quarters and continuing up to seven quarters during the period.However, the causality at the sectoral level diverges across the consuming sectors of electricity.A neutral effect exists between industrial (medium voltage (MV)) electricity consumption and economic growth.A bidirectional causality, in the long run, is found for both commercial electricity consumption and domestic electricity consumption, warranting an energy conservation policy for those two sectors in the short run only.Further, a unidirectional causality running from electricity consumption to economic growth is observed in the short run and the long run for the industrial (high voltage (HV)) and agricultural sectors, which supports the advancement of a promotional policy for energy consumption in these two sectors.
The remainder of the study is organised as follows: Section 2 reviews the literature on the causal relationship between electricity consumption and economic growth.Section 3 discusses the data and methodology.Section 4 analyses the empirical results.Section 5 concludes and provides policy implications.

Literature Review
The existing general literature on electricity consumption and economic growth can be presented in four different categories based on the corresponding individual hypotheses we put forth and the corresponding results they support.First, we present the feedback hypothesis, which states a bidirectional causality between the consumption of electricity and economic growth.This hypothesis states that a decline in the domestic consumption of electricity leads to a reduction in economic growth, and lower economic growth results in less power consumption.The empirical results validating this hypothesis are those supported by many studies, such as Masih and Masih (1996), Costantini and Martini (2010), Tang et al. (2013), Polemis and Dagoumas (2013), Bélaïd and Abderrahmani (2013), Nasreen and Anwar (2014), Mutascu (2016), andSarwar et al. (2017), among others.The policy implication of this hypothesis suggests that any policy measures aiming at reducing the use of energy or promoting energy efficiency have a detrimental effect on economic growth and vice versa.
Second, the growth hypothesis sanctions a unidirectional causality running from electricity consumption to economic growth.That means an increase in the consumption of electricity will enhance economic growth via a higher level of production.The empirical results supporting this hypothesis are confirmed by studies such as Murry and Nan (1994), Khan et al. (2007), Pradhan (2010), Ahamad and Islam (2011), Das et al. (2012), Tang and Shahbaz (2013), Wolde-Rufael (2014), Iyke (2015), Acaravcı et al. (2015), and He et al. (2017), among others.
The feedback and growth hypotheses emphasize the role of the consumption of energy (electricity) in economic development.However, the flip side of energy consumption is its effect on environmental pollution, which is also a cause of concern for policy makers.In the case of the two hypotheses, any policy aiming at conserving energy will hurt economic growth on the one hand, and reduce environmental degradation on the other hand.However, the result also suggests that for an energy-led growth economy, such as in India, policy measures should be directed towards the promotion of the use of renewable energy, which will take care of economic growth and environmental concerns as well.
Third, the conservation hypothesis supports a growth-led economy.This hypothesis is revealed through a unidirectional causality running from economic growth to electricity consumption, which reinforces the fact that electricity consumption has no causation to the growth of the economy.The empirical studies that support the conservation hypothesis include Cheng and Lai (1997), Aqeel and Butt (2001), Narayan and Singh (2007), Ho and Siu (2007), Hu and Lin (2008), Narayan and Prasad (2008), Ghosh (2009), Narayan et al. (2010), Mahmoodi and Mahmoodi (2011), Dogan (2014), Kasman and Duman (2015), Fang and Chang (2016), etc.Thus, the policy implication of the conservation hypothesis entails that instead of focusing on economic growth, policy measures should be oriented toward energy conservation and efficiency, which brings about a reduction in CO 2 emissions and an improvement in the quality of the environment (Huang et al. 2008).
Fourth, the neutral hypothesis affirms no causality between power consumption and economic growth.The results of this empirical work which finds no causal relationship are found by Wolde-Rufael (2006), Chontanawat et al. (2008), Wolde-Rufael (2009), Yoo and Kwak (2010), Ozturk andAcaravci (2011), Jafari et al. (2012), Śmiech and Papie ż (2014), etc.The policy ramification of this hypothesis affirms that limiting the use of energy in those countries will not hurt their economic growth.Therefore, a desirable reconciliation between the protection of the climate and economic competitiveness could be achieved for those countries ( Śmiech and Papie ż 2014).
In terms of the estimation techniques predominant in the existing literature, the energyeconomic growth nexus is investigated by applying time series and panel data sets for the short-run and the long-run relationship via Granger causality and cointegration methods, respectively.A survey of the literature on the causal relationship between electricity consumption and economic growth is conducted by Payne (2010) and Tiba and Omri (2017), with a focus on the different model specifications, variables used, hypotheses tested, and methodological issues.Payne (2010) attributed the mixed results of the empirical studies to the time period, model specifications, variable selection, and econometric methods used by the authors.Moreover, Ozturk (2010) reported inconsistent empirical findings in the existing literature on the nexus between energy consumption and economic growth.These kinds of inconsistencies in the findings of empirical studies are not at all conducive for policy making to use energy consumption as an economic tool for sustainable economic development (Payne 2010).
Further, giving an extensive survey of the existing literature on the energy-environmenteconomic growth link for specific-and multi-country studies covering the period from 1978 to 2014, Tiba and Omri (2017) underscore the fact that there is a lack of consensus about the direction of causality between those three variables which those authors attribute to the data availability, modelling methodology, time span, chosen measures and sample used in the study.Thus, there is a need for a further study of the dependence between the variables, using a novel methodology with a new set of data and a new set of variables.
However, a few studies are found to be related to our study concerning the relationship between sectoral electricity consumption and economic growth but not for India.Zamani (2007) examines only two electricity-consuming sectors, including the industrial and agricultural sectors in Iran, using the Engle-Granger VEC model.This author finds a bidirectional causality between industrial electricity consumption and economic growth and a unidirectional causality running from agricultural electricity consumption to economic growth.In the US, Thoma (2004) examines the causality between four electricity consuming sectors and economic growth, apart from total electricity consumption.A unidirectional causality is also reported from industrial production (a proxy for economic growth) to total commercial and industrial electricity consumption, but no causality is found for residential electricity usage and other sectors.Soytas and Sari (2007) investigate the causal relationship between industrial electricity consumption and manufacturing value added in Turkey and report a unidirectional causality running from the former to the latter.
Looking at the summary of the literature presented in the Indian context in Table 1, the results show divergences across the studies on the causality between energy consumption and economic growth.The results also diverge with regard to the time period of the study, the type of data and the methodology.To the best of our knowledge, no study has been conducted on the frequency domain causality, describing the causality at different time scales, which is more meaningful for a policy maker than for just examining the causality in the short run and long run only.Further, almost all the studies have been performed at the aggregate level, and the exceptions include Abbas and Choudhury (2013), who examined the relationship between only the agricultural electricity consumption and the economic growth nexus, and Ahmad et al. (2016), who analysed the relationship at the disaggregated level on the supply side of electricity and economic growth.In the current study, we use demand-side disaggregated data for five electricityconsuming sectors of the economy as these sectors consume the majority of electricity (see Figure 1).The quarterly data are used, which suit the results better.1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Data Description
The data are annual and cover twenty years between 1995 and 2014 for per capita power consumption by only utility sectors (lnPC) and per capita GDP (lnGDP) in the constant prices of 2004-2005.They were sourced from the EPW (India's Economic and Political Weekly Foundation) database.The data for per capita sectoral electricity consumption are obtained for the domestic/household, commercial, industrial HV (High Voltage), industrial MV (Medium Voltage) and agriculture (AGRI) sectors.Electricity consumption is in kilowatts hours (KWh).
The annual data are converted into quarterly frequency by taking the averages following the Chow-Lin method 3 , resulting in 80 quarterly data points for each variable 4 .The advantage of the conversion to quarterly data is that it increases the power of the statistical tests by using more observations, and also due to the insufficiency of annual data to enable us to achieve robust results (Zhou 2001).The quarterly interpolated data are widely accepted in empirical studies (Baxter and King 1999;Romero 2005;Mcdermott and McMenamin 2008;Tang and Chua 2012;Rashid and Jehan 2013).Investment (I) on completed projects is taken as a control variable in the study.Khan and Reinhart (1990) reported on the larger impact of private investment on economic growth.Milbourne et al. (2003) found a significant contribution of public investment to economic growth.
The descriptive statistics of the log difference of the variables are presented in Table 2.As evident from the Jarque-Bera test, the log-returns of DOM (DlnDOM), INDUHV (Dl-nINDUHV), and INDUMV (DlnINDUMV) are non-normal, while the log-returns of COM (DlnCOM), AGRI (DlnAGRI), GDP (DlnGDP), PC (DlnPC), and I (DlnI) are normal.However, their relationship with the dependent variable, per capita GDP (lnPGDP), is non-linear as per the BDS test of Independence 5 .It denounces the applications of linear time series causality models.The evidence of non-linearity could be attributed to the complexity of the economic system, thus motivating the use of non-linear methods for studying the causal relationship.

HEGY Seasonal Unit Root Tests
To capture the seasonality in the unit root process, several unit root tests have been proposed in the literature, such as by Hylleberg et al. (1990); Canova and Hansen (1995); Caner (1998); and Shin and So (2000) for quarterly and monthly data.As we have quarterly data, we follow a seasonal unit root test discussed by Franses (1990).This test is based on Hylleberg et al. (1990) (HEGY), which has the advantage that appropriate transformations follow directly from the procedure itself and do not have to be implemented a priori in order to remove possible (seasonal) unit roots.This test shows that testing for seasonal unit roots amounts to testing for the significance of the parameters of an auxiliary regression, which may also contain deterministic elements, such as constant, trend, and seasonal dummies.The results are presented in Table 3.The results show that the null of a seasonal unit is rejected at the 1% level.However, the non-seasonal unit at frequency zero cannot be rejected.This implies that all the variables are found stationary in the first difference, and we may conclude that the variables are integrated at the I(1).Our method of frequency domain causality is an appropriate way for the current dataset as the variables are stationary in the first difference.

The Frequency Domain Causality
The traditional approach to Granger causality tacitly ignores the possibility that the strength and/or direction of Granger causality (if any) can vary over different frequencies (Lemmens et al. 2008).Following the suggestion of Granger (1969), we used the Breitung and Candelon (2006) approach to Granger causality in the frequency domain, which is based on a spectral-density approach.Specifically, Breitung and Candelon (2006) proposed an approach which is based on Granger's (1969) and Geweke's (1982) suggestion, and it decomposes the total spectral interdependence between the two series into a sum of 'instantaneous', 'feed-forward', and 'feedback' causality terms.The innovativeness of this measure of Granger causality is that one can know exactly for which periodicity one variable can Granger-cause the other, which the popular one-shot linear or non-linear Granger causality tests fail to measure.Following Breitung and Candelon (2006)  6 , we can present this test by reformulating the relationship between x and y in the VAR(p) equation: The null hypothesis tested by Geweke (1982), H 0 : M y→x (ω) = 0, corresponds to the null linear restriction: where β is the vector of the coefficients of y and The ordinary F statistic for Equation ( 2) is approximately distributed as F (2, T − 2p) for ω ∈ (0, π).It is interesting to consider the frequency domain Granger causality test within a cointegrating framework.To this end, Breitung and Candelon (2006) suggest replacing x t in the regression in Equation ( 2) by ∆x t , with the right-hand side of the equation remaining the same. 7

Empirical Analysis
For comparison against a benchmark, we have estimated the causality in the time domain by using the Vector Error Correction 8 method for both the aggregate level of power consumption and the disaggregated level as well (i.e., at the sectoral level of power consumption).The results are presented in Table 4.No causality is evident from electricity consumption to economic growth (DLPGDP).However, a unidirectional causality is found running from economic growth to electricity consumption at the aggregate level and in the case of commercial (DLCOM) and industrial HV electricity (DLNDUHV) consumption.That means this average snapshot of the causality is assumed to be the same across the time scales.Again, the result being the outcome of a linear time series model ignores the existence of non-linearity between the dependent and independent variables.Therefore, in order to discern the causality at different time scales between power consumption and economic growth, we have applied the frequency domain causality test.Apart from considering the total electricity consumption at the aggregated level, to gain more insight into the nexus between power consumption and economic growth for robust policy making, the frequency domain causality is estimated for the different individual electricity-consuming sectors of the economy, that is, at the disaggregated level.The results of the frequency domain causality are presented in Figures 2-7.
The upper panel of each figure presents the unconditional frequency domain causality between the power consumption variables and the per capita GDP at different frequencies, starting from 0.00 (the lowest frequency) and through to 3.2 (the highest frequency) 9 .The lower panel shows the conditional (conditional on investment) frequency domain causality, as in this study, the value of an investment is taken as a conditioning variable that impacts the GDP apart from power consumption.Furthermore, the figures on the left of each panel present the causality from power consumption to per capita GDP and vice-versa through the figures on the right of each panel.
Figure 2 presents the frequency domain causality between aggregate power consumption (DLPPC) and economic growth (DLPGDP).In Figure 2, both the results of conditional and unconditional frequency domain causality are similar, defying the role of investment in the causal dynamics between total power consumption (DLPPC) and economic growth (DLPGDP).Our null hypothesis states that there is no business cycle causality existing between power consumption and economic growth.Our results are interesting and contrasting in comparison to the causality estimated using the VEC method and are presented in Table 4.The VEC causality shows there is unidirectional causality from economic growth (DLPGDP) to power consumption (DLPPC).However, in contrast, the frequency domain results show that a unidirectional causality exists and runs from power consumption to economic growth only for the frequencies level (omega) between 0.60 and 3.32, which correspond to the cycle length of 10.47 10 and three quarters, respectively.That means power consumption does Granger-cause economic development after two quarters and up to 10.47 quarters and not beyond that.The null hypothesis of the business cycle causality from economic growth to power consumption cannot be rejected at the 5% level of significance.Thus, power consumption is an important indicator of economic development.Our findings are consistent with studies by Ghosh and Kanjilal (2014) and Gupta and Sahu (2009), where energy consumption was found to cause economic growth.However, in the India-based studies by Paul and Bhattacharya (2004), they found a bi-directional causality between energy consumption and economic growth.Further, our findings support the growth hypothesis, which states a unidirectional causality running from electricity consumption to economic growth, as examined in Pradhan (2010) and Ahamad and Islam (2011).
Long-run (i.e., 15-62 quarters at the frequency level of 0.10-0.40)bidirectional causality between power consumption in the commercial sector (DLCOM) and economic growth (DLPGDP) is evidenced in Figure 3. Similar long run (i.e., quarters at the frequency level of 0.30-0.80),bidirectional causality is also observed between domestic sector consumption of electricity and economic growth, as shown in Figure 4.In addition to this, a short-run unidirectional causality runs from domestic sector consumption of electricity and economic growth.Our results are contradictory to the findings of Thoma (2004), who found no causality between residential electricity usage and economic growth.
As far as industrial power consumption (DLINDUHV) is concerned, Figure 5 shows unidirectional causality runs from industrial power consumption to GDP growth both in the short-and long-run ranging between the 2nd to the 12th quarter at the frequencies of 0.05-3.2.Thus, the policy makers should pursue a promotional instead of adopting a conservation policy for the industrial sector.However, this is in contrast to the findings of Thoma (2004) in the context of the US, where causality is reported from industrial production (a proxy for economic growth) to total commercial and industrial electricity consumption.Zamani (2007) finds bidirectional causality between industrial electricity consumption and economic growth, which is in contrast to our findings.However, our study is consistent with Soytas and Sari (2007), who reported a unidirectional causality running from industrial electricity consumption and manufacturing value added in Turkey.Nevertheless, in the context of industrial power consumption (DLINDUMV), unidirectional causality runs from economic growth (DLPGDP) to power consumption in the medium term, i.e., between 3-5 quarters (between 1.3-1.7 omega) as evidenced from Figure 6.Therefore, in contrast to DLINDUHV, the policy should be reversed for this industrial sector consumption (DLINDUMV).
It is apparent from Figure 7 that there is a unidirectional causality running from agricultural electricity consumption to economic growth both in the short-and the longrun, starting from the 3rd quarter and throughout the time period (i.e., at all frequencies).This supports the findings of Zamani (2007), who reports a unidirectional causality running from agricultural electricity consumption to economic growth.There is also a bidirectional causality in the very short period up to the second quarter at the frequencies between 2.80-3.2.Similar to the industrial (HV) sector, the policy makers should bring about a promotional policy of power consumption for the agricultural sector.
On the whole, the impact of using investment as a control variable is not apparent.As far as policy implications of our findings are concerned, one single policy is not the right solution for economic growth because the sectoral impact of power consumption on GDP growth is heteroscedastic.Thus, policy makers should bring about sector-specific policy initiatives which could contribute better to the economic development of the country.reported in the upper and lower panel respectively.Frequency is represented by omega (ω) on the horizontal axis, while the corresponding p-values are on the vertical axis.The label of the graph to the left of the vertical axis is in the form of (var i -var j ) where var i is the dependent variable and var j is the independent variable.The figures on the left (right) illustrate the causality from DLPPC to DLPGDP (DLPGDP to DLPPC).The null hypothesis is H 0 : There is no causality at the frequency Omega.The black horizontal line represents 5% level of significance.1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 % cosumption of electricity %Domestic %Commercial %Industrial (LV & MV) %Industrial (HV) %Agricultural Figure 2. The unconditional and conditional on investment i.e., DLINV frequency domain causalities are reported in the upper and lower panel, respectively.Frequency is represented by omega (ω) on the horizontal axis, while the corresponding p-values are on the vertical axis.The label of the graph to the left of the vertical axis is in the form of (var i − var j ) where var i is the dependent variable and var j is the independent variable.The figures on the left (right) illustrate the causality from DLPPC to DLPGDP (DLPGDP to DLPPC).The null hypothesis is H0: There is no causality at the frequency Omega.The black horizontal line represents 5% level of significance.

Conclusions and Policy Implications
India is the sixth largest and the fastest-growing large economy in the world.Further, the country being an energy-led-growth economy with a chronic deficit of power, it needs efficient allocations of resources and utilisation of power to achieve better economic growth.Furthermore, the inconclusive empirical results in the literature fail to help policy makers to come up with an appropriate energy policy for the economy in this regard.With this background, we have investigated the causality between power consumption and economic growth both at the aggregate and sectoral level using the frequency domain causality method.
The results show that total consumption of power Granger causes economic growth but not vice versa, starting from the second quarter through to the seventh quarter.It is projected that India's energy consumption will grow the fastest among all the major economies by 2035.Therefore, having more detailed insights on the sectoral level consumption causalities with GDP growth would induce more efficient policy making.No frequency domain causality is observed between the industrial medium voltage (MV) power consumption and GDP growth, thereby allowing the adoption of an energy conservation policy for the industrial sector.It is possible that this sector uses new productivity-enhancing technologies, which curb the causality.For the commercial consumption sector, since a bidirectional causality exists after the 15th quarter, then a short-term energy conservation policy could be initiated in this sector, but in the long run causality kicks in in this sector which may suggest that it uses more energy-intensive processes.
As far as domestic power consumption is concerned, the impact on economic growth is observed in the long run only.Therefore, short-term (long-term) household power policy measures may be initiated to make it productive (sustainable) in terms of economic growth.For the high voltage (HV) industrial sector, the presence of unidirectional causality running from industrial power consumption to economic growth throughout the period incentivises policy makers to have promotional initiatives in both the long run and the short run that spur more industrial power consumption for better economic development.The manufacturing sector generates 17% of GDP and 15% of the total employment in India and supports a diverse spectrum of industries which may account for the unidirectional causality.Therefore, India understands the importance of manufacturing in the country's growth strategy.
Similar to the industrial sector, policy makers should also bring about promotional policies of power consumption in the agricultural sector as a causality exists from agricultural electricity consumption to economic development in the short run and long run, but vice versa holds only in the very short period.Power consumption in the agricultural sector has increased by more than six-fold during the period 1980-2007.Thus, the government should be cautious with the implementation of its Ag DSM programme, which holds a promise for improved energy efficiency in groundwater irrigation in order not to hurt the agricultural sector.
In the context of the Indian economy, formulation of policies may be sought at both the aggregate and sectoral level for different periods in relation to significant cycles found in the frequency domain analysis to enhance the efficient utilisation of electricity for economic development.The government could bring out immediate policy measures to help the industrial sector and the domestic/household and commercial sectors in the short run to make those sectors more energy efficient.Energy in these sectors could be transferred to more productive sectors.
Further, for the industrial sector, an uninterrupted power supply should be ensured through the use of more appropriate policy measures.As far as the agricultural sector is concerned, a perpetual power supply will encourage farmers to opt for mechanisation and automation, which would increase the contribution of this sector to the GDP.Moreover, India is an agricultural economy, and the slow growth in this sector has been a cause of concern for the government.
Additionally, for these three sectors, policy initiatives could be pursued to control environmental pollution by incentivising those sectors to use more renewable energy sources.Overall, our findings should help policy makers in dealing with energy consumption across different sectors to promote better economic development.

Figure
Figure 1: Electricity Consumption in Major Consuming Sectors

Figure 4 :Figure 3 .
Figure 4: Causality from DLDOM → DLPGDP in the left and from DLPGDP → DLDOM in the right

Figure 4 :
Figure 4: Causality from DLDOM → DLPGDP in the left and from DLPGDP → DLDOM in the right

Figure 4 .
Figure 4. Causality from DLDOM → DLPGDP in the left and from DLPGDP → DLDOM in the right.

Figure 6 :
Figure 6: Causality from DLINDUMV → DLPGDP in the left and from DLPGDP → DLINDUMV in the right.

Figure 5 .
Figure 5. Causality from DLINDUHV → DLPGDP in the left and from DLPGDP → DLINDUHV in the right.

Figure 6 :
Figure 6: Causality from DLINDUMV → DLPGDP in the left and from DLPGDP → DLINDUMV in the right.

Figure 6 .
Figure 6.Causality from DLINDUMV → DLPGDP in the left and from DLPGDP → DLINDUMV in the right.

Figure 7 :
Figure 7: Causality from DLAGRI → DLPGDP in the left and from DLPGDP → DLAGRI in the right.

Figure 7 .
Figure 7. Causality from DLAGRI → DLPGDP in the left and from DLPGDP → DLAGRI in the right.

Table 1 .
Summary of the Indian Literature on Causality between Electricity Consumption and Economic Growth.

Table 1 .
Cont. : AGRI is agricultural electricity consumption, ELEC is electricity consumption, ELES is electricity supply, and EC-is Energy consumption.# refers to the study at the disaggregated level from the supply side, while ## refers to the study at the sectoral consumption level, focusing only on the agricultural electricity consumption.=> Unidirectional causality; <=> Bidirectional causality. Note The unconditional and conditional on investment i.e., DLINV frequency domain causalities are reported in the upper and lower panel respectively.Frequency is represented by omega (ω) on the horizontal axis, while the corresponding p-values are on the vertical axis.The label of the graph to the left of the vertical axis is in the form of (var i -var j ) where var i is the dependent variable and var j is the independent variable.The figures on the left (right) illustrate the causality from DLPPC to DLPGDP (DLPGDP to DLPPC).The null hypothesis is H 0 : There is no causality at the frequency Omega.The black horizontal line represents 5% level of significance.

Table 2 .
Descriptive Statistics of log Difference of the Variables.
Note: ** and *, respectively, denote significant at the 5%, and 10% levels.In the parenthesis, we report the p-values of each test.The symbol; " =>" denotes "does not linearly Granger-cause.
Causality from DLCOM → DLPGDP in the left and from DLPGDP → DLCOM in the right.
Causality from DLCOM → DLPGDP in the left and from DLPGDP → DLCOM in the right.