Energy consumption in South African industry: A decomposition analysis using the LMDI approach

Abstract

Improving efficiency levels of energy consumption needs the appreciation of previous trends in the consumption of energy and the assessment of factors that have contributed to the changes in energy consumption. This study assessed the South African industrial energy consumption from 1970/1971 to 2015/2016. During 2015/2016, the industrial energy consumption increased by 794.39%. In addition, the decomposition analyses proved that the activity effect played a crucial part in the country’s industrial energy consumption increase. The results clearly showed that saving techniques and industrial policies have not impacted on the country’s industrial energy. In achieving energy conservation, it is important to enforce the policies formulated should the policies have not been well implemented. On the other hand, should formulated policies be implemented, then there is a need for revision of the existing policies.

Keywords

Industrial energy consumption decomposition analyses activity effect policies energy conservation

Introduction

Climate change has made the environmental and energy policies more interconnected than before.¹ Coping with energy crisis continues to be a vital research since the 1970 oil crisis.² For a country to develop socio-economically, energy remains a fundamental resource.^3,4 Improving efficiency levels of energy consumption needs the appreciation of previous trends in the consumption of energy and the assessment of factors that have contributed to the changes in energy consumption. Decomposition techniques attempt to separate all the effects, so that relative contributions of different effects can be identified.⁵ Structural decomposition analysis (SDA) and index decomposition analysis (IDA) have remained popular among researchers in the decomposition of indicator changes.⁴ SDA is input–output based,⁶ whereas IDA is based on aggregate data.⁴ Comparison of IDA and SDA can be found in the study of Su and Ang.⁷ Due to aggregate data characteristics of IDA, it became the preferred choice of method for this study. Among the index decomposition methods for the analysis of the factors responsible for energy consumption, Laspeyres, Divisia Fisher, Marshall-Edgeworth among others are included; with the first two methods being popular in the research domain.⁸ Decomposed units from the application of IDA are mostly activity, structure and intensity using subsector data.⁹ This paper aims at analyzing those factors in the various manufacturing subsectors behind the change in aggregate energy consumption in South Africa using the Divisia method of IDA. Similar work to the present work was conducted by Inglesi-Lotz and Pouris¹⁰; however, the latter study failed to give information on the energy consumed during the apartheid era as well as the change in energy due to the growth of the economy which is the activity factor. A better understanding of the way energy was consumed in South Africa would be more appreciated if the apartheid era of energy consumption had been considered. The present study has considered the apartheid era together with the democratic and post-democratic era while including the activity factor. These trends and drivers help define the past use of South Africa’s energy, which is beneficial to improving efficiency and transforming the country’s energy system by reducing demand.⁸ The measurement of the impact on energy consumption as a result of changes in the sector’s overall activity level is termed as activity effect. Activity indicator is often given by an all-round measure, such as GDP as much as monetary measure in value added. This makes it applicable to all industry subsectors making the computation of aggregate activity level and activity structure easy.⁹ The structure and intensity effect, respectively, show changes in energy consumption arising from structure change, i.e. changes in activity mix by subsectors and from changes in subsectors energy intensities.⁹ Energy intensity study has provided various assumptions and its variations in different circumstances have served most times as the backbone to the use of energy. Its reduction has also served as a way to reduce energy use¹¹ as it promotes the growth of economy simultaneously.¹² Monitoring of sectoral and economy-wide energy efficiency trends have been widely achieved by the effect of energy intensity.⁹ IDA remains a famous tool for energy consumption changes analysis over time. It quantitatively assesses the drivers that lead to changes in an aggregate energy indicator over time.¹³ In unfolding the reasons to the trend of energy efficiency in South Africa between 1993 and 2006, the study of Inglesi-Lotz and Pouris¹⁰ made use of IDA. The study resulted in the structural change playing a two-fold role as both a significant and negative role in the country’s energy efficiency trend. With regard to the role of intensity factor, the study concluded that it was responsible for the decrease in energy efficiency. Other various applications of LMDI in other countries apart from the study of Inglesi-Lotz and Pouris¹⁰ are highlighted from literature studies.^1,14–24

In the study of Cruz and Dias,¹ energy and CO₂ intensity changes in the EU-27 were analyzed for policy implications using the IDA. Their study resulted in the reduction of total energy use in the EU-27 economies due to a shift to a less energy-intensive structures and improved sectoral energy efficiency not minding the poor activity effect encountered. In the study of Timma and Blumberga¹⁴ wanting to know the reason behind the decline of energy intensity in Latvia, it was concluded that technological improvement proved to be responsible as opposed to structural effect. Their study considered agricultural, mining and quarrying, manufacturing, transportation and storage, construction sectors and other consumers (with the exception of households and electricity, gas, steam air condition). The study of Ma and Stern¹⁵ in China’s change to energy intensity from 1980 to 2003 discovered technological change as the reason for the country’s energy intensity decline, whereas structural change contributed to its increase. Another study concerning China is the study of Zhao et al.¹⁶ regarding energy intensity from 1998 to 2006. Heavier industrial structure as well as economy expansion contributed to the increase in energy intensity. The results of their study are similar to that of Ma and Stern.¹⁵ Based on the Logarithmic Mean Divisia Index (LMDI) applied to the urban residential energy consumption in Jiangsu, the energy efficiency effect played a critical role as the urban residential energy consumption at the aggregate level decreased, whereas the population effect was the crucial factor to the decrease in rural residential energy consumption.¹⁷ From the study of Wang et al.,¹⁸ it was noted that energy intensity effect played the critical role in decreasing China’s energy consumption between 1991 and 2011. The study further confirmed investment and labor effect to be responsible for the country’s energy consumption growth.¹⁹ Zhang et al.²⁰ found out from their study that decrease in energy consumption in technology and energy structure in secondary industries contribute to the most reduction in energy consumption intensity. Their result was from the study they conducted to assess the contributing factors in 29 Chinese provinces from 1995 to 2012. Chong et al.²¹ employed LMDI to assess the influencing factors of China’s coal consumption growth from 2001 to 2011. From the study, the most influential factor responsible for the growth in coal consumption was the GDP per capita, whereas its reduction was due to improved energy efficiency of coal power generation and coal end-use combustion. In understanding the change in Canada’s energy systems, Torrie et al.²² applied the LMDI to decompose the country’s energy intensity from 1995 to 2010. The study concluded that the decline of the country’s energy intensity between that period was due to 48% inter-sector structural change in the economy, 24% faster growth of GDP than population, 22% reduction in business energy intensity, and 6.3% improved household energy intensity, majorly in residential energy use. From the study of Timma et al.,²³ decrease in energy intensity and energy use from the decomposition on Latvia’s energy consumption was as a result of the improvement in technology as opposed to structural effect. Before the country’s economic downturn of 2008, the decline of energy intensity was as a result of sector’s reduction in energy intensity. However, its increase after 2008 was as a result of the expansion of the energy demanding sectors. The study of Shahidzzaman and Alam²⁴ concentrated on the factors responsible for the changes in energy efficiency in Australia from 1978 to 2009. Through the application of LMDI, the result showed efficiency effect and sectoral composition effect to be responsible for the decrease in the country’s energy intensity, with efficiency effect being the more responsible of the two.

Following the introductory and literature review, the next section looks at the relationship between South Africa and energy. Then, the data applied for this study’s investigation are presented. The ‘Methodology’ section, details the approach for adoption of this study. Then, the result and discussion following the application of the method to the data are presented. Finally, conclusion is presented.

South Africa and energy

The growth and development of Africa has been largely dependent on affordable and reliable energy.²⁵ Chronic under-investment in the electricity sector has led to escalating power prices and a shortage of capacity during peak demand periods, leading to demand rationing and blackouts.¹⁹ South Africa is one of the most industrialized countries in Africa, and energy plays a very important role in the country’s production process.²⁶ The Department of Minerals and Energy ranked coal to be the most primary energy supply by source in 2004, with 68%, next to crude oil (19%), renewable (8%), nuclear (3%) and hydro (0%).^26,27 Energy challenge is one of the many issues a developing country like South Africa is facing.¹⁹ This has led to various policies. The country’s energy policy can be viewed under three periods. The first period was in the time of apartheid, from 1948 to 1994. The second period was after the apartheid era, in the time of democratic elections of 1994–2000 and the third from 2000 afterwards was the post democratic era.²⁵ The policies at the time of apartheid were focused on security. After apartheid, policies were directed at energy equity and third period was and still on achieving government’s targets and deadlines.²⁵ Energy’s contribution to Africa’s growth is noticeable; evidence can be seen in South Africa.²⁵ However, compared with BRIC countries (Brazil, Russia, India, China), South Africa’s economy has receded back to the 2006 levels.¹⁹ Socio-economic growth remains a key force to the country’s energy demand increase.²⁸ Two major forces driving the country’s energy sector are economic and political.²⁵ The report of International Energy Agency (IEA)²⁹ noted energy to be an engine of inclusive economic and social growth. South Africa is highly energy intensive, with a large consumption of energy mostly coal for the country’s economic activities.²⁵ South Africa experienced not so long shortage of electricity across the country leading to an epileptic economic activity.³⁰

Data

Industries are arguably the largest consumer of energy and the highest emission of greenhouse gas among the major sectors in an energy consuming economy. This characteristic has led to various mechanisms to understanding the change of energy consumption in the industry.³¹ To understand this change, data from 1970 to 2016 from the various subsectors of manufacturing industry as regards the energy consumed and GDP are analyzed on five sectors namely basic chemicals, non-metallic minerals, basic iron and steel, basic non-ferrous metals and other manufacturing for this study. Figures 1 and 2 represent the energy consumed for the subsectors under investigation and GDP for the subsectors under investigation.

Figure 1.

Energy consumed for the subsectors under investigation.

Figure 2.

GDP for the subsectors under investigation.

Data were provided by Quantec, a private company in South Africa. The productions are all given by the gross domestic product (GDP) output expressed in R million current prices. A close correlation is present between the energy consumed and the total GDP in the South African industrial sectors. This correlation is near to linearity, which is identified by the formula of energy = 19,135 + 2GDP with correlation coefficient of 0.9595. This confirms the nexus that exists between the South African industrial energy consumption and GDP.

Methodology

Decomposition of energy consumption

Energy consumed in this study is analyzed according to the periods through the changes in activity, structural and intensity (efficiency). Here, the change in energy consumption from period 0 to period T is decomposed into three parts that refer to activity (the Q-term, which captures a given sector/or department’s contribution to the overall gross domestic product (GDP)), energy intensity (the I-term, refers to the actual change in energy efficiency), and the structural effect (the S-term, which denotes the shifts in the mix of products or activities).

Each of the respective terms captures how much of the change in energy consumption during the periods observed that can be assigned to changes in the respective variables.

The decomposition analysis variables are given below:

$E_{i}$ – Total energy consumed in sector/or department $i$

$E$ – Total energy consumed ( $E = Σ_{i} E_{i}$ $)$

$Q_{i}$ – The production value in sector/or department $i$

$Q$ – Total production value ( $Q = Σ_{i} Q_{i}$ $)$

$S_{i}$ – Share of production in sector/or department i ( $S_{i} = \frac{Q_{i}}{Q}$ )

$I_{i}$ – Intensity of energy consumed in sector/or department ( $I_{i} = \frac{E_{i}}{Q_{i}}$ )

E = \sum_{i} E_{i} = \sum_{i} Q \frac{Q_{i}}{Q} \frac{E_{i}}{Q_{i}} = \sum_{i} Q S_{i} I_{i}

(1)

E^{T} - E^{0} = Δ E_{t o t} = Δ E_{a c t} + Δ E_{s t r} + Δ E_{i n t}

(2)

\frac{E^{T}}{E^{0}} = D_{t o t} = D_{a c t} D_{s t r} D_{i n t}

(3)

The choice between multiplicative and additive decomposition is methodologically inconsequential, and the main consideration is the ease of result presentation and interpretation as indicated by Fengling.³² When decomposition is performed on a yearly basis using time-series energy and industrial production data, it is more convenient to use the multiplicative approach, based on experience, as the results given in indices can be conveniently plotted over time. However, if decomposition is performed based on the data of two benchmark years, additive decomposition maybe adopted because the decomposition results, which can either be given in the original unit of measurement of the aggregate or in terms of percentage change, can be more easily understood. For this study, additive was employed.

The purpose of the IDA is to disentangle and identify the most important factors explaining the overall change in energy consumption in the industrial sectors for the period 0–T.

It is important to emphasize that decomposition analysis is a descriptive method only where from no conclusion for future development may be inferred, as it does not involve any statistical estimation of relevant parameters. The advantage of such a decomposition analysis is that it gives a detailed and transparent assessment of key drivers underlying past development of consumption based on very sparse assumptions only.³³

Analysis of the decomposition technique

Assume that aggregate $U$ is composed of $m$ factors $(x_{1}, \dots, x_{m})$ , i.e. $U = Σ_{i} U_{i}$ and $U_{i} = x_{1, i} x_{2, i} \dots x_{m, i}$ . Assume further that from period 0 to T the aggregate changes from $U^{0} to U^{T}$ . The objective of IDA is to derive the contributions of the $m$ factors to the change in the aggregate which can be expressed as

Additive form

Δ U_{t o t} = U^{T} - U^{0} = Δ U_{x 1} + Δ U_{x 2} + \dots + Δ U_{x n}

(4)

General formulae of LMDI

Δ U_{x k} = \sum_{i} L (U_{i}^{T}, U_{i}^{0}) ln [\frac{x_{k, i}^{T}}{x_{k, i}^{0}}]

(5)

D_{x k} = exp [\sum_{i} \frac{L (U_{i}^{T}, U_{i}^{0})}{L (U^{T}, U^{0})} \ln [\frac{x_{k, i}^{T}}{x_{k, i}^{0}}]]

(6)

where L (a, b) = \frac{a - b}{ln a - ln b} and L (a, a) = a

(7)

LMDI (Additive)

Δ E_{a c t} = Σ_{i} w_{i} \ln [\frac{Q^{T}}{Q^{0}}]

(8)

Δ E_{s t r} = Σ_{i} w_{i} \ln [\frac{S_{i}^{T}}{S_{i}^{0}}]

(9)

Δ E_{i n t} = Σ_{i} w_{i} \ln [\frac{I_{i}^{T}}{I_{i}^{0}}]

(10)

where

w_{i} = \frac{E_{i}^{T} - E_{i}^{0}}{\ln E_{i}^{T} - \ln E_{i}^{0}}

(11)

Proof of perfect decomposition for both Additive and Multiplicative can be found from Ang,³⁴ a simple guide.

Results and discussion

Matlab (matrix laboratory) software was used to perform the decomposition analysis. This software is a numerical computing environment of which the mathematical equations for the decomposition were developed. The analysis of this study is to identify the yearly changes in the South African industry from 1970 to 2016. The application of additive LMDI in equations was applied successfully in the assessment of the changes in South Africa’s energy consumption. The results are depicted in Figures 3 to 5. From Figure 3, it can be seen that the increase in energy is due to the high economic activity compensated by intensity effect as well as negative structural effect indicating mixed changes for the period under investigation. From Figure 3, it can be seen that the consumption of energy from period 1971 to 1972 has increased from 27.0182 to 2.15 × 10⁴ in the period 2015–2016. For the activity effect, changes in the level of activity between 1970 and 2016 are considered, keeping the intensity and share of the sector in value addition unchanged in the initial year values. This implies that if the activity would have changed alone, the energy demand for all the industrial activities considered would have increased by about 501166.44. For the structural effect, the structural change within the period is considered while keeping the other two factors unchanged. This suggests that the share of the industrial activities in the industrial output has reduced and if this only had changed, the energy demand would have reduced by 1013.814. Finally, for the intensity effect, we look at the changes in energy intensity within the period under investigation and keep the other factors at their initial values. This suggests that the intensity in the industrial sectors considered in this research has reduced and this intensity would have reduced the energy demand by 120329.1132 between 1970 and 2016 if other things did not change.

Figure 3.

Summary of the factors responsible for energy consumption.

For all the subsectors, it could be seen from Figure 5 that the use of energy fluctuated in a similar pattern periods from 1970 to 2016 in different rates for all industries concerned. The LMDI decomposition results depicted in Figures 4 and 5 show the variation in the total energy consumed (Dtot) and how the variations of the explanatory effects of activity (Dact), structure (Dstr), an intensity (Dint) contributed to its continuous consumption in the period investigated.

Figure 4.

Factors responsible for the energy consumed during the period of study.

Figure 5.

Decomposition of total energy consumption for the period of study.

From Figure 4, it can be noted that 2005–2006 experienced a significant improvement in energy efficiency, as this can be seen in the highest reduction of intensities compared to the other years. It is rather unfortunate to say the least that following a good efficiency status from 2004–2005 to 2013–2014, the following years of 2015–2016 experienced its worst efficient use of energy. Energy increased gradually from 1971–1972 to 1980–1991 as seen in Figure 5 and decreased the following year which later picked up again gradually until 1988–1989. Similar fluctuation continued seeing energy use increase for a certain period then decreases. The period 2008–2009 however took a different twist reducing to the lowest consumption of energy as compared to the other periods.

With the intensity result, it will be best for the government of the country to bring to effect policies appropriate for the industry in the form of eliminating backward production. The improvement in the energy efficiency as a result of the decrease in intensity can be driven further should the government improve policies that will support improvement in the country’s technologies. Strengthening international exchange will also be a boost to improve exchange in energy efficient technologies. However, increase in the output (GDP) of the industry should also be a consideration by the government as this will improve the industrial structure significantly.

The structural changes during the period of study influenced the ratio negatively as well as the changes in intensity. However, the activity effect was the only difference influencing the ratio positively. For clarity, without changes in the activity and intensity of the country’s economy there would be 3.7 × 10⁴ units lower as well as 8.36 × 10³ units lower without changes in the activity and structure. However, without intensity and structural changes, the overall ratio would have been 2.66 × 10⁵ units higher.

The activity effect is almost proportional to the energy consumed throughout the whole period of study, except for 2009–2010 and 2014–2015 periods. The proportionality of economic activity to energy consumed supports the notion of “no economic growth without energy”. The activity factor dominated both the structural and intensity factors making the energy increase linear with the level of output in the South Africa’s economy. From the figures, it could be seen that the energy increase within the period of study is largely due to the expansion of the activity effect. To clearly explain the trend of energy consumption, the changes in industrial energy consumption can be divided into the periods of policy: the apartheid days, the democratic era and the post democratic era.

Period 1: apartheid days – around 5968.6 of energy was consumed with 1988–1989 being the highest consumption during the first policy regime. The activity effect during that period edged a little less than the amount of energy consumed. During this period, however, it reported an average efficiency coupled with a very minute structural effect.

Period 2: the democratic era – during the period 1999–2000 – consumed the most of the energy. The activity effect was as close to the amount of the energy consumed. However, the structural effect is not anywhere close to the average of the energy consumed and structural effect for that same year resulting in little or no effect within that period.

Period 3: post democratic era – 2001–2002 – consumed the most energy as well as when compared to the apartheid days as well as the democratic era. The country during this period consumed six times that of the apartheid days and one plus half that of the democratic era.

The trends in the periods follow the policies that were in place. With focus on the industrial sector, the apartheid era concentrated on policies that rallied round economic and political security. After the 1994 elections, various energy policies were formulated, such as the Accelerated Electrification, the National Electrification Programme and the White Paper on Energy. With respect to the industry, one of the goals of the White Paper was to provide greater energy efficiency to the industry for its environmental and cost benefits. During the post democratic era, there were reforms in the energy sector. Most prominent was the Integrated Energy Plan of 2003 that focused on decisions for the development of different energy sources and technologies. It is noticeable from Figure 4 that efficiency increased as depicted from the intensity effect after 2003 Integrated Energy Plan was put in place. However, the turnaround in 2014–2015 is not anywhere encouraging for the industrial sector.

Conclusion

With the various policies developed and implemented in South Africa, the country’s industrial energy consumption managed to increase by 794.39% within the space of 46 years as recorded during 2015–2016 (Figure 3). The result from the decomposition analysis showed activity effect to be the most important factor among the three factors considered in this study. The result of this study coincides with the study of Cruz et al.³⁵ In their study, energy consumption and carbon emissions in developing countries were compared through a decomposition analysis from 1980 to 2003. Among the developing countries was South Africa. Their study confirmed that activity effect dominated the energy dynamics in South Africa, resulting in intensity and structural effect which exert pressure on the country’s energy consumption. It should be noted that as much as it seemed like policies were not efficiently implemented, activity effect played a huge role in the increase of the country’s industrial energy consumption. For the country’s continuous growth, energy will continue to play a huge role; however, there is need for an energy-saving and environment-conscious mode³⁶ to solve the energy crisis. According to the study of Inglesi-Lotz and Pouris,¹⁰ the addition of renewable energy as well as differentiated energy pricing are ways to contribute to solve the country’s energy crisis. No energy policy choices available are as attractive and necessary as energy efficiency and conservation. Unlike many other energy policy choices, which involve long-term investments and technology development, increased emphasis on efficiency and conservation can deliver results in the short to medium term.³⁷

Policy implications for this study are concerned with the industrial policy and energy-conserving policy. Based on the findings of this study, it is suggested that the country’s industries need a more aggressive energy-conservation policy to reduce its energy consumption. This study clearly showed that activity is very informative in the South African industry with respect to the amount of energy consumed. Energy-conservation policy can be improved by considering the information provided by the three factors represented in this study. This study is for the South African policy makers to reconsider industrial energy policy according to this finding. The industry’s energy challenges are rooted in two facts: it consumes a lot of energy and it continues to consume more. Unless there are new energy conservation policies or behavior changes, the high industrial energy consumption rate will continue.

The analysis of the varying patterns of the causative variables, through which energy consumption can potentially be explained, will assist to note the possible approaches to energy consumption management and conservation. The application of LMDI made it possible to disentangle and identify the various factors to explain the total change in energy consumed in the South African industrial sectors for the period 1970–1971 to 2015–2016. This will help to point energy policy in the right direction. The significance of this study is to provide policy makers with information to enable the revision and formulation of the country’s industrial energy policy.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Oludolapo A Olanrewaju is an established researcher in the field of energy. His specializations include applications of Data Management and Analysis, Optimization, and Artificial Intelligence on energy systems with general research interests in energy and greenhouse gas potentials.

References

Cruz

and Dias

Energy and CO₂ intensity changes in the EU-27: decomposition into explanatory effects. Sustainable Cities and Society, 2016; 26: 486–495.

Dai

and Gao

HO.

Energy consumption in China's logistics industry: a decomposition analysis using the LMDI approach. Transp Res Part D 2016; 46: 69–80.

Tolon-Becerra

Lastra-Bravo

and Botta

GF.

Methodological proposal for territorial distribution of the percentage reduction in gross inland energy consumption according to the EU energy policy strategic goal. Energy Policy 2010; 38: 7093–7105.

Balezentis

and Streimikiene

The energy intensity in Lithuania during 1995-2009: a LMDI approach. Energy Policy 2011; 39: 7322–7334.

Reddy

and Ray

BK.

Understanding industrial energy use: physical energy intensity changes in Indian manufacturing sector. Energy Policy 2011; 39: 7234–7243.

Hoekstra

and van den Bergh

JJCJM.

Comparing structural and index decomposition analysis. Energy Econ 2003; 25: 39–64.

and Ang

Structural decomposition analysis applied to energy and emissions: some methodological developments. Energy Econ 2012; 34: 177–188.

Yang

and Mingguang

Analysis on the degree of the industrial structure's impact on the energy consumption – based on empirical study of Guandong province. Energy Proc 2011; 5: 1488–1496.

Ang

and Xu

XY.

Tracking industrial energy efficiency trends using index decomposition analysis. Energy Econ 2013; 40: 1014–1021.

10.

Inglesi-Lotz

and Pouris

Energy efficiency in South Africa: a decomposition exercise. Energy 2012; 42: 113–120.

11.

Farla

and Blok

The quality of energy intensity indicators for international comparison in the iron and steel industry. Energy Policy 2001; 29: 523–543.

12.

Wang

Changing energy intensity of economies in the world and its decomposition. Energy Econ 2013; 40: 637–644.

13.

Ang

and Wang

Index decomposition analysis with multidimensional and multilevel energy data. Energy Econ 2015; 51: 67–76.

14.

Timma

and Blumberga

Index decomposition analysis for energy sectors in Latvia. Energy Proc 2014; 61: 2180–2183.

15.

and Stern

DI.

China's changing energy intensity trend: a decomposition analysis. Energy Econ 2008; 30: 1037–1053.

16.

Zhao

and Hong

Why did China's energy intensity increase during 1998-2006: decomposition and policy analysis. Energy Policy 2010; 38: 1379–1388.

17.

Zhang

et al . Study on affecting factors of residential energy consumption in urban and rural Jiangsu. Renewable Sustainable Energy Rev 2016; 53: 330–337.

18.

Wang

et al . Using a new generalized LMDI (Logarithmic Mean Divisia Index) method to analyse China's energy consumption. Energy 2014; 67: 617–622.

19.

Pollet BG, Staffell L and Adamson KA. Current energy landscape in the Republic of South Africa. Int J Hydrogen Energy 2015; 40: 16685–16701.

20.

Zhang

et al . Decomposition of intensity of energy-related CO₂ emission in Chinese provinces using the LMDI method. Energy Policy. 2016; 92: 369–381.

21.

Chong

et al . Logarithmic Mean Divisia Index (LMDI) decomposition of coal consumption in China based on the energy allocation of coal flows. Energy 2015; 85: 366–378.

22.

Torrie

Stone

and Layzell

DB.

Understanding energy systems change in Canada: 1. Decomposition of total energy intensity. Energy Econ 2016; 56: 101–106.

23.

Timma

Zoss

and Blumberga

Life after the financial crisis. Energy intensity and energy use decomposition on sectorial level in Latvia. Appl Energy 2016; 162: 1586–1592.

24.

Shahiduzzaman

and Alam

Changes in energy efficiency in Australia: a decomposition of aggregate energy intensity using logarithmic mean Divisia approach. Energy Policy 2013; 56: 341–351.

25.

Kumar

and Kumar

Effects of energy consumption on per worker output: a study of Kenya and South Africa. Energy Policy 2013; 62: 1187–1193.

26.

Zirimba

Disaggregate energy consumption and industrial production in South Africa. Energy Policy 2009; 37: 2214–2220.

27.

DME, Department of Minerals and Energy. 2006.

28.

Tomaschek

et al . Energy systems modelling and GIS to build an integrated climate protection Conceot for Gauteng province, South Africa. Energy Policy 2016; 88: 445–455.

29.

IEA. Africa energy outlook: a focus on energy prospects in Sub-Saharan Africa. World Energy Outlook Special Report. OECD/IEA, 2014

30.

Roula I. Aggregate electricity demand in South Africa: conditional forecasts to 2030. Appl Energy 2010; 87: 197–204.

31.

Liu

and Ang

BW.

Factors shaping aggregate energy intensity trend for industry: energy intensity versus product mix. Energy Econ 2007; 29: 609–635.

32.

Fengling

Decomposition analysis applied to energy: some methodological issues. PhD Thesis, National University of Singapore, Singapore, 2004.

33.

Lofgren

and Muller

Swedish CO₂ emissions 1993-2006: an application of decomposition analysis and some methodological insights. Environ Resource Econ 2010; 47: 221–239.

34.

Ang

BW.

Index decomposition analysis using LMDI: a simple guide. National University of Singapore, 2008.

35.

Cruz

and Fujita

Decomposition analysis of energy consumption and carbon emissions in developing economies. 2005, pp.1–28.

36.

Wang

et al . Study and analysis of energy consumption and energy-related carbon emission of industrial in Tianjin, China. Energy Strategy Rev 2016; 10: 18–28.

37.

Dernbach

Stabilizing and then reducing U.S energy consumption: legal and policy tools for efficiency and conservation. Washington, DC: Environmental Law Reporter 2007; 37: 8–26