Energy consumption and economic growth: causality ...

15

Energy consumption and economic growth: causality relationship for Nigeria opec_173

15..24

Suleiman Sa’ad Department of Economics and Management Sciences, Nigerian Defence Academy, PMB 2109, Kaduna, Nigeria. Email: [email protected]

Abstract In this paper, cointegration and causality between energy consumption and gross domestic product (GDP) in Nigeria over the period 1971–2006 were analysed. The study used vector error correction models to test the relationships. The results of these tests showed the existence of cointegration as well as unidirectional causality-running from GDP to energy in Nigeria. The conclusion from these results was that decline in GDP or economic recession may have an adverse effects on energy consumption in Nigeria. However, energy conservation policies could be effectively implemented without any negative consequence on the growth of GDP in Nigeria.

1. Introduction The study of the long-run equilibrium and causality relationship between energy consumption and other economic variables such as income, employment and energy prices, are the cornerstone of energy demand studies. This area has gained overwhelming attention from energy demand modellers and policy-makers. Hence, it is well researched and the literature in this area is inexhaustible. Different studies have focused on different countries over different time periods to examine the relationships between energy and other macroeconomic variables. The empirical findings from these studies have been varied and sometimes conflicting. Their results differ on both the direction of the causality and its short and long-run impacts on energy policy. Depending upon which kind of causal relation exists, and the policy implications of these relationships can be significant (Masih and Masih, 1996). Evidence of unidirectional causality, emanating from energy consumption to income or employment, suggests that the country concerned is an energy-dependent country. Therefore, energy disruptions or conservation policies such as energy pricing or rationing may have an adverse effect on economic growth or employment. This is because economic growth depends, to a large extent, on energy consumption. © 2010 The Author. Journal compilation © 2010 Organization of the Petroleum Exporting Countries. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

16

Suleiman Sa’ad

On the other hand, a direction of causality running from income or employment to energy consumption suggests that the country concerned is less energy dependent. Therefore, energy shortages or conservation policies may not necessarily have negative consequences on economic growth. Finally, evidence of no correlation between the two implies a neutrality hypothesis. Therefore, energy and income are not mutually dependent and conservation policies may be implemented without any adverse effects on economic growth. Given the importance of this issue, it is not surprising that it received overwhelming attention from energy demand modellers, leading to a plethora of studies on this phenomenon. However, most of these studies focused on developed Organisation for Economic Cooperation and Development and Asian developing countries. Despite the fact that Nigeria has been endowed with huge energy resources and is a major player in global energy market [except for three studies conducted by Obas John Ebohon (1996), Wolde-Rufael (2005) and Mehrara (2007)] research on the relationship between energy consumption and other economic variables has not been given the desired attention. This study, therefore, will add to few existing studies in Nigeria. The objective of the study is to examine the causality relationship between energy consumption and gross domestic product (GDP) in Nigeria and draw a policy implication from the result of the study. The remainder of the paper is structured as follows: Section 2 presents a brief profile of the trend of energy consumption and GDP in Nigeria during the period of the study. This is followed by a brief review of past studies on the causality relationship between energy and income. Section 4 shows the methodology of the study, and Section 5 presents the empirical findings and conclusions. 2. Profile of energy consumption and GDP in Nigeria Nigeria is richly endowed with energy resources; both hydrocarbon and renewable. The energy sector has been the backbone of the Nigerian economy since the first global oil shock of 1973. Over the years, a substantial part of the national income of Nigeria has been derived from the energy sectors. For example, the contribution of oil to the total income of Nigeria increased from 26.3 per cent in 1970 to 69.0 per cent in 1983 and further to 86.2 per cent in 1992. However, the revenue from oil declined to 62.5 per cent in 1998 (Central Bank of Nigeria, 2000). Similarly, energy consumption continued to grow substantially during that period because of factors such as increase in population and urbanisation, as well as rapid growth in per capita income and industrial activities. Figure 1 presents relative trends in per capita energy consumption in tons of oil equivalents and per capita real GDP in thousands of US dollars at 2000 constant prices for Nigeria during the period 1971–2006. From the figure, the energy consumption in tons of oil equivalents (Toe) are plotted real GDP in thousands of US dollars 2000 constant prices. From the relative trends, OPEC Energy Review March 2010

© 2010 The Author. Journal compilation © 2010 Organization of the Petroleum Exporting Countries

17

Energy consumption and economic growth

18000

140

Energy Consumption Real GDP

120

16000

Ktoe

14000

100

12000 10000

80

8000 6000

60

4000 2000 1971

1976

1981

1986

1991

1996

2001

2006

Bilions of US$ 2000 PRICES

20000

40

Figure 1 Real gross domestic product (GDP) and energy consumption. Sources: International Energy Agency 2008.

the aggregate commercial energy consumption for Nigeria was 1838.116 Ktoe in 1971. It increased more than four-fold to 8745.679 Ktoe in 1981, to 9758.395 Ktoe in 1991, and to 19,330.01 toe in 2006. Similarly, in 1971, the real GDP was 49.382 billion dollars. It increased to 68.689 billion dollars in 1980 and to 80.023 billion dollars in 1990. Finally, the real GDP rose to 131.939 billion dollars in 2006. From the figure, it is clear that both GDP and energy consumption are highly correlated; however, energy consumption grew larger than real GDP during the period of study. The current international concerns about global warming and energy security calls for policies that can ensure that economic growth in developing countries can be achieved with less dependence on the use of fossil fuels. Arguably, this can be achieved through a mix of demand and supply policies. First, rise in energy prices would encourage more parsimonious and efficient use of energy. Second, encouraging the use of other sources of energy such as renewable energy in the primary energy mix. These two factors, if complemented by other factors (better insulations, use of minimum energy efficiency standards for vehicles and enlightenment campaigns), can help to reduce the wastages in energy use and over dependence on fossil fuels. In Nigeria, despite the huge potential for renewable energy, majority of the population still rely on traditional biomass and fossil fuels as sources of energy.1 This, which continues growth in the consumption of fossil fuels, is a source of concern for the government of Nigeria. Recently, the Nigerian government is making some effort to include renewable energy such as wind and solar energy as part of its energy mix. Furthermore, the government is making some effort to deregulate the prices of petroleum products; this will increase efficiency of energy use as well as generate more revenue for other social services. © 2010 The Author. Journal compilation © 2010 Organization of the Petroleum Exporting Countries

OPEC Energy Review March 2010

18

Suleiman Sa’ad

3. Review of previous studies on developing countries One prominent study which attempted to test the causality and cointegration relationship between energy consumption and GDP is that of Masih and Masih (1996). This investigation used Johansen’s multivariate tests and variance decomposition to test for the cointegration and causality relationship between energy consumption, real GDP and energy prices in certain Asian countries. The results of their study show that it is only in India, Pakistan and Indonesia that cointegration exists between energy consumption and GDP. Energy consumption and GDP remained non-integrated in Malaysia, Singapore and the Philippines. In India, there was evidence of causality running from energy consumption to GDP. However, in Indonesia and Pakistan, the results provided evidence of mutual causality between energy and income. A similar study was conducted by Soytas and Sari (2003) to re-examine the issue of causality between energy consumption and income in G7 economies and in emerging markets countries. This study also employed Johansen’s multivariate test. The results of this study provide evidence of cointegration between energy consumption and GDP in seven out of the 16 countries. In addition, they indicate a long-run unidirectional causality running from energy consumption to income in Turkey, France, West Germany and Japan. However, there is no evidence of causality in the cases of Italy and South Korea. There is also evidence of bidirectional causality in Argentina in both the short and the long-run. Asafu-Adjaye (2000) also tested causality relationships between energy consumption and income in selected Asian developing countries. His findings show evidence of casualty running from income to energy. These findings are consistent with the earlier work of Masih and Masih (1996). Chien-Chiang Lee (2005) used a cointegrated panel approach to analyse the cointegration and causality between energy consumption and GDP in developing countries for the period 1975–2001. After allowing for the heterogeneous country effects, the findings of his study provide support for a long-run cointegration relationship. Furthermore, his findings establish evidence of long-run and short-run causalities running from energy to income. The findings of Chien-Chiang Lee (2005) revalidated the earlier findings of Masih and Masih (1996) and Asafu-Adjaye (2000), but are not consistent with earlier findings by Glasure and Lee (1997) and Soytas and Sari (2003). Some studies also attempt to examine this phenomenon in sub-Saharan African countries. In this respect, the works of Obas John Ebohon (1996) who studied the relationships between energy and GDP as well as energy consumption and GNP for Nigeria and Tanzania, are notable. The result of his study indicated a unidirectional causality between energy and GDP for Nigeria. There was, nevertheless, no evidence of causality between energy and GNP. On the other hand, the results for Tanzania suggest that the direction OPEC Energy Review March 2010



19

of the causality runs from energy to GNP, suggesting that energy disruptions will probably hinder economic activity in Tanzania. Jumbe (2004) studied cointegration and causality between electricity consumption (kWh) and agricultural GDP (AGDP) and non-agricultural GDP (NGDP) in Malawi for the period 1979–1999. The results of his study suggest that electricity consumption is cointegrated with both GDP and NGDP. However, there is no long-run relationship between electricity consumption and AGDP. The causality test suggests bidirectional causality between electricity consumption and GDP, implying that kWh and GDP are jointly determined. On the other hand, the results of the test between kWh and NGDP indicate a one-way causality, running from NGDP to kWh and suggesting that electricity is not a stimulus for non-agricultural GDP in Malawi. Wolde-Rufael (2005) also investigated the relationships between energy and income in 19 Africa countries, including Nigeria. The results of his study are heterogeneous; they show evidence of a Cointegration relationship between energy consumption and income in eight of the 19 countries, and causality for 12 countries. For Nigeria there was a negative causality relationship running from energy consumption to income. This implied that energy use negatively impacts on economic growth. The results suggest that electricity disruptions and shortages of petroleum products during the period of study had adverse effects on economic growth in Nigeria. Finally, Mehrara (2007) is one of the few available studies that attempt to study the relationship between energy and income in oil exporting countries (including Nigeria). The paper studied the relationship between energy consumption and income in 11 oil exporting countries, using annual data over the period 1971–2002. The findings of his study suggest a one-way strong causality running from income to energy consumption in all 11 countries. These results suggest that in these countries, it is GDP that drives energy consumption, with no feedback effects. The policy implications of the results are that, for these countries, energy conservation is feasible with no damaging repercussions on economic growth.

4. Methodology and model specification In the course of searching for cointegration and Granger causality relationships among the variables, we used the vector error correction approach in order to test for the short and long run as well as the direction of the causality relationships between the variables. Through this approach, the problem of estimating a spurious regression will also be avoided. This is done first by testing for the properties of the variables through the augmented Dickey and Fuller (1979) and Phillips and Perron (1988) tests; followed by testing for cointegration by Johansen maximum likelihood approach. The dynamic © 2010 The Author. Journal compilation © 2010 Organization of the Petroleum Exporting Countries


20

Suleiman Sa’ad

Granger causality can be captured from the vector error correction model derived from the long-run cointegration relationship (Granger 1969). n

n

i =1

i =1

ΔLECt = α1 + ∑ β1ΔLEC1 + ∑ δ1ΔLGDP1 + ECT−1 + ε t n

n

i =1

i =1

(1)

ΔLGDPt = α 2 + ∑ β 2 ΔLGDP2 + ∑ δ 2 ΔLEC2 + ECT−1 + ε t .

(2)

Where D is the difference operator, LGDP is the natural log of per capita real GDP; LEC is the natural log of per capital energy consumption et is the error term which is identically normally distributed. 4.1. Estimation results Data The data used in this study are sourced from the International Energy Agency (IEA) beyond 2020 Energy Statistics and Balances for non-Member Countries [27]. The data on per capita energy consumption refers to commercial energy consumed by end-user sectors during the period 1971–2005, divided by the population to give the per capita series. On the other hand, the data on real per capita GDP is the real GDP at 2000 constant prices in US dollars, divided by the total population to give the per capita annual GDP covering the period 1971–2005. Unit root tests The necessary condition against estimation of a spurious regression is for the series to share the same intergrational properties. Table 1 presents the result of unit root tests for

Table 1 Augmented Dickey–Fuller and Phillip–Perron unit root tests

Variable LEC LGDP DLEC DLGDP Critical values: 5%

tt(ADF)

tm(ADF)

t(ADF)

tt(PP)

tm(PP)

t(PP)

Remarks

-2.29 5.18 -3.38 -12.59 -3.21 -3.55

-2.41 0.52 -3.36 -13.31 -2.61 -2.95

0.53 2.13 -2.63 -7.75 -1.61 -1.95

-2.32 4.41 -3.32 -11.11 -3.21 -3.55

-2.41 0.71 -3.38 -11.69 -2.61 -2.95

0.54 2.44 -3.37 -6.47 -1.61 -1.95

I(1) I(1) I(0) I(0)

ADF, augmented Dickey–Fuller; PP, Phillip–Perron; LEC, log of per capital energy consumption; LGDP, log per capita real gross domestic product. OPEC Energy Review March 2010


21


Table 2 Johansen maximum likelihood cointegration test (1971–2004) R : number of cointegrating vectors

lMAX

Critical values 95%

P-values

lTRACE

Critical values 95%

P-values

r=0 rⱕ1

25.44 6.66

21.13 14.26

0.02 0.18

31.84* 11.40

29.79 15.49

0.02 0.18

* Indicating cointegration at 5% level of significance. Table 3 Results of Granger causality Dependent variable

c2 statistics

DLEC DLGDP

DLEC – 22.00(0.00)

DLGDP 1.66(0.433) –

ECTt-1 -0.312[-2.99] 0.025[0.56]

Their asymptotic t-statistics in squared brackets. Numbers in brackets are the P-values corresponding to the c2 statistics. LEC, log of per capital energy consumption; LGDP, log per capita real gross domestic product; ECTt-1, one period lagged error correction term from cointegration equation.

the series used in this study based on ADF and PP tests. In testing for the unit root, both trend and intercepts are included to give power to alternative hypothesis of trend stationarity in the series. From the table, the results are clear-cut, that all the series are nonstationary at their levels forms and stationary at their first differences, therefore the null hypothesis of unit root cannot be rejected at both 5 per cent and 10 per cent levels of significance. Based on the result of the unit root, a test for Johansen’s cointegration are carried out based on maximum eigenvalues and trace statistics. In conducting the test, optimal laglengths of two are used based on Akaike information criteria and Schwarz Bayesian information criteria. From the result in Table 2, both eigenvalues and trace statistics reject the null hypothesis suggesting that there is one cointegration equation at 5 per cent level of significance. After the testing for cointegration, the study proceeds by testing for Granger causality through vector error correction model. The result of the test presented in Table 3 shows that the coefficient of the error correction term (ECTt-1) normalizing DLEC is statistically significant, suggesting a long run causality running from DLGDP to D. The result of the c2 test for a joint significance of b1, and b2, which also indicates the direction of the short-run causality, suggests a short run unidirectional causality running from DLGDP to DLEC. The result of Granger causality from DLGDP to DLEC is consistent with both theory and previous studies. However, the result does not show any causality © 2010 The Author. Journal compilation © 2010 Organization of the Petroleum Exporting Countries


22

Suleiman Sa’ad

from DLEC to DLGDP. The finding of Granger causality running from income to energy in Nigeria is consistent with earlier findings in the studies conducted by Mehrara (2007), which found a direction of causality from income to energy in 11 oil producing countries (including Nigeria). However, the findings of this study are inconsistent with the earlier study conducted by Wolde-Rufael (2005). This found a negative causation running from energy to income in Nigeria. However, Obas John Ebohon (1996) found bidirectional causality between energy and income in the country. 5. Conclusions and policy implications This paper used the bivariate vector error correction models to investigate the long-run cointegration and Granger causality between per capita GDP and per capita energy consumption in Nigeria. The result of the test shows that both Engel values and trace statistics rejected the null hypothesis of no cointegration relationships between per capita GDP and per capita energy consumption. Similarly, the result of Granger causality established a unidirectional causality running from LGDP to LEC in both short and long run, without any feedback. The findings suggest that real GDP has a major role to play in energy consumption in Nigeria. Therefore, fall in income or economic recession in Nigeria will have adverse effects on the energy consumption. However, energy conservation policies, such as rationing and market based pricing for energy, can be implemented without any adverse effects on real GDP growth. Note 1. The prices of energy (gasoline, diesel and electricity) are heavily subsidised in Nigeria.

References Asafu-Adjaye, J., 2000. The relationship between energy consumption, energy prices and economic growth: time series evidence from Asian developing countries. Energy Economics 22, 615–625. Central Bank of Nigeria, 2000. Changing Structure of Nigerian Economy. Dickey, D.A. and Fuller, W.A., 1979. Distribution of the estimators for auto-regressive time series with a unit root. Journal of American Statistical Association 74, 427–431. Ebohon, O.J., 1996. Energy, economic growth and causality in developing countries: a case study of Tanzania and Nigeria. Energy Policy 24, 447–453. Glasure, Y.U. and Lee, C.C., 1997. Co-integration, error correction, and the relationship between GDP and energy: the case of South Korea and Singapore. Resource and Energy Economics 20, 17–25. OPEC Energy Review March 2010



23

Granger, C.W.J., 1969. Investigating causal relations by econometrics models and cross spectral methods. Econometrica 37, 424–438. International Energy Agency (IEA), 2008. Beyond 2020 Energy Statistics and Balances for non-Member Countries. IEA/OECD, Paris. Jumbe, C.B.L., 2004. Cointegration and causality between electricity consumption and GDP: empirical evidence from Malawi. Energy Economics 26, 61–68. Lee, C.C., 2005. Energy consumption and GDP in developing countries: a co-integrated panel analysis. Energy Economics 27, 415–427. Masih, A.M.M. and Masih, R., 1996. Energy consumption, real income and temporal causal: results from multi-country study based on co-integration and error correction modelling techniques. Energy Economics 18, 165–183. Mehrara, M., 2007. Energy consumption and economic growth: the case of oil exporting countries. Energy Policy 35, 2939–2945. Oh, W. and Lee, K., 2004. Causal relationships between energy consumption and GDP revisited: the case of South Korea 1970–1999. Energy Economics 26, 51–59. Paul, S. and Bhattacharya, R.N., 2004. Causality relationships between energy consumption and economic growth in India: a note on conflicting results. Energy Economics 26, 977–983. Phillips, P.C.B. and Perron, P., 1988. Testing for unit root in time series regression, Biometrica 75, 335–346. Soytas, U. and Sari, R., 2003. Energy consumption and GDP: causality relationships in G-7 countries and emerging markets. Energy Economics 25, 33–37. Wolde-Rufael, Y., 2005. Energy demand and economic growth: the African experience. Journal of Policy Modeling 27, 891–903. Yu, E.S.H. and Choi, J.Y., 1985. The causal relationship between energy and GNP: an international comparison. Journal of Energy and Development 10, 2, 249–272.



24

Suleiman Sa’ad

Appendix 1: The empirical results from causality tests for developing countries Authors

Empirical method

Yu and Choi (1985) Standard Granger test Glasure and Lee, (1997) Masih and Masih ECM (1996)

Masih and Masih (1996) Yang (2000) Asafu-Adjaye, (2000)

ECM

Period

Country

Causal relationships

1954–1976

South Korea

Income → Energy

Philippines

Energy → Income

1961–1999

South Korea and Singapore 1955–1990 Malaysia Singapore, and the Philippines India Indonesia Pakistan 1955–1990 Sri Lanka and Thailand 1954–1997

ECM

Soytas and Sari, (2003)

ECM

Oh and Lee (2004) Paul and Bhattacharya (2004) Jumbe (2004) Ebohon, (1996)

ECM Granger

1950–1992

Argentina South Korea Turkey Indonesia Poland 1970–1999 South Korea 1950–1996 India

Granger 1970–1999 Granger(1969 1960–1975


Taiwan India and Indonesia, Thailand and the Philippines Turkey

Malawi Nigeria Tanzania

Energy → Income Non-Cointegrated Energy → Income Income → Energy Energy → Income Energy → Income Energy → Income Energy → Income Energy → Income Energy → Income Energy → Income Energy → Income Income → Energy Energy → Income Non Cointegrated Energy → Income Energy → Income

Income → Energy Bi-directional
