Vous êtes sur la page 1sur 20

THE JOURNAL OF ENERGY AND DEVELOPMENT

Dawit Diriba Guta, Energy Demand in Rural Ethiopia from a

Household Perspective: A Panel Data Analysis,

Volume 35, Number 2


Copyright 2011

www.iceed.org

ENERGY DEMAND IN RURAL ETHIOPIA FROM A HOUSEHOLD PERSPECTIVE: A PANEL DATA ANALYSIS
Dawit Diriba Guta*

Introduction

his paper analyzes household energy demand in the context of rural households in Ethiopia. It focuses exclusively on energy use and household energy consumption patterns for household use purposes (i.e., for lighting, cooking, heating, etc.). To study demand for each of the energy end-uses services, data were not available in a disaggregated form. However, local experience indicates that advanced fuels are mainly used for light purposes and traditional biomass fuels (in our case, fuel wood and charcoal) are consumed in cooking and heating. More specifically, the aim of this research is to estimate household demand for different sources of energy.
*Dawit Diriba Guta is Junior Researcher in the Department of Economics and Technological Change at the Center for Development Research (ZEF), University of Bonn, Germany. He received his bachelors degree in economics and his M.Sc. in natural resource and environmental economics from Addis Ababa University, Ethiopia. The author was a lecturer at Ambo University before undertaking his Ph.D. study at the University of Bonn. His fields of interests are natural and environmental economics, energy economics, development economics, and institutional economics. Among the authors current areas of research are biomass-based energies, rural livelihood, energy security, and energy-sector models. The author would like to thank Dr. Mahmud Yesuf for his comments. Data used for this study were panel data collected by collaboration with Addis Ababa Universitys Department of Economics, the Centre for Studies of African Economies of Oxford University, and the International Food Policy Research Institute (IFPRI). The author received financial support from the African Economic Research Consortium to conduct the study. The Journal of Energy and Development, Vol. 35, Nos. 1 and 2 Copyright 2011 by the International Research Center for Energy and Economic Development (ICEED). All rights reserved.

195

196

THE JOURNAL OF ENERGY AND DEVELOPMENT

Very few studies exist on energy-use optimization behavior of rural households of Ethiopia. There is significant literature focused on urban household energy demand analysis.1 Early energy economists proposed what is termed as the energy ladder hypothesis according to which households switch from consumption of inferior fuels, like fuel wood and dung, toward modern energy sources with improvements in their income.2 In the past decade fuel-stacking or multiple fuel use theory has been dominant in energy economics.3 This study was motivated by three objectives: (1) to estimate the rural household energy demand, (2) to compute income/expenditure elasticity of energy demand for traditional fuels (fuel wood and charcoal) and advanced fuels (kerosine, candles, matches, batteries), and (3) to analyze the dynamics of demand for the period covered. For this reason, the econometric model of the Linear Approximation Almost Ideal Demand System (LAAIDS) was adopted to compute income or expenditure elasticities. The paper estimates the income effect on the energy demand system with household-level panel data. However, lack of information on prices hindered the computation of price elasticities. In the existing economic literature, although household energy demand might appear intuitive enough, there is little emphasis on rural households. Moreover, it is fair to say that in the field, many important issues remain controversial. There is also a debate in the energy literature about the impact of energy prices on low-income rural energy consumers, and the efficiency and efficacy of different policies. Part of the reason for this lack of consensus is that conclusions in the literature generally have been based on extrapolating the results from individual studies of a single city, village, or region or of a few cities, villages, or regions. Moreover, in the field, rigorous micro-econometric studies on the determinants of energy demand at the micro-household level have been scarce, with few recent exceptions.4 In this regard, household energy demand in Ethiopia has been examined by a number of empirical studies, but with little attention given to rural energy consumers.5 Moreover, the use of panel data in estimating household energy demand in rural Ethiopia is a first attempt. Use of panel data allowed for control of individual heterogeneity and provides more robust estimates of variation both cross-sectionally and across time. The contribution of this study is to provide a coherent view of rural household energy demand and energy transitions and the associated policy options for intervening in the rural energy setting. Theoretical Review Economic development and energy demand interact and co-evolve together over time. There is great disparity between energy demands across countries at a national level and between the relatively poor and rich households within a given country. Disparities in household energy use exist between rural and urban

ENERGY DEMAND IN RURAL ETHIOPIA

197

populations, between high-income and low-income groups within a nation and among countries.6 Moreover, a large number of studies have focused on the existence of the huge gap in energy consumption across the world and have argued that the richer nations are continuing to use both more and more advanced energy while the poor are confined to the consumption of inefficient fuel as a source of energy. A recent work by F. Barnes, K. Krutilla, and W. Hyde supported this view and argued that the poorer half of the worlds populace has long relied on fuel wood for their energy needs.7 Deforestation is an important issue in the discussion of fuel consumption of rural households. Various studies have focused on this subject. Firewood gathered from common forests is the major source of domestic energy in rural areas of many poor countries.8 Demand for fuel wood by subsistence agricultural households may be the leading cause of the worlds deforestation. Recent empirical research on firewood consumption and forest degradation has focused mostly on India, Nepal, and China; to our knowledge, only two papers have dealt with this issue in rural regions of Ethiopia.9 Household heavy reliance on traditional energy sources in Ethiopia is the focus of a number of studies.10 For example, B. Jargstorf noted that Ethiopia is the third largest user in the world of traditional fuels for household energy consumption, with 96 percent of the population dependent on traditional biomass (e.g., fuel wood and dung) to meet their energy needs.11 The agricultural households are the predominant rural energy consumer. Energy use of a rural household is described by using a household economic framework. The opportunity costs to a household for firewood collection are determined by factors such as household wealth, labor, and availability of wood stock. Because fuel wood is used mainly at the household level, modeling household energy-demand behavior is a critical issue in energy economics. The energy crisis of rural households is rooted in their dependence on few and low-grade energy sources. To mitigate the problem, fuel diversification and interfuel substitution can help to discourage the use of traditional sources and to optimize fuel mix in the energy portfolio. In fact, a study by the International Energy Agency argued that the major goal of fuel diversification is to reduce dependence on traditional fuel and promote the switch to modern sources as well as to increase fuel efficiency as an alternative or to complement use.12 It also is noted that energy efficiency improvements enhance both energy security and environmental protection. In addition to energy security, environmental concerns may provide a large part of the impetus for the introduction of alternative fuels in rural household energy consumption. Fuel choices therefore need to be understood in terms of relative household resource scarcities. Although the use of low-grade fuels (such as dung and residue) may be less detrimental to forests, there is a tradeoff between using them as agricultural inputs and burning them as fuel.13 However, recent studies of household energy use in developing countries show that the energy ladder theory is too simplistic and that there are many additional

198

THE JOURNAL OF ENERGY AND DEVELOPMENT

factors other than income that determine household fuel choice.14 The household sector is the major consumer of rural energy. Household energy demand in the past often has been understood through the concept of the energy ladder. The energy ladder model has been shown to be strong in its emphasis on the role of income in determining fuel choices. It states that households switch their fuel use from biomass to modern energy sources as a country develops and incomes increase, implying that firewood is an inferior good.15 Similarly, other studies by the Energy Sector Management Assistance Program (ESMAP) and G. Leach envision the energy ladder model as a three-stage fuel-switching process.16 The first stage is marked by universal reliance on biomass. In the second stage, households move to transition or intermediate fuels such as kerosine, coal, and charcoal in response to higher incomes and factors such as deforestation and urbanization. In the third phase, households switch to liquefied petroleum gas (LPG) and electricity once their income is sufficient. Empirical Review A panel data model was used for Peruvian households to compare point estimates of price and income elasticities with demand for clean fuels with pooled estimates.17 With the application of the AIDS (Almost Ideal Demand System) to analyze urban fuel (particularly firewood) demand in Harare, Zimbabwe, in an energy mix context, M. Chambwera found a strong substitution effect.18 That study used a multi-stage budgeting approach in the model and estimated the share of energy in total household expenditure and the shares of firewood, electricity, and kerosine in the total energy expenditure. The author made a policy recommendation that decreasing prices of alternative fuels, notably kerosine, could induce households to shift from firewood to kerosine and thus ease the problem of environmental degradation. Moreover, the Energy Sector Management Assistance Programs study suggested that price competition between fuel wood and petroleum is likely to be particularly intense where wood is relatively scarce or distant from the city.19 The budget and income constraints of households are important determinants of the overall pattern of consumption expenditures. In line with this argument, B. Campbell et al. conducted a study of energy transition and indicated that a household switch from wood through kerosine to electricity occurred with a rising household income.20 In addition, household energy surveys have found income to be a major determinant of the energy transition.21 Those with higher income can switch to other fuels. Thus, while the households with higher incomes are expected to switch to advanced fuels, the poor move down the ladder to the relatively less-efficient fuels such as crop residue and animal dung. This strong link between energy demand and income in the energy ladder framework also was observed in a study by the Energy Sector Management Assistance

ENERGY DEMAND IN RURAL ETHIOPIA

199

Program.22 However, the heavy income dependence of household energy demand has been emphasized by many researchers.23 Higher incomes induce households to switch away from traditional fuels, such as cow dung and firewood, to higher quality but more expensive substitutes.24 In contrast, the increase in incomes gives rise to a decline in the percentage of consumers who choose to consume traditional fuels and kerosine and an increase in the percentage of consumers who consume LPG and electricity.25 In the case of Ethiopia, poverty appears to be the driving force behind the problem. Electricity consumption is highly unaffordable for urban households in Ethiopia.26 Thus, from the viewpoint of economic theory and with no exception for Ethiopia, a high level of income is expected to induce increased consumption per capita and a switch toward more advanced energy sources. Poor households rely more on forest firewood than modern fuel substitutes purchased from the market, because the lower cost of their time (the chief cost of collecting forest wood) dominates income effects (which may be positive, as higher food expenditures associated with higher income raise demand for cooking energy). The results of a previous study indicated that for Nepal, the income and the substitution effects neutralize each other, so firewood collections are essentially inelastic with respect to improvements in living standards.27 Some studies do not include income or consumption expenditures and rely on proxies such as landholdings.28 Empirical Methodology This study developed a household energy-mix model based on the fact that in any one period, rural households use multiple sources of energy. Using a multistage budgeting approach, the share of energy in total household expenditure and the shares of traditional fuel and advanced fuel in the total energy expenditure of rural households are estimated using a panel data method. This study places heavy emphasis on income or expenditure in explaining household energy demand. Moreover, it incorporates a multitude of explanatory variables. The variables that were used for estimation of household energy demand include household total expenditure, assets (such as land-holding size, total number of livestock owned), and household level characteristics like family size, education of the head, gender of the head, a time dummy, and total time spent on fuel collection.
The Model: For this paper, we developed an econometric model for computing

expenditure elasticity. We adopted a simple model of economic behavior called the AIDS (Almost Ideal Demand System) and its linear approximationLinear Approximation Almost Ideal Demand System (LAAIDS). The AIDS budget uses equations for estimation of household energy demand for different energy types and was first introduced into economic literature by A. Deaton and J. Muellbauer.29 The AIDS model provides insight into the share of a given energy source in

200

THE JOURNAL OF ENERGY AND DEVELOPMENT

a household energy budget outlay and estimation of the expenditure elasticity of a household for advanced and inferior fuels. Because information on the price of rural energy was not in our data, a framework of a normalized AIDS or LAAIDS model was employed to examine the income or expenditure elasticities of household energy demand. The nonexistence of a monetized market for rural energy (traditional fuels) necessitated estimation of a shadow price or opportunity cost for fuel collection. The LAAIDS model with normalized price was used to estimate separately each of the two energy types. Two panel data models with fixed effect and random effect estimates were conducted using pooled ordinary least square (OLS) and generalized least square (GLS) estimations, respectively. The Haussmann test was employed to make comparisons between the two parameter estimates. Based on the test result, the fixed effect was found to be more appropriate for empirical estimation. Using a multi-stage budgeting approach, the paper estimated the share of energy in the total household expenditure and the share of inferior fuels and advanced fuels in the total energy expenditure of rural households. Different energies consumed by households were segregated into two types: the traditional or inferior fuels that in this study include fuel wood and charcoal and modern or advanced fuel sources that mainly include kerosine, matches, batteries, and candles. In general, the AIDS model is given as: Wit = ai +
J X j =1

xt g ij lnpjt + bi ln + uit pt

where Wit = QitPit/x is the budget share of good i, uit is a random disturbance, lnpjt is the price of commodity j in period t, xt is total expenditure in period t, and Pt is a price index in period t. For the index, Pt, Pt = ao +
K X k=1

ak lnpkt +

J K 1X X lnpjt lnpkt: 2 j =1 k =1

To obtain the LAAIDS, we use Stones price index (p*). P* =


K X k=1

wk lnpk

where P = zP* and z is the proportionality constant. Then, the LAAIDS model can be written (in terms of the AIDS model parameters) as:   J X x wi = ai bi lnj + 4 g ij lnpj + bi ln * : p j=1

ENERGY DEMAND IN RURAL ETHIOPIA When prices are normalized, AIDS can be written as: wit = ai + bi lnxt bi a0 where a0 = lnz. Similarly, at the point of price normalization LAAIDS can be written as: wit = a* + bi lnxt = ai + bi lnxt bi a0 : i

201

At the point of price normalization, where prices are set to unity, the uncompensated price elasticity for the AIDS model is given as: eit = d +
J X g ij b i aj + gij lnpjt vij vit j=1

where d is the Kroenecker delta. At the point of price normalization, and when a0 is set equal to expenditure in the base period such that ai equals the predicted budget share, equation (7) reduces to: eijt = d + eijt = d + g ij b i wjt wit wit
J X g ij b i wjt + gij lnpjt 1 + bi lnpjt wit : wit wit j=1

8 9

However, at the point of price normalization, equation (9) reduces to equation (8). The formulas for the uncompensated elasticities are equal for both the AIDS model and its linear approximation if a0 is set equal to expenditure in the base period. Finally, the reduced expenditure elasticity for the AIDS and LAAIDS model at the point of price normalization is given as: hit = 1 + bi : wit 10

Empirical Analysis
Data: The empirical data for developing this paper were compiled from the

Ethiopian rural household panel survey conducted in collaboration with the Department of Economics of Addis Ababa University, the Centre for Studies of African Economies of Oxford University, and the International Food Policy Research Institute. They were collected from all regions of the country and possessed all the desirable properties of panel data. To conduct an empirical investigation,

202

THE JOURNAL OF ENERGY AND DEVELOPMENT

the study used the 2000 and 2004 surveys. The total number of households interviewed and included in this survey in the year 2004 is 1,381 households, and in the year 2000, the number of respondents was 1,415 households.
Descriptive Statistics of Variables: The average percentage of inferior fuel ex-

penditures slightly increased from 36 percent in the year 2000 to 42 percent in the year 2004, while the share of advanced fuel declined from 64 percent to 52 percent over the years from 2000-2004. The proportion of households that used advanced fuels decreased from 99 percent in 2000 to 96 percent in the year 2004. Similarly, the proportion of households that consumed inferior fuels also declined slightly from 99.9 percent to 98 percent over that same period. In contrast, the proportion of households that consumed a mix of inferior and advanced fuels concurrently increased from 92 percent in 2000 to 94 percent in 2004. This indicates the increasing tendency of households energy diversification and reduced dependence on either inferior or advanced fuel type. This tendency of households toward the use of a mix of advanced and inferior fuels may be attributed to various factors. These include economic status (welfare change) of households, social (familiarity and loyalty especially for fuel wood), and environmental (deforestation, forest degradation, soil erosion, climatic change, and ecological effects) causes. Other factors that account for household fuel stacking include the end use of the fuel, cooking or consumption habits, affordability of fuels or dependency on cost, taste or preferences of household, and numerous other influences. Average energy expenditure of households assumed a higher share of total monthly household expenditures while exhibiting a declining trend by about 10 percent over the period range. This may perhaps be attributed to shifting to lowgrade fuels like animal dung and crop residue, which were not included in the data. Elasticity of Energy Demand To examine the income or wealth effect of household energy demand, we computed the expenditure elasticities. The parameter estimates of household demand for each fuel were obtained by regressing the log of the fuel expenditure share of the total household energy expenditure on the log of the households total monthly expenditure. As seen in the appendices, the coefficient of the total household expenditure was negative for traditional/inferior fuels and positive for advanced fuels. This implies that traditional fuels are inferior goods, and advanced fuels are normal goods. Normal goods are those for which expenditures rise as household total expenditures increase. When the economic welfare of a household improves, they switch their energy consumption in favor of modern and cleaner fuel alternatives. This result lends support to the energy ladder hypothesis because households respond to changes in their income or expenditures by adopting modern energy forms into their energy mix.

ENERGY DEMAND IN RURAL ETHIOPIA Table 1


DESCRIPTIVE STATISTICS OF VARIABLES

203

Variable label

Year 2004(N=1,381) Year 2000(N=1,415) Mean Std. dev. Mean Std. dev. 47.26 574.45 0.15 43.61 17.12 0.27 0.27 0.19 0.15 0.24 0.47 42.96 188.12 0.27 28.34 15.07 0.64 0.36 0.99 0.999 0.92 0.3 15.33 465.36 0.75 13.57 8.2 0.34 0.34 0.34 0.2 0.4 0.46 28.14 5.93 1.15 3.05 48.96 15.58 1,664.65 0.43 0.48

Total energy expenditure per month (in birr) 46.82 Total expenditure per month (in birr) 455.64 Share of energy in total expenditure 0.17 Expenditure on advanced fuels per month (in birr) 28.88 Expenditure on inferior fuel per month 17.93 Share of advanced fuel in total energy 0.58 Share of inferior fuel in total energy 0.42 Use advanced fuel (yes=1, else=0) 0.96 Use inferior fuel (yes=1, else=0) 0.98 Use mix of advanced and inferior fuel (yes=1, else=0) 0.94 Education of household head (never any school=0, else=1) 0.33 Total time spent on fuel collection per month (in minutes) 24.61 Total number of live stock owned by household 9.76 Total land owned by household (in hectors) 1.41 Family size 7.01 Family size square 58.28 Age of household head (in years) 50.63 Age square 2,842.43 Sex of household head (male=1, else=2) 1.25 Male headed household (yes=1, else=0) 0.74

28.78 50.92 9.45 8.36 1.2 1.41 3.01 6.88 54.64 59.76 16.72 49.25 1,855.1 2,667.16 0.44 1.23 0.44 0.71

Source: Computed from the Ethiopian Rural Household Survey, 2000 and 2004, Department of Economics of Addis Ababa University, in collaboration with the Centre for Studies of African Economies of Oxford University and the International Food Policy Research Institute.

Applying AIDS or LAAIDS with normalized price in equation (10) to our regression results, the expenditure or income elasticity of demand was calculated as 0.72 in 2000 and 0.76 in 2004 for inferior fuel, and 1.14 in 2000 and 1.15 in 2004 for advanced fuels. This implies that demand was expenditure inelastic for inferior fuel and expenditure elastic for advanced fuel. Since expenditure elasticity is greater than 1 for advanced fuels, they are found to be luxury goods. Firewood collection is essentially inelastic with respect to an improvement in household living standards in rural areas. Interestingly, this result is consistent with the finding for Nepali households that firewood collection is essentially inelastic with respect to improvement in the living standard, which was approximated in the study with total household income or expenditure.30

204

THE JOURNAL OF ENERGY AND DEVELOPMENT Table 2


EXPENDITURE ELASTICITY OF INFERIOR AND ADVANCED FUELS Year
a

Energy Category

Parameters wit bi hit wit bi hit

2000 0.36 -0.099 0.72 0.64 0.087 1.14

2004 0.42 -0.099 0.76 0.58 0.087 1.15

Traditional fuels

Advanced fuels
a

Where the parameters are defined as: wit is the budget share of energy type i in total household energy expenditure; bi is the coefficient of total household expenditure as predictor in estimation of demand for inferior fuels and advanced fuel in the fixed effect model; and hit is the elasticity of energy demand for each of the fuel categories. Source: Computed from the Ethiopian Rural Household Survey, 2000 and 2004, Department of Economics of Addis Ababa University, in collaboration with the Centre for Studies of African Economies of Oxford University and the International Food Policy Research Institute, and based on descriptive statistics given in table 1 and in the appendices.

Conclusion and Recommendations The environment matters greatly to the poor because they directly or indirectly derive their livelihood from it. Low-income characteristics of rural Ethiopian households hinder their fuel diversification to advanced fuel alternatives. The subject of household fuel stacking has not been adequately explored for subSaharan African countries in general and rural Ethiopian households in particular. Computed elasticities in this study imply that households are inclined to choose more clean, efficient, and advanced fuels than inferior fuel in their energy mix as their total expenditure increases. This implies that policy intervention should concentrate on empowering rural residents economically to substitute advanced fuels for biomass. Thus, with improvement in the quality of life, households exhibit an increasing desire for comfort and discretion in their energy consumption or fuel choice. However, rural households have little incentive to diversify their energy consumption from firewood to available alternatives only because of income or expenditure growth due to the free availability of fuel wood. Therefore, policy intervention is vital to devise appropriate strategies that can actively address this issue. A complete energy transition from inferior fuel use to modern fuels requires holistic social, economic, cultural, and even ideological changes of the whole society. For sub-Saharan African countries, it may take a long time to make the complete switch to modern fuels. This is attributable to various factors: the use of

ENERGY DEMAND IN RURAL ETHIOPIA

205

simple technologies characterized by low energy efficiency and harmful emissions; multifaceted links between energy poverty and improving the quality of the environment; and the inability of households to afford the advanced fuels. Tradition, religion, lack of stoves, and other elements also have a critical role. Policies should assure balanced growth of demand and supply in the rural energy sector. In Ethiopia there is enormous potential to generate hydroelectric power that could avert the deepening crisis in energy. Nevertheless, poor countries like Ethiopia are encumbered by financial, infrastructure, and technological challenges and are not able to mobilize resources and realize potentialities. Thus, building technological capabilities and innovations are key issues to success. Moreover, encouraging afforestation, tree-planting practices by rural households, and enacting forest protection laws are sound policy initiatives. Policies should be geared towards promoting interfuel substitution and interfuel switching by supporting households financially and technically. For example, policy makers can support households to adopt improved fuel stoves. In Ethiopia, energy-saving stoves like Mirte and Lakech have been shown to reduce the amount of fuel consumed and the health hazards that arise from smoke emissions. However, distribution of these stoves to large segments of rural society demands a huge capital investment. Generally, the design of appropriate policy instruments and their implementation are critical for relieving the deepening energy crisis of rural households. This finding has relevant insights for the United Nation millennium project that recommends halving the number of households using traditional biomass for cooking by 2015 and would involve about 1.3 billion people switching to other fuel alternatives. More studies therefore are needed in the areas of household fuel stacking. In summary, policy intervention is deemed critical to help improve household welfare and assure sustainable availability of diverse energy sources for rural households.
NOTES
1 Abebaw D. Ejige, Household Determinants of Fuelwood Choice in Urban Ethiopia: A Case Study of Jimma Town, Journal of Developing Areas, vol. 41, no. 1 (2007), pp. 117-26; M. Alemu, and G. Kohlin, Determinants of Household Fuel Choice in Major Cities in Ethiopia, Environ ment for Development (EdF), EdF Discussion Paper no. 08-18, with Resources for the Future, Washington, D.C, August 2008, pp. 8-18; B. Campbell, S. Vermeulen, J. Mangono, and R. Mabugu, The Energy Transition in Action: Domestic Fuel Choices in a Changing Zimbabwe, Energy Policy, vol. 31, no. 6 (2003), pp. 553-62; and A. Foster and M. Rosenzweig, Economic Growth and the Rise of Forests, Journal of Economics, vol. 118, no. 2 (May 2003), pp. 601-37.

O. Masera, B. Saatkamp, and D. Kammen, From Linear Fuel Switching to Multiple Cooking Strategies: A Critique and Alternative to the Energy Ladder Model, World Development, vol. 28, no. 12 (2000), pp. 2083-103, and Samuel F. Faye, Household Consumption Pattern and Demand for Energy in Urban Ethiopia, World Bank Technical Paper no. 304, Energy Series (Washington, D.C.: The World Bank, 2002).

206

THE JOURNAL OF ENERGY AND DEVELOPMENT

3 A. Ejige, op. cit.; M. Alemu and G. Kohlin, op. cit.; R. Heltberg, Factors Determining Household Fuel Choice in Guatemala, Environment and Development Economics, vol. 10, no. 3 (2005), pp. 337-61; D. Kammen and D. Lew, Review of Technologies for the Production and Use of Charcoal, Renewable and Appropriate Energy Lab Report, Energy and Resources Group and the Goldman School of Public Policy, University of California, Berkeley, California, 2004; and G. Van der Horst and A. J. Hovorka, Reassessing the Energy Ladder: Household Energy Use in Maun, Botswana, Energy Policy, vol. 36, no. 9 (2008), pp. 3333-344.

Le Chen, Nico Heerink, and Marrit van den Berg, Energy Consumption in Rural China: A Household Model for Three Villages in Jiangxi Province, Ecological Economics, vol. 58, no 2 (2006), pp. 407-20; A. Foster and M. Rosenzweig, op. cit.; R. Heltberg, op. cit.; and B. S. Reddy, Overcoming the Energy Efficiency Gap in Indias Household Sector, Energy Policy, vol. 31, no. 14 (2003), pp. 1117-127.
5 A. Ejige, op. cit.; M. Alemu and G. Kohlin, op. cit.; S. Faye, op. cit.; and K. Bereket, A. Bekele, and E. Kedir Can the Urban Poor Afford Modern Energy? The Case of Ethiopia, Energy Policy, vol. 30, no. 11-12 (2002), pp. 1029-45.

O. Chipman, Trends in Consumption and Production: Household Energy Consumption, ST/ ESA/1999/DP.6, DESA Discussion Paper no. 6, United Nations Department of Economic and Social Affairs, New York, 1999, available at www.un.org/esa/desa/papers/discussionserie.html. D. Barnes, K. Krutilla, and W. F. Hyde, The Urban Household Energy Transition: Social and Environmental Impacts in Developing Countries, Resources for the Future and the Energy Sector Management Assistance Program, World Bank, Washington, D.C., 2005. M. Alemu, Rural Household Biomass Fuel Production and Consumption in Ethiopia: A Case Study, Journal of Forest Economics, vol. 5, no. 1 (2009), pp. 69-97; R. Heltberg, T. Arndt, and N. Sekhar, Fuel Wood Consumption and Forest Degradation: A Household Model for Domestic Energy Substitution in Rural India, Land Economics, vol. 76, no. 2 (2000), pp. 213-32; and B. Reddy, op. cit.
9 Articles dealing with India, Nepal, and China are Bhim Adhikari, Salvatore Di Falco, and Jon C. Lovett, Household Characteristics and Forest Dependency: Evidence from Common Property Forest Management in Nepal, Ecological Economics, vol. 48, no. 2 (2004), pp. 245-57; L. Chen et al., op. cit.; R. Heltberg et al., op. cit.; and B. Reddy, op. cit. The article dealing with Ethiopia is M. Alemu, op. cit. 10 B. Reddy, op. cit.; K. Bereket et al., op. cit.; S. Faye, op. cit.; A. Ejige, op. cit.; and M. Alemu and G. Kohlin, op. cit. 8 7

B. Jargstorf, Renewable Energy & Development (Addis Ababa, Ethiopia: Deutsche Gesellschaft Technische Zusammenarbeit and Ethiopian Rural Energy Development and Promotion Centre, 2004).
12 13 14

11

International Energy Agency (IEA), World Energy Outlook (Paris: IEA, 2005). R. Heltberg et al., op. cit.

Mark Davis, Rural Household Energy Consumption: The Effects of Access to ElectricityEvidence from South Africa, Energy Policy, vol. 26, no. 3 (1998), pp. 207-17, and O. Masera et al., op. cit.

ENERGY DEMAND IN RURAL ETHIOPIA

207

15 Michael J. Arnold, G. Kohlin, and R. Persson, Wood Fuels, Livelihoods, and Policy In terventions: Changing Perspectives, World Development, vol. 34, no 3 (2006), pp. 596-611.

Energy Sector Management Assistance Program (ESMAP), Household Fuel Use and Fuel Switching in Guatemala, Joint United Nations Development Program/World Bank Energy Sector Management Assistance Program, Washington, D.C., 2003, and G. Leach, The Energy Transition, Energy Policy, vol. 20, no. 2 (1992), pp. 116-23.
17 D. Jack, Income, Household Energy and Health (Cambridge, Massachusetts: Kennedy School of Government, Harvard Press, 2004).

16

M. Chambwera, Economic Analysis of Urban Fuel Wood Demand: The Case of Harare in Zimbabwe, World Wide Fund for South Africa Regional Programme Office (WWF- SARPO), Wageningen, Netherlands, 2004.
19 20

18

Energy Sector Management Assistance Program, op. cit. B. Campbell et al., op. cit.

21 M. Alam, J. Sathaye, and D. Barnes, Urban Household Energy Use in India: Efficiency and Policy Implications, Energy Policy, vol. 26, no. 11 (1998), pp. 885-91; B. Campbell et al., op. cit.; M. Davis, op. cit.; and Boukary Ouedraogo, Household Energy Preferences for Cooking in Urban Ouagadougou, Burkina Faso, Energy Policy, vol. 34, no. 18 (2006), pp. 3787-795. 22 23

Energy Sector Management Assistance Program, op. cit.

S. Demurge and M. Fournier, Poverty and Firewood Consumption: A Case Study of Rural Households in Northern China, Working Paper no. 1020, Groupe dAnalyse et de Theorie Economique (GATE), French Centre for Research on Contemporary China, September 2010.
24 25 26 27

M. Arnold et al., op. cit. D. Barnes et al., op. cit. B. Campbell et al., op. cit.

J. Baland, P. Bardhan, S. Das, D. Mookherjee, and S. Sarkar, The Environmental Impact of Poverty: Evidence from Firewood Collection in Rural Nepal, Indian Statistical Institute, Delhi, India, 2007.
28 29

R. Heltberg et al., op. cit.

A. Deaton and J. Muellbauer, An Almost Ideal Demand System, American Economic Review, vol. 70, no. 3 (1980), pp. 312-26.
30

J. Baland et al., op. cit.

208

THE JOURNAL OF ENERGY AND DEVELOPMENT Appendix 1


FIXED EFFECT ESTIMATES FOR INFERIOR FUELS
a

Fixed effects (within) regression Group variable (i): householdh;e R-square: within = 0.4878 between = 0.6888 overall = 0.4978 corr(u_i, Xb) = 0.0518

Number of observations = 2,574 Number of groups = 230 Observations per group: Minimum = 1 Average = 11.2 Maximum = 36 F(12,2332) = 185.11 Prob > F = 0.0000 Ln share of inferior fuels Coefficient Standard error -.5293739 -.0992429 -.0530543 -.0015909 -.0169841 .0007815 .0230885 .0120507 -.004594 .0000332 -.2140095 1.260466 .2895455 .0224785 .0147846 .0328038 b .0050436 c .016097 b .0008952 b .0447186 .0003532 .001736 b .0000461 .0160571 .0367022 .1630363

Explanatory variables Ln monthlyenergy Ln totalexpenditure Male headed Age of household head Household family size Household family size square Neducation Time for fuel collection Total livestock Age square Land owned Dummy 2004 Constant

sigma_u j .33833361 sigma_e j .68962965 rho j .19493722 (fraction of variance due to u_i) F test that all u_i=0: F(229, 2332) = 0.82 Prob > F = 0.9742

a Ln share of inferior fuels = logarithm of share of expenditure on inferior fuels out of total household energy expenditure; Ln monthlyenergy = logarithm of total monthly expenditure on energy out of total household energy expenditure; Ln totalexpenditure = logarithm of household total monthly expenditure; Male headed = households that are headed by a male; Neducation = households whose head is literate; Time for fuel collection = time spent by household on fuel collection; Total livestock = total number of livestock possessed by household; Land owned = land holding of households; Age square = age of household head square; and Dummy 2004 = dummy for households that used inferior fuel in year 2004. b Significance at a 5-percent level. c Significance at a 1-percent level.

ENERGY DEMAND IN RURAL ETHIOPIA Appendix 2


RANDOM EFFECT ESTIMATES FOR INFERIOR FUELS Random-effects GLS regression Group variable (i): householdh;e R-square: within = 0.4873 between = 0.6984 overall = 0.4983 Random effects u_i ; Gaussian corr(u_i, X) = 0 (assumed)
a

209

Number of observations = 2,574 Number of groups = 230 Observations per group: Minimum = 1 Average = 11.2 Maximum = 36 Wald chi2(12) = 2544.12 Prob > chi2 = 0.0000 Ln share of inferior fuels Coefficient Standard error -.5482823 -.0947086 -.0488898 -.0008888 -.0144359 .0005589 .0101746 .0122888 -.0029072 -.2095486 .000022 1.261316 .2874214 .0208406 .0140152 c .0311056 b .0047632 c .0150301 b .0008404 b .0421348 .0003358 .0016185 .01518 b .0000437 .035495 .1544288

Explanatory variables Ln monthlyenergy Ln totalexpenditure Male headed Age of household head Household family size Household family size square Neducation Time for fuel collection Total livestock Land owned Age square Dummy 2004 Constant sigma_u j 0 sigma_e j .68962965 rho j 0 (fraction of variance due to u_i)

a Ln share of inferior fuels = logarithm of share of expenditure on inferior fuels out of total household energy expenditure; Ln monthlyenergy = logarithm of total monthly expenditure on energy out of total household energy expenditure; Ln totalexpenditure = logarithm of household total monthly expenditure; Male headed = households that are headed by a male; Neducation = households whose head is literate; Time for fuel collection = time spent by household on fuel collection; Total livestock = total number of livestock possessed by household; Land owned = land holding of households; Age square = age of household head square; and Dummy 2004 = dummy for households that used inferior fuel in year 2004. b Significance at a 5-percent level. c Significance at a 1-percent level.

210

THE JOURNAL OF ENERGY AND DEVELOPMENT Appendix 3


HAUSSMANN TEST FOR INFERIOR FUELS Coefficient (b-B) Difference -.004534 .0189084 -.0041644 -.0007022 -.0025482 .0002225 .0129139 -.0002381 -.0016868 -.0044608 .0000112 -.0008503 Sqrt (diag(V_b-V_B)) S.E. .0047074 .0084233 .0104178 .0016583 .0057627 .0003085 .0149804 .0001095 .0006278 .0052343 .0000146 .0093357
a

Explanatory variables Ln monthlyenergy Ln totalexpenditure Male headed Age of household head Household family size Household family size square Neducation Time for fuel collection Total livestock Land owned Age square Dummy 2004

(b) Fixed -.5293739 -.0992429 -.0530543 -.0015909 -.0169841 .0007815 .0230885 .0120507 -.004594 .0000332 -.2140095 1.260466

(B) Random -.5482823 -.0947086 -.0488898 -.0008888 -.0144359 .0005589 .0101746 .0122888 -.0029072 -.2095486 .000022 1.261316

chi2(12) = (b-B)[(V_b-V_B)^(-1)](b-B) = 22.15 Prob>chi2 = 0.0359

b = consistent under Ho and Ha (obtained from xtreg); B = inconsistent under Ha, efficient under Ho (obtained from xreg); Test: Ho difference in coefficients not systematic; Ln monthlyenergy = logarithm of total monthly expenditure on energy out of total household energy expenditure; Ln totalexpenditure = logarithm of household total monthly expenditure; Male headed = households that are headed by a male; Neducation = households whose head is literate; Time for fuel collection = time spent by household on fuel collection; Total livestock = total number of livestock possessed by household; Land owned = land holding of households; Age square = age of household head square; and Dummy 2004 = dummy for households that used inferior fuel in year 2004.

ENERGY DEMAND IN RURAL ETHIOPIA Appendix 4


FIXED EFFECT ESTIMATES FOR ADVANCED FUELS Fixed effects (within) regression Group variable (i): householdh;e R-square: within = 0.2893 between = 0.3880 overall = 0.2920 corr(u_i, Xb) = 0.0430
a

211

Number of observations = 2,548 Number of groups = 229 Observations per group: Minimum = 1 Average = 11.1 Maximum = 35 F(12,2307) = 78.25 Prob > F = 0.0000 Ln share of advanced fuels Coefficient Standard error .1209656 .0874288 .037241 .0002461 .0047989 -.0001132 -.0029132 -.005057 .0027259 .1006283 -.0000101 -.5150081 -1.174782 .0144727 .0095619 .0212061 b .0032305 b .0103561 b .0005739 b .0286686 .0002258 .0011193 .0103862 b .0000295 .0231099 .1052705

Explanatory variables Ln monthlyenergy Ln totalexpenditure Male headed Age of household head Household family size Household family size square Neducation Time for fuel collection Total livestock Land owned Age square Dummy 2004 Constant

sigma_u j .29405992 sigma_e j .44034194 rho j .30841581 (fraction of variance due to u_i) F test that all u_i = 0: F(228, 2307) = 1.07 Prob > F = 0.2475

a Ln share of advanced fuels = logarithm of share of expenditure on advanced fuels out of total household energy expenditure; Ln monthlyenergy = logarithm of total monthly expenditure on energy out of total household energy expenditure; Ln totalexpenditure = logarithm of household total monthly expenditure; Male headed = households that are headed by a male; Neducation = households whose head is literate; Time for fuel collection = time spent by household on fuel collection; Total livestock = total number of livestock possessed by household; Land owned = land holding of households; Age square = age of household head square; and Dummy 2004 = dummy for households that used inferior fuel in year 2004. b Significance at a 5-percent level.

212

THE JOURNAL OF ENERGY AND DEVELOPMENT Appendix 5


RANDOM EFFECT ESTIMATES FOR ADVANCED FUELS
a

Random-effects GLS regression Group variable (i): householdh;e R-square: within = 0.2879 between = 0.4242 overall = 0.2934 Random effects u_i ; Gaussian corr(u_i, X) = 0 (assumed)

Number of observations = 2,548 Number of groups = 229 Observations per group: Minimum = 1 Average = 11.1 Maximum = 35 Wald chi2(12) = 1052.65 Prob > chi2 = 0.0000 Ln share of advanced fuels Coefficient Standard error .1389205 .0830401 .0326566 -.0007469 .0049582 -.0000478 .0090065 -.0053275 .0019204 .0946593 -1.24e-07 -.5203008 -1.169512 .0136935 .0091666 c .0202562 b .0030987 b .0097561 b .0005441 b .0272193 .0002171 .0010577 .0099488 b .0000284 .0230657 .1012165

Explanatory variables Ln monthlyenergy Ln totalexpenditure Male headed Age of household head Household family size Household family size square Neducation Time for fuel collection Total livestock Land owned Age square Dummy 2004 Constant sigma_u j 0 sigma_e j .44040619 rho j 0 (fraction of variance due to u_i)

a Ln share of advanced fuels = logarithm of share of expenditure on advanced fuels out of total household energy expenditure; Ln monthlyenergy = logarithm of total monthly expenditure on energy out of total household energy expenditure; Ln totalexpenditure = logarithm of household total monthly expenditure; Male headed = households that are headed by a male; Neducation = households whose head is literate; Time for fuel collection = time spent by household on fuel collection; Total livestock = total number of livestock possessed by household; Land owned = land holding of households; Age square = age of household head square; and Dummy 2004 = dummy for households that used inferior fuel in year 2004. b Significance at a 5-percent level. c Significance at a 1-percent level.

ENERGY DEMAND IN RURAL ETHIOPIA Appendix 6


HAUSSMANN TEST FOR ADVANCED FUELS Coefficient (b-B) Difference -.6682944 -.1822829 -.0857108 -.0008441 -.0219423 .0008293 .014082 .0173782 -.0065143 -.3086688 .0000333 1.780767
a

213

Explanatory variables Ln monthlyenergy Ln totalexpenditure Male headed Age of household head Household family size Household family size square Neducation Time for fuel collection Total livestock Land owned Age square Dummy 2004

(b) Fixed .1209656 .0874288 .037241 .0002461 .0047989 -.0001132 -.0029132 -.005057 .0027259 .1006283 -.0000101 -.5150081

(B) Random .1389205 .0830401 .0326566 -.0007469 .0049582 -.0000478 .0090065 -.0053275 .0019204 .0946593 -1.24e-07 -.5203008

Sqrt (diag(V_b-V_B)) S.E. .0178262 .0115999 .0258026 .0039794 .0128036 .0007109 .0354805 .0002787 .0013766 .0126037 .0000362 .0285486

chi2(12) = (b-B)[(V_b-V_B)^(-1)](b-B) = 7401.01 Prob>chi2 = 0.000

b = consistent under Ho and Ha (obtained from xtreg); B = inconsistent under Ha, efficient under Ho (obtained from xreg); Test: Ho difference in coefficients not systematic; Ln monthlyenergy = logarithm of total monthly expenditure on energy out of total household energy expenditure; Ln totalexpenditure = logarithm of household total monthly expenditure; Male headed = households that are headed by a male; Neducation = households whose head is literate; Time for fuel collection = time spent by household on fuel collection; Total livestock = total number of livestock possessed by household; Land owned = land holding of households; Age square = age of household head square; and Dummy 2004 = dummy for households that used inferior fuel in year 2004.

Vous aimerez peut-être aussi