International Journal of Sustainable Development & World Ecology ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tsdw20 Sustainable electricity generation: the possibility of substituting fossil fuels for hydropower and solar energy in Italy Sakiru Adebola Solarin, Mufutau Opeyemi Bello & Festus Victor Bekun To cite this article: Sakiru Adebola Solarin, Mufutau Opeyemi Bello & Festus Victor Bekun (2021) Sustainable electricity generation: the possibility of substituting fossil fuels for hydropower and solar energy in Italy, International Journal of Sustainable Development & World Ecology, 28:5, 429-439, DOI: 10.1080/13504509.2020.1860152 To link to this article: https://doi.org/10.1080/13504509.2020.1860152 Published online: 20 Dec 2020. Submit your article to this journal Article views: 866 View related articles View Crossmark data Citing articles: 24 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=tsdw20 INTERNATIONAL JOURNAL OF SUSTAINABLE DEVELOPMENT & WORLD ECOLOGY 2021, VOL. 28, NO. 5, 429–439 https://doi.org/10.1080/13504509.2020.1860152 Sustainable electricity generation: the possibility of substituting fossil fuels for hydropower and solar energy in Italy Sakiru Adebola Solarina, Mufutau Opeyemi Bellob and Festus Victor Bekunc,d aSchool of Economics, University of Nottingham, Malaysia Jalan Broga, Malaysia; bFaculty of Social Sciences, Department of Economics, University of Ilorin, Ilorin, Nigeria; cFaculty of Economics Administrative and Social Sciences, Istanbul Gelisim University, Istanbul, Turkey; dDepartment of Accounting, Analysis and Audit, School of Economics and Management, Chelyabinsk, Russia ABSTRACT ARTICLE HISTORY Electricity remains the most important form of end-use energy consumption and an important Received 12 November 2020 factor for economic growth and development. However, electricity generation also constitutes Accepted 2 December 2020 a great source of concern for global warming and climate change with threats to sustainable KEYWORDS development as fossil fuels dominate electricity generation fuel mix for most economies of the Interfuel substitution; world. This is exactly the case for Italy with fossil fuels dominating electricity generation fuel electricity; solar energy; mix over the last couple of decades. Thus, there is a need to reverse this trend by increasing the hydropower; translog shares of renewable energy sources in the fuel mix. This study, therefore, investigates the production function; Italy potentials for renewable energy sources of hydropower and solar energy to substitute the fossil fuels of coal and natural gas in electricity generation for Italy. Adopting the ridge regression procedure to obtain the parameter estimates, the results provide evidence that substitution is possible among all fuels. While both hydropower and solar energy are found to be substitutable for the fossil fuels, solar energy is found to provide more substitutability than hydropower. This implies that Italy has the potential to gradually move away from the carbon- intensive fossil fuels of coal and gas to more environmental-friendly solar energy and hydro- power in the process of generating electricity. 1. Introduction features are beneficial in meeting abrupt changes in electricity demand and to augment supply from inflex- Electricity generation is one of the main sources of ible sources of electricity. The water stored in reservoirs emission across the globe. According to the of dams can be utilized as a key source of providing International Energy Association (2020), electricity water for diverse purposes such as irrigation for agri- and heat production generated 13 billion tonnes of cultural products (Sachdev et al. 2015). carbon dioxide emissions or 41% of the total carbon The use of solar energy confers several advantages dioxide emissions from fuel combustion in the globe in in the economy. For instance, the cost of solar energy is 2017. The major reason for this situation is because usually negligible, beyond the initial cost outlay. The fossil fuels dominate the electricity mix. About 16,947 operational labour requirement of solar energy is enor- terawatt-hours or 63% of the total electricity was gen- mously lower than the conventional power plant. Since erated through fossil fuel sources. Only 4,222 terawatt- it is a locally sourced form of energy, solar power can hours or 16% of the total electricity was generated improve energy security. Therefore, solar energy plays through hydropower and 724 terawatt-hours or 3% a crucial role by supporting the economy of a nation. of the total electricity was generated through solar Energy security ensures that a nation is less susceptible energy (British Petroleum 2020). In addition to the to external events that might raise the price of domes- fact that the use of these two sources of electricity tic energy products. A particularly advantageous and will lead to less pollution, they also have several advan- relevant characteristic of solar energy production is tages. Unlike fossil fuels such as gas, coal or oil, hydro- that it generates employment opportunities. In 2018, power is environmental-friendly as it releases very the solar photovoltaic industry was able to support small amounts of greenhouse gases (GHG). Beyond more than over 3.6 million jobs globally (International its function of promoting energy security and decreas- Renewable Energy Agency 2019). ing a country’s reliance on fossil fuels, hydropower In order to reap the benefits of these two sources of provides the prospects for poverty alleviation. The electricity, their usage must be increased and to success- development of hydropower infrastructures is inti- fully increase their usage, these two sources of electricity mately associated with local and global development should have the potential to substitute the fossil fuels, policies. Its quick response storage and capacity which are currently dominant in the electricity sector. The CONTACT Sakiru Adebola Solarin sakirusolarin@gmail.com Centre for Globalisation and Sustainability Research, Multimedia University, Melaka 75450, Malaysia © 2020 Informa UK Limited, trading as Taylor & Francis Group 430 S. A. SOLARIN ET AL. feasibility of such substitution can be examined within profiles of the major economies of the world- the inter-fuel substitution framework as exemplified in a scenario that meant that most countries were adversely Berndt and Wood (1975). However, interfuel substitution affected by the sudden oil glut of the 1973 to 1974 per- studies that incorporate hydropower and solar energy iod. This development awakened the interest of stake- into the analyses are rare with most studies focusing on holders including policymakers and researchers on examining the substitutability relationships between the exploring the possibility of substituting oil for other alter- four popular energy sources of coal, gas, oil and electri- native sources of energy. Consequently, Berndt and city. Besides, these studies have also been largely con- Wood (1975) produced a novel paper on the subject. ducted within the framework of the aggregate energy Generally speaking, the research efforts are of two sector with limited attention on the electricity sector and strands, namely, inter-factor substitution and inter-fuel interfuel substitution studies on Italy as a country has not substitution. Inter-factor substitution deals with investi- been adequately considered. gating the possibility of substituting energy for other The aim of the paper is, therefore, to contribute to the primary factors of production such as labour and capital existing literature by investigating the potential of sub- while inter-fuel substitution investigates the substitution stituting the fossil fuels – coal and gas for both hydro- possibilities among competing energy fuels. Berndt and power and solar energy in the process of generating Wood (1975) paper was essentially an inter-factor sub- electricity in Italy. We have selected Italy because of stitution exploration as it investigated the substitution several reasons. Firstly, with a real GDP of US$2.1 billion possibility between energy and the primary factors of (constant 2010 prices), Italy was fourth largest economy labour and capital. Their study, which focused on the in Europe after Germany, United Kingdom and France in US manufacturing sector for the period 1947–1971, 2019 (World Bank Group 2020). Secondly, by consuming found substitutability relationships between energy and 6.37 exajoules of energy, the country accounted for 7.6% labour and complementary relationships between of the total energy consumed in the continent and was energy and capital. the fifth in the continent after Germany, France, United Since then, majority of the interfuel substitution Kingdom and Turkey (British Petroleum 2020). Thirdly, by studies have focused on investigating the relationships generating 283 terawatt-hours of electricity in 2019, Italy between the four main energy fuels of oil, natural gas, was the fifth largest electricity generator in Europe, after electricity and coal. Some of the notable studies in this Germany, France, United Kingdom and Turkey. Fourthly, regard include Lin et al. (2016) on Ghana, Ma and Stern by producing 325 million tonnes of carbon dioxide in (2016) on Chinese provinces, Lin and Atsagli (2017) on 2019, the country has the fourth biggest carbon dioxide Nigeria, Lin and Tian (2017) on China, Wesseh and Lin in Europe, after Germany, United Kingdom and Turkey (2018) on Egypt and Considine (2018) on the and accounts for 8% of the total carbon dioxide emitted US. However, in more recent times, authors are in the continent (British Petroleum 2020). Fifthly, similar expanding the scope of analysis by exploring other to the situation in several nations, fossil fuels dominate areas of research in interfuel substitution. For instance, both the energy mix and electricity mix in the country Li et al. (2019) examined the underlying dynamics in (Güney and Kantar 2020; Smith and Archer 2020; the coal market in China. The study also reveals that Sudsawasd et al. 2020). Hence, this has made the elec- inter-fuel substitution is a major issue in the coal mar- tricity sector as one of the major causes of the country’s ket in China. emissions. According to the International Energy Contributing to the research, Serletis and Xu (2019) Association (2020), electricity and heat production gen- employed the Markov Switching Minflex Laurent erated 109 million tonnes of carbon dioxide emissions or demand system to investigate interfuel substitution 34% of the total carbon dioxide emissions from fuel in the United States for the period 1919 to 2012. combustion in the country in 2017. Lastly, the use of Their results provided strong evidence in support of hydropower and solar energy account for 16% and 9% substitutability relationships between all energy pairs. of the total electricity (British Petroleum 2020). Lin and Abudu (2020) estimated a translog production The other parts of the paper include Section 2 that function with the ridge regression approach to mea- involves the existing literature, while Section 3 con- sure energy intensity and inter-fuel substitution for tains the methodology used in this study. Section 4 Ghana for the period spanning 2000 to 2015. They presents the empirical findings and Section 5 involves also provided evidence for substitution among the the conclusion of the paper. energy resources. However, due to the exacerbating problem of glo- bal warming and the concerns for resource sustainabil- 2. Literature review ity, several variants of renewable energy sources are Research into the substitutability relationships between increasingly being included in the analyses. For different energy fuels has a rich history. It dates back to instance, Jones (2014), introduced biomass as the the end of the first post-oil shock of the 1970s. Prior to fifth fuel alongside the traditional fuels of coal, natural this period, oil dominated the energy consumption gas, oil and electricity for the US into the analysis and INTERNATIONAL JOURNAL OF SUSTAINABLE DEVELOPMENT & WORLD ECOLOGY 431 found substantial evidence that biomass can substi- concept of sustainable development in a study that tute natural gas in US energy profile. In a similar study investigate possibilities of substitution between biomass for the US, Suh (2016) was also able to establish sub- and fossil fuels in Brazil for the period 1980–2015. The stitutability between coal and biomass and between study provided strong evidence of substitutability natural gas and biomass while a complementary rela- between biomass and fossil fuels and that sustainable tionship between biomass and electricity was estab- development index can reveal some of the inherent lished. Kumar et al. (2015), in a study involving 12 negative effects of fossil-fuel consumption in the designated industries from the countries belonging economy. to the Organisation for Economic Cooperation and The above survey of the related literature reveals Development (OECD), examined the nature of the sub- two important points. First, most of the studies on stitutability relationships between the renewable and interfuel substitution have been conducted within the non-renewable energy sources. The result pro- the context of aggregate energy sector with little duced a negative substitution elasticity estimate attention being accorded to the electricity sector which supports the complementarity relationship despite its important not only to the energy sector between the two energy sources. but also to the aggregate economy. For instance, In another study conducted by Wesseh and Lin (2016), while other sub-sectors, such as the manufacturing, to investigate the relationships between renewable and transportation, iron and steel and chemical sectors, non-renewable energy sources in the economic commu- have been given much attention to in the literature, nity for West African countries (ECOWAS) member coun- as far as we know, only few studies such as Bello et al. tries, the ridge regression procedure was employed to (2018), (2020) have specifically focused on interfuel obtain the parameters of a translog production function. substitution solely within the electricity sector and Their results provided evidence in support of substitut- none of these studies has simultaneously considered ability between renewable and non-renewable energy for two renewable energy sources such as hydropower ECOWAS member countries. Focusing on a component of and solar energy. Secondly, with specific reference to renewable energy, Bello et al. (2018) introduced hydro- Italy, there is a dearth of empirical study on interfuel power in a study on three transition economies, namely, substitution. While Italy has been included in several Malaysia, China and Thailand. Using the ridge regression studies such as Griffin (1977), Hall (1986), Jones (1996), procedure, the results show evidence of strong substitu- Renou-Maissant (1999), Morana (2000), and Serletis tion between renewable energy represented by hydro- et al. (2010), (2011), that focus on OECD countries, power and non-renewable energy represented by coal, single country-specific studies with Italy as the main gas and oil that are employed for generating electricity in focus is rare. As far as we know, Bardazzi et al. (2015) is those countries. A similar study was also conducted by perhaps the only study to have singularly conducted Bello et al. (2020) on Malaysia using the translog cost a full single-country specific interfuel substitution function framework where hydropower was found to be study focusing on Italy. It is important to conduct substituted for the fossil fuels of coal, and natural gas used a single-country specific study of interfuel substitution in the generation of electricity for the country. The study on Italy as such a study is likely to generate a more also showed that trading off the fossil fuels for hydro- accurate policy inference by taking into considerations power in electricity has the potential to reduce carbon the distinctive characteristics of the country. This pre- dioxide emissions. sent study is a departure from Bardazzi et al. (2015) in Contributing to the research while also factoring in the sense that it is more comprehensive as it con- the dynamics of renewable energy source, Suh (2019) ducted within the context of the aggregate economy examined the interfuel substitution impacts of biofuel while Bardazzi et al. (2015) was essentially a firm-level usage on carbon dioxide emissions in the transportation analysis. Thus, we attempt to fill this gap in the litera- sector of the US. Using a dynamic linear logit approach, ture by obtaining the estimate of substitution elasti- the results reveal substitutability relationships between cities for Italy using macro-level data while ethanol and petroleum, complementary relationships incorporating variants of renewable energy sources between ethanol and natural gas, while natural gas has into the analysis. the potentials to substitute petroleum. Lin and Ankrah (2019) incorporated the dynamics of renewable energy into the analysis for Nigeria. Using the ridge regression 3. Methodology: model, data and estimation procedure, the authors found the existence of substitut- technique ability relationship between renewable and non- 3.1 Model renewable energy while noting that output is primarily driven by capital and labour with both renewable and A linear production function relating output to input is non-renewable playing insignificant roles in output gen- specified as follows: eration. In a somewhat different study, Solarin and Bello Q ¼ qðK; L; EÞ: (1) (2019) extended the analysis by further introducing the 432 S. A. SOLARIN ET AL. where output level Q is dependent on the amount of @LogðQtÞη ¼ total capital stock, numbers of labours employed and Ht @LogðHhtÞ the quantity of energy resources. The energy input is ¼ βH þ βCHLogðCtÞ þ βGHLogðGtÞ þ βHSLogðStÞ weakly separable and homothetic in its different forms þ 2βHHLogðHtÞ; (8) of coal energy (C), natural gas (G), hydropower (H) and solar energy (S). Thus, Equation (1) is further re- @LogðQ η t Þ specified as: St ¼ @LogðSstÞ ¼ β þ β LogðCtÞ þ β LogðGtÞ þ β LogðHtÞ Q ¼ qðK; L; E C G H S S CS GS HSð ; ; ; ÞÞ: (2) þ 2βSSLogðStÞ: (9) Equation (2) is then transformed into a twice differenti- The calculated output elasticity estimates are then able transcendental logarithm (translog) production used to generate the substitution elasticity estimates function of the form: between two energy pairs using the following formula: X XX LogðQtÞ ¼ β0 þ βi LogðX h i itÞ þ 0:5 βij LogðXitÞLogðXjtÞ: (3) � � � � β 1 2 β β η η β η η η η 1 1 i i j ij ¼ þ ij iið j= iÞ jjð i= jÞ : i þ j ; where Q is the output and the Xs are the various units ði�j;¼ C;G;H; SÞ: of input combinations capital stock, labour, coal, nat- (10) ural gas, hydropower and solar energy with subscripts i and j representing such combinations. Period, which is Equation (10) gives the symmetric substitution elasti- annual in this instance, is denoted by subscript t and city between two energy pairs, that is the elasticity the β are the estimable parameters. estimates are symmetry, i.e. ðβij ¼ βjiÞ. Thus, the sub-s Equation (3), expressed in the logarithm form, is the stitution elasticity estimate between the respective general form of a second order Taylor Series approx- energy pairs is obtained as follows: imation. The specific form of the model is specified as Coal and Natural Gas : σCG ¼ ½1þ 2½βCG βCCðηG=ηCÞ follows: 1 1 LogðQtÞ ¼ β0 þ βK LogðKtÞ þ βL LogðLtÞ þ βC LogðCtÞ þ βG LogðG Þ þ β LogðH Þ βGGðηC=ηGÞ� � ½η þ ηt H t C G� � ; (11) þ βSLogðStÞ þ βCGLogðCtÞLogðGtÞ þ βCH LogðCtÞLogðHtÞ þ βCS LogðCtÞLogðStÞ þ βGHLogðGtÞLogðHtÞ þ βGSLog 2 ðGtÞLogðStÞ þ βHS LogðHtÞLogðStÞ þ βCC LogðCtÞ Coal and Hydropower : σ ¼ ½1þ 2½β β 2 2 2 CH CH CCðηH=ηCÞ þ βGG LogðGtÞ þ βHH LogðHtÞ þ βSS LogðStÞ : (4) β η η η η 1 1 ð = Þ� � ½ þ � � ; (12) As the core aim of this paper is to investigate the HH C H C H substitutability between the energy inputs, we have Coal and Solar Energy : σ ¼ ½1þ 2½β β ðη =η Þ only included their translog terms in Equation (4) to CS CS CC S C prevent over-parameterization. Using the parameter 1 1 estimates in Equation (4), it is feasible to compute the βSSðηC=ηSÞ� � ½ηC þ ηS� � ; (13) estimate of the output of elasticity which is required to obtain the substitution elasticity estimates between Gas and Hydropower :σGH ¼ ½1þ 2½βGH βGGðηH=ηGÞ the energy pairs. The output elasticity estimate of 1 1 each of the energy series is obtained as follows: βHHðηG=ηHÞ� � ½ηG þ ηH� � ; (14) @LogðQtÞ Xηit ¼ ¼ β þ β LogðX Þ > 0; (5) Gas and Solar Energy :σjt GS¼½1þ 2½βi ij GS βGGðηS=ηGÞ @LogðXitÞ j β 1 1 SSðη =η Þ��½η þη � � ; (15) where ηitis the output elasticity of an input i at time t. G S G S Thus, for each of the energy series coal, natural gas, Hydro and Solar Energy :σHS¼½1þ 2½β αHHðη =η Þ hydropower and solar energy, the output elasticities HS S H are calculated, respectively, as follows: β 1 1SSðηH=ηSÞ��½ηHþηS� � ; (16) @LogðQ η t Þ Ct ¼ In Equations (11) to (16), the decision rule is that posi- @LogðCctÞ tive estimates imply substitutability while negative ¼ βC þ βCGLogðGtÞ þ βCHLogðHtÞ þ βCSLogðStÞ 2β Log C (6) estimates suggest complementarity between two þ CC ð tÞ; energy inputs. @LogðQ η t Þ Gt ¼ @LogðGgtÞ 3.2 Data ¼ βG þ βCGLogðCtÞ þ βGHLogðHtÞ þ βGSLogðStÞ þ 2β LogðG Þ; (7) The dataset entails annual time series data of Italy’s GG t real gross domestic product (GDP), gross fixed capital INTERNATIONAL JOURNAL OF SUSTAINABLE DEVELOPMENT & WORLD ECOLOGY 433 formation (GFCF), labour, coal, natural gas, hydro- variables exceeds 100 thereby establishing the exis- power and solar energy consumption for the period tence of an extreme multicollinearity problem and 1989 to 2018. Data on real GDP and GFCF series were rendering the application of the OLS technique unsui- generated from the World Bank Group (2020)’ World table in this circumstance. Development Indicators at constant 2010US$ to control To remedy this problem, Hoerl (1962) developed for inflationary trend. The labour series, calculated as a special regression procedure known as the ridge a number of persons employed (in thousand per per- regression procedure. The procedure is a slight mod- sons), were obtained from The Conference Board ification of the ordinary least squares regression esti- (2020). The British Petroleum (2020)’s Statistical mate with the introduction of a ridge parameter called Review of World Energy provided the data on the a biasing constant (p). Therefore, the original matrix for energy series and are stated in million tonnes of oil the OLS coefficient estimate β 0 1OLS¼ ðX XÞ X0Yis mod- equivalent (MTOE). In order to prevent double count- ified into a ridge specification as ing and data overstatement, we have only included β ¼ ðX0XþpI 1Þ X0R Ywhere p is the penalty parameter only the volume of coal, natural gas, hydropower and with a range of values from zero to unity and Iis an solar energy used in the generation of electricity for identity matrix. The ordinary least squares estimates Italy. correspond to the ridge regression estimates with a penalty parameter of zero. The target is to select an optimum p value for which the mean squared error of 3.3 Estimation technique: Ridge regression the ordinary least squares estimator is more than ridge Translog models involving squared polynomial terms regression estimator, implies a lesser bias in the such as the one specified in Equation (4) are suscepti- estimation. ble to severe multicollinearity problem can cause To achieve this, Hoerl and Kennard (1970) proposed a serious problem in model estimation. In cases of the use of the ridge trace as a systematic way of extreme multicollinearity, model estimation is severely determining the optimum value of p. The ridge trace marred by deflated t-statistics due to exaggeration of plots the ridge regression parameters as a function of the standard errors, thereby leading to not only non- p and the value of p for which the regression para- significant probability values but also misleading para- meters stabilize is selected as the optimum. A penalty meter estimates. Under this circumstance, the adop- parameter of 0.874 has been chosen as the optimum tion of the usual ordinary least squares estimates is no value of p based on the ridge trace plot shown in longer consistent. Figure 1 as the parameter estimates seem to stabilise Thus, we commence the analysis by first testing for around this value. the extent of multicollinearity in the model by examin- Furthermore, Table 2 is also used to show the effect ing the variance inflation factors (VIFs) of the regres- of the ridge regression procedure on the variance sors and the condition number of the eigenvalues of inflation factors. As can be seen, varying the penalty correlation of the variables. The outputs of the multi- parameters reduces the variance inflation factors. The collinearity analysis, available in Table 1, show that not zero value of the penalty parameter corresponds to the only is the variance inflation factors for each of the variance inflation factors for the OLS estimates which regressors significantly exceed 10 but also the condi- are very large, but gradual increment in the penalty tion number of the eigenvalues of correlation of some parameter continues to decrease the variance inflation Table 1. Least squares multicollinearity test result. Independent Variable Variance Inflation Factors Eigenvalues Condition Number Log (K) 50.929 10.292 1.0000 Log (L) 71.2276 3.32031 3.1000 Log (C) 9949.142 1.431199 7.1900 Log (G) 23,600.59 0.713818 14.420 Log (H) 8229.186 0.120297 85.560 Log (S) 18,418.81 0.089578 114.90 Log (C)*Log (G) 31,826.77 0.026957 381.83 Log (C)*Log (H) 9973.269 0.00204 5046.6 Log (C)*Log (S) 3292.604 0.001658 6206.39 Log (G)*Log (H) 7526.564 0.000485 21,232.23 Log (G)*Log (S) 5169.585 0.000403 25,570.43 Log (H)*Log (S) 5110.836 0.000162 63,536.7 Log (C)*Log (C) 15,464.56 0.000081 126,515.81 Log (G)*Log (G) 17,655.13 0.00006 171,827.6 Log (H)*Log (H) 7396.401 0.000018 579,516.26 Log (S)*Log (S) 20.1889 0.000015 685,464.84 Multicollinearity is severe as the variance inflation factors exceed 10 and some condition numbers of the eigenvalues of correlations of some variables are more than 100. 434 S. A. SOLARIN ET AL. Figure 1. The ridge trace plot for the selection of optimum penalty parameter. factors until the value of 0.874 where the variance fossil fuels of coal and natural gas have higher output inflation factors for all variables have come under 10 elasticities than the renewable energy sources of and multicollinearity effectively addressed. hydropower and solar energy considered in the study. This is probably due to the fact that the non- renewable sources currently dominate Italy electricity 4. Result and discussion sector. Among the four fuels considered in this study, the non-renewable energy sources of coal and natural Following the determination of the optimum p value, gas accounted for more than 70% of the fuel mix over the results of the ridge regression procedure are pre- the course of the sample period. It is also noted that sented in Table 3 which also displays the variance solar energy has the least output elasticity amongst all inflation factors along with the parameter estimates. the energy input considered, a possible reflection of We now present the results of the parameter estimates the fact that it has the least share in the electricity of the ridge regression procedure in Table 3. The table generation fuel mix, about 2.5%, over the course of displays the values of the variance inflation factor for the sample period. each of the parameters and as can be seen these From the estimates of the output of elasticity, values are below 10 thereby establishing the fact that the estimates of the substitution elasticities the problem of multicollinearity has been effectively between the energy pairs are calculated and the solved. In addition to this, the f-ratio is significant at 1% resulted are presented in Table 5. The results show level with an R-squared of 83.4% indicating a strong that all energy pairs considered are substitutes. The goodness of fit and explanatory power of the para- highest substitutability estimate occurred between meters in the model. This is also reflected in the sig- hydropower and solar energy, followed by between nificance levels of the parameters with the majority coal and solar energy and then between gas and being significant at the 1% level. solar energy. This implies that the highest substitu- From the parameter estimates of the ridge regres- tion estimate occurs between solar energy and each sion available in Table 3, the output elasticities of each of the energy input. of the energy inputs are obtained using Equations (6), The plausible logic for this outcome is seen in her (7), (8) and (9), respectively, for coal, natural gas, hydro- economic strategic drive for increase and sustainable power and solar energy and the results are presented energy as the energy inputs show a positive relation- in Table 4. The result shows that the average output ship with output (GDP). This result further gives cre- elasticities, over the sample period, for all energy series dence to the energy-induced growth hypothesis, are positive thus satisfying the positivity condition which is indicative for Italy economy given that her imposed by Equation (3). It is also noted that both energy mix is currently driven by fossil fuel. However, INTERNATIONAL JOURNAL OF SUSTAINABLE DEVELOPMENT & WORLD ECOLOGY 435 Table 2. Effect of penalty parameter (p) on the variance inflation factor. p Log (K) Log (L) Log (C) Log (G) Log (H) Log (S) Log (C)*Log G) Log (C)*Log H) Log (C)*Log (S) 0.000 50.903 71.228 9949.142 23,600.590 8229.18618,418.808 31,826.770 9973.269 3292.604 0.001 23.1341 13.9012 64.1715 56.8151 60.0758 90.8676 67.581 56.488 125.074 0.002 20.5174 12.0711 23.3704 21.5744 21.631 33.4558 29.117 21.821 55.278 0.003 18.8021 11.2322 12.7099 12.1443 11.6209 17.8725 17.245 12.400 32.743 0.004 17.4512 10.6485 8.2021 8.1812 7.4148 11.3317 11.684 8.212 22.0226 0.005 16.3099 10.1757 5.8378 6.1136 5.2221 7.9419 8.5636 5.9293 15.9686 0.006 15.3125 9.7673 4.4323 4.8837 3.9256 5.9464 6.6167 4.5329 12.1832 0.007 14.4245 9.4029 3.5241 4.0836 3.0922 4.6658 5.3114 3.6113 9.6463 . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.874 0.0943 0.1053 0.0748 0.0465 0.1064 0.0377 0.0189 0.0703 0.0415 (contd.) p Log (G)*Log (H) Log (G)*Log (S) Log (H)*Log (S) Log (C)*Log (C) Log (G)*Log (G)Log (H)*Log (H)Log (S)*Log (S) 0.000 7526.564 5169.585 5110.836 15,464.556 17,655.128 7396.401 20.189 0.001 128.542 135.511 150.378 74.820 91.185 63.037 9.046 0.002 51.001 71.698 68.257 31.842 35.048 20.989 8.460 0.003 28.425 45.135 39.894 18.544 19.491 10.709 8.110 0.004 18.545 31.2958 26.3987 12.3984 12.8544 6.6194 7.8478 0.005 13.2752 23.1281 18.8593 8.996 9.3642 4.5741 7.6299 0.006 10.1084 17.8882 14.2014 6.8988 7.2797 3.4018 7.4374 0.007 8.0455 14.3165 11.115 5.5087 5.9223 2.6658 7.2618 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.874 0.0408 0.0501 0.0384 0.0753 0.0417 0.106 0.133 The first row where p is the OLS’s Variance Inflation Factors. The last row where p is 0.874 gives the optimal VIF values for the ridge regression where the problem of severe multicollinearity has been corrected. Table 3. Ridge regression parameter estimates. Independent Variable Parameter Estimates t-Stat Variance Inflation Factor Constant 23.985 Log (K) 0.092a*** 4.023 0.094 Log (L) 0.094* 1.661 0.105 Log (C) 0.008 0.992 0.075 Log (G) 0.021*** 6.101 0.047 Log (H) 0.003 0.140 0.106 Log (S) 0.0005 0.978 0.038 Log (C)*Log (G) 0.0027*** 6.819 0.019 Log (C)*Log (H) 0.0014 0.723 0.070 Log (C)*Log (S) 0.00008 0.557 0.042 Log (G)*Log (H) 0.0045*** 5.663 0.041 Log (G)*Log (S) −0.00005 −0.342 0.050 Log (H)*Log (S) 0.0001 0.755 0.038 Log (C)*Log (C) 0.0012 0.965 0.075 Log (G)*Log (G) 0.0024*** 6.184 0.042 Log (H)*Log (H) −0.0005 −0.197 0.106 Log (S)*Log (S) −0.00103*** −2.685 0.133 R Squared: 0.834 F-ratio: 4.0834 (0.007) ***implies 1% level of significance, * implies 10% level of significance. Figures in parenthesis are probability values. the current energy-induced growth is not sustainable from the consumption of solar and hydropower energy. as it is depleting the quality of the environment from This outcome resonates with the study of Gyamfi et al. plants with a combustion engine that emits carbon (2020) for emerging (E7) countries, where hydropower dioxide emissions. Additionally, the United States energy significantly contributes to E7 economies. Energy Information Administration Agency (Energy Furthermore, in our empirical analysis, it is seen that Information Administration 2018) alluded to the pivo- the reported elasticity of output estimates and elasticity tal role of energy in driving economic growth. The of substitution estimates are insightful, where we see unanswered puzzle of alternative, clean and sustain- a substantial degree of sustainability among the differ- able energy sources still applies to Italy. ent sources of energy investigated. Interestingly, renew- Interestingly, our empirical results reveal that both able energy from hydropower and solar energy shows hydropower and solar energy can substitute other non- strong evidence to substitute fossil-fuel energy over the renewable energy sources like (coal and natural gas examined period. It is clear that the substitutability of energy). The possible intuition for this is due to the solar energy is greater than hydropower energy in Italy. clean nature and its environmental advantage derived For instance, the substitutability between natural gas 436 S. A. SOLARIN ET AL. Table 4. Elasticity of output estimates. Year ηƐC ηƐG ηƐH ηƐS 1989 0.030 0.064 0.014 0.014 1990 0.031 0.064 0.014 0.012 1991 0.031 0.065 0.014 0.012 1992 0.030 0.064 0.013 0.011 1993 0.030 0.064 0.013 0.010 1994 0.030 0.065 0.014 0.010 1995 0.031 0.065 0.015 0.010 1996 0.031 0.066 0.015 0.010 1997 0.031 0.066 0.016 0.010 1998 0.032 0.067 0.017 0.010 1999 0.033 0.069 0.017 0.009 2000 0.033 0.070 0.018 0.009 2001 0.034 0.071 0.019 0.009 2002 0.034 0.070 0.019 0.009 2003 0.034 0.071 0.020 0.009 2004 0.035 0.072 0.020 0.008 2005 0.035 0.072 0.021 0.008 2006 0.036 0.072 0.021 0.008 2007 0.036 0.072 0.022 0.008 2008 0.036 0.073 0.022 0.004 2009 0.036 0.073 0.021 0.002 2010 0.036 0.073 0.021 0.000 2011 0.036 0.073 0.021 −0.004 2012 0.036 0.072 0.021 −0.005 2013 0.035 0.072 0.020 −0.005 2014 0.035 0.071 0.019 −0.005 2015 0.035 0.071 0.020 −0.005 2016 0.035 0.071 0.021 −0.005 2017 0.035 0.070 0.021 −0.006 2018 0.035 0.071 0.020 −0.006 Average 0.034 0.069 0.018 0.005 Table 5. Elasticity of substitution estimates. Year σCG σGH σCS σGH σGS σHS 1989 1.018 0.913 1.005 0.853 0.902 0.891 1990 1.017 0.918 0.978 0.859 0.888 0.883 1991 1.017 0.913 0.981 0.855 0.883 0.879 1992 1.018 0.911 0.965 0.852 0.867 0.868 1993 1.018 0.912 0.955 0.853 0.861 0.864 1994 1.018 0.914 0.953 0.855 0.861 0.864 1995 1.017 0.919 0.936 0.861 0.855 0.861 1996 1.017 0.920 0.932 0.862 0.853 0.859 1997 1.017 0.924 0.922 0.867 0.850 0.857 1998 1.016 0.927 0.914 0.872 0.847 0.855 1999 1.016 0.931 0.906 0.876 0.845 0.853 2000 1.016 0.934 0.899 0.880 0.842 0.851 2001 1.015 0.935 0.896 0.882 0.840 0.849 2002 1.015 0.935 0.891 0.882 0.837 0.846 2003 1.015 0.938 0.880 0.885 0.831 0.841 2004 1.015 0.939 0.870 0.888 0.823 0.834 2005 1.015 0.942 0.863 0.890 0.820 0.830 2006 1.014 0.942 0.857 0.891 0.814 0.825 2007 1.014 0.944 0.849 0.893 0.809 0.820 2008 1.014 0.943 0.724 0.893 0.695 0.710 2009 1.014 0.941 0.488 0.890 0.474 0.484 2010 1.014 0.942 −0.184 0.891 −0.186 −0.184 2011 1.014 0.942 2.591 0.891 2.256 2.833 2012 1.014 0.941 1.973 0.890 1.773 2.179 2013 1.014 0.938 1.914 0.886 1.713 2.156 2014 1.015 0.936 1.912 0.883 1.702 2.190 2015 1.014 0.939 1.871 0.886 1.682 2.107 2016 1.015 0.940 1.877 0.888 1.692 2.098 2017 1.015 0.942 1.816 0.889 1.649 2.020 2018 1.015 0.940 1.856 0.888 1.673 2.082 Average 1.016 0.932 1.143 0.878 1.042 1.164 and solar is higher than between natural gas and hydro- environmental sustainability. This position aligns with energy source. This suggests the need for more invest- the findings of Serletis et al. (2011) for OECD countries. ment into solar energy plants and photovoltaic energy The results of the possibility of substitutability in the country to drive clean energy targets and among these highlighted variables have its INTERNATIONAL JOURNAL OF SUSTAINABLE DEVELOPMENT & WORLD ECOLOGY 437 environmental implication in Italy’s economy energy found to provide more substitutability than hydro- mix. This assertion is timely and worthy of further power. Interestingly, from the ridge regression, while exploration by stakeholders, energy experts, practi- most of the variables are statistically insignificant tioners who design and formulate energy strategies under consideration of their isolated forms, they in Italy. As outcomes offer policy insights for the need became statistically significant when each variable for more alternative and clean energy sources is interacts with one of the other variables in the model. encouraged. Given that the current county’s (energy) Specifically, the interaction of coal energy with natural electrification is seen as a key driver for sustainable gas and the interaction of natural gas with hydropower growth due to its ability to drive other sectors of the has a positive and significant impact on economic economy like industrial layout and small and medium growth. enterprise. However, there is a trade-off for the quality From a policy perspective, our analysis for Italy has of the environment as non-renewable energy drives by insightful outcomes for policymakers and stakeholders environmental pollution. in terms of her energy mix. We observe a significant Given the aforementioned dynamics of Italy energy relationship between growth drivers (capita and sector and her energy portfolio mix, it is imperative for labour and energy sources both renewable and non- fuel interchange to cleaner electrification sources. renewable energy). This study thus validates the Italy’s is reputed as the fourth largest energy consumer energy-induced growth hypothesis. However, the con- in the European blocs, where the bulk of her energy consumption comprises petroleum and natural gas cern for policymakers is that economy driven by fossil- sources. Interestingly, the government of Italy has fuel energy sources, which are dirty and not environ- had deliberate strides to increase her share of renew- mentally sustainable as seen in Italy where over 70% of able energy consumption from hydroelectricity. For her electrification (energy) is driven from fossil-fuel instance, in 2005 hydroelectricity consumption sources. These non-renewable forms of energy pro- increased from a record of less than 2% to approxi- duce carbon dioxide emissions. Thus, it is imperative mately 10% over a decade. Furthermore, commitment for Italy to adopt the expansion of both solar and of the government includes reinforcing the fourth hydropower generation plants to avoid pollutant emis- national efficiency action plans (NEEAP) to foster sions in the economy. This also establishes the need for national energy innovation, efficiency and energy a paradigm shift in Italy’s energy portfolio mix from security without trade-off for quality of the environ- ment in the economic growth trajectory. non-renewable to renewables, which also includes the construction of clean energy infrastructures. For instance, given that hydropower and photovoltaic 5. Conclusion and policy implications (solar) energy shows substitutability for fossil fuel. United Nations Sustainable Development Goals (UN- This necessitates the need to increase government SDGs) that address pertinent issues across the globe by increasing energy investment (renewables) on 2030 motivate this country-specific study. More speci- a gradual basis not to jeopardize her economic growth, fically this study focuses on access to clean and respon- which is currently based on fossil-fuel sources. sible energy interfuel consumption (SDGs-7, 12) and Conclusively, this study suggests there is a need for climate change mitigation (SDG-13), on the need for a gradual transition from conventional energy sources substituting fossil-fuel base energy sources for renew- able energy mix for the case of Italy for annual fre- (fossil-fuel base) to renewable (clean) energy sources in quency data from1989-2018. This study thereby Italy for electricity generation. These clean energy explores the possibility of substitutability of coal, nat- sources are known to be more environmentally and ural gas energy option for hydropower, and solar ecosystem friendly especially in an era where there are energy sources using a translog production function serious concerns for green or clean energy. The Italian framework. To this end, we employ the use of ridge governments, both central and regional, have made regression procedure and circumvent for possible mul- several efforts to improve the share of renewable ticollinearity among the investigated variable energy sources, not only in electricity generation but parameter. This study also ameliorates for omitted variable bias also in the overall energy profile of the country. Such by the inclusion of renewable energy sources (hydro- initiatives include the feed-in tariff for all renewable energy and solar energy sources) to model framework. energy producers, the feed-in premium for electricity Empirical evidence gives credence to the possibility of produced by photovoltaic plants, and the award of substitution between non-renewable energy sources green certificates to the producers of renewable and renewable energy sources. Furthermore, we energy resources. However, these initiatives are mostly observe that both hydropower and solar energy are targeted to encourage and promote the use of found to be substitutable for fossil fuels, solar energy is 438 S. A. SOLARIN ET AL. renewable energy resources. For maximum results, Hall VB. 1986. Major OECD country industrial sector interfuel efforts should also be made in regards to the discour- substitution estimates, 1960–1979. Energy Econ. 8 (2):74–89. doi:10.1016/0140-9883(86)90032-0. agement of fossil fuels through the introduction and Hoerl AE. 1962. Applications of ridge analysis to regression implementation of deterrent policies in the form of problems. Chem Eng Prog. 58(3):54–59. imposition of carbon tax to the producers of fossil Hoerl AE, Kennard RW. 1970. Ridge regression: biased estima- tion for nonorthogonal problems. Technometrics. 12 fuels. This will serve the dual purpose of providing (1):55–67. doi:10.1080/00401706.1970.10488634. additional source of revenue to the government and International Energy Association. 2020. IEA world energy importantly, a source of disincentive to produce more balances 2020. [Accessed 2020 Jul 25]. https://www.iea. org/reports/world-energy-balances-overview fossil fuels. This will go a long way in complementing International Renewable Energy Agency. 2019. Renewable the incentive programs and policies on renewable energy and jobs annual review 2019. [Accesed 2020 Jul energy and help in the process of increasing the 25] https://www.irena.org/publications/2019/Jun/ Renewable-Energy-and-Jobs-Annual-Review-2019 share of renewable energy not just in electricity gen- Jones CT. 1996. A pooled dynamic analysis of interfuel sub- eration but in the overall energy profile of the country. stitution in industrial energy demand by the G-7 countries. Appl Econ. 28(7):815–821. doi:10.1080/000368496328254. Jones CT. 2014. The role of biomass in US industrial interfuel substitution. Energy Policy. 69(1):122–126. doi:10.1016/j. Disclosure statement enpol.2014.02.038. Kumar S, Fujii H, Managi S. 2015. Substitute or complement? No potential conflict of interest was reported by the authors. Assessing renewable and nonrenewable energy in OECD countries. Appl Econ. 47(14):1438–1459. doi:10.1080/ 00036846.2014.997922. References Li J, Xie C, Long H. 2019. The roles of inter-fuel substitution and inter-market contagion in driving energy prices: evi- Bardazzi R, Oropallo F, Pazienza MG. 2015. Do manufacturing dences from China’s coal market. Energy Econ. 84:104525. firms react to energy prices? Evidence from Italy. Energy doi:10.1016/j.eneco.2019.104525. Econ. 49:168–181. doi:10.1016/j.eneco.2015.01.014. Lin B, Abudu H. 2020. Impact of inter-fuel substitution on Bello MO, Solarin SA, Yen YY. 2018. Hydropower and poten- energy intensity in Ghana. Front Energy. 14(1):27–41. tial for interfuel substitution: the case of electricity sector doi:10.1007/s11708-019-0656-5. in Malaysia. Energy. 151(1):966–983. doi:10.1016/j. Lin B, Ankrah I. 2019. On Nigeria’s renewable energy pro- energy.2018.03.055. gram: examining the effectiveness, substitution potential, Bello MO, Solarin SA, Yen YY. 2020. Interfuel substitution, hydro- and the impact on national output. Energy. 167 electricity consumption and CO2 emissions mitigation in (1):1181–1193. doi:10.1016/j.energy.2018.11.031. Malaysia: evidence from a transcendental logarithm Lin B, Atsagli P. 2017. Energy consumption, inter-fuel substi- (trans-log) cost function framework. Environ Sci Pollut Res. tution and economic growth in Nigeria. Energy. 27:1–13. 120:675–685. doi:10.1016/j.energy.2016.11.115. Berndt ER, Wood DO. 1975. Technology, prices, and the Lin B, Atsagli P, Dogah KE. 2016. Ghanaian energy economy: derived demand for energy. Rev Econ Stat. 57 inter-production factors and energy substitution. (3):259–268. doi:10.2307/1923910. Renewable Sustainable Energy Rev. 57:1260–1269. British Petroleum. 2020. Statistical review of world energy. doi:10.1016/j.rser.2015.12.160. [Accessed 2020 Jul 25]. https://www.bp.com/en/global/cor Lin B, Tian P. 2017. Energy conservation in China’s light porate/energy-economics/statistical-review-of-world- industry sector: evidence from inter-factor and inter-fuel energy.html substitution. J Clean Prod. 152:125–133. doi:10.1016/j. The Conference Board. 2020. Economic data and analysis. jclepro.2017.03.099. [Accessed 2020 Jun 26] https://www.conference-board. Ma C, Stern DI. 2016. Long-run estimates of interfuel and org/data/ interfactor elasticities. Resour Energy Econ. 46:114–130. Considine TJ. 2018. Estimating concave substitution possibi- doi:10.1016/j.reseneeco.2016.09.003. lities with non-stationary data using the dynamic linear Morana C. 2000. Modelling evolving long-run relationships: logit demand model. Econ Model. 72:22–30. doi:10.1016/j. an application to the Italian energy market. Scott J Polit econmod.2017.12.021. Econ. 47(1):72–93. doi:10.1111/1467-9485.00154. Energy Information Administration. 2018. [Accessed 2020 Renou-Maissant P. 1999. Interfuel competition in the indus- Aug 22]. https://www.eia.gov/conference/2018/ trial sector of seven OECD countries. Energy Policy. 27 Griffin JM. 1977. Inter-fuel substitution possibilities: (2):99–110. doi:10.1016/S0301-4215(99)00006-3. a translog application to intercountry data. Int Econ Rev Sachdev HS, Akella AK, Kumar N. 2015. Analysis and evalua- (Philadelphia). 18(3):755–770. doi:10.2307/2525959. tion of small hydropower plants: A bibliographical survey. Güney T, Kantar K. 2020. Biomass energy consumption and Renewable Sustainable Energy Rev. 51:1013–1022. sustainable development. Int J Sustainable Dev World doi:10.1016/j.rser.2015.06.065. Ecol. 27(8):762–767. doi:10.1080/13504509.2020.1753124. Serletis A, Timilsina G, Vasetsky O. 2011. International evidence Gyamfi BA, Bein MA, Bekun FV. 2020. Investigating the nexus on aggregate short-run and long-run interfuel substitution. between hydroelectricity energy, renewable energy, non- Energy Econ. 33(2):209–216. doi:10.1016/j.eneco.2010.05.013. renewable energy consumption on output: evidence from Serletis A, Timilsina GR, Vasetsky O. 2010. International evi- E7 countries. Environ Sci Pollut Res. 27(20):25327–25339. dence on sectoral interfuel substitution. Energy J. 31(4):4. doi:10.1007/s11356-020-08909-8. doi:10.5547/0195-6574-EJ-Vol31-No4-1. INTERNATIONAL JOURNAL OF SUSTAINABLE DEVELOPMENT & WORLD ECOLOGY 439 Serletis A, Xu L. 2019. Interfuel substitution: evidence from Suh DH. 2016. Interfuel substitution and biomass use in the the markov switching minflex laurent demand system US industrial sector: A differential approach. Energy. 102 with BEKK errors. Energy J. 40:6. doi:10.5547/ (1):24–30. doi:10.1016/j.energy.2016.02.012. 01956574.40.6.aser. Suh DH. 2019. Interfuel substitution effects of biofuel use on Smith GR, Archer R. 2020. Climate, population, food security: carbon dioxide emissions: evidence from the transporta- adapting and evolving in times of global change. tion sector. Appl Econ. 51(31):3413–3422. doi:10.1080/ Int J Sustainable Dev World Ecol. 27(5):419–423. 00036846.2019.1581906. doi:10.1080/13504509.2020.1712558. Wesseh PK, Lin B. 2016. Output and substitution elasticities of Solarin SA, Bello MO. 2019. Interfuel substitution, biomass energy and implications for renewable energy expansion consumption, economic growth, and sustainable develop- in the ECOWAS region. Energy Policy. 89:125–137. ment: evidence from Brazil. J Clean Prod. 211 doi:10.1016/j.enpol.2015.11.007. (1):1357–1366. doi:10.1016/j.jclepro.2018.11.268. Wesseh PK, Lin B. 2018. Energy consumption, fuel substitu- Sudsawasd S, Charoensedtasin T, Pholphirul P. 2020. Does tion, technical change, and economic growth: implications international trade enable a country to achieve sustain- for CO2 mitigation in Egypt. Energy Policy. 117 able development goals? Empirical findings from two (1):340–347. doi:10.1016/j.enpol.2018.03.022. research methodologies. Int J Sustainable Dev World World Bank Group. 2020. World development indicators Ecol. 27(5):405–418. doi:10.1080/13504509.2019.1709913. 2020. Washington, DC: World Bank Publications.