Electricity demand, GDP and employment: evidence from Italy

Abstract

This paper applies time series methodologies to examine the causal relationship among electricity demand, real per capita GDP and total labor force for Italy from 1970 to 2009. After a brief introduction, a survey of the economic literature on this issue is reported, before discussing the data and introducing the econometric techniques used. The results of estimation indicate that one cointegrating relationship exists among these variables. This equilibrium relation implies that, in the long-run, GDP and labor force are correlated negatively, as well as GDP and electricity. Moreover, there is a bi-directional Granger causality flow between real per capita GDP and electricity demand; while labor force does not Granger-cause neither real per capita GDP nor electricity demand. This implies that electricity demand and economic growth are jointly determined at the same time for the Italian case. The forecast error variance decomposition shows that forecast errors in real per capita GDP are mainly caused by the uncertainty in GDP itself, while forecast errors in labor force are mainly resulted from the labor force itself, although aggregate income and electricity are important, too.

Graphical abstract

Keywords

energy policies / electricity demand / GDP / labor force / stationarity / structural breaks / cointegration / causality / Italy

1 Introduction

The study of the causal relationship between electricity demand and GDP fosters a better understanding of the role electricity has played in Italy’s economic growth. The results of causality tests can be useful in shaping future electricity policies, such as conservation programs, capacity expansion planning, and construction of nation-wide interconnections of power networks. Electricity demand, supply and pricing have an impact on socio-economic development, living standards and the overall quality of people’s [1]. On the other hand, a higher level of economic development could induce more electricity demand.

If electricity consumption causes economic growth, then policies encouraging a reduction in electricity consumption will have an effect on growth. On the other hand, if electricity consumption does not cause economic growth or economic growth causes consumption, then electricity conservation policies will have no impact on growth [2].

Over the past three decades, many studies, using co-integration and Granger causality, have focused on different countries and time periods. Since Kraft and Kraft’s [3] pioneering study, empirical research has produced mixed and, for some countries, controversial results [4]. These results differ also for the direction of causality and the short-term versus long-term effects on energy policies. Depending upon what kind of causal relationship exists, its policy implications may be significant.

Moreover, multiple causality studies have been performed in many countries in the world; however, very few studies have been devoted to the analysis of the Italian case [2,5−9].

This paper examines the nexus between real per capita GDP and per capita electricity demand in Italy from 1970 to 2009, using time series methodologies. The results could be used to define and implement appropriate energy development policies in Italy. The data used are obtained from total economy database (TED) and organization for economic cooperation and development (OECD) database.

The cost of energy in Italy, in particular electricity prices, is among the highest within the EU. Rents due to the lack of competition in sheltered sectors imply higher prices for a range of intermediate goods and services [10]. After a timid initial recovery during 2010, the drop of electricity markets trading in Europe caused by the economic crisis continued also in 2011. In the second half of 2012, the greatest retail price increase in EU was observed in Italy (20%), while in France prices fell by 13% [11]. Italy remains very dependent on imports because of limited available capacity and a lack of siting approvals for new power plants, having undertaken partial privatization (Table 1).

More than 80% of Italy’s electricity supplies are purchased in the deregulated market. Electricity demand in Italy increased by 0.56% during 2011. The modest upswing in electricity consumption is still below the 2007 − 2008 level, indicating that the slump of the previous year has not yet been overcome. Currently 57.9% of all electricity consumption is traded on Italy’s electricity exchange (IPEX) with the remaining amount being consumed through bi‐lateral contracts [12].

2 The nexus between electricity demand and GDP

In the last two decades, owing to the strong and constant increase in electricity consumption, which imposed an accurate planning in order to avoid electricity shortage and guarantee adequate infrastructures, a load of study focused on the relationship between economic growth and electricity.

The directions of causality between electricity demand and aggregate income could be based on four different hypotheses, each giving rise to different and important implications for energy policy [13].

As explained in Ref [14], there are:

1) Neutrality hypothesis: if no causality exists between GDP and electricity demand. This implies that energy consumption is not correlated with GDP. The absence of Granger-causality supports the neutrality hypothesis as documented by Refs [13,15−20].

2) Conservation hypothesis: the unidirectional causality runs from GDP to electricity demand. If there is unidirectional causality running from economic growth to electricity consumption, electricity conservation policies through changes in the tariff structure, energy efficiency improvements and other demand side managements, aimed at curtailment of wastage of electricity and reduction of electrical consumption without affecting the end-use benefits, can be initiated without deteriorating a country’s economic growth [21]. This hypothesis was empirically supported by Refs [3,15−17,22−30].

3) Growth hypothesis: the unidirectional causality runs from electricity demand to GDP. If causation is found to run from electricity consumption to economic growth with no feedback, planners can justify prioritization of more resources to boost a country’s electrical network. This hypothesis is in line with empirical findings in Refs [13,15,17,20,21,28,31−41].

4) Feedback hypothesis: if there is a bi-directional causality flow between GDP and electricity demand. If results suggest that there is a mutual relationship between electricity and GDP, then any global policy to reduce electricity consumption in order to reduce emissions would have an impact on the GDP of all countries. The feedback hypothesis was documented in Refs [15,16,20,28,42−53].

Table 2 presents concisely the main findings on causality between aggregate income and electricity demand discussed in several studies on this topic.

3 Econometric methodology, data, and empirical results

According to Ref [54], a linear combination of two or more non-stationary series (with the same order of integration) may be stationary. If such a stationary linear combination exists, the series are considered to be co-integrated and therefore long-run equilibrium relationships exist. Incorporating these co-integrated properties, an error-correction model (ECM henceforth) could be constructed to test Granger causation of the series in at least one direction. In this study, the ECM is specifically adopted to examine the Granger causality between real GDP and electricity demand.

To investigate the stationarity properties of the series considered, the augmented tests have been performed in Refs [55−58]. A time series integrated of order 1, said I(1), requires the first differencing filter to remove the stochastic trend.

To investigate the stationarity properties of the variables considered, this paper employs the Zivot and Andrews (ZA henceforth) test [59] to account for structural breaks, while it adopts the Clemente et al. (henceforth CMR) procedure [60] to check multiple break points via a gradual shift in the analysis.

When both series are integrated of the same order, the cointegration can be tested. The Johansen maximum likelihood procedure [61] is used for this purpose. Any long-run co-integrating relationship found between the series will contribute an additional error-correction term to the ECM.

In the case under discussion in this paper, all the variables analyzed have been expressed in a logarithmic scale. The empirical study uses time-series data for real per capita GDP, electricity demand and labor force for the 1970 − 2009 period in Italy. Data are obtained from the Total Economy Database (2010) maintained and updated by the Conference Board of the Groningen Growth and Development Centre, and from OECD. In this paper electricity demand is expressed in terms of GW·h, per capita GDP in constant 1990 US$, and the total labor force in 1000 unit. The choice of the starting period is constrained by data availability on electricity demand.

In Table 3, the variables used are described. All series contain yearly data for the real value of the variables.

Figure 1 shows, in a log-scale, the historical trends of real per capita GDP and electricity demand for Italy.

Descriptive statistics of the samples are listed in Table 4.

The series are strongly correlated, since the correlation coefficients are r _PCGDPGK,ED = 0.9912, r _PCGDPGK,TLF = 0.9740, r _ED,TLF = 0.9673.

First, the log-transformations of the time-series are obtained. As a preliminary analysis, inter-quartile range suggests the absence of outliers in the samples. A more accurate description of the data is presented in Table 5 and Fig. 2. Table 6 tabulates the results of common unit root tests.

In more detail, from the third column to the last in Table 6, the ADF, ERS, PP, and KPSS tests are respectively reported. The results show that the series are non-stationary according to their levels, but stationary in first differences. This suggests that the PCGDPGK, ED and TLF variables are individually integrated for order 1, or I(1).

The Zivot and Andrews’s results unit root test are summarized in Table 7, which indicates that the null hypothesis of a unit root cannot be rejected in levels for the series. Examining the results in first differences, the null hypothesis can be rejected at a 1% level of significance for GDP and electricity demand, and at a 5% level for labor force. Therefore, each of these series can be characterized as an I(1) process.

An examination of Table 8 suggests the breaks detected by Ref [60] roughly corresponds to the timing of the second oil shock and the signing of the Maastricht Treaty for PCGDPGK, and to the second oil shock and the starting of EMU for TLF. Here, the results are in line with those found applying the ZA test. Despite the structural break, it is impossible to reject the null hypothesis of a unit root in these series; yet, if the test is performed at the first differences, the series become stationary: so, it can be concluded again that GDP, electricity demand and labor force are I(1) processes.

Since the series examined have the same order of integration, this paper can apply the Johansen and Juselius co-integration procedure. Therefore, to conduct the test, an assumption has to be made regarding the trend underlying the data. In this paper, the level data are assumed to have no deterministic trends and the co-integrating equations have intercepts. The choice of this specification is based on the investigation of the graphs of the two series and the unit root tests, both of which suggest that the two series do not have a common deterministic trend. The lag-order selection has been chosen according to the final prediction error (FPE), Akaike’s information criterion (AIC), Schwarz’s Bayesian information criterion (SBIC), and the Hannan and Quinn information criterion (HQIC): two lag intervals in first differences, for both series.

The null hypothesis of no co-integration, H₀: r₀ = 0, is rejected at the 5% level of significance (Table 9). Hence, the results of both tests imply that the hypothesis of no co-integrating equation is rejected at the 5% significance level. Turning to the maximal eigenvalue, the statistic is 25.4204, which is above the 5% critical value of 22.00. Hence, the null hypothesis of r₀ = 0 is rejected at the 5% level of significance. However, under H₀: r₀ = 1, the trace and maximum eigenvalue statistics are equal to 8.8966 and 6.3374 respectively, which are below the 5% critical values of 19.96 and 15.67. Hence, the null hypothesis is accepted at the 5% significance level. These results imply that the three ED, PCGDPGK and TLF series have one co-integrating equation; in other words, there is a long-run relationship between real per capita GDP, per capita electricity demand and total labor force for Italy. The co-integration implies the existence of Granger causality; however, it does not indicate the direction of causality.

Since the two series are co-integrated, a vector error correction model (VECM) is set up for investigating short- and long-run causality. The lag of the system is set to two following Hannan-Quinn Information Criterion (HQIC). The results of VECM estimate produce a co-integrating vector equals (1, − 0.68, − 0.92, and 2.18). This equilibrium relation implies that, in the long-run, GDP is negatively correlated with labor force and electricity demand. The ECT is found to be statistically significant in the three equations at the 1% level after normalizing the co-integrating equation with respect to the PCGDPGK coefficient. This result suggests a bi-directional long-run causality (feedback effect) between real per capita GDP, electricity demand and labor force.

Table 10 shows the results of tphe Granger causality tests, which suggest a bi-directional flow for real per capita GDP and electricity demand; while TLF is not the Granger-cause to PCGDPGK or ED.

As regards the robustness of the VECM, for all the equations, a Lagrange-multiplier (LM) test for autocorrelation in the residuals of VECM clarifies at the 5% significance level, the null hypothesis cannot be rejected that there is no serial correlation in the residuals for the orders 1, …, 5 tested. Using the Portmanteau autocorrelation test, the Box-Pierce and Ljung-Box Q-statistics are equal to 73.94 and 90.73, respectively. Thus, the null hypothesis of no serial correlation up to lag 12 cannot be rejected. Checking the eigenvalue stability condition in a VECM, the eigenvalues of the companion matrix lie inside the unit circle, and the real roots are far from 1. With reference to the Wald lag-exclusion statistics, the hypothesis is strongly rejected that the coefficients either on the first lag or on the second lag of the endogenous variables are jointly zero in all two equations. The Jarque and Bera normality test results present statistics for each equation and for all equations jointly against the null hypothesis of normality. For the models, the results suggest normality: in fact, the Jarque-Bera statistics are equal to 3.96 in the PCGDPGK equation, to 4.99 in the ED equation and to 8.95 in the TFL equation, suggesting that the null hypothesis of normality of the residuals cannot be rejected. The joint test statistics of the white homoskedasticity test with the no cross terms is 75.40, with a P-value of 0.02, so the null hypothesis of non-heteroskedasticity at a 1% confidence level is not rejected. Hence, the model passes all the tests successfully and the residuals are Gaussian white noise. Finally, the analysis of auto-regressive conditional heteroskedasticity (ARCH) effects shows the absence of this problem in the case.

The long-run impact results are illustrated in Table 11 by the forecast error variance decompositions (FEVDs) for the three variables based on a VECM model. The first (left-hand side) panel shows that forecast errors in real per capita GDP are mainly caused by the uncertainty in GDP itself, but a few years of shocks in labor force may have a significant effect on GDP series. The second panel for electricity demand shows that these errors also result increasingly from labor force, while in a first stage GDP shocks are more relevant. Finally, the third panel shows that forecast errors in labor force mainly stem from labor force itself, although aggregate income and electricity are important as well. This suggests that labor force and real per capita GDP are very much related.

4 Conclusions and policy implications

The purpose of this paper is to contribute to the literature on the nexus among GDP, electricity demand and total labor force, using recent econometric techniques. It analyzed this relationship in Italy, using annual data covering the 1970−2009 time period. The properties of the series were assessed using several unit root tests (ADF, DF-GLS, PP, and KPSS). Furthermore, in order to evaluate the presence of eventual structural breaks, ZA and CMR tests were performed. Empirical findings indicated that both series were clearly non-stationary, as a I(1) process.

Co-integration analysis revealed that there was a long-run relationship between GDP, electricity demand and total labor force. Based on a VEC model after testing for multivariate co-integration, it was found that energy entered significantly into the co-integration space. The coefficient of the ECT was found to be statistically significant in the three equations at the 1% level. Granger causality tests suggested a bi-directional flow for real per capita GDP and electricity demand; while labor force was not the Granger-caused to the real per capita GDP or electricity demand; thus, there was a long-run bi-directional causal relationship (or feedback effect) between PCGDPGK and ED. Yet, if there was a bi-directional causal relationship, then economic growth might demand more electricity, whereas more electricity demand might also induce economic growth. Therefore, electricity demand and economic growth complemented each other such that radical electricity conservation measures might significantly hinder economic growth [42,62]. Finally, the forecast errors in real per capita GDP are mainly due to uncertainty in GDP itself, with an increasing effect for the shocks in labor force. The errors in electricity demand also become increasingly due to labor force; while the forecast errors in labor force are mainly due to labor force itself, although aggregate income and electricity are important as well. This can be interpreted as that labor force and real per capita GDP are very much related.

As shown in Ref. [63], the estimation of GDP and per capita GDP elasticities revealed higher values with respect to price elasticities, demonstrating that the consumption response to GDP and GDP per capita changes was relevant. Therefore, it was necessary to assure an appropriate level of electricity supply to sustain the economic growth in Italy.

Power prices in Italy were higher than those in other regions in the last three years, though during the last couple of months the Italian price premium to other European markets significantly decreased 11]. This was mainly caused by the increasing share of renewables in the country’s power generation mix and lower gas prices on the Italian PSV hub, converging with other European gas hub prices. With an eye to the growing integration between the Italian and neighboring markets, it should be noted that the coupling project with Slovenia became operational in 2011. Thanks to this project, cross-border capacity was allocated more efficiently vis-à-vis explicit auctions; also, it was constantly consistent with the price spread developing along the border.

5 Suggestions for future researches

Further analysis may be conducted to study the nexus between different sources of energy and aggregate income in Italy. This could be of help to the debate on Italy’s return to nuclear power. Conclusions for Italy may be relevant for a number of countries that have to go through a similar development path of increased pressure on already scarce energy resources.

References

[1] Iwayemi A. Energy sector development in Africa. 1998

[2] Ciarreta A, Zarraga A. Economic growth-electricity consumption causality in 12 European countries: a dynamic panel data approach. Energy Policy, 2010, 38(7): 3790–3796

[3] Kraft J, Kraft A. On the relationship between energy and GNP. Journal of the Energy Division, 1978, 3(2): 401–403

[4] Ozturk I. A literature survey on energy-growth nexus. Energy Policy, 2010, 38(1): 340–349

[5] Guttormsen A G. Causality between energy consumption and economic growth. 2004

[6] Bohm D C. Electricity consumption and economic growth in the European Union: a causality study using panel unit root and cointegration analysis. In: 5th International Conference on European Electricity Market, Lisboa, Portugal, 2008, 1–6

[7] Narayan P K, Smyth R, Prasad A. Electricity consumption in G7 countries: a panel cointegration analysis of residential demand elasticities. Energy Policy, 2007, 35(9): 4485–4494

[8] Narayan P K, Prasad A. Electricity consumption-real GDP causality nexus: Evidence from a bootstrapped causality test for 30 OECD countries. Energy Policy, 2008, 36(2): 910–918

[9] Magazzino C. On the relationship between disaggregated energy production and GDP in Italy. Energy & Environment, 2012, 23(8): 1191–1208

[10] Franco D. Indagine conoscitiva sulle caratteristiche e sullo sviluppo del sistema industriale, delle imprese pubbliche e del settore energetico. 2012-09-26

[11] European Commission. Quarterly report on european electricity markets (First quarter, 2013). 2013

[12] NUS Consulting Group. 2011–2012 International Electricity & Natural Gas Report & Price Survey. 2012-06

[13] Chen S T, Kuo H I, Chen C C. The relationship between GDP and electricity consumption in 10 Asian countries. Energy Policy, 2007, 35(4): 2611–2621

[14] Ozturk I, Acaravci A. The causal relationship between energy consumption and GDP in Albania, Bulgaria, Hungary and Romania: Evidence from ARDL bound testing approach. Applied Energy, 2010, 87(6): 1938–1943

[15] Murry D A, Nan G D. A definition of the gross domestic product-electrification interrelationship. The Journal of Energy and Development, 1996, 19(2): 275–283

[16] Squalli J, Wilson K. A bounds analysis of electricity consumption and economic growth in the GCC. Zayed University EPRU Working Papers, 2006, Paper No. 06–09

[17] Wolde-Rufael Y. Electricity consumption and economic growth: a time series experience for 17 African countries. Energy Policy, 2006, 34(10): 1106–1114

[18] Zachariadis T, Pashourtidou N. An empirical analysis of electricity consumption in Cyprus. Energy Economics, 2007, 29(2): 183–198

[19] Acaravci A, Ozturk I. Electricity consumption-growth nexus: evidence from panel data for transition countries. Energy Economics, 2010, 32(3): 604–608

[20] Yoo S H, Kwak S. Electricity consumption and economic growth in Seven South American countries. Energy Policy, 2010, 38(1): 181–188

[21] Abosedra S, Dah A, Ghosh S. Electricity consumption and economic growth, the case of Lebanon. Applied Energy, 2009, 86(4): 429–432

[22] Ghosh S. Electricity consumption and economic growth in India. Energy Policy, 2002, 30(2): 125–129

[23] Ghosh S. Electricity supply, employment and real GDP in India: evidence from cointegration and Granger-causality tests. Energy Policy, 2009, 37(8): 2926–2929

[24] Narayan P K, Smyth R. Electricity consumption, employment and real income in Australia: evidence from multivariate Granger causality tests. Energy Policy, 2005, 33(9): 1109–1116

[25] Yoo S H. The causal relationship between electricity consumption and economic growth in the ASEAN countries. Energy Policy, 2006, 34(18): 3573–3582

[26] Yoo S H, Kim Y. Electricity generation and economic growth in Indonesia. Energy, 2006, 31(14): 2890–2899

[27] Mozumder P, Marathe A. Causality relationship between electricity consumption and GDP in Bangladesh. Energy Policy, 2007, 35(1): 395–402

[28] Squalli J. Electricity consumption and economic growth: bounds and causality analyses for OPEC members. Energy Economics, 2007, 29(6): 1192–1205

[29] Jamil F, Ahmad E. The relationship between electricity consumption, electricity prices and GDP in Pakistan. Energy Policy, 2010, 38(10): 6016–6025

[30] Lean H H, Smyth R. Multivariate Granger causality between electricity generation, exports, prices and GDP in Malaysia. Energy, 2010, 35(9): 3640–3648

[31] Aqeel A, Butt M S. The relationship between energy consumption and economic growth in Pakistan. Asia-Pacific Development Journal, 2001, 8(2): 101–110

[32] Shiu A, Lam P L. Electricity consumption and economic growth in China. Energy Policy, 2004, 32(1): 47–54

[33] Wolde-Rufael Y. Disaggregated industrial energy consumption and GDP: the case of Shanghai. Energy Economics, 2004, 26(1): 69–75

[34] Altinay G, Karagol E. Electricity consumption and economic growth: evidence from Turkey. Energy Economics, 2005, 27(6): 849–856

[35] Halicioglu F. Residential electricity demand dynamics in Turkey. Energy Economics, 2007, 29(2): 199–210

[36] Ho C Y, Siu K W. A dynamic equilibrium of electricity consumption and GDP in Hong Kong: an empirical investigation. Energy Policy, 2007, 35(4): 2507–2513

[37] Narayan P K, Singh B. The electricity consumption and GDP nexus for the Fiji Islands. Energy Economics, 2007, 29(6): 1141–1150

[38] Yuan J, Zhao C, Yu S, Hu Z. Electricity consumption and economic growth in China: co-integration and co-feature analysis. Energy Economics, 2007, 29(6): 1179–1191

[39] Acaravci A. Structural breaks, electricity consumption and economic growth: evidence from Turkey. Journal for Economic Forecasting, 2010, (2): 140–154

[40] Chandran V G R, Sharma S, Madhavan K. Electricity consumption-growth nexus: the case of Malaysia. Energy Policy, 2010, 38(1): 606–612

[41] Sarker A R, Alam K. Nexus between electricity generation and economic growth in Bangladesh. Asian Social Science, 2010, 6(12): 16–22

[42] Yang H Y. A note of the causal relationship between energy and GDP in Taiwan. Energy Economics, 2000, 22(3): 309–317

[43] Jumbe C B L. Cointegration and causality between electricity consumption and GDP: empirical evidence from Malawi. Energy Economics, 2004, 26(1): 61–68

[44] Morimoto R, Hope C. The impact of electricity supply on economic growth in Sri Lanka. Energy Economics, 2004, 26(1): 77–85

[45] Yoo S H. Electricity consumption and economic growth: evidence from Korea. Energy Policy, 2005, 33(12): 1627–1632

[46] Cifter A, Ozun A. Multi-scale causality between energy consumption and GNP in emerging markets: evidence from Turkey. MPRA Papers, 2007, Paper No. 2483

[47] Aktaş C, Yilmaz V. Causality between electricity consumption and economic growth in Turkey. Journal of the Social Sciences, 2008, 4(8): 45–54

[48] Tang C F. A re-examination of the relationship between electricity consumption and economic growth in Malaysia. Energy Policy, 2008, 36(8): 3077–3085

[49] Odhiambo N M. Electricity consumption and economic growth in South Africa: a trivariate causality test. Energy Economics, 2009, 31(5): 635–640

[50] Lean H H, Smyth R. On the dynamics of aggregate output, electricity consumption and exports in Malaysia: evidence from multivariate Granger causality tests. Applied Energy, 2010, 87(6): 1963–1971

[51] Lorde T, Waithe K, Francis B. The importance of electrical energy for economic growth in Barbados. Energy Economics, 2010, 32(6): 1411–1420

[52] Ouédraogo I M. Electricity consumption and economic growth in Burkina Faso: a cointegration analysis. Energy Economics, 2010, 32(3): 524–531

[53] Tang C F, Shahbaz M. Revisiting the electricity consumption-growth nexus for Portugal: evidence from a Multivariate framework analysis, MPRA Papers, 2011, Paper No. 28393

[54] Engle R E, Granger C W J. Co-integration and error correction: representation, estimation and testing. Econometrica, 1987, 55(2): 251–276

[55] Dickey D A, Fuller W A. Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association, 1979, 74(366): 427–431

[56] Phillips P C B, Perron P. Testing for a unit root in time series regression. Biometrika, 1988, 75(2): 335–346

[57] Elliott G, Rothenberg T J, Stock J H. Efficient tests for an autoregressive unit root. Econometrica, 1996, 64(4): 813–836

[58] Kwiatkowski D, Phillips P C B, Schmidt P, Shin Y. Testing the null hypothesis of stationarity against the alternative of a unit root: How sure are we that economic time series have a unit root? Journal of Econometrics, 1992, 54(1–3): 159–178

[59] Zivot E, Andrews D W K. Further evidence on the Great Crash, the oil price shock, and the unit-root hypothesis. Journal of Business & Economic Statistics, 1992, 10(3): 251–270

[60] Clemente J, Montañés A, Reyes M. Testing for a unit root in variables with a double change in the mean. Economics Letters, 1998, 59(2): 175–182

[61] Johansen S, Juselius K. Maximum likelihood estimation and inference on cointegration with applications to the demand for money. Oxford Bulletin of Economics and Statistics, 1990, 52(2): 169–210

[62] Belloumi M. Energy consumption and GDP in Tunisia: cointegration and causality analysis. Energy Policy, 2009, 37(7): 2745– 2753

[63] Bianco V. Manca O. Nardini S. Electricity consumption forecasting in Italy using linear regression models. Energy, 2009, 34(9): 1413– 1421

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg