On the adoption of electricity as a domestic source by Mozambican households

On the adoption of electricity as a domestic source by Mozambican households

Energy Policy 38 (2010) 7235–7249 Contents lists available at ScienceDirect Energy Policy journal homepage: www.elsevier.com/locate/enpol On the ad...

3MB Sizes 1 Downloads 13 Views

Energy Policy 38 (2010) 7235–7249

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

On the adoption of electricity as a domestic source by Mozambican households Maria de Fatima S.R. Arthur n, Sammy Zahran, Gabriela Bucini Electricidade de Mozambique, Av Zedequias Manganhela, 267, Jat IV Building, 4th floor, Maputo, Mozambique

a r t i c l e in fo

abstract

Article history: Received 5 November 2009 Accepted 30 July 2010 Available online 21 August 2010

In Mozambique, domestic energy is often composed of a mix of sources, primarily used for lighting and cooking, with biomass and kerosene as more common sources. Electrification programs, intended to connect new consumers countrywide, have not significantly contributed either to the intensification of electricity consumption or to the reduction of the use of biomass in households. The choice of energy source is dependent on price and on the capability of the household to invest in energy-consuming appliances. Based on the data from a household survey carried out in Mozambique during 2002/03, this paper analyzes the geographic differences in unit expenditures for domestic energy and finds evidence of an inverted energy ladder with prices of useful energy units. The data show that biomass sources are often more expensive per unit of useful energy than higher-grade sources, supporting arguments favoring electrification as a poverty alleviation strategy. In addition, this study estimates the likelihood of poor households transitioning from biomass to electricity consumption based on various factors. Results indicate that income is not a determining factor in the transition, but wealth and the level of the Primary Energy Consumption Share (PECS) are as important factors as the nature of the energy mix. & 2010 Elsevier Ltd. All rights reserved.

Keywords: Domestic electricity Likelihoods Energy ladder

1. Introduction Mozambican households rely mostly on firewood and charcoal as sources of domestic energy (Brouwer and Falcao, 2004). Although the national public electricity company (EdM) has invested heavily in electrification programs in the past 30 years,1 the connection rate was only at 8.2% of the population in 2006, and the average monthly domestic consumption at the level of 89 kWh per month per household connected to the national grid (EDM, 2007), constituting about a third of the current US domestic energy consumption (EIA, 2001). The expansion of electrical grids supports future economic and social developments (see Mulder and Tembe, 2006) and remains a priority in the company’s agenda. However, the low connection and consumption rates of domestic energy raise doubts on the validity of accelerated electrification as a poverty alleviation strategy targeting households and rural communities. Although the number of connected households in the country has grown at an average rate of 15% per year between 2000 and 2007, connection rates, discounting for population changes, have only increased by 5.4% of the population. It is obvious that increased access, by extension of electrical networks, is insufficient to transition households from lower grade energy sources such as biomass and kerosene to electricity. n

Corresponding author. Tel.: +258 82 313 9200. E-mail addresses: [email protected], [email protected] (M. de Fatima S.R. Arthur). 1 In the US, government also intervened to expand residential electricity in the late 1940s–1950s (Morton, 2002). 0301-4215/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2010.07.054

Typically, the price of electricity also constrains adoption by poor families. For this reason and in the spirit of equal opportunity, the Mozambican Electricity Law (Mozambique, 1977) establishes that (a) The electricity rates at the low-voltage distribution level (retail distribution) cannot be geographically differentiated, i.e. electricity prices are the same whether the consumer lives in a northern province (farthest from the existing generating sources) or in the south; (b) There must be a fixed rate called a ‘social tariff’ that is significantly lower than the average rate for domestic consumers—this rate is currently applicable to all households consuming up to 100 kWh per month (EDM, 2007) and after having proved their condition of low income.2 The retail prices of electricity increased by 18% between 2005 and 2007, however the social tariff, specially designed to favor low-income households, was kept at the level of 0.14 $PPP/kWh.3 This (subsidized) rate is much lower than market prices for charcoal recorded at 0.17 $PPP/kWh in 1997 (Falca~ o, 1999), and 2 To benefit from the ‘social tariff’ the consumer must prove their condition of low income (a laborious, subjective and difficult procedure by which the Welfare Service Centers (Acc- a~ o Social) issue a Certificate of Poverty) and that the electricity consumed is purely for domestic consumption and not for any productive activity, including self-employment and domestic production of any sort. This requirement makes the social tariff virtually inaccessible by poor domestic consumers. 3 Conversion rates published by the United Nations (2007).

7236

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Nomenclature AEU Adult Equivalent Units EDM or EdM National Public Electricity Company (Electricidade de Mozambique) HH household IAF or IAF2002/03 National Household Survey (Inquerito aos Agregados Familiares) in 2002/3 INE National Institute of Statistics (Instituto Nacional de Estatistica)

for kerosene and liquefied petroleum gas (LPG), recorded both at 0.59 $PPP/kWh4 in 2006 (Mozambique, 2007). Less than 1% of the domestic consumers take advantage of the social tariff (EDM, 2005, 2007.5 When planning for domestic electricity, three important issues must be resolved to increase the number of electrical consumers and to intensify the electricity consumption per household: (a) Access: the extension of medium and low voltage networks in rural and urbanized areas must continue, so that the grid may reach a larger number of the population; (b) Capital investment: the cost (investment) of an electrical connection to the grid6 and acquiring electricity consuming appliances must be within the reach of potential new or existing electricity consumers; (c) Affordability: the electricity price must be competitive with other sources and affordable by domestic consumers. The assumption is that households will consume electricity if made affordable and accessible, as electricity is a preferred source when no budget or capital constraints limit the choice of the domestic source7. This assumption fits the principle of an energy ladder, which establishes that as a household increases its income it will transition from cheap, low-efficiency biomass sources to more costly, more efficient and more wide-ranging-in-use sources such as kerosene and electricity. The energy ladder, first postulated by Hosier and Dowd (1987), is a criterion for ranking preferences for domestic energy sources based on the households’ incomes, with biomass sources at the bottom and electricity at the top. Conventionally, the energy 4 Cockburn and Low (2005) place domestic kerosene costs at 0.40 $/kWh versus electricity costs at 0.07 $/kWh, though not converted to $PPP. Electricity prices are (cross) subsidized because electricity is generated in Mozambican soil (in Cahora Bassa and Revue hydro-stations mostly), and it is seen as a necessary input for economic development and social equality. Only off-grid electricity generation is allowed to charge prices differentiated from the unique national electricity tariff, applied by EDM in the national grid. Kerosene is an imported source and its prices are regulated at the port of entry (Maputo, Beira and Nacala). The process, transportation and distribution from the port to the final customers are run partly by private operators and in a de-regulated market. Consequently, kerosene prices show market-driven fluctuations and are sensitive to international oil-prices fluctuations. 5 In 2004 EDM introduced the Quadrelec meter, which corresponds to a onephase meter with a consumption board that allows grass huts to benefit from electricity. Still, the attribution of the social tariff for these consumers is not automatic under the assumption that they may use electricity for productive purposes and consume more than 100 kWh per month. As a consequence, in 2009 still only about 1% of the domestic population benefited from the social tariff. 6 Only in 2010 was reduced the price of electrical installations for those consumers that adopted the pre-paid metering system, from 3500 to 800 MTn (published in ‘‘Canal de Moc-ambique’’, 20/April/2010, www.canalmoz.com). 7 Previous research indicates that investment in energy appliances and electricity prices are limiting factors in the transition from low-grade sources to electricity by households (Hosier and Dowd, 1987; Reddy, 1995; Masera et al., 2000; Tewari and Shah, 2003).

kWh kilowatt-hour (energy unit) kWh-eq kilowatt-hour-equivalent (useful energy unit, equivalent to electricity consumption) Kg kilogram L liter LPG Liquefied petroleum gas MTn Mozambican currency (Metical) $PPP Purchasing-power-parity dollar Un Units of candles

ladder places the cheapest sources at the bottom (preferred by the low-income families) and the more expensive at the top, also called ‘modern fuels’ (Kebede et al., 2002; Howells et al., 2005) in recognition of their superiority in convenience, cleanness, safety (Masera et al., 2000) and on the variety of end-uses. Although electricity is required for economic, technological and social developments, it is still a rare commodity for the majority of the African population (Wolde-Rufael, 2006). Urbanization is a major factor in electrification (Karekezi and Majoro, 2002) as houses are more compact and closer to services and industries. In fact, a secondary benefit of electrification for profit (industry, commercial and services enterprises, public or private) is a lower grid investment per domestic connection. Evidence shows that in developing countries higher-income families retain some level of consumption of biomass fuels, even when combined with electricity consumption (Campbell et al., 2003; Brouwer and Falcao, 2004). The energy transition is not a progressive adoption of higher-grade and more expensive fuels, but rather a combination of fuels from the lower and the top levels of the ladder. Biomass is still the preferred energy source for cooking (Madubansi and Shackleton, 2007) even if lighting is obtained from candles, batteries, kerosene8 or electricity (Davis, 1998; Howells et al., 2005). The energy ladder concept, in its initial formulation, does not account for the mix of sources in both extremes of price and efficiency ranges: for example, why do wealthy consumers maintain charcoal as a cooking source? Hughes-Cromwick (1985) has surveyed the reasons behind the selection of primary fuels in Kenya. The survey indicates that economics is the most important decision factor for poor households, while factors such as convenience and availability play an important role in wealthier households. Similarly, Gupta and Kohlin (2006) ranked sources for cooking activities based on price, availability, ease of use, capital cost of appliance (oven) and pollution level and asked the households to rank the cooking sources based on these criteria. Not surprisingly, most of the households ranked electricity as the most expensive and the cleanest, while fuel wood was classified as the dirtiest and the cheapest source. Green and Erskine (1999), on the other hand, surveyed the inconvenience ranking for several sources measured by the time taken or the distance travelled to secure a specific source. Surprisingly, households perceive fuel wood as an inconvenient energy source, similar in inconvenience as the energy drawn from a petrol generator, a gas generator or from a car battery. Green’s results indicate that sources not readily available at the household location are undesirable, no matter their relative position in the energy ladder. Households value the time required in securing

8 Transitional sources such as candles and kerosene are mostly used for lighting (Brouwer and Falcao, 2004; Gupta and Kohlin, 2006), while the household still maintains biomass as a cooking source. However, some urban communities use kerosene or LPG, rather than firewood, for cooking (Anozie et al., 2007; Troncoso et al., 2007), so transitional sources may be different for different geographical locations.

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

the source more than any classification of grade in a ladder that has no practical meaning in itself, but to satisfy a researcher’s need to categorize and isolate causes and effects. Although there are differences among communities, the factors that determine the choice of fuel for domestic use and the intensity of consumption are very similar, and poverty is the major determinant of household preferences. For example, in Maputo, the majority of the (sampled) population burns charcoal for cooking and firewood usage is in a descending trend (Brouwer and Falcao, 2004), as opposed to Ouagadougou, where firewood is the most common source (Ouedraogo, 2006). The household size and educational levels are, in both cases, significant in the choice of fuel for cooking. Some authors (Reddy, 1995; Masera et al., 2000; Brouwer and Falcao, 2004) emphasize taste (palate) and tradition to explain behaviors, which are discordant with the energy ladder expectation. However, notwithstanding the uniqueness of the individual’s preferences in choice, self-interest pushes the household to progressively adopt the higher-grade sources9. Thus, new energy sources will be adopted if the households can access and afford them, if they are adequately informed of the benefits and gains from this adoption, and if the opportunity costs of the energy transition are not too high. Factors such as the cost of the appliances for the specific energy source (Reddy, 1995) and the compatibility of the cost schedules (tariffs) with the household earnings (income) have shown strong correlations with household choices of fuel (Masera et al., 2000). The affordability of energy consuming appliances can accelerate household adoption of electricity (Karekezi and Majoro, 2002; Pachauri et al., 2004). Note that a wealthy status, or simply the ability to invest after paying for the consumption costs, is necessary for the acquisition of energy-consuming appliances. So far, no quantitative assessment has been made to demonstrate that wealth is determinant in adopting electricity as a domestic source. The ownership of appliances as a necessary condition to adopt higher-grade source has been suggested (Elkan, 1988; Reddy, 1995; Tyler, 1996; Gupta and Ravindranath, 1997; Tiwari, 2000), without a quantification of possible effects. This study investigates energy consuming behaviors, recorded in a household survey from 2002/3 collected by the National Statistics Institute in the whole of the Mozambican territory (INE, 2007). The goal is to clarify factors that determine choices in the household energy consuming behaviors and what may determine the adoption of electricity as a domestic source. A brief description of the data set and the transformations made on the data can be found in the Appendix. This work has two specific objectives: (1) To present the current unit expenditures on energy sources in Mozambique and analyze the relative position of charcoal, kerosene and electricity in the Mozambican domestic energy ladder, by price-per-unit-of-useful-energy. (2) To investigate and quantify the likelihood of transitioning from biomass to higher-grade sources such as kerosene and electricity as a function of income levels, wealth ownership10 and the level of Primary Energy Consumption Share11 (PECS). 9 Jenkins and Scott (2007) show that where self interest is served, together with accessibility and affordability of the technology, it is adopted by the household. 10 There sometimes is some confusion between income and wealth. An easy distinction is that Income is a flow and Wealth is a stock. Note however that only households with income surplus may invest by acquiring wealth and those households with productive stocks (wealth) may be able to generate higher incomes. 11 This quantity is calculated by dividing the amount of kWh-equivalent consumed of the primary source, over the total kWh-equivalent consumed in the household, during the same period.

7237

This paper is organized in the following manner: firstly, we analyze the recorded energy consumption per district and discuss how the domestic energy ladder for Mozambique is inverted in terms of price-per-unit-of-useful-energy. Secondly, we calculate the likelihood of adopting electricity as a domestic source and discuss the results and their implications for policy strategies. Finally, we summarize the findings and discuss some policy implications.

2. Study area and the household survey data set 2.1. Location Mozambique, located in the southeast Africa at the coordinates 18:15 South and 35:00 East, with an extension of 799.38 km2, has a population of 20.4 million people of which 51% are women (INE, 2007). Although a country of many resources, in agriculture, mining and energy, and with an economic growth rate of 6.2% in 2005, Mozambique still ranks 172 in the Human Development Index, with a GDP of only 1242 $PPP/capita in 2005.12 The northern part of the country is more densely populated than the south. The north is also where major agricultural, forestry and mining resources exist. In general the south is more developed (better infrastructure and more industry and services) than the north and has a better-trained workforce because of the traditional migration to South Africa.13 The population is mostly concentrated on the coast, along which the main national (EN1) road runs. The Maputo province is a route of transportation (between the continent and the Maputo seaport), trade, tourism and migration to and from South Africa and constitutes one of the main development corridors, named the ‘Maputo Corridor’. The center of the country is only about 100 km wide from East to West, and it contains the ‘Beira Corridor’, with a system of roads (EN6) and railway lines that links the Zimbabwean hinterland with the Beira seaport. Along the Beira corridor there is an oil pipeline to transfer oil into Zimbabwe. In this route, there is also a relatively well-developed electrical distribution system (previously known as SHER) and several communities that provide services and trade. In the north, the ‘Nacala Corridor’ is a system of roads and rail tracks connecting Malawi and Zambia to the Nacala seaport. In these development corridors, urbanization and higher living standards are generally observed.14 In the spirit of the energy ladder concept, the development corridors should record higher energy consumption levels and the use of higher-grade sources (electricity for example) in the domestic settings.15 2.2. Observations on the energy data The energy unit values, recorded per source (per useful energy unit, see details in the Appendix), were plotted by the household consumption to demonstrate the typical behavior of the ‘demand curves’ per individual source. The ‘demand curves’ (Fig. 1) indicate that the energy source consumption decreases as its unit value 12 Source: UNDP, Human Development Report 2007/2008 at:http://hdrstats. undp.org/indicators/5.html 13 The high concentration in the south, of families whose head is a woman, is a consequence of male migration to the South African gold mines in search of work. 14 The revitalization of transport networks has the secondary effect of cheapening the trade of agricultural goods and the provision of services (Simler et al., 2004). See Tarp et al. (2002) for an historic overview of the political and economic policies in Mozambique. 15 Simler et al. (2004) studied the determinants of poverty in terms of food consumption and nutrition. They calculated lifelines of consumption per household and demonstrated that rural households have significantly lower lifeline consumption levels, particularly on the non-food items (transport, energy, education, health, clothes and other perishable items such as soap).

7238

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Charcoal $PPP/kWh proportional to exp (-0.050831 * kWh/day) , R2 = 0.0012743

Firewood $PPP/kWh proportional to exp (0.14391 * kWh/day) , R2 = 0.0004986

30 25 20 15 10 5

10 8 6 4 2 0

0 0

0.2

0.4

0.6

0.8

1

1.2

0

1

2

3

4

5

6

7

8

Consumed kWh-eq per Day

Consumed kWh-eq per Day

Kerosene $PPP/kWh proportional to exp (-0.14847 * kWh/day), R2 = 0.31121

LPG $PPP/kWh proportional to exp (-0.14067 * kWh/day) , R2 = 0.5944

5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0

9

0

3 2.5 2 1.5 1 0.5 0

5

10

15

20

25

1.5

2

6

12 10 8 6 4 2

5 4 3 2 1 0

0

0

1

Electricity $PPP/kWh proportional to exp (-0.044689 * kWh/day ) , R2 = 0.67491 Average paid $PPP/kWh-eq

Average paid $PPP/kWh-eq

3.5

0.5

Consumed kWh-eq per Day

14

4 Average paid $PPP/kWh-eq

Average paid $PPP/kWh-eq

12 Average paid $PPP/kWh-eq

Average paid $PPP/kWh-eq

35

Candles $PPP/kWh proportional to exp (-1.3018 * kWh/day) , R2 = 0.21096

0

5 10 15 20 25 30 35 40 45 50

Consumed kWh-eq per Day

0

Consumed kWh-eq per Day

5

10

15

20

25

30

35

Consumed kWh-eq per Day

Fig. 1. Household expenses in energy (IAF 2002/03 survey).

Table 1 Number of records and households (HHs) surveyed on energy consumption, IAF 2002/03. Auto-consumption

Firewood Charcoal Candles Kerosene LPG Electricity Total records/HHs

Daily expenses

Monthly expenses

Total energy expenses

Records

HHs

Records

HHs

Records

HHs

Records

HHs

22,626 99.9% 21 0% 0 0% 4 0% 0 0% 0 0% 22,651

4727 54% 16 0% 0 0% 4 0% 0 0% 0 0% 4747

2,690 23% 4,171 36% 629 5% 4,132 36% 0 0% 3 0% 11,625

882 10% 1053 12% 263 3% 1638 19% 0 0% 1 0% 3837

973 16% 755 13% 346 6% 2,685 45% 213 4% 986 17% 5958

973 11% 754 9% 345 4% 2685 31% 213 2% 986 11% 5956

26,289 65% 4,947 12% 975 2% 6,821 17% 213 1% 989 2% 40,234

6,582 76% 1,823 21% 608 7% 4,327 50% 213 2% 987 11% 14,540

In italic are the percents of the number of records and the number of households, with regards to the total of records in each consumption category and the total of households surveyed (8700).

increases for most energy sources. Firewood, charcoal and kerosene are sufficiently represented in the recorded consumption of domestic sources (Table 1),16,17 the same does not happen

16 The quantities of firewood and charcoal, candles and kerosene were recorded in local units (such as ‘bundle’) and then converted by the surveyors into standard units such as kilograms and liters. 17 As shown in Table 1, ‘auto-consumption’ is recorded almost entirely for firewood, gathered by poor families on their own or on unsupervised public land. ‘Daily expenses’ data indicate high usage of low-grade energy sources such as firewood, charcoal and kerosene. To conform with the households’ schedule of earnings (retail economy), the electric public utility has installed and expanded

with candles, LPG and electricity, for which data are limited. Electricity is of higher importance in the consumption corresponding to ‘monthly expenditures’, although the record of 17% of households declaring monthly consumption of electricity corresponds to only 2% of the total number of households surveyed. This does not conform to the electric company’s statistics for

(footnote continued) the prepaid system for electricity consumers. However, at the time of the survey the coverage rate was much less than 30% of the consumer population and the sample records electricity consumption only as a monthly expenditure, not as a daily expenditure.

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

7239

Table 2 Recorded value of assets (owned) in the sampled households. Sources

Charcoal Kerosene LPG Electricity

Total value MTn

9.49E+ 04 5.78E+ 04 6.43E+ 07 3.51E+ 08

Total energy expenses [MTn/HH/day]

Average value per HH MTn/HH

$PPP/HH

1899 3211 2,297,071 67,680

0.33 0.57 405.2 11.94

Income (revenue) levels [MTn/HH/day]

# of HHs declaring assets

% share on 8700

50 18 28 5188

0.6 0.2 0.3 59.6

Total energy consumption [kWh/HH/day]

Fig. 2. Provincial energy expenses, consumption and income (scale lighter to darker ¼ smaller to higher values).

2003, which indicate a coverage of 5.3% of the total population (EDM, 2007). Out of 8700 households surveyed, 76% consume firewood, 54% of which were classified as auto-consumption (assumed selfgathered); 50% consume kerosene (19% daily, 31% with monthly expenses); 21% consume charcoal (12% daily, 9% with monthly expenses) and 11% consume electricity (only monthly expenses). The auto-consumption is 99.9% firewood—other (higher grade) sources are mostly acquired (bought) on a daily or a monthly basis. The usage of low-grade sources, firewood, charcoal and kerosene, respectively, 76%, 21% and 50% of households recording expenses in energy sources (Table 1) contradicts the ownership of assets (recorded at 0.6% for charcoal and 0.2% for kerosene of all households, Table 2). The percentage count of owners of electricity consuming appliances is surprisingly high (59.6%, Table 2) when compared with only 11% of households recording expenses with electricity. This discrepancy may be due to recording the possession of electrical appliances even when the household does not consume electricity, which makes the population’s asset records unrepresentative of the national average on electricity connection rates. This discrepancy is unlikely to result from electrical supply of private (isolated) generators, as they are estimated to cover only between 0.18% and 0.34% of the population (Mulder, 2007). Brouwer and Falcao (2004) showed that even in Maputo in the year 2000, where access to electricity was the highest in the country, firewood and charcoal were still sources for about 21% and 74% of the population, respectively. The expectation is to have

higher utilization of firewood in rural areas. The statistics for kerosene consumption in Maputo (49%) are in line with the total expenses recorded per source at the national level, 50% in the IAF2002/3 survey.

2.3. Income versus energy expenditures in the districts Income per day per capita,18 varying in the range 0.31–8.80 $PPP/ day-capita, with a mean of 2.39 $PPP/day-capita and a standard deviation of 2.07 $PPP/day-capita, record higher values in Maputo and in the central and some northern districts (Fig. 2). A comparison of the income levels with total energy consumption and expenditure per household indicates that the higher energy consumers are at the higher levels of income but that not all high income levels spend more in energy. Energy expenses are not the households’ highest priority: the mean energy expenditure is only of 0.24 $PPP/day-capita (0.14 $PPP/day-capita standard deviation), i.e. only 10% of the mean recorded income. The income-share of expenses in nonenergy items (food, clothing, etc) and in leisure tends to be higher in the north, while the share of expenses in public services (education, transport, health, etc) tends to be higher in the more urbanized south. Energy expenditure also seems to take higher shares of the income (budget) in the northern districts, though in 18 Data per capita were calculated by applying the IAF 2002/03 data on household earnings and expenses to the counting of households and population of the latest 2007 census (INE, 2007).

7240

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Fig. 3. Urbanization and development corridors correlate with higher-grade sources.

Table 3 Energy consumption and expenditures per district.

Unit values for in quantiles MTn/kWh Aggregated

Firewood

Charcoal

Kerosene

Electricity

Aggregate

kWh/day-capita Mean Std. deviation Median

0.13 0.03 0.13

0.05 0.13 0.00

0.77 0.92 0.44

0.71 2.03 0.00

1.78 2.88 0.71

$PPP/day-capita Mean Std. deviation Median

0.16 0.10 0.15

0.03 0.06 0.00

0.03 0.03 0.02

0.01 0.04 0.00

0.24 0.14 0.22

absolute values higher energy consumption coincide with high incomes, differences being attributable to unit (energy) expenditures.19 Higher budget-shares for energy expenses may signify lower budgets (poorer families) or higher energy prices, which are typical conditions of the northern districts. The households’ statements regarding their main sources for cooking and lighting can be used as proxy for asset ownership. Data conform to the concept of the energy ladder (higher-grade sources at higher incomes and vice versa). Although firewood is also heavily used in the south for lighting, its higher usage is observed in the center–northern inland districts, where lower income levels are also recorded. The higher-grade sources such as charcoal, kerosene and electricity record higher usage (for cooking and lighting) in the development corridors and urban areas (see Fig. 3). The study of the distribution of the sources and their prices and expenditures across the country may clarify the correlation factors between high-income districts and the levels of domestic energy consumption and energy expenditure recorded, calculated respectively at 0.27 (income  kWh/day) and 0.25 (daily income  daily energy expenses), both with p-values virtually null. 2.4. Energy in the districts Districts in the southern provinces of Maputo, Gaza and Inhambane, along the Beira Corridor and in the northwest of Mozambique, have higher levels of energy consumption, in kWh 19 The consumption of energy, in kWh units, almost perfectly fits the pattern of income in the provinces (Fig. 2).

Legend

128.5 - 476.9 477.0 - 1004 1005 - 2101 2102 - 3384 3385 - 11560

Fig. 4. Unit values per district, in useful energy units.

per household per day. The consumption of firewood is fairly even across the country, with a countrywide mean of 0.13 kWh/day per capita (0.03 standard deviation), while charcoal, kerosene and electricity are higher at the south, Beira corridor and north-west districts. In summary, in Mozambique the consumption of highergrade domestic sources corresponds to higher energy intensity; it is used for cooking and lighting and occurs in the south, Beira corridor and north-west districts. Energy expenditure amounts to a mean of 0.24 $PPP/day per capita, about 10% of the mean

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Firewood consumption kWh/HH-day

Charcoal consumption kWh/HH-day

7241

Kerosene consumption kWh/HH-day

Fig. 5. Energy consumption at the district level highlighted the development corridors (scale lighter to darker ¼ smaller to higher values).

recorded income. Table 3 reports the means and standard deviations of the consumption and the expenditures for individual energy sources. The data indicate that the energy expenditure corresponds to a higher share of the household budget in the north, partially explained by lower incomes. It is important to look at the cost of the more common sources used in the various districts: the calculation of the unit value (expenditure divided by quantity) of energy consumption, in aggregated form, indicates higher unit expenditures in the center–north western districts (Fig. 4). Most households consume firewood at a constant energy level, which results in high unit values of firewood for many districts.20 Charcoal consumption is the highest in the south, although the unit values are not the highest. Large expenditures are recorded in the central districts in association with high unit values and in the south in association with the highest consumption. Although most households in the sample consume firewood, its low energy levels result in a lower share of the total domestic energy consumed. Families, for which firewood is the only energy source, are well represented in this sample, particularly in the center–northern districts. Firewood consumption is predominant as a domestic source in the poorer districts (center and north), although it is present in households all over the country. By contrast, charcoal consumption occurs mostly in urbanized areas (higher income), possibly because firewood sources are distant and unavailable, i.e. charcoal consumption replaces firewood consumption (see Fig. 5). In the case of kerosene, the highest unit values are recorded in the center and the expenditures are at the highest levels in most of the districts, including some coastal areas. Kerosene is widely used, with higher consumption shares in the southern districts. Although kerosene may be up in the energy ladder because of its high efficiency and production process complexity, it is more affordable and widely accessible than charcoal. Coastal districts,

20 This constant level of energy from firewood is a result of standardizing the piles of firewood sold in retail markets and ignoring differences in calorific values of different types of wood and possible varying volumes of woodpiles. This approximation is necessary because data are not available, at the district level, establishing average woodpile volumes and their calorific content.

the country south, northwest and the Beira corridor, reserve a higher share of their budgets for kerosene (lighting) consumption. Electricity has significant consumption levels along the development corridors of Maputo and Beira. Electricity expenditures are at the highest in the south and some coastal northern districts, where it constitutes an alternative source also for poor families. Higher electricity consumption corresponds to higher electricity expenditures as the price is geographically uniform and only varies by consumption levels.

3. The domestic energy ladder: price-inverted? Fig. 3 shows maps of kerosene, charcoal and electricity consumption and the urban household subset. Kerosene consumers are clearly present in the development corridors though more widespread than other high-grade source consumers. Households consuming charcoal and electricity visibly correlate with urban status. Records for electricity, kerosene and charcoal consumption in the development corridors21 districts overlap records of higher energy consumption and higher income levels, indicating that high-grade sources are the preferred one by high-income families. The energy source choice for consumption in the domestic setting aligns with the traditional (expected) order of the energy ladder: firewood, charcoal, kerosene and electricity. However, the unit values (prices per useful energy, recorded per district) do not follow the same pattern as the consumption.22 The firewood unit values are higher on the western districts. 21 Electricity shows significant consumption levels along the development corridors of Maputo and Beira. Electricity expenditures are the highest in the south and some coastal northern districts, where it constitutes an alternative source also for poor families. Higher electricity consumption corresponds to higher electricity expenditures, as the price is geographically uniform and only varies by consumption levels. As the consumption increases, the expenditure of the household per electric kWh will increase, pulling electricity up the energy ladder of domestic sources. 22 The unit values are calculated by dividing the household’s expenditure (per source or aggregated for all sources) per day by the equivalent kWh of useful energy consumed, using conversion factors first published by the World Bank (1987), see the Appendix for more details on the data transformations.

7242

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Fig. 6. Unit values of domestic sources converted to $PPP: (A) across districts and (B) across the households.

Charcoal and kerosene’s unit values are higher in the central and northern districts. In other words, firewood, charcoal and kerosene unit values seem to be higher in the districts with lower income households than in the districts with higher development levels (where retail markets are better developed). So are poor households consuming energy (for their domestic needs) at higher prices than high-income households? An analysis of the unit values of the total domestic energy consumption (including all sources) shows that households in the poorer districts are spending more per unit of useful energy (Fig. 4) than those in wealthier districts. If the poor households are willing to pay such high prices for the consumption of biomass energy, they certainly would adopt a cheaper source if they could recognize the gains from this adoption (assuming rational behavior). Countrywide, from the district recorded totals, domestic energy (aggregated) records a mean of 0.41 $PPP/kWh and a standard deviation (SD) of 0.44 $PPP/kWh. This high price is a result of the prevalence of firewood consumption with a unit value of 1.29 $PPP/ kWh and SD of 0.69 $PPP/kWh. Higher energy sources show lower prices because of their higher efficiency: 0.30 $PPP/kWh (SD¼0.60 $PPP/kWh) for charcoal, 0.13 $PPP/kWh (SD¼0.23 $PPP/kWh) for kerosene and 0.12 $PPP/kWh (SD¼0.40 $PPP/kWh) for electricity. The unit values of the individual sources consistently show that firewood is more expensive than electricity (see Fig. 6). Unit values of electricity consumption do not reflect geographic variation because electricity prices are the same throughout Mozambique by law (Mozambique, 1977). Rather, the unit prices variation between districts show some districts have a larger number of big domestic consumers (high unit-values) than others. Based on these results, the ordering of sources by their energy price would be first electricity, followed by kerosene, charcoal and firewood. This is an inverted energy ladder where the prices are calculated per useful energy unit, i.e. the energy ladder for Mozambican households per price-of-useful-energy-units has the following order: electricity, kerosene, charcoal and firewood (Fig. 6). The puzzle is why does the consumption pattern follow the traditional order of the energy ladder when the sources’ prices indicate an inverted order? Electricity is cheaper than biomass and in the same price order as kerosene. It is also a more diverse, cleaner and safer source. Why is it not the choice of the poorer families? The records for household expenditures (Table 1) show that the relative importance of the energy sources in the average Mozambican household: at the national level, firewood and other low-grade sources are the most common domestic energy sources (72.4%), followed by kerosene (47.7%), charcoal (19.9%) and

electricity (11.3%). This ranking differs from the ranking of the unit (daily) cost of the sources, by which firewood is the most expensive, followed by charcoal, electricity and kerosene (Fig. 6B). Falca~ o (1999) has collected average prices for firewood, charcoal and kerosene and corrected them by the consumer price index as an inflation measure. The data show higher prices for kerosene, per unit consumed. If, however, these prices are converted to a comparable unit of kWh-equivalent, kerosene then is cheaper than charcoal and firewood (see Fig. 7), because of its higher use-efficiency and calorific content. The similarity of the sample’s unit cost ranking (Fig. 4) with Falca~ o’s (1999) price ranking (Fig. 7) provides validation of our results, despite the fact that the unit values (Fig. 6) are a combination of real prices and households’ perceptions of the energy sources quality and they are contaminated by measurement errors. There is evidence that electricity consumers are actually spending less than charcoal consumers in southern Mozambique (Cockburn and Low, 2005). Furthermore, Falca~ o’s (1999) price ranking, which our results corroborate, shows that electricity is cheaper than charcoal. This suggests an inverted energy ladder for domestic consumption. The categorization of the households by wealth and the composition of the domestic energy is important, in that other studies recognize that household wellbeing is linked to acquisition of higher grade sources: see examples in the models of McKenzie (2005) and Larsen and Nesbakken (2004). Location variables per household can also be indicators of life quality, as some areas of the country are more developed than others. The evidence of an energy ladder inverted by price of useful energy units is strong. From a policy standpoint, the poor need not to be deprived of more efficient, more convenient, more diverse and even cleaner domestic sources. By not consuming electricity, the poor households are actually paying more for their domestic source and being exposed to ground-level emissions of smoke, with negative health implications. The next section analyzes the factors determining the likelihood of Mozambican households adopting electricity as a main domestic energy source.

4. Likelihood of consuming electricity, kerosene or charcoal 4.1. Explanatory and response variables Electricity has cheaper prices than other sources. Under the concept of the energy ladder, it should thus be a source preferred

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

7243

Fig. 7. Prices of energy sources in Mozambique, 1985–1997.

by the low income families. However, it is a source preferred by high-income families. Why don’t the poor households adopt electricity if it is cheaper than biomass sources? We formulated the following hypotheses as possible causes: 1. Inadequate access in rural and peri-urban areas, where the poor live; 2. Initial high capital costs to connect to the grid and acquire electric appliances; 3. Misinformation and mismatched payment schedules. The first hypothesis is built from the fact that the electrical networks do not extend to all villages and all districts of Mozambique. The evaluation of household choices of domestic source needs to take into account the electricity accessibility. Other sources use the transportation routes and retail markets that reach wherever there is demand for the product, so their consumption is not constrained by access. Based on correlation observed between electricity consumption and urban status (Fig. 3), the location of the households (urban or rural) was used as an indicator of accessibility to electrical supplies. The second hypothesis relates to a shortage of funds available to people to invest in an electrical connection and appliances. In the absence of a rigorous survey on households’ wealth, the study used the characteristics of the household dwelling as indicators of wealth. These include whether the dwelling is self-owned, whether it has concrete walls or not, the number of rooms in the dwelling, and finally, whether the water for drinking consumed by the household is obtained from a piped system or not. These variables are binary, with exception to the ‘number of rooms’. The third hypothesis accounts for unfamiliarity with electricity as a domestic source and ignorance that it is cheaper per useful energy unit than other sources. In addition, a mismatch between the schedule of earnings and the schedule of electricity costs could influence the preference for a particular energy source. Poor consumers living on day-to-day earnings are unable to save to pay a monthly electricity bill, all at once. This problem was addressed with the use of prepaid meters (‘Credelec’), increasingly the preferred option even for urban consumers.23 To represent 23 Mozambique adopted pre-paid metering after the success of their use in the domestic market of South Africa (Tewari and Shah, 2003). In 2004, 26,247 connections of new clients in Credelec were connected, against only 5197 new

Table 4 Explanatory variables used in logistic regression. Explanatory variables

Type

Description

Intersect Income Urban Self-owned Concrete walls No of rooms Drinking water PECS

Double Double Binary Binary Binary

To account for a constant effect The households’ daily income $PPP/day-HH Location: Urban (1) or rural (0) Dwelling: Self-owned (1) or rented/borrowed (0) Dwelling: with Concrete walls (1) or otherwise (0)

Cash OnTotal Size Illiterate

Double The number of rooms in the dwelling Binary Drinking Water: from a piped system (1) or not (0) Double The % share of kWh on the total consumption, of the predominant domestic energy source Double The % share of cash on total earnings Double Size of the household, in AEU (adult-equivalentunits) Binary Head of the household: Illiterate (1) or educated (0)

the income and the cultural effects, two variables are used: the recorded daily income per household and the Primary Energy Consumption Share (PECS). The PECS variable represents the effect of households being heavily anchored in a predominant domestic source, and as such likely to resist transition to a new (unknown) source. This variable was calculated by dividing the maximum consumption in any one source (in units of useful kWh/day per household) by the total energy consumption (in units of useful kWh/day per household). We modeled the likelihood that a household consumes electricity as a function of income, PECS and wealth indicators. In the sample of 8377 households that recorded energy consumption, some declare expenses in $PPP or consumption in kWh of electricity, and some declare electricity to be their main lighting or cooking source. These households are assumed to be electricity consumers and are given the value 1 for variable ‘ElectricYes’. Those who neither record any expense or consumption of electricity, nor declare it to be their main source of cooking or lighting are assumed not to be consumers of electricity and are

(footnote continued) clients with conventional meters. In 2005, 43,530 new clients in Credelec were connected, against only 5609 new clients with conventional meters (EDM, 2007).

7244

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Table 5 Coefficients and odds on being a ‘source’ consumer (standard errors in brackets). Explanatory Variables

Charcoal model Coefficient a

Intersect

3 (0.200)  0.0001 (0.000) 2.73 (0.092)  0.75 (0.090) 1.01 (0.074) 0.002 (0.025)  0.17 (0.096)  0.0005 (0.147) 0.43 (0.093) 0.1 (0.017)  0.43 (0.074)

Income Urban Self  owned Concrete walls No of rooms Drinking water PECS CashOnTotal Size Illiterate

Kerosene model Odds 0.05 0.9999 15.29 0.47 2.75 1.002 0.85 0.9995 1.54 1.11 0.65

Electricity model

Coefficient a

Odds

Coefficient a

 0.47 (0.145)  0.0028 (0.001) 1.08 (0.058) 0.03 (0.080)  0.38 (0.070) 0.08 (0.019)  1.77 (0.098) 0.22 (0.104) 0.19 (0.065)  0.02 (0.013) 0.13 (0.071)

0.63

 5.56 (0.350) 0.0075 (0.001) 2.71 (0.231) 0.004 (0.129) 1.47 (0.104) 0.26 (0.033) 2.12 (0.112)  0.42 (0.218)  0.08 (0.148) 0.09 (0.023)  1.14 (0.100)

0.9972 2.93 1.03 0.68 1.08 0.17 1.24 1.2 0.98 1.14

Odds 0.004 1.0076 14.99 1.004 4.34 1.29 8.29 0.66 0.93 1.1 0.32

The estimated odds (Table 5) indicate that

given the value 0 for variable ‘ElectricYes’. Variables ‘KeroseneYes’ and ‘CharcoalYes’ are similarly constructed. The logistic regression was ran for the response variables ‘ElectricYes’, ‘KeroseneYes’ and ‘CharcoalYes’ and the explanatory variables were listed in Table 4, through function glmfit of MATLAB in ‘binomial mode’24, in two arrangements named ‘first’ and ‘second’ runs, respectively.

 Income levels have a small effect on the choice of energy

4.2. The first run of a logistic regression



The same procedure is applied separately to the variables ‘KeroseneYes’ and ‘CharcoalYes’ to estimate the maximum likelihood of a household adopting, respectively, kerosene and charcoal as an energy source.

source, being slightly positive for electricity (1.008) consumption and slightly negative for charcoal (0.999) and kerosene (0.997). This result is not surprising as the energy consumption in the sampled households is small and mostly at its lower limit: even poor households need some energy for cooking and lighting.25 Urban households are 15 times more likely to adopt electricity and charcoal and 2.9 more likely to adopt kerosene as domestic sources than rural households. This confirms that urbanization is an important drive in the transition to more efficient, higher-grade energy sources. Interestingly, households living in self-owned dwellings are 0.47 times less likely to adopt charcoal but as likely to adopt kerosene and electricity as domestic sources than those living in rented or borrowed dwellings. The majority of self-owning-dwellers (7458 households) are rural (57%), while only 43% are urban. An analysis of the subsamples of only rural and only urban households indicates that rural households are 1.60 times more likely to adopt electricity when they have self-owned dwellings, while in urban settings households are 0.98 times less likely to adopt electricity if they live in self-owned dwellings. This slight difference may be a result of the self-owned dwellers constituting the wealthier in rural areas, but being the poorest (living in suburbs) in urban areas. In addition, the proximity to firewood sources in rural settings may be the discouraging factor for the adoption of charcoal as a domestic source. Households living in dwellings with concrete walls are 4.3 times more likely to adopt electricity, 2.8 times more likely to adopt charcoal and 0.7 times less likely to adopt kerosene as domestic sources. Assuming that having concrete walls in the dwelling is an expression of wealth, then wealthy households will favor electricity and reject kerosene (out of 2000 recorded

24 The functions used are part of the standard library of MATLAB, http://www. mathworks.com/

25 The OLS-regression of Income on all five Wealth measures shows a R2 of only 0.07. Collinearity between Income and Wealth can be ruled out.

Eq. (1) gives the logistic function for the odds of a household being an electric consumer: OddsElectric Consumer ¼

pElectric Yes ¼ eP 1pElectric Yes



where

P ¼ a0 þ aI þ a3 V3 þ a4 V4 þ a5 V5 þ a6 V6 þ a7 V7 þ    þ aE þ aC þ aS þ aH

ð1Þ

where I is ‘Income’, V3 defines the ‘Urban Status’, V4 equals 1 when the ‘Dwelling is SelfOwned’, V5 indicates the presence of ‘Concrete Walls’, V6 is the ‘Number of Rooms’ in the dwelling, V7 indicates if ‘Drinking Water’ is from a piped source, E stands for ‘PECS’, C stands for ‘CashOnTotal’, S stands for the household ‘Size’ in AEU and H stands for ‘Illiterate’. The probability of the household adopting electricity as a domestic source is therefore: pElectric Yes ¼

1 1 þeP

ð2Þ



M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249











households living in dwellings with concrete walls, 87% are urban). In rural settings, concrete walls are rarer. Consequently, having concrete walls in rural areas increases the odds of adopting electricity as a domestic source by 21 times, while in urban areas only by 3.9 times. The number of rooms does not seem to have an effect in determining the likelihood of adopting kerosene or charcoal as domestic sources. However, for electricity the odds are increased by 1.3 times per each new room in the dwelling, i.e. larger houses (wealthier) will be more likely to adopt electricity as a domestic source than smaller houses. Households with piped drinking water are 8.3 times more likely to consume electricity. However, these households are, respectively, 0.85 and 0.17 times less likely to consume charcoal and kerosene as domestic sources. Urban households (98.6% of 794 households that have piped drinking water) have their odds increased only by 8.1 times while rural households have their odds increased by 13.4 times. Finally, each percentage point of PECS will decrease the consumption of electricity by a factor of 0.66 and only lightly affects the consumption of kerosene (1.24 times more likely) and charcoal (0.99 times less likely). This result indicates that households will transition to electricity more easily if starting at a lower PECS, i.e. if there are more than one equally predominant energy sources in the energy mix. On contrary, households with one predominant source will more easily transition to kerosene than those with a more even mix. Rural households have a larger decrease in the odds of consuming electricity, by a factor of 0.57, while urban households are within the national average (0.66). Cash earners are more likely to increase the consumption of charcoal and of kerosene, respectively, by 1.5 and 1.2 times per each percent of earnings in cash, while their likelihood of being electricity consumers is reduced by a factor of 0.93. This effect is particularly accentuated for rural households, where each percent of cash earnings reduces the likelihood of being an electricity consumer by a factor of 0.59. Bigger households will have a higher likelihood of consuming charcoal and electricity (1.1 times) and a lower likelihood of consuming kerosene (reduced by a factor of 0.98). On the contrary, illiterate household heads will increase the chance for the household being a kerosene consumer by 1.14 times but reduce its likelihood of being a charcoal and an electricity consumer, by factors of 0.65 and 0.32, respectively.

These results unmistakably indicate that urbanization and infrastructure (drinking water from piped sources) increase the odds of adopting electricity as a domestic source, i.e. infrastructure programs and urbanization favor the adoption higher-grade sources. The results also show that even in urban areas, the probability of being an electricity consumer drastically reduces to 50% or less if the dwelling does not have concrete walls and piped water. Comparing probabilities of being a user of a particular source for the cases of being wealthy (the dwelling has concrete walls and piped water) with an income of 2 $PPP/day, a self-owned dwelling with three rooms and an PECS varying between 5% and 100%, we obtain the following outputs at less than 1% standard deviation:

 For wealthy households: 57% probability of being a charcoal



user in an urban setting and 8% in a rural setting. For nonwealthy households: 36% probability of being a charcoal user in an urban setting and 4% in a rural setting. Conclusion: charcoal is an urban source, preferred by wealthier households. For wealthy households: 25% probability of being a kerosene user in an urban setting and 10% in a rural setting. For



7245

non-wealthy households: 74% probability of being a kerosene user in an urban setting and 49% in a rural setting. Conclusion: kerosene is an urban source, preferred by the poorer. For wealthy households: 81% probability of being a electricity user in an urban setting (3% standard deviation) and 22% in a rural setting (5% standard deviation). For non-wealthy households: 11% probability of being an electricity user in an urban setting and 1% in a rural setting. Conclusion: electricity is an urban source, preferred by the wealthier.

Charcoal is confirmed to be above kerosene in the energy ladder, in terms of price and access, although kerosene is of higher efficiency and of more complex production process. Electricity, shown to be cheaper in useful energy units, is still preferred by the wealthier who can afford the acquisition of electric appliances. This study indicates clearly that the wealth of a household is a determining factor in the choice of domestic source. 4.3. The second run of a logistic regression In the above result the higher the PECS level, the lower the odds of being an electricity consumer, which would point to the incentive to a better domestic energy mix as a means to promote the adoption of electricity by the poorer families. To clarify this effect of the energy mix, the logistic regression was run to estimate the likelihood of being an electricity consumer, including the ‘CharcoalYes’ and ‘KeroseneYes’ variables to the above explanatory variables, namely: pElectric Yes ¼ eF 1pElectric Yes where F ¼ a0 þ aC þ aK þ aI þ    þ a3 V3 þ a4 V4 þ a5 V5 þ a6 V6 þ a7 V7 þ aE þ    þ aC þ aS þ aH

ð3Þ

The results shown in Table 6 indicate that being a charcoal consumer reduces the odds of using electricity to 0.56 and being a kerosene user reduces the odds of using electricity to almost zero (0.02). A wealthy 100% charcoal user, with 2 $PPP/day income in an urban environment has 96% probability of becoming an electricity user, but a 100% kerosene user in the same settings will have only 40% of probability of becoming an electricity user. Moving to a rural setting, the charcoal wealthy user will only have the probability of 38% of transitioning to electricity, while the wealthy kerosene user will have only a 2% probability. In conclusion, the consumption of kerosene is not favorable to the adoption of electricity, even for wealthy households. There is

Table 6 Logistic regression: coefficients and odds on being an electric consumer. Variables

Electricity

Odds

Standard error

Intersect Consumer of Charcoal? Consumer of Kerosene? Income Urban Self-owned Concrete walls No of rooms Drinking water PECS CashOnTotal Size Illiterate

 4.4304  0.5711  4.1044 0.0056 3.6513 0.0844 1.5796 0.2891 2.1499  0.5711 0.1465 0.0904  1.3296

0.0119 0.5649 0.0165 1.0056 38.5258 1.0881 4.8529 1.3352 8.584 0.5649 1.1577 1.0946 0.2646

0.4111 0.1462 0.16 0.0012 0.2474 0.1767 0.1418 0.0425 0.166 0.2903 0.1881 0.0308 0.1397

7246

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

evidence that budget constraints may force poor communities to consume more in less efficient lighting energy such as kerosene (van der Plas, 1988), even though electricity is the preferred lighting source. This study shows that kerosene is a source competing with electricity in urban and rural households: both sources are mostly used for lighting26 and are in the same priceof-useful-energy-units range (13 and 12 cents $PPP/kWh, respectively, see Fig. 6A). However, kerosene does not require any appliance (investment or wealth ownership) to be used as a lighting source. The acquisition of wealth, more specifically the ability to acquire electric appliances, is necessary to adopt electricity as a domestic source. Both simulations indicate that even urban households will be significantly less likely to adopt electricity, in the absence of wealth.

5. Conclusions This work analyzed the energy consumption patterns in Mozambique, from a sample of 8377 energy-consuming households surveyed during 2002/03 (INE, 2007). Results indicate that urban high-income households are the consumers of electricity, while poor rural households rely mostly only on firewood. In other words, the energy ladder concept, associating high incomes to high-grade sources (electricity) and low incomes with lowgrade sources, (biomass) is applicable. However, the data also show that expenditures per unit of useful-energy are higher for lower-grade sources than for higher-grade sources. Records on energy unit values indicate that electricity, kerosene and charcoal are cheaper by this order than firewood, i.e. the energy ladder is inverted on price-per-useful-units. There is a contradiction in the relation between the household income level, the source’s price and the source’s rank in the traditional energy ladder: high-grade sources, consumed by highincome households are cheaper on price-per-useful-energy-units than low-grade sources (the main choice of low-income households). In other words, those who earn less are paying more per unit of useful-energy. This suggests that lowering prices is not necessarily effective in widening the domestic access to electricity. Other factors may also influence the choice of the domestic source. A likelihood analysis (logistic regression) was made to identify them, with the following results: (a) Income is not a strong determining factor in the adoption of electricity, kerosene or charcoal as domestic sources: each dollar of new income increases the likelihood of being an electricity consumer only by 0.76% (0.56% after discounting the presence of charcoal and kerosene in the domestic energy mix). Urban households have a slightly better rate (0.99%) than rural (0.08%). This result is aligned with the previous findings that low-income households are paying more per unit of useful energy than the high-income ones. So, income levels are not the restraining factor in the adoption of electricity as a domestic source. (b) The ownership of wealth favors the consumption of charcoal and electricity and reduces the probability of being a kerosene user. Wealthy households are those that have income surplus in their budgets and can consequently invest (or mobilize credit to invest) in better dwellings and a piped water 26 In Tanzania and in South Africa, kerosene is the main cooking energy in urban areas. However, in Mozambique, the main cooking source, in urban areas, is still charcoal (Hosier and Kipondya, 1993; Davis, 1998; Brouwer and Falcao, 2004). The preferred cooking source in most of the southern Africa is charcoal even when electricity is available (Madubansi and Shackleton, 2007).

connection. Households are more likely to be consumers of electricity by 4+ times when they live in a dwelling with concrete walls, and 8+ times when they access drinking water from a piped system. Even if households access electricity at low prices, they will not adopt it if they don’t own wealth, because households need to first invest in an electrical connection and electrical appliances to consume electricity. (c) Households with high PECS (commonly consume firewood) have a reduced probability of adopting electricity as a domestic source. However, the use of kerosene for lighting deters the adoption of electricity in the domestic setting, even if it reduces the household PECS. Charcoal and electricity users are mostly urban. The national transmission and distribution grid is not very dense in Mozambique, resulting in a direct link between electricity access and urbanization. EDM has made a massive electrification effort in the period 2005–2009, and the electricity access jumped to 14.4% by the end of 2009. The cost of electrification is estimated at a rate of 300 USD per 100 m of low-voltage lines and higher for mediumand higher voltages. This cost is recovered in the cross subsidies incorporated in the electricity tariff, so that all consumers of electricity in the country pay for electrification. Still, domestic consumers must pay for the electrical connection which costs about 100 USD. Starting from 2010, the connection fee is subsidized by 75% hoping to increase the number of connections per line extension and the optimization of electrification investments. Charcoal replaces firewood (for cooking) in urban settings and is often associated with electricity use (mostly for lighting). Therefore, a policy that facilitates the investment in electric appliances will reduce the burden of energy consumption in the budget of poor households (through lower prices) and improve their life quality by allowing them to benefit from a more diverse end-uses and cleaner and safer source. Furthermore, environmental gains can be achieved when firewood is replaced with charcoal, or kerosene is replaced with hydro-generated electricity. The findings of this study show that electricity pricing is not supporting the adoption of electricity as a domestic source because (1) the cheap electricity rate instituted to support the low-income consumers (the social tariff) is not benefiting the majority of domestic consumers, probably due to an unviable eligibility process and (2) although electricity is price competitive with biomass fuels and kerosene, it cannot be consumed without capital expenditure in electrical appliances, which only the wealthy can afford. Consequently, electricity intensification programs should review the conditions of access to cheap electricity (use of the social tariff) and implement a credit system to support consumers of electricity in acquiring electrical devices (facilitate wealth accumulation).

6. Policy recommendations A strategic objective of the energy sector in Mozambique 201427 is to increase electricity access from the recorded 14% of the total population in 2009 to 23% in 2014. By the end of 2014, all the 128 district capitals should be electrified (in 2009 only 94 district capitals were connected to the national transmission grid). Energy sector officials recognize the importance of electricity as a high quality and reliable energy source. Initiatives to install off-grid solar panels and develop off-grid hydroelectric 27 ‘‘Plano Quinquenal do Governo’’ (http://www.me.gov.mz/prt/downloads/ box1/PQGSectordeEnergia2010_2014.pdf).

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

schemes are ongoing, but their contribution to increase in the electricity access (by Mozambican families) is not very significant. Overall, the pace of electrification has moved slower than expected. Subjacent to this slow pace of electrification, is the notion that electricity is desirable as a domestic source but not as a basic need for a decent, safe and productive life. If this is the case, access to electricity is directly linked to the amount of investment funds service providers can mobilize for their electrification projects. No quantification is made for the cost of not providing electricity access to everyone: costs of deforestation around urban areas, cooking fires’ pollution affecting the health of city dwellers and the cost of having limited resources for self-employment and income generation with a consequent cost of unresolved poverty, among others. Electricity is considered an expensive source and as such only appropriate for those with higher income. The introduction of a social tariff is not really contributing to alleviating poverty, because the poor cannot benefit from it. Our results suggest that poor households in Mozambique are locked by a classic poverty trap. To realize the economic and social benefits of electricity, households must purchase electrical appliances. To purchase electricity powered appliances, households must save sufficient monies. Households are less able to save monies for electrical appliances if they consume lower grade energy sources. So, although electricity is desirable, clean and price-competitive, it will remain a ‘high-income’ source unless policies focus on increasing access, not just by extending the electric lines, but also by facilitating the energy transition that requires initial credit for electric appliances. If we recognize that electricity access is a basic need rather than just a desirable asset/resource, a more forward and efficient policy is necessary. The high-income families need electricity to live, produce and be happy, but so do the low-income families. As a society, we have recognized that sustenance, education and health are basic human rights, but we have failed to acknowledge the importance of energy in people’s lives, regardless of their income status. The current research confirms the inequality in access and price for energy sources, which result in inequalities of opportunities for a healthy and productive life, thus violating the Constitution Law of Mozambique. The Universal Declaration of Human Rights, from 10th December 1948, declares that ‘‘all human beings are (y) equal in dignity and rightsy’’ (Art. 1), all have a ‘‘right to life’’ (Art. 3), all have a ‘‘right of equal access to public servicey’’ (Art. 21) and all have ‘‘the right to worky’’ (Art. 23). The effective access to electricity as domestic source responds to these rights by creating the opportunity for individuals to make the most of their potential and to achieve an adequate standard of living, for themselves and their families. In this way the effective access to electricity as domestic source is a human right on its own. Some initiatives towards declaring the access to electricity, a basic human right, have been initiated28,29; however, it is far from being realized. A recent report by the U.N. Secretary-General’s Advisory Group on Energy and Climate Change30 recognizes that poverty can only be fought through an increased access to modern energy services, and set, as a key objective for 2030, the universal basic access to modern energy systems. The revision of the Electricity Act (Law 21/97 of 1st October), currently undergoing, should recognize access to electricity as a ‘‘basic human need’’ and declare the Universal Access to Electricity as an objective for the sector’s

28 In 2000, the NGO ‘‘Droit a l’energy, SOS Futur’’ was created to move towards the recognition of the right to energy for all (http://www.droitalenergie.org/). 29 ‘‘The Nepalese government has issued a notification that grants the right to energy to all the citizens’’, 10 May 2010 (http://southasia.oneworld.net/today sheadlines/nepal-first-to-exercise-right-to-energy). 30 This report calls the universal access to modern energy a ‘‘basic human need’’, http://ipsnews.net/news.asp?idnews=51239.

7247

development. The implementation of such a policy would be difficult (South Africa’s plan for universal access to electricity by 2012/13 is fraught with uncertainties31). Still establishing the goal of Universal Access to Electricity in the Law could trigger a faster pace on electrification programs and the provision of effective electricity services to everyone. The following changes in the Mozambican Energy Policy are also recommended: (1) Declare that the first 100 kWh consumed are charged by the social rate, regardless of who, when and where, thus ensuring that all consumers will be supported in the satisfaction of their basic need for domestic electricity; (2) Review the limit of 100 kWh per month as the consumption limit that satisfies the basic need for domestic electricity, for an average Mozambican household; (3) Require that electrification programs couple the line extensions with credits for domestic connections, and possibly basic electric appliances and lighting systems. The viability of domestic electricity should be calculated based on units of useful-energy, taking into consideration the social cost of using wood fuels for lighting and cooking, and of not using modern sources in rural or family enterprises. In this way, the competitiveness of electricity, when compared with other sources, can be demonstrated and the argument pro-electrification as a poverty alleviation strategy can be made.

Appendix. The household survey—transformations of the data set The National Institute of Statistics of Mozambique conducted the household survey of 2002/03 (INE, 2007), by selecting randomly 9–12 households per strata, corresponding to ‘Localidades’ (villages, communities) that are nested in administrative divisions. In total, 8700 households were successfully surveyed. In addition to energy use and expenditure data, the survey collected information on household demographic characteristics such as family size, gender, employment, education and dwelling characteristics. The survey also recorded the sources and levels of income (earnings in cash, in species and government transfers), household expenses (daily, monthly, yearly and in species) and the ownership of a variety of assets (see final report for more detail on survey design and execution; Zacarias et al., 2004). The survey data were arranged in a matrix 8700  50, from which some columns are described in Table 7. These variables were studied using Arc Map and the shape file for territorial division obtained from the ‘Centro Nacional de Cartografia e Teledetecc- a~ o’ of Mozambique (Teledetecc- a~ o, 2008). This study uses the data on income levels and consumption of energy sources to analyze trends and correlations concerning domestic energy behavior. The survey data are cross-sectional and spatially organized, with households classified as urban or rural. The variables representing total revenue32 (income) in MTn/ day-HH33 and total expenses in MTn/day-HH for the 8700 households interviewed fit lognormal distributions almost perfectly. Altogether, about 450 product codes were registered in 31 http://www.gsb.uct.ac.za/files/UncertaintieswithinSouthAfricasgoalofuni versalaccesstoelectricity.pdf. 32 In the text, revenue, income and earnings will be alternatively used to designate the total amount, in cash or equivalent in species that households earn per day to support their livelihood. 33 MTn is the Mozambican currency; day-HH is ‘per day per household’. The value of expenditures and income was converted into $PPP (DollarPower-Purchasing-Parity) of 5669.1 MTn/$PPP, corresponding to the year 2003 (United-Nations, 2007).

7248

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Table 7 Some data used for the analysis of the households.

Table 9 Conversion factors for consumption of energy sources.

Column

Field name/description

Type

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 30 34 35 36 37 38 39 40 41 42 43 44 45 46 47

Province Cluster Is the household located in an urban environment? Is the dwelling self-owned? Are the walls made of concrete or clay blocks? Number of rooms in dwelling Is the drinking water piped? Main cooking source: firewood, biomass, dung and others Main cooking source: charcoal Main cooking source: kerosene Main cooking source: LPG Main cooking source: electricity Main lighting source: firewood and other Main lighting source: candles Main lighting source: batteries Main lighting source: kerosene Main lighting source: LPG Main lighting source: electricity, solar/diesel generator Household total revenue (MTn/day) Household Expenses in firewood (MTn/day) Household Expenses in charcoal (MTn/day) Household Expenses in candles (MTn/day) Household Expenses in kerosene (MTn/day) Household Expenses in LPG (MTn/day) Household Expenses in electricity (MTn/day) Household consumption in firewood (KWH/day) Household consumption in charcoal (KWH/day) Household consumption in candles (KWH/day) Household consumption in kerosene (KWH/day) Household consumption in LPG (KWH/day) Household consumption in electricity (KWH/day) Household total expenses in energy (MTn/day) Household total consumption in energy (KWH/day)

Double Double Binary Binary Binary Double Binary Binary Binary Binary Binary Binary Binary Binary Binary Binary Binary Binary Double Double Double Double Double Double Double Double Double Double Double Double Double Double Double

Table 8 Domestic energy in consumption and expenditures. Class Name ‘E1’ ‘E2’ ‘E3’ ‘E4’ ‘E5’ ‘E6’ ‘E7’

‘Energy other’ ‘Energy ‘Energy ‘Energy ‘Energy ’Energy ‘Energy

Description sources—firewood and sources—charcoal’ sources—candles’ sources—batteries’ sources—kerosene’ sources—LPG’ sources—electricity’

Energy other Energy Energy Energy Energy Energy Energy

sources—firewood and sources—charcoal sources—candles sources—batteries sources—kerosene sources—LPG sources—electricity

expenses, generating more than 400,000 records. Revenues were registered in more than 25,000 records. Monthly valuations (of expenses or earnings) were converted to daily values by multiplying with 12 and dividing by 365. Yearly valuations were converted to daily values by division of 365. Given the variety of products registered in the household expenses in energy, these were reclassified into seven types of energy sources, as described in Table 8. Revenues were converted into currency units per day per household and summed to obtain a total value of daily earnings per household. For energy source expenditures, energy ‘prices’ (or unit values) were calculated, per household, per day, per kWh; this was done by dividing the recorded expenditure in energy by the standard quantity recorded, and then applying conversion factors detailed in Table 9.   Unit_valuei MTn=day=kWh-eq ¼ Recorded expenditurei ½MTn=day Recorded consumptioni ½units  Net Conversioni ½kWh-eq=unit

Source’s Unit Conversion kWh/Unit ‘Burn’ efficiency ‘Net’ conversion [kWh-eq./ unit]

E1

E2

E3

E4

E5

E6

Kg 4.07 0.1 0.626

Kg 8.14 0.2 2.505

Un 9.72a 0.1 0.449

kWh – 0.65b 1

L Kg 9.72 12.55 0.3 0.45 4.486 8.689

E7 kWh – 0.65 1

a Assumed the same as kerosene calorific content, at a 0.1 burn efficiency (similar to firewood, an open fire). b Assumed the same as the electrical supply form the grid (in alternate current).

where   Net conversioni to electricity kWh-eq=unit ¼ Calorific capacity½kWh=unit  Efficiency of burn source i Effciency of usage of electricty These conversion factors were used by the World Bank (1987) in the first and more complete energy study ever done for Mozambique, by Hosier and Kipondya (1993) in Tanzania and by Kebede (2006) for Ethiopia. The net conversion unit applies the ‘burn’ efficiencies to reflect the equivalent kWh consumption (from an electric supply) corresponding to each source’s unit of consumption.

References Anozie, A.N., Bakare, A.R., Sonibare, J.A., Oyebisi, T.O., 2007. Evaluation of cooking energy cost, efficiency, impact on air pollution and policy in Nigeria. Energy 32, 1283–1290. Brouwer, R., Falcao, M.P., 2004. Wood fuel consumption in Maputo, Mozambique. Biomass and Bioenergy 27, 233–245. Campbell, B.M., Vermeulen, S.J., Mangono, J.J., Mabugu, R., 2003. The energy transition in action: urban domestic fuel choices in a changing Zimbabwe. Energy Policy 31, 553–562. Cockburn, M., Low, C., 2005. Output-Based Aid in Mozambique. Private Electricity Operator Connects Rural Households. The Global Partnership on Output-Based Aid. World Bank. Davis, M., 1998. Rural household energy consumption: the effects of access to electricity—evidence from South Africa. Energy Policy 26 (3), 207–217. EDM , 2005. Master Plan de Electricidade 2005–2020, EDM. EDM , 2007. Source of data Electricidade de Moc-ambique, Statistical Reports 1985–2006. EIA, 2001. U.S. Household Electricity Data: A/C, Heating, Appliances. Elkan, W., 1988. Alternatives to fuelwood in African Towns. World Development 16 (4), 527–533. Falca~ o, M.P., 1999. Price analysis of fuelwood and charcoal in markets of Maputo City, Chaposa Research Project. Green, J.M., Erskine, S.H., 1999. Solar (photovoltaic) systems, energy use and business activities in Maphephethe, KwaZulu-Natal. Development Southern Africa 16, 2. Gupta, G., Kohlin, G., 2006. Preferences for domestic fuel: analysis with socioeconomic factors and rankings in Kolkata, India. Ecological Economics 57, 107–121. Gupta, S., Ravindranath, N.H., 1997. Financial analysis of cooking energy options for India. Energy Conversion and Management 38 (18), 1869–1876. Hosier, R.H., Dowd, J., 1987. Household fuel choice in Zimbabwe. An empirical test of the energy ladder hypothesis. Resources and Energy 9, 347–361. Hosier, R.H., Kipondya, W., 1993. Urban household energy use in Tanzania : prices, substitutes and poverty. Energy Policy 21 (5 May), 454–473. Howells, M.I., Alfstad, T., Victor, D.G., Goldstein, G., Remme, U., 2005. A model of household energy services in a low-income rural African village. Energy Policy 33, 1833–1851. Hughes-Cromwick, E.L., 1985. Nairobi households and their energy use. An economic analysis of consumption patterns. Energy Economics (October), 265–278. INE, 2007. Source of data Instituto Nacional de Estatı´stica/Comissa~ o Geral de Recenseamento. Jenkins, M.W., Scott, B., 2007. Behavioral indicators of household decision-making and demand for sanitation and potential gains from social marketing in Ghana. Social Science & Medicine 64 (12), 2427–2442. Karekezi, S., Majoro, L., 2002. Improving modern energy services for Africa’s urban poor. Energy Policy 30, 1015–1028.

M. de Fatima S.R. Arthur et al. / Energy Policy 38 (2010) 7235–7249

Kebede, B., 2006. Energy subsidies and costs in urban Ethiopia: the cases of kerosene and electricity. Renewable Energy 31, 2140–2151. Kebede, B., Bekele, A., Kedir, E., 2002. Can the urban poor afford modern energy? The case of Ethiopia. Energy Policy 30, 1029–1045. Larsen, B.M., Nesbakken, R., 2004. Household electricity end-use consumption: results from econometric and engineering models. Energy Economics 26, 179–200. Madubansi, M., Shackleton, C.M., 2007. Changes in fuelwood use and selection following electrification in the Bushbuckridge lowveld, South Africa. Journal of Environmental Management 83, 416–426. Masera, O.R., Saatkamp, B.D., Kammen, D.M., 2000. From linear fuel switching to multiple cooking strategies: a critique and alternative to the energy ladder model. World Development 28 (12), 2083–2103. McKenzie, D.J., 2005. Measuring inequality with asset indicators. Journal of Population Economics 18 (2), 229–260. Morton, D.L., 2002. Reviewing the history of electric power and electrification. Endeavour 26 (2), 60–63. Mozambique, EMO, 2007. Statistical databases. Mozambique, RO, 1977. Boletim da Republica, I Se´rie, no. 74, 20 Suplemento, Conselho de Ministros, Decreto-Lei 38/77 de 27 de Agosto. Mulder, P., 2007. Perspectivas da Energia em Moc-ambique Direcca~ o Nacional de Estudos e Analise de Politicas. MPD, Discussion Papers (Junho, 53p). Mulder, P., Tembe, J., 2006. Rural Electrification in Mozambique. Is it Worth the Investment?. Ministry of Planning and Development Mozambique. Ouedraogo, B., 2006. Household energy preferences for cooking in urban Ouagadougou, Burkina Faso. Energy Policy 34, 3787–3795. Pachauri, S., Mueller, A., Kemmler, A., Spreng, D., 2004. On measuring energy poverty in Indian households. World Development 32 (12), 2083–2104.

7249

Reddy, B.S., 1995. A multilogit model for fuel shifts in the domestic sector. Energy Economics 20 (9), 929–936. Simler, K.R., Mukherjee S., Dava G.L., Datt G., 2004. Rebuilding after war: microlevel determinants of poverty reduction in Mozambique. International Food Policy Research Institute Research Report 132. Tarp, F., C. Arndt, H.T. Jensen, S. Robinson, R. Heltberg, 2002. Facing the development challenge in Mozambique. An economywide perspective. International Food Policy Research Institute Research Report 126. Teledetecc- a~ o, C.N.d.C.e., 2008. Maps of Mozambique /http://www.cenacarta.com/S. Tewari, D.D., Shah, T., 2003. An assessment of South African prepaid electricity experiment, lessons learned, and their policy implications for developing countries. Energy Policy 31 (9), 911–927. Tiwari, P., 2000. Architectural, Demographic, and Economic Causes of Electricity Consumption in Bombay. Housing Development Finance Corporation. Troncoso, K., Castillo, A., Masera, O.R., Merino, L., 2007. Social perceptions about a technological innovation for fuelwood cooking: case study in rural Mexico. Energy Policy 35, 2799–2810. Tyler, S.R., 1996. Household energy use in Asian cities: responding to development success. Atmospheric Environment 30 (5), 809–816. United-Nations, 2007. The Millennium Development Goals Report. van der Plas, R., 1988. Domestic lighting. WB, 1987. Mozambique: issues and options in the energy sector (Report 6128MOZ). Wolde-Rufael, Y., 2006. Electricity consumption and economic growth: a time series experience for 17 African countries. Energy Policy 34 (10), 1106–1114. Zacarias, F., C.S. Chipembe, E. Mazive, E. Triebkorn, P. DuceC. Creva, 2004. Relatorio final do inquerito aos agregados familiares sobre o orcamento familiar, 2002/03.