Download Aggarwal et al Carbon pricing and household welfare Evidence from Uganda draft submission

Carbon pricing and household welfare: Evidence from Uganda
Raavi Aggarwal,1,2 Sinem H. Ayhan,1 Michael Jakob1 and Jan Christoph Steckel1
Abstract
Policymakers frequently voice concerns that carbon pricing could impair economic
development in the short-run, especially in low-income countries such as Uganda. We estimate
a consumer demand system for energy and food items to assess how households’ welfare, and
demand for food and energy, can be expected to react to a carbon price of US$40/ton. The
results show modest and overall progressive welfare effects of carbon pricing, in the range of
0.2 – 1.8%, across the population. We observe declines of 15 – 59% in the demand for
electricity, kerosene and charcoal, with concomitant shifts within food consumption baskets,
due to complementarities with cooking fuels and negative income effects. Heterogeneous
demand responses across expenditure terciles reveal disparities in biomass and food
consumption, as well as nutritional intake, due to carbon pricing. Around half of the Ugandan
population experiences declines of up to 2.4% in food and nutrient consumption with disparate
effects within income groups, revealing important “horizontal” equity considerations.
Complementary social protection policies in conjunction with carbon pricing could potentially
ease adverse effects on economic development in Uganda.
Keywords: Consumer demand system, censored EASI, carbon pricing, Sub-Saharan Africa,
household welfare, distribution
JEL Codes: D12, Q41, Q56, R15
Declarations of interest: None
Funding Source: German Federal Ministry of Education and Research (BMBF) (funding code
011A1807A, DECADE)
1
Mercator Research Institute on Global Commons and Climate Change, Torgauer Str. 19, 10829 Berlin, Germany.
Corresponding Author. Email: [email protected]. Contact No.: +49 30 33 85 537 -217. Postal Address:
MCC Berlin, Torgauer Str. 19, 10829 Berlin, Germany.
2
1
1. Introduction
Carbon pricing (by means of a carbon tax or a cap-and-trade system) is an efficient instrument
to reduce emissions (Somanathan et al. 2014) and avoid future lock-in into carbon-intensive
energy systems (Mattauch et al. 2015). The High-Level Commission on Carbon Prices further
recommends a global carbon price starting at US$40/t CO2 for efficient emissions reduction
(World Bank 2017). While absolute emissions are arguably low, countries in Sub-Sahara
Africa (SSA) carbonise rapidly, with 8 of 10 of the world’s fastest carbonising countries located
in the region (Steckel et al. 2020). Some SSA countries including South Africa, Senegal and
Côte d’Ivoire have already implemented or are currently considering carbon prices (World
Bank 2021a). Uganda’s Intended Nationally Determined Contribution (INDC) to the 2015
Paris Agreement similarly aims to implement mitigation measures that could result in GHG
emissions reduction of up to 22% by 2030 relative to the business-as-usual scenario
(Government of Uganda 2015). It further emphasizes decarbonisation of the energy,
infrastructure and transport sectors (UNFCCC 2016).
Higher prices for fossil fuels resulting from a carbon price however impose greater costs on
households, and in SSA, could lead to unintended welfare outcomes, such as promoting a shift
toward traditional biomass like firewood (Jewell et al. 2018; Greve and Lay 2021). Biomass
use could well be problematic for the climate, e.g. because of increased deforestation resulting
from the unsustainable use of forests (Masera et al. 2015). Increasing use of biomass could also
affect socio-economic development, such as health (through indoor air pollution) (Pratiti et al.
2020), food consumption (due to commodity price hikes) (Fuje 2019) and female labour force
participation (due to time spent to gather firewood) (Köhlin et al. 2011). We investigate the
potential impacts of a carbon price on household welfare in Uganda, examining the link
between energy use and food consumption, and accounting for substitution within households’
consumption baskets. We first estimate price elasticities of demand for several food groups and
energy sources, and then simulate the second-order welfare effects of a US$40/ton price on
carbon emissions.
Previous studies of energy demand in SSA find mixed patterns of substitution among fuels in
response to fossil fuel price increases. Olabisi et al. (2019) estimate a censored quadratic almost
ideal demand system (QUAIDS) for Tanzania, and find charcoal price increases to raise
kerosene demand, with heterogeneous effects across rural and urban areas. On the contrary,
fossil fuel price increases induce substitution toward traditional biomass in Senegal (Yaméogo
2
2015). Similar analyses for South Africa and Ethiopia highlight the role of energy access,
availability of cook-stoves and connections to the electric grid, in determining energy demand
(Okonkwo 2021; Alem and Demeke 2020; and Guta 2012).
The distributional welfare impacts of carbon pricing in advanced countries are typically
regressive (Sterner 2012). Cross – country analyses in low and middle-income countries
(LMICs), however, suggest a high likelihood of progressive welfare outcomes, due to the
relatively low energy expenditure shares of poorer households in LMICs (Ohlendorf et al.
2021; Dorband et al. 2019). Country-specific evidence for Asian and Latin American
economies further confirms progressive distributional effects of energy taxes, but also reveals
large indirect welfare consequences of food price increases (Renner et al. 2018; Bhuvandas
and Gundimeda 2020; and Irfan et al. 2018).
The Almost Ideal Demand (AID) System of Deaton and Muellbauer (1980), its quadratic
version (Banks et al. 1997), and the more recently developed Exact Affine Stone Index (EASI)
demand system (Lewbel and Pendakur 2009), have been extensively applied to models of food
and energy demand in advanced countries (van der Ploeg et al., 2021, Eisner et al., 2021,
Reaños and Wölfing, 2020, and Zhen et al. 2013). In low and middle-income countries, the
QUAIDS demand system has been estimated for patterns of food consumption in Mexico,
Uganda, Malawi and others (Attanasio et al. 2013; Attanasio et al. 2011; Boysen 2016, and
Ecker and Qaim 2011).
We extend the literature by implementing an approximately linear EASI model for Ugandan
households, accounting for censored observations in reported commodity expenditures. We
estimate a system of correlated Tobit equations, which model zero consumption of specific
commodities, in a joint framework via the Generalized Method of Moments (GMM), taking
into account cross-equation error correlations, and imposing parameter restrictions from
consumer theory. We thus build on advancements made by Meyerhoeffer et al. (2005),
Jakubson (1988) and Chamberlain (1980, 1984) in efficient estimation of censored models for
a nonlinear equation system.
Drawing on household-level expenditure data from the Uganda National Household Survey
(UNHS), 2016-17, conducted by the Uganda Bureau of Statistics (UBOS), and commodity
price data from the World Bank’s Living Standards Measurement Study (LSMS) for 2015-16,
we estimate the uncompensated and compensated price elasticities of demand for 12
commodities. These include electricity, kerosene, traditional biomass (charcoal and firewood),
3
and several food groups. We then combine the estimated elasticities with product-level data on
CO2 emissions from the Global Trade Analysis Project (GTAP 10), to examine the first- and
second-order welfare impacts of a carbon pricing policy in Uganda. We investigate the
consequent shifts in energy and food demand, in particular calorie and nutrient intake, across
the income distribution. Lastly, we analyse potential increases in biomass consumption due to
carbon pricing, and evaluate revenue redistribution schemes needed to mitigate trade-offs
between climate policy and economic development.
The results show clear patterns of substitution among fuels, with a 1% increase in electricity
and kerosene prices raising firewood demand by 1% and charcoal demand by 0.5%,
respectively. We find complementarities between food and energy items, with households
raising their consumption of vegetables and meat in response to electricity price hikes, while
reducing cereal consumption, synonymously with increases in biomass demand. A carbon price
of US$40/t CO2 generates modest welfare losses in the range of 0.2 – 1.8% of expenditure.
The consequent energy and food price increases lead to significant disparities in food, calorie
and nutrient consumption across the population. Given the prevalence of cooking in household
patterns of energy use, we identify cooking fuels as additional channels of policy impact. In
conjunction with carbon pricing policies, social welfare programmes such as lump-sum
transfers, could therefore create important pathways for long-term sustainable development in
Uganda and the broader SSA region.
This paper is structured as follows. Section 2 outlines the methodology, and section 3 discusses
the data and presents descriptive statistics. Section 4 outlines the econometric methods and
section 5 presents the results. Section 6 discusses the results and limitations, and section 7
concludes.
2. Methodology
2.1. Exact Affine Stone Index (EASI) demand system
We estimate the approximate EASI implicit Marshallian demand system developed by Lewbel
and Pendakur (2009), which is derived from consumer theory and expresses budget shares in
linear forms. The EASI log of cost function 𝐶(. ) for the exact model for J goods, is expressed
in terms of utility u, a J- vector of log prices p and a J- vector of errors 𝜀:
4
1
𝐶(𝑝, 𝑢, 𝜀) = 𝑢 + 𝑝′ (𝑏0 + 𝑏1 𝑢) + 𝑝′ 𝐴𝑝 + 𝑝′𝜀
2
(1)
where A is a J×J matrix of parameters with elements [𝐴]𝑗𝑘 = 𝑎𝑗𝑘 , while 𝑏1 is a J- vector of
coefficients. Applying Shephard’s Lemma to the cost function, we obtain Hicksian demands,
and via duality, the vector of Marshallian budget shares in implicit form:
𝑤 = 𝑏0 + 𝑏1 𝑦 + 𝐴𝑝 + 𝜀
(2)
where y is interpreted as a measure of implicit utility and the log of real household expenditure.
Engel curves are modelled as approximately linear, while the 𝜀 terms are random utility
parameters reflecting individual preference heterogeneity. Solving for y by equating the cost
function 𝐶(𝑝, 𝑢, 𝜀) to the log of nominal household expenditure x and substituting for 𝑤 yields
the following expression, with 𝑝′ 𝑤 equal to the log Stone price index (Stone 1954):
1
𝑦 = 𝑥 − 𝑝′ 𝑤 + 2 𝑝′𝐴𝑝
(3)
This expression however relies on the matrix of estimable parameters A, and is endogenous
due to its dependence on the budget shares 𝑤. We therefore estimate the approximate model
(Lewbel and Pendakur 2009), with 𝑦̃ = 𝑥 − 𝑝′𝑤
̅, and deflate nominal expenditure x by the
Consumer Price Index, whose commodity weights are derived from national household
expenditure surveys (UBOS 2018).
The following parameter restrictions are imposed to satisfy consumer theory, in particular
Walras’ law, homogeneity of the cost function of degree one in prices and symmetry of the
Slutsky matrix: 1𝐽′ 𝑏0 = 1, 1𝐽′ 𝑏1 = 0, 1𝐽′ 𝐴 = 0, and 𝑎𝑗𝑘 = 𝑎𝑘𝑗 , ∀ 𝑗 ≠ 𝑘 (Lewbel and Pendakur
2009). However, as inequality constraints cannot be imposed, the model does not ensure
negative semi-definiteness of the Slutsky matrix.
The compensated price elasticity of the budget share for good j in response to an increase in
the price of good k, is expressed as follows (Lewbel and Pendakur 2009):
𝑐
𝑒𝑤
=
𝑗 ,𝑝𝑘
𝑎𝑗𝑘
𝑤𝑗
(4)
where 𝑤𝑗 is the observed budget share of good j. To obtain price elasticities of demand, we
note that the cross-price budget share elasticities equal the corresponding price elasticities of
demand.
5
The own-price elasticities are related to the budget share elasticities as follows:3
𝑐
𝑐
𝑒𝑗,𝑘
= 𝑒𝑤
−1
𝑗 ,𝑝𝑘
(5)
𝑐
where 𝑒𝑗,𝑘
is the compensated price elasticity of demand for good j with respect to the price of
good k. The expenditure elasticities of demand are derived as follows:
𝑦̃
𝑒𝑗,𝑦̃ = 𝑏1𝑗 𝑤 + 1
(6)
𝑗
We then derive the uncompensated (Marshallian) price elasticities of demand from the Slutsky
equation in budget share form, as follows:
𝑢
𝑐
𝑒𝑗,𝑘
= 𝑒𝑗,𝑘
− 𝑒𝑗,𝑦̃ 𝑤𝑘
(7)
2.2. Welfare impacts
The first-order (FO) welfare effect of a price change is the additional expenditure incurred by
households for the original consumption bundle to be affordable. In budget share form, the FO
effect is expressed as:
0
𝐹𝑂 = ∑𝐾
𝑗=1 𝑤𝑗
∆𝑝𝑗
(8)
𝑝𝑗0
where 𝑤𝑗0 and 𝑝𝑗0 are the initial budget share and price for good j, respectively, with
expenditures aggregated across the K goods that exhibit price increases. For goods unaffected
by carbon pricing (in this study, charcoal and firewood), the welfare effect is zero.
To account for household demand changes in response to relative price hikes and shifts within
the consumption basket, we analyse the second-order (SO) welfare effects. Because SO
calculations account for substitution effects, they are typically smaller than first-order
calculations, which overestimate welfare losses. Banks et al. (1996) derive second order
approximations to the welfare loss of an indirect tax on a single item in the consumption basket.
Renner et al. (2018) extend this framework to multiple simultaneous price changes through a
second-order Taylor series expansion of the expenditure function. We follow their approach to
3
This can be obtained by differentiating the budget shares 𝑤𝑖 =
𝑝 𝑖 𝑓𝑖
𝑥
, where 𝑓𝑖 is the quantity of good 𝑖 demanded,
with respect to prices. These formulae hold for both uncompensated and compensated demand. The theory of
duality shows that at consumers’ optimal bundles, Hicksian and Marshallian demands (and budget shares) are
equal.
6
compute SO welfare effects, or the compensating variation, which measures the amount of
compensation households require to achieve their initial utility levels at the new set of prices.
As a proportion of household expenditure, the approximate second-order welfare effect is:
𝑆𝑂 ≈ ∑𝐽𝑖=1 𝑤𝑖
∆𝑝𝑖
𝑝𝑖0
1
𝑐
+ 2 ∑𝐽𝑖=1 ∑𝐽𝑗=1(𝑤𝑖 𝑒𝑖,𝑗
∆𝑝𝑖 ∆𝑝𝑗
𝑝𝑖0
𝑝𝑗0
)
(9)
The first term represents the first-order effect, while the second-term measures changes in
consumption due to multiple price changes, captured by the compensated price elasticities of
𝑐
demand (𝑒𝑖,𝑗
) and expenditure shares (𝑤𝑖 ). As the carbon pricing policy leaves the prices of
biomass (charcoal and firewood) unaffected, any cross-price effects involving solid fuels will
result in no additional welfare losses, despite non-zero demand changes.
2.3. Changes in final consumption
2.3.1. Commodity demand
We compute the total change in quantity demanded for each item, in response to a carbon price,
based on the estimated uncompensated price elasticities of demand. Consider a Marshallian
demand function 𝑓𝑖 for commodity i, as a function of the J-vector of prices q,4 and total
household expenditure x:
𝑓𝑖 = 𝑓𝑖 (𝑞1 , 𝑞2 , … 𝑞𝑖 , … , 𝑞𝐽 ; 𝑥)
The proportionate change in quantity demanded is then approximately equal to:5
∆𝑓𝑖
𝑓𝑖
≈ ∑𝐽𝑗=1 𝑒𝑓𝑢𝑖 ,𝑞𝑗
∆𝑞𝑗
𝑞𝑗
(10)
2.3.2. Food and nutrient consumption
We additionally compute the change in total food consumption across the population. We first
recalculate budget shares with respect to households’ total food expenditure, and use these
modified budget shares as weights for the demand changes derived in equation (10) for the k
4
5
We distinguish prices q from the log of prices p, considered earlier.
The derivation is presented in Appendix D.
7
food groups to compute the aggregate change in food consumption. Hence, the proportionate
change in total food demand due to carbon pricing is:
∆𝐹
𝐹
= ∑𝑘𝑖=1
∆𝑓𝑖
𝑓𝑖
𝑤𝑓𝑖
(11)
where 𝑤𝑓𝑖 is the modified expenditure share for food group i, for a total of k food groups.
Despite lack of data on quantities, budget shares can be used to analyse shifts in the
composition of food consumption, while holding total expenditure constant.6
Lastly, we calculate the percentage change in consumption of calories, protein and
micronutrients due to carbon pricing. Following Ecker and Qaim (2011), we analyse changes
in five key micronutrients: Iron, Vitamin C, Riboflavin, Folate and Vitamin B12. Using a food
composition table for Uganda from HarvestPlus,7 we first individually calculate the average
level of nutrients, 𝑛𝑖 for calories (kcal), protein (grams) and micronutrients (grams) in each
food category i per kilogram, using food item-specific consumption shares as weights for the
∆𝑁
broad food category. We then obtain the percentage change in total nutrient intake, ( 𝑁 ) due to
the carbon price, as follows:
∆𝑁
𝑁
=
∑𝑘
𝑖=1 𝑛𝑖 𝑤𝑓
∆𝑓𝑖
𝑖 𝑓𝑖
∑𝑘
𝑖=1 𝑛𝑖 𝑤𝑓
(12)
𝑖
3. Data and Descriptive Statistics
3.1. Expenditure and price data
Household expenditure data are drawn from the Uganda National Household Survey (UNHS),
conducted by the Uganda Bureau of Statistics (UBOS), for the latest year available, 2016-17.
The UNHS is nationally representative, with a sample size of 15,912 households for an
estimated population of 38 million (UBOS 2018). The survey comprises detailed consumption
modules on expenditures incurred for food items and energy sources, with respective weekly
and monthly recall periods. Following Deaton and Zaidi (2002), we scale expenditures to
annual levels. For each item, we aggregate the expenditures incurred on market purchases and
the expenditure values of goods produced at home, collected from villages/forests or items
6
We further simulate national price increases for each commodity, so that prices do not exhibit differential
increases across regions, due to carbon pricing.
7
The Food Composition Table for Uganda from HarvestPlus is available here.
8
received in-kind. Accounting for missing data on commodity expenditures, we obtain a sample
of 15,682 observations.
Individual food items are categorised into commodity groups based on likely
complementarities between items, in order to satisfy the assumption of multi-stage budgeting
and weak separability between the demand for food and fuel, from the demand for other nondurable and durable items. We model demand for 12 consumption groups including 4 energy
sources – electricity, kerosene, charcoal and firewood, and 8 food categories – cereals; fruits;
vegetables; meat & fish; milk & eggs; cheese, oils & fats; alcohol & tobacco, and other food
& drink.
The demand system excludes other non-durable and durable goods due to lack of commoditylevel price data. We further exclude transport due to inadequate expenditure and price data.8
However, the non-durable items considered account for 65 per cent of households’ mean total
spending. To curtail the effects of outliers, we omit observations whose budget shares for
individual items exceed the corresponding 99th percentile values, and budget shares that are
censored at 1. We further trim the top and bottom 1% of observations of the real expenditure
distribution, resulting in a final sample of 13,582 households, which covers 85% of the
Ugandan population.
Data on commodity prices are drawn from two sources. First, energy prices for kerosene,
biomass and food items are obtained from the Uganda National Panel Survey, conducted by
the World Bank’s Living Standards Measurement Study (LSMS), for the closest year available,
2015-16.9 We take simple averages of item-specific market prices reported by households and
construct average annual prices for each of the 12 categories of the demand system, at the
district – rural/urban level, to capture spatial heterogeneity, and mitigate seasonal influences in
biomass availability. This level of spatial aggregation controls for potential endogeneity of
prices due to household production, and mitigates concerns of quality differentiation in unit
prices (Capéau and Dercon 2006). For districts with missing price data in the UNPS, we impute
the corresponding national, annual average price for each commodity. We then inflate all
8
Only 7% of sample households report spending on motor fuels.
We find insufficient commodity price data in the Uganda National Household Survey and thus rely on the UNPS
for prices.
9
9
commodity prices to 2016-17 levels using the Consumer Price Index for Uganda, to match the
household expenditure data.10
Second, we obtain national electricity tariffs for Uganda from the Electricity Regulatory
Authority (ERA).11 These are available at a quarterly frequency, with the pricing structure
involving a two-block tariff, around a threshold consumption level of 15 kWh units per month.
Data on electricity consumption from the UNPS for 2015-16 suggest that around 90% of
households with positive reported expenditures for electricity consume more than 15 units per
month. Therefore, we apply only the higher marginal tariff as a uniform price for electricity for
all households.
3.2. CO2 Emissions Data
To analyse the distributional welfare impacts of a national carbon price, we combine the
household-level expenditure survey with the Global Trade Analysis Project (GTAP 10)
database for 2014, which provides carbon dioxide emissions levels for 65 sectors, based on an
environmentally-extended multiregional input-output table (Aguiar et al. 2019). Since the
GTAP database excludes emissions from biomass, there are no resulting price changes for
charcoal or firewood (Dorband et al. 2019).
Based on the conversion table provided for Uganda, the 140 consumption items from the
UNHS are mapped to 16 corresponding sectors in GTAP 10 (documented in Appendix Table
A7). The CO2 intensity for a specific item can either be attributed to direct emissions due to
energy consumption (e.g. electricity and kerosene), or indirect emissions due to food
consumption (Renner 2018). We compute the weighted carbon intensities for the 12 categories
of the demand system based on the mean consumption shares of individual items within each
category, in the UNHS. This yields a household-level dataset with expenditures and budget
shares for various commodity groups, and corresponding weighted carbon intensities for each
category. We deflate expenditures and prices to 2014 values using the consumer price index
and subsequently calculate price changes for all commodities due to the carbon price.
10
11
Additional details of the commodity price data are provided in Appendix A.
The Schedules of End-User Tariffs are available here.
10
3.3. Descriptive Statistics
Summary statistics for expenditure shares, annual household expenditure on food and energy,
and prices used in estimation of the EASI demand system are presented in Table 1. The mean
budget shares (Column 1) suggest that households’ consumption baskets for food and fuel are
largely composed of vegetable consumption, followed by cereals, meat & fish, and other foods,
while energy comprises less than 5% of expenditure. For the sub-sample of electricity users
(Column 3), the average budget share is 7.3%, compared to mean budget shares of 0.6% for
kerosene, 2.3% for charcoal and 3.3% for firewood, for the respective users. A substantial
proportion of households further report zero expenditures for various items. For instance, only
around 27% of households report positive expenditures on electricity (Column 4), with 73% of
the sample reporting zero spending on electricity, which is largely due to lack of widespread
access to the electric grid (IEA et al. 2020).12
Table 1. Expenditure shares for the full sample (% of expenditure on food & fuel)
Item
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
Annual Expenditure on
Food & Energy (UgX)
N
(1)
(2)
Mean (%)
Std. Dev.
1.95
0.19
0.62
1.98
17.7
2.96
42.51
10.86
2.74
2.09
2.98
13.41
4.516
0.335
1.236
2.393
14.708
4.589
21.974
11.227
4.551
2.09
6.067
20.7
3,338,494
2,012,756
13,582
13,582
(3)
Mean if exp. >
0 (%)
7.32
0.57
2.29
3.30
20.46
6.11
44.56
15.96
6.70
2.92
8.89
13.46
(4)
Percent with positive
exp.
26.65
33.43
27.28
59.86
86.51
48.48
95.38
68.05
40.95
71.69
33.48
99.65
-
13,582
Note: The sample sizes for items with positive reported expenditures in Column 3 are
distinct for each item and hence not reported.
12
The reporting of zero expenditures partly reflects lack of access to modern energy sources, but also differential
fuel and food consumption patterns across the population. Additionally, the 7-day recall period used in the survey
may lead to exclusion of certain food groups not consumed in the week preceding the survey (Ecker and Qaim
2011).
11
Households’ budget shares for different energy items by expenditure quintile are displayed in
Figure 1.13 Patterns of energy use show a gradual decline in the share of biomass from 3.4% to
2%, and that of kerosene use from 0.3% to 0.08%, across expenditure quintiles. The share of
electricity in households’ energy mix however rises sharply from 0.3% to 4.8%, along the
expenditure distribution. While the Engel curves are nonlinear for several items (Appendix
Figure A1), the EASI model specification we employ is parsimonious and linearly
approximates Engel curves for model tractability, and is a limitation of the analysis.
Figure 1. Energy expenditure shares by expenditure quintile
Note: The X-axis represents expenditure quintiles.
Biomass comprises charcoal and firewood.
Summary measures for commodity prices are presented in Table 2. We observe significant
price variation for most items, in particular kerosene, charcoal, meat & fish and cheese, oils &
fats, which ensures reliable econometric identification.
The mean expenditure shares and standard deviations for all items by expenditure tercile – rural/urban groups,
are presented in Appendix Table A1.
13
12
Table 2. Summary statistics for item-specific prices
Item (Unit)
Electricity (kWh)
Kerosene (L)
Charcoal (Kg)
Firewood (Bundle)
Cereals (Kg)
Fruits (Kg)
Vegetables (Kg)
Meat & Fish (Kg)
Milk & Eggs (Kg)
Cheese, Oils & Fats (Kg)
Alcohol & Tobacco (L)
Other Food & Drink (Kg/L)
Mean
657.77
2700.67
466.54
77.24
1865.68
955.52
1292.93
5921.61
1163.79
2151.41
2886.99
1590.09
Std. Dev.
33.63
374.66
211.10
21.83
396.93
341.74
415.96
1436.84
320.83
1171.74
653.74
235.21
Min.
623.60
105.68
147.95
35.23
528.40
42.27
211.36
422.72
369.88
132.10
158.52
1044.12
Max.
696.90
4227.19
5529.34
399.23
3593.11
4720.36
2792.96
13209.96
2656.08
13611.54
5748.21
3579.90
Note: Prices are in Ugandan shillings (UgX) per unit. Litre is abbreviated as ‘L’.
4. Econometric Methods
We estimate the EASI model as a system of 12 correlated Tobit equations efficiently via the
Generalized Method of Moments (GMM) (Hansen 1982; Wooldridge, 2010). A significant
proportion of households in the sample report zero consumption expenditure for individual
items, such as electricity, due to factors related to energy access, cost of electricity grid
connections and availability of cook-stoves. This motivates the use of a corner solution
response model.
The latent budget share equation, based on Eqn. (2), for good j and household i is:
𝑤𝑖𝑗∗ = 𝑏0𝑗 + 𝑏1𝑗 𝑦̃𝑖 + ∑𝐽𝑘=1 𝑎𝑗𝑘 𝑝𝑘𝑑 + 𝜀̃𝑖𝑗
(13)
where 𝜀̃𝑖𝑗 ~𝑁(0, 𝜎𝜀2 ), with the corresponding Tobit Type I model:
𝑤𝑖𝑗 = max(0, 𝑤𝑖𝑗∗ )
where 𝑤𝑖𝑗 is the observed budget share, 𝑦̃𝑖 is the log of real expenditure (nominal expenditure
x, deflated by the Consumer Price Index), 𝑝𝑘𝑑 are log prices for commodity k and district d,
and 𝜀̃𝑖𝑗 is a normally distributed error term. In compact form, let 𝑤 denote the vector of
observed budget shares (𝑤𝑖𝑗 ), X the vector of covariates (𝑦̃, 𝑝𝑘𝑑 ), 𝛽 the vector of coefficients
(𝑏0𝑗 , 𝑏1𝑗 , 𝑎𝑗𝑘 ), and let 𝑋𝛽 be the expected value of the latent budget share.
13
Assuming weak exogeneity between the covariates and the error terms, i.e. 𝐸(𝜀̃𝑖𝑗 |𝑋𝑖 ) = 0, we
apply the pooled Tobit estimator, which maximizes the partial likelihood for each equation.
Given potential correlation between error terms across equations, we estimate the equation
system efficiently via GMM, using all covariates as instruments. As each of the covariates is
plausibly exogenous, the number of instruments equals the number of regressors, and the model
is exactly identified.
We define each moment condition as follows:
𝐸[𝑔(𝑤, 𝑋, 𝛽)] = 𝐸[𝑤𝑖𝑗 − 𝐸(𝑤𝑖𝑗 |𝑋)] = 0
The expected value of the observed budget share given covariates X is obtained from the Tobit
model as:
𝑋𝛽
𝑋𝛽
𝜀
𝜀
𝐸(𝑤𝑖𝑗 |𝑋) = Ф ( 𝜎 ) 𝑋𝛽 + 𝜎𝜀 𝜑( 𝜎 )
where Ф(. ) is the standard normal cumulative distribution function and 𝜑(. ) is the
corresponding probability density function (Wooldridge 2010, pp. 672).
Identification of the coefficient vector 𝛽 thus requires a preliminary estimator for 𝜎𝜀 . We
estimate each commodity equation distinctly using pooled Tobit and obtain an estimator 𝜎̂𝜀 .
From the Law of Large Numbers, the Continuous Mapping Theorem and the Slutsky Theorem,
𝜎̂𝜀 consistently (albeit not efficiently) estimates 𝜎𝜀 . We then implement two-step GMM, taking
advantage of cross-equation correlations in the error terms, and impose cross-equation
parameter restrictions from consumer theory, to obtain a consistent and efficient estimator for
the coefficient vector 𝛽, to solve the minimization problem:
min𝑏∈Θ 𝑔̅ (𝑤, 𝑋, 𝑏)′ 𝑊𝑔̅ (𝑤, 𝑋, 𝑏)
where 𝑔̅ are the sample analogues of the population moment conditions and 𝑊 is the weight
matrix.
5. Results
This section presents the estimated elasticities, the welfare impacts of carbon pricing and
demand shifts in households’ consumption baskets due to energy and food price increases. We
assess all effects for a counterfactual carbon price of US$40/t CO2. This price is the lower
14
bound of the range recommended by the High-Level Commission on Carbon Prices to meet
the temperature targets of the 2015 UNFCCC Paris Agreement (World Bank 2017).
5.1. Price elasticities of demand
The compensated and uncompensated price elasticities of demand for the full sample are
presented in Tables 3 and 4, respectively, while Table 5 displays compensated elasticities for
energy items for the rural – urban sub-samples. Each cell (row i, column j) represents the
percentage change in demand for good i due to a 1% increase in the price of good j. The EASI
model coefficients, with z-statistics and standard errors are presented in Appendix Table A2.14
Additional robustness checks for the demand system using an instrumental variable approach
for commodity prices, and an alternative pooled cross-sectional dataset from the Uganda
National Panel Survey (2009-2020), are presented in Appendix Tables A8 and A9.
5.1.1. Energy items
The results show negative own-price compensated elasticities of demand for all items across
the sample, with several items like electricity, kerosene and charcoal, being highly price elastic
in both rural and urban areas. A 1% increase in the price of charcoal lowers its demand by 3.3%
for the average household, and by a considerable 9.6% among rural households. Similarly, a
1% increase in the price of electricity lowers its demand by 12% and 3% in rural and urban
areas, respectively, with all effects statistically significant at the 99% confidence level. The
uncompensated own-price elasticities of demand (Table 4) are negative for all items and
smaller in absolute value than the corresponding compensated elasticities for most items,
except for inferior goods (for e.g. kerosene and firewood15).
The cross-price elasticities of demand reveal important substitution effects between energy
sources. A 1% increase in electricity prices raises kerosene demand by 3.5% in urban areas,
while it remains unaffected in rural areas. Similarly, higher electricity prices raise firewood
14
Expenditure elasticities of demand are presented in Table A3, while the full matrices of compensated price
elasticities for rural and urban areas are presented in Appendix Tables A4 and A5. Elasticities for the full sample
and the rural/urban sub-samples are estimated through nonlinear combinations of the sample-specific estimated
regression coefficients, using the ‘nlcom’ command in statistical software Stata.
15
The estimated elasticities may not accurately reflect substitution behavior for firewood due to potentially
imprecise measurements of collections from forests and wood-fuel received in-kind.
15
demand across rural and urban regions. On the other hand, firewood is complementary to
kerosene use, particularly in rural areas, with the former primarily used for cooking, and the
latter for lighting. Charcoal is however substitutable for kerosene, with a 1% increase in
kerosene prices raising charcoal demand by 0.14% on average and by 0.1% in urban areas.
Charcoal and firewood are net substitutes, with percent increases in charcoal prices increasing
firewood demand by 0.1 – 0.5%, across regions. Uncompensated elasticities largely mirror the
observed patterns of substitution among fuels. The results broadly corroborate the energy
ladder hypothesis, wherein households sequentially “step down the ladder” from modern
energy sources such as electricity and kerosene, to traditional biomass in the event of fuel price
increases.
16
Table 3. Compensated price elasticities of demand (Full sample)
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils &
Fats
Alcohol &
Tobacco
Other Food &
Drink
E
K
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-5.205***
(0.351)
0.112
(0.235)
-1.740***
(0.236)
0.941***
(0.114)
-0.084***
(0.020)
0.202***
(0.058)
0.070***
(0.008)
0.414***
(0.044)
0.098
(0.092)
0.011
(0.023)
-1.596***
(0.125)
0.144***
(0.035)
-0.039***
(0.013)
0.003**
(0.001)
0.008**
(0.004)
0.000
(0.000)
0.006***
(0.002)
-0.006
(0.006)
-0.556***
(0.076)
0.470***
(0.114)
-3.275***
(0.203)
0.099***
(0.027)
-0.017***
(0.005)
0.113***
(0.017)
0.043***
(0.002)
-0.008
(0.007)
0.17***
(0.028)
0.954***
(0.115)
-0.400***
(0.135)
0.315***
(0.085)
-1.325***
(0.066)
-0.041***
(0.006)
-0.046**
(0.018)
-0.009***
(0.002)
-0.008
(0.007)
-0.186***
(0.029)
-0.759***
(0.184)
0.243**
(0.111)
-0.472***
(0.140)
-0.371***
(0.057)
-1.456***
(0.033)
0.311***
(0.073)
0.162***
(0.010)
0.083***
(0.031)
0.054
(0.082)
0.307***
(0.088)
0.120**
(0.059)
0.537***
(0.079)
-0.069**
(0.027)
0.052***
(0.012)
-1.821***
(0.048)
-0.034***
(0.004)
0.058***
(0.013)
0.224***
(0.048)
1.534***
(0.164)
-0.065
(0.089)
2.944***
(0.138)
-0.191***
(0.052)
0.388***
(0.025)
-0.493***
(0.06)
-1.181***
(0.014)
-0.366***
(0.030)
0.448***
(0.075)
2.302***
(0.246)
0.318***
(0.108)
-0.135
(0.127)
-0.042
(0.040)
0.051***
(0.019)
0.213***
(0.048)
-0.094***
(0.008)
-1.19***
(0.029)
0.385***
(0.072)
0.138
(0.129)
-0.091
(0.093)
0.746***
(0.122)
-0.258***
(0.041)
0.008
(0.013)
0.208***
(0.044)
0.029***
(0.005)
0.097***
(0.018)
-2.767***
(0.099)
0.334***
(0.056)
-0.206***
(0.045)
0.265***
(0.053)
0.087***
(0.021)
-0.074***
(0.005)
0.060***
(0.016)
-0.009***
(0.002)
-0.030***
(0.006)
0.108***
(0.025)
0.400**
(0.166)
0.916***
(0.084)
1.047***
(0.103)
-0.068
(0.119)
-0.223***
(0.017)
0.082
(0.058)
0.002
(0.007)
0.075***
(0.022)
-0.006
(0.065)
-0.459
(0.354)
-0.819***
(0.187)
-1.377***
(0.232)
0.234
(0.234)
0.393***
(0.029)
0.162
(0.107)
0.021*
(0.012)
-0.131***
(0.045)
0.478***
(0.130)
0.312***
(0.052)
-0.019***
(0.004)
0.079***
(0.016)
0.082***
(0.019)
-0.627***
(0.044)
0.086***
(0.022)
-0.186***
(0.035)
-0.156***
(0.031)
0.142***
(0.032)
-1.17***
(0.016)
0.179***
(0.032)
0.278***
(0.065)
0.262**
(0.109)
0.059***
(0.005)
0.220***
(0.022)
-0.045
(0.079)
-1.325***
(0.101)
0.082
(0.057)
0.026
(0.094)
0.274***
(0.080)
-0.005
(0.060)
0.126***
(0.022)
-2.392***
(0.094)
1.720***
(0.162)
-0.067
(0.051)
-0.012***
(0.003)
-0.064***
(0.011)
0.035
(0.034)
0.518***
(0.039)
0.036
(0.024)
0.068*
(0.037)
-0.106***
(0.036)
0.098***
(0.027)
0.043***
(0.010)
0.382***
(0.036)
-1.930***
(0.104)
Note: The items are: Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me), Milk & Eggs (Mi), Cheese,
Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Robust standard errors in parentheses.
N = 13,582. *** p < 0.01, ** p < 0.05, * p < 0.1.
17
Table 4. Uncompensated price elasticities of demand (Full sample)
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils
& Fats
E
-5.501***
(0.347)
0.364
(0.234)
-1.744***
(0.235)
1.101***
(0.114)
-0.092***
(0.020)
0.073
(0.058)
0.078***
(0.008)
0.263***
(0.044)
-0.096
(0.092)
K
-0.018
(0.023)
-1.572***
(0.125)
0.144***
(0.035)
-0.023*
(0.013)
0.002
(0.001)
-0.005
(0.004)
0.000
(0.000)
-0.009***
(0.002)
-0.025***
(0.007)
C
-0.651***
(0.075)
0.551***
(0.115)
-3.276***
(0.202)
0.151***
(0.027)
-0.019***
(0.005)
0.072***
(0.017)
0.046***
(0.002)
-0.056***
(0.007)
0.108***
(0.028)
Fw
0.654***
(0.117)
-0.144
(0.136)
0.311***
(0.086)
-1.163***
(0.067)
-0.050***
(0.007)
-0.177***
(0.019)
-0.001
(0.003)
-0.160***
(0.008)
-0.382***
(0.031)
Ce
-3.443***
(0.236)
2.533***
(0.144)
-0.509***
(0.164)
1.078***
(0.072)
-1.535***
(0.041)
-0.862***
(0.095)
0.232***
(0.017)
-1.285***
(0.044)
-1.706***
(0.109)
Fr
-0.143
(0.092)
0.503***
(0.061)
0.531***
(0.082)
0.174***
(0.028)
0.039***
(0.013)
-2.018***
(0.049)
-0.023***
(0.005)
-0.171***
(0.014)
-0.070
(0.050)
V
-4.911***
(0.372)
5.433***
(0.233)
2.856***
(0.260)
3.289***
(0.123)
0.198***
(0.067)
-3.310***
(0.165)
-1.013***
(0.037)
-3.651***
(0.084)
-3.776***
(0.187)
Me
0.655**
(0.269)
1.723***
(0.124)
-0.158
(0.142)
0.847***
(0.049)
0.002
(0.025)
-0.507***
(0.063)
-0.051***
(0.011)
-2.029***
(0.036)
-0.695***
(0.084)
Mi
-0.278**
(0.130)
0.264***
(0.094)
0.740***
(0.123)
-0.033
(0.041)
-0.004
(0.013)
0.026
(0.045)
0.040***
(0.005)
-0.115***
(0.019)
-3.040***
(0.099)
Ch
0.017
(0.058)
0.064
(0.046)
0.261***
(0.054)
0.258***
(0.021)
-0.083***
(0.006)
-0.078***
(0.018)
-0.001
(0.002)
-0.192***
(0.007)
-0.100***
(0.026)
Al
-0.051
(0.171)
1.301***
(0.088)
1.041***
(0.105)
0.176
(0.120)
-0.236***
(0.018)
-0.115*
(0.059)
0.014*
(0.007)
-0.155***
(0.023)
-0.302***
(0.067)
O
-2.492***
(0.376)
0.916***
(0.196)
-1.404***
(0.242)
1.332***
(0.235)
0.332***
(0.034)
-0.727***
(0.116)
0.074***
(0.015)
-1.168***
(0.051)
-0.855***
(0.140)
0.317***
(0.052)
-0.018***
(0.004)
0.081***
(0.016)
0.087***
(0.020)
-0.582***
(0.055)
0.093***
(0.023)
-0.079
(0.089)
-0.128***
(0.038)
0.149***
(0.033)
-1.165***
(0.017)
0.187***
(0.033)
0.312***
(0.069)
Alcohol &
0.183*
0.051*** 0.194***
-0.126
-2.048***
-0.039
-1.711***
-0.170*
-0.118*
0.040
-2.513***
1.172***
Tobacco
(0.108)
(0.005)
(0.022)
(0.080)
(0.140)
(0.059)
(0.257)
(0.101)
(0.061)
(0.025)
(0.097)
(0.174)
Other Food &
-0.002
-0.005** -0.043*** 0.100*** 1.108*** 0.135***
1.484***
0.256***
0.189*** 0.113***
0.481***
-1.483***
Drink
(0.052)
(0.003)
(0.011)
(0.035)
(0.062)
(0.025)
(0.122)
(0.048)
(0.027)
(0.012)
(0.037)
(0.106)
Note: The items are: Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me), Milk & Eggs (Mi), Cheese,
Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Robust standard errors in parentheses.
N = 13,582. *** p < 0.01, ** p < 0.05, * p < 0.1.
18
Table 5. Compensated price elasticities of demand for energy sources (Rural vs. Urban)
Electricity
Kerosene
Charcoal
Firewood
Panel A: Rural
Electricity Kerosene Charcoal
-11.845***
-0.057
-0.041
(1.027)
(0.054)
(0.163)
-0.287
-1.834***
0.296
(0.274)
(0.183)
(0.196)
-0.164
0.233
-9.645***
(0.653)
(0.155)
(0.66)
0.612***
-0.036*** 0.108***
(0.126)
(0.012)
(0.026)
Firewood
1.391***
(0.286)
-0.413***
(0.141)
0.982***
(0.237)
-1.466***
(0.057)
Panel B: Urban
Electricity Kerosene
-3.108***
0.133***
(0.319)
(0.029)
3.54***
-2.037***
(0.763)
(0.264)
-0.065
0.089***
(0.191)
(0.034)
-0.0300
2.422***
(0.435)
(0.062)
Firewood
0.614***
(0.11)
-0.203
(0.42)
0.324***
(0.071)
-0.986***
(0.319)
N = 9,185
Electricity
Kerosene
Charcoal
Firewood
Charcoal
-0.023
(0.069)
0.852***
(0.324)
-1.400***
(0.094)
0.462***
(0.101)
N = 4,397
Note: Robust standard errors in parentheses.
*** p < 0.01, ** p < 0.05, * p < 0.1.
5.1.2. Food items
Energy price hikes have implications for food demand due to complementary patterns of
cooking fuel use, and budgetary reallocations within consumption baskets. Electricity price
increases simultaneously raise the demand for firewood and for most food groups including
vegetables, meat & fish, and fruits, while reducing cereal and charcoal consumption.
On the other hand, increases in kerosene prices raise the demand for charcoal, cereals, meat
and fish, while reducing the demand for firewood, cheese, oils & fats, and other food & drink.
Given the symmetry of substitution effects, food price increases similarly have consequences
for biomass demand and create additional shifts in food baskets. Higher vegetable prices
simultaneously raise the demand for cereals and charcoal.
The increase in vegetable and meat consumption due to higher electricity prices is
complementary to firewood use, owing to interdependencies between cooking fuels and the
relatively longer cooking times of these items (Treiber et al. 2015). Kerosene price increases,
19
however, modestly raise the demand for cereals and meat & fish, revealing smaller
compensating effects for households’ food consumption relative to electricity price increases.
These substitution patterns suggest trade-offs between food security and clean fuel use, as fossil
fuel price increases could affect household nutrition.
5.2. Welfare effects of carbon pricing
The percent price increases in energy and food items resulting from a carbon price of US$40/t
CO2 are presented in Table 6. Kerosene and electricity exhibit the largest price hikes (16.4%
and 10.2%, respectively),16 while prices for food items rise by less than 1%.
Table 6. Price and demand changes due to carbon pricing (US$40/tCO2)
Category
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other food & drink
Price Increase (%)
10.2
16.4
0
0
0.25
0.17
0.19
0.39
0.22
0.2
0.4
0.25
Demand Change (%)
-58.9
-18.8
-14.7
12.6
-1.3
-1.0
0.7
0.3
-3.9
2.7
1.0
0.5
Note: Demand changes (%) are based on the uncompensated price
elasticities of demand estimated for the full sample (Table 4).
The first- and second-order welfare effects of a carbon price of US$40/t CO2 are highlighted
in Figure 2. The average first-order welfare loss is estimated at 0.46% of households’ total
expenditure on food and fuel, with progressive distributional effects in the range of 0.17% to
3.5%. Despite the large price increases for electricity and kerosene, their relatively low budget
shares generate overall modest welfare effects.17 The average second-order welfare loss is
estimated at 0.35% of households’ expenditure on food and fuel, rising steadily from 0.17% to
16
The 10% price increase for electricity is comparable to tariff changes of approximately 8% implemented by the
Electricity Regulatory Authority, Uganda between 2017 and 2020.
17
A scatterplot correlating item-wise expenditure shares and percent increases in prices (Appendix Figure A2)
reflects this insight.
20
1.8% across the expenditure distribution. The second-order effects are smaller than the firstorder effects at each percentile of the distribution, and drop sharply at the upper tails, reflecting
larger substitution possibilities in households’ consumption baskets.18
Figure 2. First- and second-order welfare effects of a US$40/ton carbon price
Note: The Y-axis represents the percent change in households’ expenditure on food and fuel.
18
This is due to the combined effect of larger budget shares for fossil fuel-based energy sources among richer
households and sizeable compensated price elasticities of demand.
21
Figure 3. Welfare effects of carbon pricing for energy & select food items by expenditure
quintile
Note: We calculate the average welfare effect for each item, at each percentile, and plot the variation in average
effects within expenditure quintiles. Expenditure percentiles are calculated based on households’ total real
expenditure. The boxes represent the interquartile range (25th percentile to 75th percentile) and the horizontal line
in each box represents the median value. The upper and lower whiskers capture the upper (lower) adjacent values,
which are the largest (smallest) observations equal to the value at the third (first) quartile plus (minus) 1.5 times
the interquartile range, while all remaining observations, i.e. those above (below) the upper (lower) adjacent
values, are excluded from the plot.
Decomposing the welfare effects by consumption categories, Figure 3 highlights significant
heterogeneity across and within expenditure quintiles for energy and select food items. We find
welfare losses along the vertical (i.e. across-quintile) and horizontal (i.e. within-quintile) equity
dimensions, with the largest impact for electricity among households in the top quintile, despite
overall progressive effects. The indirect welfare loss due to electricity consumption ranges
from 0 – 1.6% over the sample, while kerosene consumption generates regressive, albeit
modest welfare effects, which is analogous to the decline in budget shares for kerosene across
expenditure quintiles (Figure 1). While the welfare loss due to consumption of meat & fish is
progressive, the effect for vegetable consumption is largely regressive, as households in the top
quintile exhibit proportionately smaller welfare losses.
22
5.3. Shifts in the consumption basket
Carbon pricing further creates shifts in the consumption basket, with a staggering reduction in
electricity consumption by up to 60%, and declines in kerosene and charcoal demand by 19%
and 15%, respectively (Table 6). This is due to the highly price elastic nature of electricity
demand (with an own-price uncompensated elasticity of -5.5) and important cross-price effects
due to multiple, simultaneous commodity price hikes.
However, firewood demand exhibits a 13% increase as a result of a US$40/ton price on carbon
emissions, and suggests substitutability between readily available, low-grade biomass, and
fossil-based energy sources. The demand for several food items including vegetables, meat &
fish, and cheese, oils & fats, similarly increases, by 0.3 – 2.7% for the average household, while
consumption of cereals, fruits and milk & eggs declines by 1 – 4% on average.
We find disparities in food consumption and nutritional intake due to carbon pricing across
expenditure tercile – rural/urban groups, resulting from differences in the composition of
households’ food consumption (Figure 4, panels A and B).19 On average, food consumption
rises by 0.1%, but ranges from declines of up to 2.3%, to increases of up to 1.7%. Patterns for
nutrient intake mirror those for food consumption, with a rise in calorie and micronutrient
consumption of 0.15% and 0.23% respectively, for the average household. On the contrary,
protein intake declines by 0.1% for the mean household, with impacts ranging from -2.4 to
2.14% across the sample.
Impacts vary substantially within expenditure terciles, reflecting the importance of horizontal
equity effects. Increases in food consumption and nutrient intake among urban households
occur largely due to increased demand for meat & fish and vegetables, concomitantly with
greater firewood use. Households in the bottom third of the distribution exhibit large variations
in impacts on food security and nutrition. The majority of households in the top tercile in urban
areas experience declines in food and protein intake, thus highlighting the role of other
demographic and socio-economic characteristics in determining heterogeneous impacts of
carbon pricing.
Demand changes by expenditure tercile – rural/urban groups are computed from group-specific elasticities,
which are derived from the estimated regression coefficients (for the full sample) and the group-specific budget
shares and real expenditure for all items. These are presented in Appendix Tables A6.1 – A6.3.
19
23
Figure 4. Percent changes in A) Food and Calorie consumption and B) Protein and Micronutrient intake, due to carbon pricing (US$40/t CO2)
Panel A
Panel B
Note: T1, T2, and T3 refer to the first, second, and third expenditure terciles, respectively, while R and U represent respective rural and urban areas. The boxes represent the
interquartile range (25th percentile to 75th percentile), with the horizontal line measuring the median value. The upper and lower whiskers capture the upper (lower) adjacent
values, which are the largest (smallest) observations equal to the value at the third (first) quartile plus (minus) 1.5 times the interquartile range, while all remaining observations,
i.e. those above (below) the upper (lower) adjacent values, are excluded from the plot. The Y-axis measures the percentage change in household demand due to carbon pricing,
relative to baseline demand observed in the survey.
24
6. Discussion
Carbon pricing is considered an effective instrument for climate change mitigation and longterm sustainable development. In the SSA context, it could stimulate investments in renewable
energy and reduce dependence on carbon-intensive energy infrastructure in the long run. In
Uganda alone, a carbon price of US$40/t CO2 could provide revenues to cover one-fifth of the
financing requirement for the SDG 2030 Agenda (Franks et al. 2018). However, when
designing carbon pricing schemes, potential trade-offs with economic development objectives
particularly in low-income countries, need to be taken into account.
6.1. Welfare impacts
Previous studies analysing the first-order welfare effects of a carbon price find progressive
distributional outcomes for low- and middle-income countries (Dorband et al. 2019; Ohlendorf
et al. 2021). Using first- and second-order welfare measures, our results similarly show modest
and progressive welfare outcomes across the Ugandan distribution, broadly reinforcing the
conclusions of the literature. The paper additionally underscores the impacts of carbon pricing
on multidimensional measures of household welfare including energy demand, biomass
consumption, and consequences for food and nutrient intake, revealing important interactions
between climate policies and economic development. These dimensions have hitherto been
neglected in the literature, but are important considerations in the design of socially just and
efficient sustainable development policies in the SSA region.
Average welfare impacts across the distribution mask substantial heterogeneity of effects
within income groups. The recent literature distinguishes between the horizontal (within
income) and vertical (across income) equity effects of carbon taxes/cap-and-trade systems, and
finds substantial welfare losses and greater variation of impacts within consumption deciles
relative to across deciles, albeit with a focus on the United States (Cronin et al. 2019; Fischer
and Pizer 2019). In Uganda, accounting for the direct and indirect effects of carbon pricing
points to welfare losses along both equity dimensions. Given the prevalence of these hardship
cases in all expenditure quintiles, further research is necessary to investigate the factors that
determine specific patterns of energy use, consumption expenditures and demand shifts along
the distribution. While differential elasticities across expenditure terciles allow for differing
demand responses, further disaggregation in estimating price elasticities of demand is
important to understand substitution behaviour across households within specified income
groups.
25
6.2. Revenue recycling
Potential adverse welfare effects can typically be ameliorated through compensation payments
and revenue recycling schemes (Douenne 2020). The existing literature finds that the
implementation of cash transfer programmes in combination with carbon pricing enhances the
progressivity of welfare outcomes in middle-income countries like Mexico and Ecuador
(Renner et al. 2018; Schaffitzel et al. 2020). Similar evidence on the direct and indirect welfare
impacts of energy subsidy reforms in Latin America and the Caribbean suggests that negative
welfare outcomes for the bottom 40 per cent of the population can be alleviated through
government transfers utilising only 19 per cent of the collected tax revenue (Feng et al. 2018).
For the Ugandan case, we investigate a basic revenue recycling model with the total generated
revenue redistributed equally on a per-capita basis to all households via lump sum transfers.
Taking into account the short-run demand changes due to energy and food price hikes, the total
revenue generated from a US$40/t CO2 price amounts to UgX 90 billion (US$35 million). For
8.5 million Ugandan households, the lump sum transfer is approximately UgX 10,575 (US$3)
per household. We find this rebate compensates close to three-quarters of the population for
welfare losses accruing due to carbon pricing, yielding net welfare gains of up to 1.6% for these
households. On average, households in the bottom expenditure quintile observe the largest net
welfare gains of 0.6% of expenditure on average, while households in the top quintile
experience welfare losses of 0.25% of expenditure on average.
26
Figure 5. Mean welfare effects of carbon pricing with lump-sum transfers by
expenditure quintile
Note: The graph depicts average welfare effects for households by expenditure quintile, measured as the percent
change in expenditure on food and fuel, due to carbon pricing. In the second scenario, revenues from carbon
pricing are redistributed to households via lump-sum transfers, and the scheme generates net average welfare
gains for the bottom 60% of the distribution.
In addition to alleviating negative welfare effects, cash transfer programmes in SSA countries
can further have positive impacts such as stimulating agricultural households to undertake
productive activities, fostering arrangements for sharing of food and production inputs within
communities, and especially in reducing market failures typically faced by low-income
households in SSA (Daidone et al. 2019). In the context of LMICs, providing subsidies for
cleaner cooking fuels, such as LPG (Irfan et al. 2018), may prove beneficial in promoting a
clean energy transition and reducing indoor air pollution, while ameliorating potential adverse
effects on food consumption. Evidence from Indonesia finds that LPG subsidies along with
quantity restrictions on kerosene consumption within districts can effectively stimulate largescale fuel transitions. The LPG program lowered the demand for kerosene by 80% over the
2007-2012 period and created positive spillovers on maternal and child health (Imelda 2020).
Although LPG use is less prominent in Uganda, expansions in access to fuel efficient cookstoves, for instance via the World Bank-led clean cooking supply chain expansion project,
27
could alleviate pressures on deforestation (World Bank 2021b). The use of charcoal briquettes
for fuel, processed from agricultural residue, can similarly reduce the overall demand for
charcoal and promote efficient energy use (Kwesiga 2019).
6.3. Substitution and complementary effects in energy use
Carbon pricing generates significant shifts in the composition of energy demand in Uganda,
with households substituting solid fuels like firewood for electricity. Based on the final changes
in demand due to carbon pricing for the average household (Table 6), we compute the potential
change in short-run aggregate demand for biomass in Uganda, using household survey weights.
Relative to annual production levels (WAVES 2019), we find an aggregate decline in charcoal
demand by 2.8% and a 1.6% increase in total firewood consumption.20 These are interpretable
as lower bound estimates, given a potentially large unmeasured increase in home produced
firewood. Fossil fuel subsidy removal and introduction of a carbon price could thus have
implications for the environment through increased fuelwood demand, and potentially
aggravate the public health burden due to resulting increases in indoor air pollution (IAP) from
biomass burning.21
6.4. Heterogeneous effects on food consumption
Energy demand is closely linked to food consumption and dietary composition through the
complementary use of cooking fuels. Items with long cooking times such as vegetables and
meat & fish are typically cooked with firewood or charcoal, while those with shorter cooking
times like milk and tea/coffee could be complementary to kerosene use (Treiber et al. 2015).
These complementarities differ by income groups due to differential prices and the availability
of fuels in rural vs. urban regions. The relatively higher incomes of urban households enable a
large increase in firewood use and increased consumption of nutrient-rich foods like
vegetables. In rural areas, the availability of lower grade biomass and consumption of possibly
cheaper varieties of meat and fish, ameliorates the negative effects of carbon pricing on food
consumption for some households. Nevertheless, around half of the population experiences
adverse effects on dietary composition and nutritional intake. This is due to multiple factors
including low budget shares for protein- and nutrient-rich food groups, and negative income
effects of food and energy price increases.
20
Bundles of firewood recorded in the data are converted to kilograms based on the average weight of a firewood
bundle, which is approximately 30 kgs (Albers 2016).
21
In 2017, 400,000 deaths were recorded due to IAP across SSA (Roth et al. 2018).
28
The results are viewed against a background of a large prevalence of nutritional deficiencies
among over a quarter of the Ugandan population, despite a marked drop in poverty levels over
the 1993-2013 period (Boysen 2016). Carbon pricing without revenue recycling therefore
creates welfare losses and could lead to a compounding of pre-existing challenges of food
insufficiency and undernutrition, particularly for households around the threshold of poverty.
In order to prevent adverse impacts of increased biomass use and stimulate a transition towards
cleaner cooking fuels, targeted food security programmes and cash transfers would be required
in conjunction with emission pricing policies in Uganda.
6.5. Limitations
Our analysis is limited in a number of ways. First, the demand system is parsimonious and does
not account for household demographics and other socio-economic characteristics, which
would better reflect preference heterogeneity at the individual level. A more granular analysis
of household demand responses is also needed to understand shifts in demand patterns and
identify households most affected by environmental policies in Uganda.
Second, the analysis relies on price data from the year preceding the household expenditure
survey, which could be subject to differential time trends across Ugandan sub-regions. While
we inflate energy and food prices by the consumer price index, we do not control for regionspecific time trends, for the purpose of model tractability, which qualifies the analysis.
Third, while we assume weak separability between broad commodity groups, this may be
violated due to complementarities between cooking fuels and certain food items. For example,
higher kerosene prices may impact the demand for certain cereals or vegetables differently than
others, due to taste preferences, cultural reasons or longer cooking times of specific items.
Further, while we assume multi-stage budgeting between consumables (food and energy),
durables and other non-durables, we do not compute shifts in the expenditures allocated to food
and energy, due to carbon pricing. Hence, for simplicity, we do not simulate carbon prices for
durable goods.
Fourth, the demand system does not account for households’ access to (and expenditure on)
electricity sourced directly from local hydropower or solar power facilities, such as mini-grids
or off-grid power, due to lack of survey data on expenditures and related investment costs. If
the use of solar/hydropower were modelled in the demand system, we could expect transitions
toward their use in the context of carbon pricing, although predictions regarding substitutions
between solar power and food consumption cannot be made ex ante.
29
Lastly, the paper does not delve into the multidimensional welfare impacts of carbon pricing,
albeit these are important questions for future research. Specifically, future work could analyse
firewood collections from forests vis-à-vis market purchases and the complementarities
between cooking fuels and food consumption. In a related manner, the time use and market
labour supply of women (Imelda 2020; Köhlin et al. 2011), are important policy questions to
understand the multifaceted implications of mitigation policies on economic development in
SSA.
7. Conclusion
This paper investigates the welfare effect on Ugandan households imposed by a carbon price
of US$40/t CO2. We estimate a consumer demand system to account for substitution effects
between traditional biomass and modern energy, and the effects of energy price changes on
food consumption. We identify trade-offs between climate change mitigation and economic
development. Our results show substantial reductions in the demand for charcoal, kerosene and
electricity by 15% – 59%, and an increase in firewood consumption by 13% on average, in
response to carbon pricing. Welfare effects are progressive and reach a maximum of 1.8% of
household expenditures on food and fuel, with heterogeneous impacts within and across
expenditure quintiles for individual items.
The demand system reveals dependencies between food consumption and energy price changes
due to the complementary use of cooking fuels. Increases in fossil fuel prices create large shifts
in food consumption baskets, with substitution toward vegetables, meat & fish and other food
groups, alongside an increase in firewood use. Commodity price increases negatively impact
the dietary composition and nutritional intake of around half of the Ugandan population, with
disparate effects within expenditure tercile groups.
Energy price hikes resulting from a carbon price reveal important trade-offs between access to
cleaner fuels and food security, and could have consequences for forest degradation. Climate
policies could potentially exacerbate undernutrition for a majority of households, particularly
those at the brink of poverty. Combinations of energy and development policies are therefore
required to ameliorate these adverse effects and could result in multiple co-benefits for
households. Carbon pricing with revenue redistribution could help achieve the twin objectives
of long-term sustainable and inclusive economic development in Uganda and the broader SubSaharan African region.
30
Author Contributions (see here)
Raavi Aggarwal: Methodology, Software, Formal Analysis, Writing – Original Draft; Sinem
Ayhan - ; Michael Jakob - ; Jan Steckel - .
References
Aguiar, A., Chepeliev, M., Corong, E. L., McDougall, R. and van der Mensbrugghe, D. 2019.
The GTAP Data Base: Version 10. Journal of Global Economic Analysis 4 (1): 1-27.
Albers, P. 2016. Sustainability: availability, distribution and consumption of firewood near
Kidepo Valley national park. Household surveys and forest inventory. Bachelor’s Thesis.
Unpublished.
Alem, Y. and Demeke, E. 2020. The persistence of energy poverty: A dynamic probit analysis.
Energy Economics 90: 104789.
Angrist, J. D. and Pischke, J-S. 2008. Mostly Harmless Econometrics: An Empiricist's
Companion. Princeton University Press.
Attanasio, O., Di Maro, V., Lechene, V. and Phillips, D. 2013. Welfare consequences of food
prices increases: Evidence from rural Mexico. Journal of Development Economics 104: 136151.
Attanasio, O., Battistin, E. and Mesnard, A. 2011. Food and Cash Transfers: Evidence from
Colombia. The Economic Journal 122 (March): 92-124.
Banks, J., Blundell, R. and Lewbel, A. 1997. Quadratic Engel Curves and Consumer Demand.
The Review of Economics and Statistics 79 (4): 527-539.
Banks, J., Blundell, R. and Lewbel, A. 1996. Tax reform and welfare measurement: Do we
need demand system estimation? The Economic Journal 106 (438): 1227-1241.
Bhuvandas, D. and Gundimeda, H. 2020. Welfare impacts of transport fuel price changes on
Indian households: An application of LA-AIDS model. Energy Policy 144: 111583.
Boysen, O. 2016. Food Demand Characteristics in Uganda: Estimation and Policy Relevance.
South African Journal of Economics 84(2): 260-293.
31
Capéau, B. and Dercon, S. 2006. Prices, Unit Values and Local Measurement Units in Rural
Surveys: an Econometric Approach with an Application to Poverty Measurement in Ethiopia.
Journal of African Economies 15(2): 181-211.
Chamberlain, G. 1984. "Panel data," Handbook of Econometrics, in: Z. Griliches & M. D.
Intriligator (ed.), Handbook of Econometrics, edition 1, volume 2, chapter 22, pages 1247-1318
Elsevier.
Chamberlain, G. 1980. Analysis of Covariance with Qualitative Data, The Review of Economic
Studies, Volume 47, Issue 1, 1980, Pages 225–238.
Cronin, J.A., Fullerton, D. and Sexton, S. 2019. Vertical and Horizontal Redistributions from
a Carbon Tax and Rebate. Journal of the Association of Environmental and Resource
Economists 6(S1): S169 - S208.
Daidone, S., Davis, B., Handa, S. and Winters, P. 2019. The household and individual-level
productive impacts of cash transfer programs in Sub-Saharan Africa. American Journal of
Agricultural Economics 101(5): 1401-1431.
Deaton, A. and Muellbauer, J. 1980. An Almost Ideal Demand System. The American
Economic Review 70 (3): 312-326.
Deaton, A. and Zaidi, S. 2002. Guidelines for Constructing Consumption Aggregates for
Welfare Analysis. World Bank LSMS Working Paper No. 135.
Dorband, I., Jakob, M., Kalkuhl, M. and Steckel, J. C. 2019. Poverty and distributional effects
of carbon pricing in low- and middle-income countries – A global comparative analysis. World
Development 115: 246-257.
Douenne, T. 2020. The Vertical and Horizontal Distributive Effects of Energy Taxes: A Case
Study of a French Policy. The Energy Journal 41(3).
Ecker, O. and Qaim, M. 2011. Analyzing Nutritional Impacts of Policies: An Empirical Study
for Malawi. World Development 39(3).
Eisner, A., Kulmer, V. and Kortschak, D. 2021. Distributional effects of carbon pricing when
considering household heterogeneity: An EASI application for Austria. Energy Policy Volume
156, September 2021, 112478.
32
Feng, K., Hubacek, K., Liu, Y., Marchán, E. and Vogt-Schilb, A. 2018. Managing the
distributional effects of energy taxes and subsidy removal in Latin America and the Caribbean.
Applied Energy 225: 424-436.
Fischer, C. and Pizer, W. 2019. Horizontal Equity Effects in Energy Regulation. Journal of the
Association of Environmental and Resource Economists 6(S1): S209 - S237.
Franks, M., Lessmann, K., Jakob, M., Steckel, J. C. and Edenhofer, O. 2018. Mobilizing
domestic resources for the Agenda 2030 via carbon pricing. Nature Sustainability 1(7): 350357.
Fuje, H. 2019. Fossil fuel subsidy reforms, spatial market integration, and welfare: Evidence
from a natural experiment in Ethiopia. American Journal of Agricultural Economics 101(1):
270-290.
Government of Uganda. 2015. Uganda’s Intended Nationally Determined Contribution
(INDC).
Ministry
of
Water
and
Environment.
Accessible
at
<
https://www4.unfccc.int/sites/ndcstaging/PublishedDocuments/Uganda%20First/INDC%20U
ganda%20final%20%2014%20October%20%202015.pdf>
Greve, H. and Lay, J. 2020. Stepping down the ladder: The unintended impacts of fossil fuel
subsidy removal in a developing country. European Association of Environmental and
Resource Economists (EAERE) Conference Paper 2020.
Guta, D. 2012. Application of an almost ideal demand system (AIDS) to Ethiopian rural
residential energy use: panel data evidence. Energy Policy 50: 528-539.
Hansen, L. P. 1982. Large Sample Properties of Generalized Method of Moments Estimators.
Econometrica Vol. 50, No. 4, pp. 1029-1054.
IEA, IRENA, UNSD, World Bank, and WHO. 2020. Tracking SDG 7: The Energy Progress
Report. World Bank, Washington DC.
Imelda. 2020. Cooking that kills: Cleaner energy access, indoor air pollution, and health.
Journal of Development Economics 147 (2020) 102548.
Irfan, M., Cameron, M. P. and Hassan, G. 2018. Household energy elasticities and policy
implications for Pakistan. Energy Policy 113: 633-642.
33
Jakubson, G. 1988. The Sensitivity of Labor-Supply Parameter Estimates to Unobserved
Individual Effects: Fixed- and Random-Effects Estimates in a Nonlinear Model Using Panel
Data. Journal of Labor Economics 6(3): Jul, 1988.
Jewell, J., McCollum, D., Emmerling, J., Bertram, C., Gernaat, D., Krey, V., Paroussos, L. et
al. 2018. Limited emission reductions from fuel subsidy removal except in energy-exporting
regions. Nature 554 (7691): 229-233.
Köhlin, G., Sills, E. O., Pattanayak, S. K. and Wilfong, C. 2011. Energy, Gender and
Development: What are the Linkages? Where is the Evidence? The World Bank, Policy
Research Working Paper 5800.
Kwesiga, C. 2019. Charcoal briquettes – affordable, clean energy for Uganda. United Nations
Conference on Trade and Development, Regional Workshop on Promoting Cotton ByProducts in Eastern and Southern Africa. Accessible at < https://unctad.org/system/files/nonofficial-document/Cotton_SA_May2019_S4_Charles_Kwesiga.pdf>
Lewbel, Arthur, and Krishna Pendakur. 2009. "Tricks with Hicks: The EASI Demand System."
American Economic Review, 99 (3): 827-63.
Masera, O., Bailis, R., Drigo, R., Ghilardi, A. and Ruiz-Mercado, I. 2015. Environmental
Burden of Traditional Bioenergy Use. Annual Review of Environment and Resources 40(1):
121-150.
Mattauch, L., Creutzig, F. and Edenhofer, O. 2015. Avoiding Carbon Lock-In: Policy Options
for Advancing Structural Change. Economic Modelling 50: 49-63.
MEMD, 2015. National Charcoal Survey for Uganda 2015. Final Report. Accessible at
<http://unreeea.org/wp-content/uploads/2018/10/National-Charcoal-Survey_uganda.pdf>
Meyerhoefer, C. D., Ranney, C. K. and Sahn, D. E. 2005. Consistent Estimation of Censored
Demand Systems Using Panel Data. American Journal of Agricultural Economics 87(3): pp.
660-672.
Ohlendorf, N., Jakob, M., Minx, J.C. et al. 2021. Distributional Impacts of Carbon Pricing: A
Meta-Analysis. Environmental and Resource Economics 78: 1-42.
Okonkwo, J. U. 2021. Welfare effects of carbon taxation on South African households. Energy
Economics 96, April 2021, 104903.
34
Olabisi, M., Tschirley, D. L., Nyange, D. and Awokuse, T. 2019. Energy demand substitution
from biomass to imported kerosene: Evidence from Tanzania. Energy Policy 130: 243-252.
Pratiti, R., Vadala, D., Kalynych, Z. and Sud, P. 2020. Health effects of household air pollution
related to biomass cook stoves in resource limited countries and its mitigation by improved
cookstoves. Environmental Research 186: 109574.
Reaños, M. and Wölfing, N. 2018. Household energy prices and inequality: Evidence from
German microdata based on the EASI demand system. Energy Economics 70: 84-97.
Renner, S., Lay, J. and Greve, H. 2018. Household Welfare and CO2 Emission Impacts of
Energy and Carbon Taxes in Mexico. Energy Economics 72: 222-235.
Renner, S. 2018. Poverty and distributional effects of a carbon tax in Mexico. Energy Policy
112: 98-110.
Roth, Gregory A, Degu Abate, Kalkidan Hassen Abate, Solomon M Abay, Cristiana Abbafati,
Nooshin Abbasi, Hedayat Abbastabar, et al. 2018. Global, regional, and national age-sexspecific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a
systematic analysis for the Global Burden of Disease Study 2017. The Lancet 392 (10159):
1736-1788.
Schaffitzel, F., Jakob, M., Soria, R., Vogt-Schilb, A. and Ward, H. 2019. Can government
transfers make energy subsidy reform socially acceptable? A case study on Ecuador. Energy
Policy 137: 111120.
Somanathan E., T. Sterner, T. Sugiyama, D. Chimanikire, N.K. Dubash, J. Essandoh-Yeddu,
S. Fifita, L. Goulder, A. Jaffe, X. Labandeira, S. Managi, C. Mitchell, J.P. Montero, F. Teng,
and T. Zylicz, 2014: National and Sub-national Policies and Institutions. In: Climate Change
2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change [Edenhofer, O., R.
Pichs-Madruga, Y. Sokona, E. Farahani, S. Kadner, K. Seyboth, A. Adler, I. Baum, S. Brunner,
P. Eickemeier, B. Kriemann, J. Savolainen, S. Schlömer, C. von Stechow, T. Zwickel and J.C.
Minx (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY,
USA.
Steckel, J. C., Hilaire, J., Jakob, M. and Edenhofer, O. 2020. Coal and carbonization in subSaharan Africa. Nature Climate Change 10(1): 83-88.
35
Sterner, T. 2012. Fuel taxes and the poor: the distributional effects of gasoline taxation and
their implications for climate policy. Resources for the Future (RFF) Press.
Stone, R. (1954) Linear Expenditure Systems and Demand Analysis: An Application to the
Pattern of British Demand. The Economic Journal, 64, 511-527.
Treiber, M. U., Grimsby, L. K. and Aune, J. B. 2015. Reducing energy poverty through
increasing choice of fuels and stoves in Kenya: Complementing the multiple fuel model.
Energy for Sustainable Development 27: 54-62.
UBOS. 2018. Uganda National Household Survey 2016/2017. Uganda Bureau of Statistics,
Kampala, Uganda.
UNFCCC. 2016. Uganda’s Intended Nationally Determined Contribution (INDC). October
2015.
Accessible
at
<https://www4.unfccc.int/sites/ndcstaging/PublishedDocuments/Uganda%20First/INDC%20
Uganda%20final%20%2014%20October%20%202015.pdf>
van der Ploeg, R., Rezai, A. and Tovar, M. 2021. GATHERING SUPPORT FOR GREEN
TAX REFORM: Evidence from German Household Surveys. European Economic Review,
Available online 17 November 2021, 103966.
WAVES. 2019. Uganda Woodfuels Overview: Technical Report. Ministry of Water &
Environment
(Republic
of
Uganda)
and
the
World
Bank.
Accessible
at
<
https://www.wavespartnership.org/sites/waves/files/kc/WAVESWoodfuel%20Overview%20-Technical%20Report%20Design%20Layout%20Final%20October.pdf>
Wooldridge, J. M. 2010. Econometric Analysis of Cross Section and Panel Data, Second
Edition. MIT Press, Cambridge, USA.
World Bank. 2021a. State and Trends of Carbon Pricing 2021. Washington, DC: World Bank.
Accessible at <https://openknowledge.worldbank.org/handle/10986/35620>
World Bank. 2021b. Quality, Efficient Biomass Stoves Reach Ugandan Households via
Strengthened
Supply
Chains.
Washington,
DC:
World
Bank.
Accessible
at
<
https://documents1.worldbank.org/curated/en/825841615961218428/pdf/Quality-EfficientBiomass-Stoves-Reach-Ugandan-Households-via-Strengthened-Supply-Chains.pdf>
36
World Bank. 2017. Report of the High-Level Commission on Carbon Prices. Washington, DC:
World Bank.
Yaméogo, N. D. 2015. Household Energy Demand and the Impact of Energy Prices: Evidence
from Senegal. African Development Bank Group, Working Paper 218.
Zhen, C., Finkelstein, E. A., Nonnemaker, J. M., Karns, S. A. and Todd, J. E. 2013. Predicting
the Effects of Sugar-Sweetened Beverage Taxes on Food and Beverage Demand in a Large
Demand System. American Journal of Agricultural Economics 96(1): 1-25.
37
Appendix
A. Data preparation and Descriptive Statistics
This section presents additional details of the commodity price data, and additional descriptive
statistics.
Market prices for food and energy items reported by households in the Uganda National Panel
Survey (UNPS, World Bank LSMS), 2015-16, are expressed in various units, including nonstandard units with no available metric conversion factors. We first convert the majority of unit
codes for which metric conversions are available, into kilograms for food items and charcoal,
bundles for firewood, and litre for kerosene. Next, we plot the mean price for each item and
unit code, to detect potential outlier unit codes, We omit unit codes with prices in the order of
5-10 times the mean price for a specific item. The remaining unit codes are not converted.
Hence, prices are reported in kilogram and litre for the majority of the sample, while the
remaining observations contain prices reported in non-standard units such as bundles and heaps
of firewood. Therefore, the price variation observed in Table 2, partly reflects differences in
unit codes of reporting. However, graphical inspection of the range of prices per item, for
different unit codes, suggests that only a handful of unit codes reflect outlier values for prices.
Further, we subsequently compute and analyse elasticities, which negate differences arising
due to measurement units.
38
Table A1: Item-wise expenditure shares (%) by expenditure tercile and rural-urban areas
Item
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
N
Tercile 1 – Rural
Mean Std. Dev.
0.28
1.790
0.29
0.413
0.21
0.840
3.09
2.817
18.28
17.311
2.51
4.619
48.26
23.644
8.25
10.265
1.59
3.784
2.10
2.449
3.31
6.634
11.84
21.411
3,638
Tercile 1 – Urban
Mean
Std. Dev.
1.36
3.559
0.29
0.436
1.03
1.683
2.06
2.791
18.40
16.330
2.02
3.895
39.59
24.702
6.92
8.913
1.80
3.971
2.52
2.533
3.03
6.676
20.98
28.461
883
Tercile 2 – Rural
Mean
Std. Dev.
0.86
3.105
0.21
0.310
0.29
0.890
2.29
2.195
17.20
14.612
3.22
4.875
47.91
20.352
11.57
11.865
2.58
4.548
2.01
1.945
2.73
5.890
9.13
15.162
3,135
Tercile 2 – Urban
Mean
Std. Dev.
2.85
4.715
0.16
0.311
1.46
1.654
0.92
1.756
18.27
14.102
2.41
3.939
34.48
21.186
10.17
10.770
2.77
4.474
2.37
2.102
2.99
6.075
21.16
26.480
1,386
Tercile 3 – Rural
Mean
Std. Dev.
2.61
5.390
0.12
0.230
0.34
0.850
1.77
1.884
17.17
12.533
3.78
4.870
43.00
18.364
13.63
11.718
3.65
4.870
1.86
1.734
2.79
5.459
9.29
13.600
2,405
Tercile 3 – Urban
Mean
Std. Dev.
5.33
6.210
0.07
0.197
1.43
1.368
0.51
1.202
17.35
11.517
3.18
4.211
30.59
17.372
13.28
10.944
4.31
4.943
2.09
1.695
2.95
5.673
18.90
21.624
2,115
39
Figure A1. Item-specific Engel Curves
Note: The X-axis represents the household’s log of Expenditure on Food and Energy items, modelled in the demand system.
While the Engel curves are nonlinear for several items, the EASI model specification we employ is parsimonious and linearly approximates Engel
curves for model tractability, and is a limitation of the analysis.
40
D. Derivation (Section 2.3): Proportionate change in quantity demanded
Consider a Marshallian demand function 𝑓𝑖 for commodity i, as a function of the J-vector of
prices q, and total household expenditure x:
𝑓𝑖 = 𝑓𝑖 (𝑞1 , 𝑞2 , … 𝑞𝑖 , … , 𝑞𝐽 ; 𝑥)
The differential of 𝑓𝑖 is expressed as:
𝜕𝑓
𝜕𝑓
𝑑𝑓𝑖 = 𝜕𝑞 𝑖 𝑑𝑞1 + ⋯ + 𝜕𝑞 𝑖 𝑑𝑞𝐽 +
1
𝜕𝑓𝑖
𝜕𝑥
𝐽
𝑑𝑥
Now we consider changes in all commodities’ prices, holding total expenditure x constant
(hence dx = 0), and compute the proportionate change in quantity demanded for good i:
𝑑𝑓𝑖
𝑓𝑖
𝜕𝑓 𝑞1 𝑑𝑞1
= 𝜕𝑞 𝑖
1 𝑓𝑖 𝑞1
𝜕𝑓 𝑞 𝑑𝑞𝐽
+ ⋯ + 𝜕𝑞 𝑖 𝑓𝐽
𝐽
𝑖
𝑞𝐽
For small price changes around the initial demand vector, this can be approximated in elasticity
form, as follows:
∆𝑓𝑖
𝑓𝑖
≈ 𝑒 𝑢𝑓𝑖 ,𝑞1
∆𝑞1
𝑞1
+ ⋯ + 𝑒 𝑢𝑓𝑖 ,𝑞𝐽
≈ ∑𝐽𝑗=1 𝑒 𝑢𝑓𝑖 ,𝑞𝑗
∆𝑞𝐽
𝑞𝐽
∆𝑞𝑗
𝑞𝑗
41
Table A2: EASI Model Regression Coefficients (Full Sample, N = 13,582)
Parameter
Coefficient
Std. Error
Z-Statistic
P>|z|
[95% Conf. Interval]
b0_1
-0.421
0.017
-25.18
0.000
-0.454
-0.388
b0_2
b0_3
0.028
-0.016
0.001
0.003
27.06
-4.93
0.000
0.000
0.026
-0.023
0.030
-0.010
b0_4
b0_5
0.225
0.297
0.017
0.026
13.59
11.66
0.000
0.000
0.193
0.247
0.258
0.347
b0_6
-0.187
0.016
-11.86
0.000
-0.218
-0.156
b0_7
b0_8
1.089
-0.579
0.034
0.020
32.11
-28.95
0.000
0.000
1.022
-0.618
1.155
-0.540
b0_9
b0_10
-0.245
0.058
0.011
0.004
-22.03
14.08
0.000
0.000
-0.267
0.050
-0.223
0.066
b0_11
b0_12
-0.049
0.800
0.017
0.041
-2.94
19.31
0.003
0.000
-0.082
0.718
-0.016
0.881
b1_1
0.028
0.002
18.05
0.000
0.025
0.031
b2_1
b3_1
-0.003
-0.001
0.000
0.000
-28.18
-1.63
0.000
0.104
-0.003
-0.001
-0.003
0.000
b4_1
b5_1
-0.019
-0.010
0.001
0.003
-36.02
-3.93
0.000
0.000
-0.020
-0.015
-0.018
-0.005
b6_1
0.017
0.001
16.03
0.000
0.015
0.019
b7_1
b8_1
-0.061
0.075
0.003
0.002
-17.93
37.42
0.000
0.000
-0.067
0.071
-0.054
0.079
b9_1
b10_1
0.025
-0.003
0.001
0.000
23.09
-6.61
0.000
0.000
0.023
-0.003
0.027
-0.002
b11_1
b12_1
0.009
-0.060
0.002
0.004
5.63
-16.19
0.000
0.000
0.006
-0.067
0.013
-0.052
a1_1
-0.082
0.007
-12.00
0.000
-0.095
-0.069
a2_2
a3_3
-0.001
-0.014
0.000
0.001
-4.76
-11.22
0.000
0.000
-0.002
-0.017
-0.001
-0.012
a4_4
a5_5
-0.006
-0.081
0.001
0.006
-4.90
-13.82
0.000
0.000
-0.009
-0.092
-0.004
-0.069
a6_6
a7_7
-0.024
-0.077
0.001
0.006
-17.04
-12.94
0.000
0.000
-0.027
-0.089
-0.022
-0.065
a8_8
-0.021
0.003
-6.63
0.000
-0.027
-0.015
a9_9
a10_10
-0.049
-0.004
0.003
0.000
-17.79
-10.39
0.000
0.000
-0.054
-0.004
-0.043
-0.003
a11_11
a12_12
-0.041
-0.125
0.003
0.014
-14.75
-8.98
0.000
0.000
-0.047
-0.152
-0.036
-0.098
a1_2
0.000
0.000
0.48
0.634
-0.001
0.001
a1_3
a1_4
-0.011
0.019
0.001
0.002
-7.37
8.28
0.000
0.000
-0.014
0.014
-0.008
0.023
a1_5
a1_6
-0.015
0.006
0.004
0.002
-4.12
3.49
0.000
0.000
-0.022
0.003
-0.008
0.009
a1_7
a1_8
0.030
0.045
0.003
0.005
9.37
9.36
0.000
0.000
0.024
0.036
0.036
0.054
42
a1_9
0.003
0.003
1.07
0.284
-0.002
0.008
a1_10
a1_11
0.007
0.008
0.001
0.003
5.99
2.42
0.000
0.016
0.004
0.001
0.009
0.014
a2_3
0.001
0.000
4.11
0.000
0.000
0.001
a2_4
a2_5
-0.001
0.000
0.000
0.000
-2.96
2.18
0.003
0.029
-0.001
0.000
0.000
0.001
a2_6
a2_7
0.000
0.000
0.000
0.000
2.03
-0.74
0.042
0.462
0.000
0.000
0.000
0.000
a2_8
0.001
0.000
2.94
0.003
0.000
0.001
a2_9
a2_10
0.000
0.000
0.000
0.000
-0.98
-4.57
0.327
0.000
-0.001
-0.001
0.000
0.000
a2_11
a3_4
0.002
0.002
0.000
0.001
10.93
3.69
0.000
0.000
0.001
0.001
0.002
0.003
a3_5
a3_6
-0.003
0.003
0.001
0.000
-3.38
6.78
0.001
0.000
-0.005
0.002
-0.001
0.004
a3_7
0.018
0.001
21.37
0.000
0.017
0.020
a3_8
a3_9
-0.001
0.005
0.001
0.001
-1.07
6.13
0.286
0.000
-0.002
0.003
0.001
0.006
a3_10
a3_11
0.002
0.007
0.000
0.001
4.97
10.20
0.000
0.000
0.001
0.005
0.002
0.008
a4_5
-0.007
0.001
-6.46
0.000
-0.010
-0.005
a4_6
a4_7
-0.001
-0.004
0.001
0.001
-2.55
-3.70
0.011
0.000
-0.002
-0.006
0.000
-0.002
a4_8
a4_9
-0.001
-0.005
0.001
0.001
-1.07
-6.32
0.287
0.000
-0.002
-0.007
0.001
-0.004
a4_10
a4_11
0.002
-0.001
0.000
0.002
4.25
-0.57
0.000
0.568
0.001
-0.006
0.003
0.003
a5_6
0.009
0.002
4.27
0.000
0.005
0.013
a5_7
a5_8
0.069
0.009
0.004
0.003
15.47
2.67
0.000
0.008
0.060
0.002
0.077
0.016
a5_9
a5_10
0.001
-0.013
0.002
0.001
0.65
-14.31
0.516
0.000
-0.003
-0.015
0.006
-0.011
a5_11
-0.039
0.003
-13.06
0.000
-0.045
-0.034
a6_7
a6_8
-0.015
0.006
0.002
0.001
-8.16
4.43
0.000
0.000
-0.018
0.004
-0.011
0.009
a6_9
a6_10
0.006
0.002
0.001
0.000
4.69
3.83
0.000
0.000
0.004
0.001
0.009
0.003
a6_11
a7_8
0.002
-0.040
0.002
0.003
1.43
-12.27
0.153
0.000
-0.001
-0.046
0.006
-0.033
a7_9
0.012
0.002
5.99
0.000
0.008
0.016
a7_10
a7_11
-0.004
0.001
0.001
0.003
-5.28
0.27
0.000
0.785
-0.005
-0.005
-0.002
0.006
a8_9
a8_10
0.011
-0.003
0.002
0.001
5.38
-5.03
0.000
0.000
0.007
-0.005
0.014
-0.002
a8_11
0.008
0.002
3.41
0.001
0.003
0.013
a9_10
0.003
0.001
4.39
0.000
0.002
0.004
43
a9_11
0.000
0.002
-0.09
0.927
-0.004
0.003
a10_11
a1_12
0.004
-0.009
0.001
0.007
5.63
-1.30
0.000
0.194
0.002
-0.022
0.005
0.005
a2_12
-0.002
0.000
-4.39
0.000
-0.002
-0.001
a3_12
a4_12
-0.009
0.005
0.001
0.005
-5.93
1.00
0.000
0.316
-0.011
-0.004
-0.006
0.014
a5_12
a6_12
0.069
0.005
0.005
0.003
13.44
1.52
0.000
0.128
0.059
-0.001
0.080
0.011
a7_12
0.009
0.005
1.85
0.065
-0.001
0.019
a8_12
a9_12
-0.014
0.013
0.005
0.004
-2.94
3.68
0.003
0.000
-0.024
0.006
-0.005
0.020
a10_12
a11_12
0.006
0.051
0.001
0.005
4.29
10.65
0.000
0.000
0.003
0.042
0.008
0.061
Table A3. Expenditure elasticities of demand (Full Sample)
Item
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
Expenditure Elasticity
15.16***
(0.784)
-12.94***
(0.494)
0.21
(0.488)
-8.19***
(0.255)
0.45***
(0.140)
6.63***
(0.351)
-0.40***
(0.078)
7.73***
(0.180)
9.94***
(0.387)
-0.25
(0.190)
4.08***
(0.548)
-3.33***
(0.268)
Note: Robust standard errors in parentheses. N = 13,582.
*** p < 0.01, ** p < 0.05, * p < 0.1.
44
Table A4: Compensated price elasticities of demand (Rural Sample)
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
E
K
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-11.845***
(1.027)
-0.287
(0.274)
-0.164
(0.653)
0.612***
(0.126)
-0.118***
(0.024)
0.159*
(0.085)
-0.024***
(0.008)
0.510***
(0.064)
0.002
(0.151)
0.898***
(0.083)
0.354*
(0.181)
0.464***
(0.098)
-0.057
(0.054)
-1.834***
(0.183)
0.233
(0.155)
-0.036***
(0.012)
0.004***
(0.001)
0.011**
(0.005)
0.001*
(0)
0.01***
(0.002)
0.027***
(0.009)
-0.018***
(0.005)
0.069***
(0.006)
-0.021***
(0.004)
-0.041
(0.163)
0.296
(0.196)
-9.645***
(0.66)
0.108***
(0.026)
0.001
(0.005)
0.071***
(0.016)
0.013***
(0.002)
-0.028***
(0.006)
0.160***
(0.044)
-0.090***
(0.015)
0.180***
(0.032)
0.077***
(0.012)
1.391***
(0.286)
-0.413***
(0.141)
0.982***
(0.237)
-1.466***
(0.057)
-0.060***
(0.007)
0.000
(0.022)
0.016***
(0.003)
0.015*
(0.008)
-0.109***
(0.038)
0.204***
(0.022)
0.323***
(0.11)
-0.146**
(0.063)
-1.921***
(0.384)
0.312***
(0.112)
0.061
(0.334)
-0.427***
(0.05)
-1.514***
(0.037)
0.283***
(0.088)
0.136***
(0.011)
0.111***
(0.035)
0.180*
(0.108)
-0.666***
(0.052)
-1.481***
(0.117)
0.879***
(0.06)
0.451*
(0.242)
0.161**
(0.07)
0.801***
(0.186)
0.000
(0.028)
0.049***
(0.015)
-2.431***
(0.061)
-0.006
(0.005)
0.118***
(0.017)
0.275***
(0.064)
0.051*
(0.028)
-0.084
(0.078)
0.125***
(0.043)
-1.048***
(0.344)
0.181*
(0.096)
2.172***
(0.305)
0.300***
(0.052)
0.360***
(0.03)
-0.092
(0.076)
-1.110***
(0.015)
-0.267***
(0.034)
0.930***
(0.112)
-0.139***
(0.047)
0.985***
(0.128)
-0.317***
(0.062)
5.063***
(0.634)
0.486***
(0.123)
-1.108***
(0.255)
0.066*
(0.034)
0.068***
(0.022)
0.412***
(0.058)
-0.062***
(0.008)
-1.362***
(0.032)
0.309***
(0.078)
-0.213***
(0.033)
0.475***
(0.097)
-0.284***
(0.066)
0.004
(0.344)
0.306***
(0.105)
1.449***
(0.404)
-0.110***
(0.038)
0.025*
(0.015)
0.221***
(0.052)
0.049***
(0.006)
0.071***
(0.018)
-4.570***
(0.138)
-0.103***
(0.04)
-0.003
(0.064)
0.455***
(0.041)
1.660***
(0.153)
-0.170***
(0.048)
-0.660***
(0.114)
0.166***
(0.018)
-0.076***
(0.006)
0.033*
(0.018)
-0.006***
(0.002)
-0.040***
(0.006)
-0.084***
(0.032)
-1.345***
(0.02)
0.077***
(0.024)
0.060***
(0.018)
0.970*
(0.497)
0.951***
(0.086)
1.970***
(0.352)
0.390***
(0.132)
-0.250***
(0.02)
-0.081
(0.075)
0.063***
(0.008)
0.131***
(0.027)
-0.003
(0.077)
0.114***
(0.036)
-2.776***
(0.097)
0.256***
(0.064)
4.373***
(0.92)
-0.989***
(0.21)
2.907***
(0.453)
-0.604**
(0.26)
0.511***
(0.035)
0.414***
(0.142)
-0.069***
(0.014)
-0.269***
(0.063)
1.885***
(0.171)
0.307***
(0.09)
0.881***
(0.219)
-2.549***
(0.179)
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me), Milk & Eggs (Mi), Cheese,
Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Robust standard errors in parentheses. N = 9,185. *** p < 0.01, ** p < 0.05, * p < 0.1.
45
Table A5: Compensated price elasticities of demand (Urban Sample)
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
E
K
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-3.108***
(0.319)
3.54***
(0.763)
-0.065
(0.191)
2.422***
(0.435)
0.119**
(0.047)
-0.116
(0.097)
0.160***
(0.015)
0.067
(0.061)
0.142
(0.133)
0.470***
(0.089)
-0.580**
(0.224)
-0.127**
(0.057)
0.133***
(0.029)
-2.037***
(0.264)
0.089***
(0.034)
-0.0300
(0.062)
-0.015***
(0.004)
0.026***
(0.008)
-0.002
(0.001)
-0.013**
(0.005)
-0.01
(0.011)
0.016*
(0.009)
0.040**
(0.017)
-0.009**
(0.004)
-0.023
(0.069)
0.852***
(0.324)
-1.400***
(0.094)
0.462***
(0.101)
-0.019
(0.013)
0.078***
(0.029)
0.054***
(0.005)
0.009
(0.016)
0.125***
(0.033)
0.175***
(0.032)
-0.232***
(0.048)
-0.091***
(0.013)
0.614***
(0.11)
-0.203
(0.42)
0.324***
(0.071)
-0.986***
(0.319)
-0.061***
(0.02)
-0.118***
(0.034)
0.007
(0.007)
-0.070***
(0.025)
-0.233***
(0.049)
-0.046
(0.04)
-0.089
(0.173)
0.017
(0.048)
0.563**
(0.221)
-1.863***
(0.45)
-0.253
(0.17)
-1.136***
(0.376)
-1.534***
(0.095)
0.728***
(0.15)
0.232***
(0.026)
-0.189**
(0.091)
-0.244
(0.159)
-0.246***
(0.092)
-1.687***
(0.286)
0.389***
(0.072)
-0.084
(0.07)
0.492***
(0.161)
0.156***
(0.057)
-0.335***
(0.095)
0.110***
(0.023)
-0.431***
(0.111)
-0.088***
(0.01)
0.028
(0.032)
-0.078
(0.064)
0.029
(0.035)
0.545***
(0.103)
-0.097***
(0.03)
1.434***
(0.138)
-0.364
(0.280)
1.327***
(0.114)
0.249
(0.240)
0.437***
(0.048)
-1.088***
(0.121)
-0.919***
(0.036)
-0.438***
(0.063)
-0.514***
(0.123)
-0.308***
(0.061)
-3.042***
(0.215)
0.066
(0.056)
0.198
(0.178)
-1.016**
(0.392)
0.076
(0.132)
-0.812***
(0.288)
-0.117**
(0.056)
0.112
(0.129)
-0.144***
(0.021)
0.186*
(0.102)
-0.035
(0.134)
-0.256***
(0.08)
-1.307***
(0.234)
-0.09
(0.062)
0.125
(0.117)
-0.232
(0.268)
0.304***
(0.08)
-0.809***
(0.171)
-0.045
(0.03)
-0.096
(0.079)
-0.051***
(0.012)
-0.011
(0.04)
0.153
(0.133)
0.308***
(0.05)
0.08
(0.147)
-0.097***
(0.036)
0.283***
(0.053)
0.258*
(0.14)
0.292***
(0.053)
-0.11
(0.095)
-0.031***
(0.012)
0.024
(0.029)
-0.021***
(0.004)
-0.053***
(0.016)
0.211***
(0.034)
-0.854***
(0.032)
0.015
(0.055)
-0.035**
(0.014)
-0.459**
(0.177)
0.832**
(0.357)
-0.508***
(0.105)
-0.278
(0.54)
-0.281***
(0.048)
0.6***
(0.113)
-0.269***
(0.019)
-0.353***
(0.063)
0.072
(0.132)
0.02
(0.072)
6.376***
(0.36)
-0.167***
(0.059)
-0.676**
(0.303)
-1.259**
(0.628)
-1.34***
(0.198)
0.362
(1.01)
0.436***
(0.08)
-0.719***
(0.225)
0.039
(0.033)
-0.164
(0.113)
-0.588***
(0.216)
-0.309**
(0.12)
-1.121***
(0.399)
-0.759***
(0.144)
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me), Milk & Eggs (Mi), Cheese,
Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Robust standard errors in parentheses. N = 4,397. *** p < 0.01, ** p < 0.05, * p < 0.1.
46
Table A6.1: Uncompensated price elasticities of demand: Expenditure Tercile - I
E
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
K
Rural (N = 3,638)
-30.871 -0.195
0.096
-1.376
-5.167
0.432
0.615
-0.012
-0.082
0.001
0.219
-0.012
0.062
0.000
0.519
-0.019
0.127
-0.056
0.311
-0.018
0.226
0.042
-0.066
-0.003
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-4.117
0.330
-7.757
0.073
-0.017
0.119
0.038
-0.030
0.261
0.079
0.190
-0.065
3.781
-0.027
0.974
-1.068
-0.055
-0.279
-0.003
-0.299
-0.802
0.087
-0.152
0.151
-22.704
1.581
-1.180
0.592
-1.532
-0.961
0.171
-1.600
-2.754
-0.592
-1.851
1.249
-0.220
0.274
1.627
0.070
0.038
-2.153
-0.026
-0.158
-0.002
0.090
-0.017
0.131
-35.030
3.703
9.338
2.069
0.136
-4.092
-1.083
-4.995
-6.740
-0.101
-1.720
1.826
8.378
0.852
-0.301
0.347
0.008
-0.348
-0.069
-2.021
-0.618
-0.141
-0.052
0.179
-0.532
0.062
2.235
-0.093
0.000
0.130
0.028
-0.020
-4.305
0.144
-0.062
0.168
0.355
0.026
0.814
0.151
-0.082
-0.081
-0.005
-0.236
-0.140
-1.165
0.037
0.125
-0.330
0.868
3.153
0.107
-0.232
-0.144
0.007
-0.211
-0.526
0.184
-2.370
0.553
-14.475
0.373
-3.946
0.687
0.321
-0.669
0.037
-1.280
-1.016
0.297
1.118
-1.625
Urban (N = 883)
Electricity
-7.314
-0.043
-1.007
0.954
-4.800
0.034
-5.779
1.911
-0.165
-0.028
-0.037
-4.889
Kerosene
0.176
-1.367
0.385
-0.105
1.552
0.231
2.956
0.732
0.077
0.056
0.828
1.054
Charcoal
-1.061
0.086
-2.383
0.179
-0.387
0.314
1.566
-0.120
0.442
0.147
0.618
-0.949
Firewood
1.002
-0.016
0.171
-1.161
0.987
0.082
2.706
0.465
-0.116
0.267
0.156
1.756
Cereals
-0.087
0.001
-0.021
-0.050
-1.531
0.040
0.176
0.014
-0.001
-0.084
-0.230
0.273
Fruits
0.177
-0.014
0.076
-0.248
-1.159
-2.383
-4.201
-0.295
0.146
-0.132
-0.146
-1.605
Vegetables
0.081
0.001
0.051
-0.001
0.249
-0.029
-1.032
-0.072
0.038
0.001
0.014
0.109
Meat & Fish
0.500
-0.023
-0.125
-0.238
-1.884
-0.130
-4.909
-2.055
-0.045
-0.323
-0.214
-2.502
Milk & Eggs
-0.038
-0.050
0.116
-0.567
-2.459
0.062
-4.785
-0.370
-3.938
-0.183
-0.428
-2.170
Cheese, Oils & Fats
0.259
-0.016
0.066
0.068
-0.524
0.071
-0.163
-0.131
0.118
-1.142
0.148
0.227
Alcohol & Tobacco
0.205
0.046
0.176
-0.124
-2.010
0.003
-1.500
0.002
-0.075
0.026
-2.483
0.881
Other Food & Drink
-0.021
-0.003
-0.024
0.055
0.627
0.055
0.680
0.044
0.092
0.068
0.293
-1.257
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me),
Milk & Eggs (Mi), Cheese, Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Elasticities are computed based on coefficients
estimated for the full sample (Table A2), combined with tercile-rural/urban specific mean budget shares and mean real expenditure on food & fuel.
47
Table A6.2: Uncompensated price elasticities of demand: Expenditure Tercile - II
E
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
K
Rural (N = 3,135)
-10.833 -0.044
0.206
-1.526
-3.730
0.311
0.874
-0.019
-0.090
0.002
0.133
-0.006
0.065
0.000
0.325
-0.010
0.014
-0.029
0.326
-0.019
0.248
0.055
-0.052
-0.006
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-1.360
0.469
-5.883
0.106
-0.018
0.086
0.039
-0.029
0.149
0.083
0.226
-0.078
1.404
-0.097
0.693
-1.122
-0.052
-0.184
-0.002
-0.175
-0.438
0.093
-0.149
0.174
-7.451
2.276
-0.892
0.880
-1.543
-0.781
0.185
-1.186
-1.755
-0.598
-2.195
1.689
-0.375
0.494
1.176
0.165
0.040
-1.956
-0.023
-0.182
-0.101
0.099
-0.052
0.226
-12.467
5.652
6.663
3.179
0.194
-3.426
-1.044
-3.862
-4.570
-0.047
-2.067
2.683
1.375
1.673
-0.208
0.771
0.002
-0.522
-0.055
-2.028
-0.810
-0.126
-0.208
0.468
-0.546
0.224
1.619
-0.043
-0.003
0.031
0.032
-0.098
-3.149
0.155
-0.119
0.283
0.088
0.050
0.584
0.216
-0.085
-0.069
-0.003
-0.176
-0.097
-1.170
0.049
0.172
-0.002
1.172
2.269
0.132
-0.241
-0.094
0.008
-0.130
-0.294
0.194
-2.635
0.708
-4.081
0.332
-2.892
0.840
0.365
-0.417
0.041
-0.794
-0.454
0.317
1.473
-1.874
Urban (N = 1,386)
Electricity
-4.184
-0.009
-0.536
0.554
-2.468
-0.047
-2.627
0.492
-0.200
-0.024
-0.045
-2.571
Kerosene
0.595
-1.707
0.811
-0.343
3.229
0.534
5.452
2.022
0.332
0.127
1.604
2.389
Charcoal
-0.764
0.061
-1.983
0.129
-0.323
0.214
1.032
-0.125
0.301
0.098
0.429
-0.729
Firewood
2.541
-0.054
0.483
-1.523
2.599
0.302
5.993
1.798
-0.037
0.626
0.410
4.429
Cereals
-0.094
0.002
-0.023
-0.044
-1.527
0.039
0.215
0.002
-0.005
-0.083
-0.230
0.281
Fruits
0.023
-0.003
0.024
-0.129
-1.058
-2.198
-3.324
-0.540
0.037
-0.112
-0.135
-1.469
Vegetables
0.107
0.001
0.064
-0.004
0.330
-0.025
-0.977
-0.043
0.055
0.006
0.024
0.177
Meat & Fish
0.209
-0.007
-0.127
-0.084
-1.403
-0.135
-3.205
-2.032
-0.122
-0.225
-0.164
-1.867
Milk & Eggs
-0.183
-0.022
0.025
-0.275
-1.744
-0.015
-2.947
-0.619
-3.026
-0.126
-0.300
-1.607
Cheese, Oils & Fats
0.278
-0.017
0.071
0.074
-0.535
0.078
-0.129
-0.127
0.128
-1.148
0.161
0.267
Alcohol & Tobacco
0.145
0.052
0.159
-0.083
-2.061
-0.017
-1.375
-0.141
-0.118
0.029
-2.507
0.854
Other Food & Drink
0.007
-0.005
-0.015
0.038
0.646
0.065
0.642
0.109
0.110
0.069
0.294
-1.222
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me),
Milk & Eggs (Mi), Cheese, Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Elasticities are computed based on coefficients
estimated for the full sample (Table A2), combined with tercile-rural/urban specific mean budget shares and mean real expenditure on food & fuel.
48
Table A6.3: Uncompensated price elasticities of demand: Expenditure Tercile - III
E
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
K
Rural (N = 2,405)
-4.467
-0.006
0.784
-1.950
-3.185
0.266
1.311
-0.032
-0.097
0.002
0.011
-0.001
0.081
0.000
0.156
-0.003
-0.137
-0.014
0.362
-0.021
0.163
0.058
0.050
-0.010
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-0.458
0.849
-5.180
0.145
-0.019
0.069
0.044
-0.029
0.100
0.091
0.219
-0.074
0.499
-0.248
0.589
-1.190
-0.050
-0.136
-0.001
-0.124
-0.283
0.101
-0.127
0.149
-2.662
4.360
-0.776
1.276
-1.539
-0.726
0.238
-1.077
-1.350
-0.621
-2.181
1.710
-0.232
1.069
1.008
0.296
0.039
-1.857
-0.017
-0.206
-0.138
0.114
-0.082
0.264
-4.093
9.815
5.645
4.023
0.228
-2.815
-0.984
-3.155
-3.144
-0.001
-1.895
2.507
0.063
3.665
-0.175
1.295
-0.002
-0.603
-0.031
-2.058
-0.814
-0.109
-0.318
0.610
-0.341
0.692
1.390
0.071
-0.006
-0.043
0.045
-0.165
-2.625
0.177
-0.169
0.346
0.023
0.092
0.498
0.281
-0.084
-0.058
-0.001
-0.148
-0.069
-1.182
0.051
0.167
-0.041
2.143
1.941
0.199
-0.241
-0.093
0.014
-0.126
-0.230
0.214
-2.610
0.708
-1.475
0.801
-2.481
1.177
0.367
-0.398
0.063
-0.723
-0.392
0.357
1.420
-1.824
Urban (N = 2,115)
Electricity
-2.884
0.000
-0.296
0.316
-1.401
-0.094
-1.420
-0.018
-0.228
-0.013
-0.044
-1.392
Kerosene
2.426
-2.620
1.867
-0.902
7.555
1.593
11.959 6.148
1.458
0.259
3.699
5.239
Charcoal
-0.793
0.062
-2.001
0.134
-0.316
0.214
1.090
-0.143
0.298
0.102
0.438
-0.720
Firewood
5.562
-0.124
0.903
-2.065
4.879
0.893
10.386 4.664
0.575
1.095
0.810
7.773
Cereals
-0.107
0.002
-0.023
-0.044
-1.536
0.040
0.272
-0.002
-0.009
-0.084
-0.239
0.324
Fruits
-0.161
0.003
0.012
-0.076
-0.844
-1.973
-2.459
-0.670
-0.088
-0.080
-0.116
-1.085
Vegetables
0.154
0.000
0.075
-0.007
0.407
-0.014
-0.931
0.009
0.085
0.009
0.033
0.228
Meat & Fish
-0.025
0.000
-0.104
-0.041
-1.115
-0.169
-2.385
-2.060
-0.214
-0.167
-0.140
-1.396
Milk & Eggs
-0.312
-0.009
0.008
-0.154
-1.183
-0.080
-1.861
-0.686
-2.428
-0.078
-0.211
-1.022
Cheese, Oils & Fats
0.329
-0.019
0.084
0.084
-0.571
0.096
-0.087
-0.113
0.156
-1.163
0.189
0.340
Alcohol & Tobacco
0.036
0.056
0.160
-0.068
-2.083
-0.054
-1.287
-0.294
-0.190
0.037
-2.532
0.926
Other Food & Drink
0.073
-0.007
-0.013
0.036
0.758
0.097
0.736
0.223
0.166
0.078
0.337
-1.235
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me),
Milk & Eggs (Mi), Cheese, Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Elasticities are computed based on coefficients
estimated for the full sample (Table A2), combined with tercile-rural/urban specific mean budget shares and mean real expenditure on food & fuel.
49
Table A7: Matching of UNHS Consumption Items with GTAP Sectors
UNHS Item Code
101
102
103
107
109
114
116
117
119
120
125
126
128
129
130
132
133
134
135
136
138
139
140
141
142
143
144
145
147
150
151
152
153
154
155
156
157
158
159
UNHS Item
Matooke (Bunch)
Matooke (Cluster)
Matooke (Heap)
Cassava (Fresh)
Irish Potatoes
Bread (wheat)
Sorghum
Beef
Goat Meat
Other Meat (eg duck, rabbit etc)
Fresh Milk
Infant Formula Foods
Ghee
Margarine
Passion Fruits
Mangoes
Oranges/Tangerines
Other Fruits
Onions
Tomatoes
Dodo/Nakati/gyobyo/Malakwang
Other vegetables
Bean( fresh)
Beans (dry)
Ground nuts (in shell)
Ground nuts (shelled)
Ground nuts (pounded)
Peas(fresh)
Sugar
Salt
Soda
Beer
Other Alcoholic drinks
Other drinks
Cigarrates
Other Tobacco
Food
in Restaurants
Soda in Restaurants
Beer Restaurants
GTAP Sector
v_f
v_f
v_f
v_f
v_f
ofd
ofd
cmt
cmt
omt
mil
ofd
mil
vol
v_f
v_f
v_f
v_f
v_f
v_f
v_f
v_f
v_f
ofd
v_f
v_f
v_f
v_f
sgr
ocr
b_t
b_t
b_t
b_t
b_t
b_t
ofd
b_t
b_t
GTAP Code
4
4
4
4
4
25
25
19
19
20
22
25
22
21
4
4
4
4
4
4
4
4
4
25
4
4
4
4
24
8
26
26
26
26
26
26
25
26
26
50
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
306
308
309
310
1051
1052
1061
1062
1063
1081
1082
1083
1101
1102
1111
1112
1121
1122
1131
1132
1151
1171
1172
1191
1192
1193
Other juice Packed in Restaurant
Other foods in Restaurants
Peas(dry)
Ground nuts (paste)
Green Pepper
Pumpkins
Avocado
Carrots
Egg plants
Watermelon
Pineapple
Pawpaw
Wheat (flour)
Chapati
Apples
Water
Electricity
Paraffin or kerosene
Charcoal
Firewood
Sweet Potatoes white/yellow(Fres
Sweet Potatoes-orange fleshed (f
Sweet Potatoes white/yellow (Dry
Sweet Potatoes-orange (Dry)
Sweet Potatoes white/yellow (flo
Cassava (Dry)
Cassava (flour)
Pancakes(Kabalagala)
Rice (white)
Rice (brown)
Maize yellow (grains)
Maize white (grains)
Maize white (cobs)
Maize yellow (cobs)
Maize white (flour)
Maize yellow (flour)
Millet flour
Beef Liver
Beef Offals
Goat Liver
Goat offals
Roasted goat meat
b_t
ofd
ofd
ofd
v_f
v_f
v_f
v_f
v_f
v_f
v_f
v_f
wht
ofd
v_f
b_t
ely
p_c
NA
NA
v_f
v_f
v_f
v_f
v_f
v_f
v_f
v_f
pdr
pdr
gro
gro
gro
gro
ofd
ofd
ofd
cmt
cmt
cmt
cmt
cmt
26
25
25
25
4
4
4
4
4
4
4
4
2
25
4
26
46
32
0
0
4
4
4
4
4
4
4
4
1
1
3
3
3
3
25
25
25
19
19
19
19
19
51
1201
1211
1212
1213
1214
1215
1221
1222
1231
1232
1234
1235
1236
1237
1241
1242
1243
1251
1252
1253
1254
1271
1272
1281
1291
1311
1312
1313
1351
1352
1353
1371
1372
1391
1461
1462
1471
1472
1481
1482
1491
1492
Roasted other meat
Chicken off-layer
Chicken Broiler
Chicken Kroiler
Chicken Local
Roasted Chicken
Fresh tilapia Fish
Fresh Nile perch
Dry/ Smoked tilapia fish
Dry/Smoked Nile perch
Dried Nkejje
Other fresh fish
Other dry/smoked fish
Silver Fish (Mukene)
Eggs (yellow yolk)
Eggs (white yolk)
Other eggs (duck, turkey etc)
Milk Powdered
Fermented milk (Bongo)
Ice-cream
Yoghurt
Cooking oil refined
Cooking oil unrefined
Cheese
Butter
Sweet Bananas-Ndiizi
Sweet Bananas-Bogoya
Plantain (gonja/kivuvu)
Garlic
Ginger fresh
Ginger powder
Cabbages – Red leaf
Cabbage – green leaf
Other spices
Simsim
Simsim paste
Honey
Jam/ Mamalede
Coffee
Coffee Other
Tea leaves
Tea bags
omt
oap
oap
oap
oap
oap
ofd
ofd
ofd
ofd
ofd
ofd
ofd
ofd
oap
oap
oap
mil
mil
mil
mil
vol
vol
mil
mil
v_f
v_f
v_f
v_f
v_f
ocr
v_f
v_f
ocr
osd
osd
oap
ofd
b_t
ocr
ocr
b_t
20
10
10
10
10
10
25
25
25
25
25
25
25
25
10
10
10
22
22
22
22
21
21
22
22
4
4
4
4
4
8
4
4
8
5
5
10
25
26
8
8
26
52
1493
1601
1602
1603
1651
1652
1653
1654
1721
1731
1732
1733
1734
1735
1741
1761
1762
Green tea
Other juice fresh
Other juice packed
Other juice Fresh in Restaurants
Pumpkin Leaves
Mushrooms
Cucumber
Okra
Macaroni/Spaghetti
Biscuits
Cakes
Doughnuts
Cornflakes
Samosas
Jackfruit (ffene)
Soya beans (fresh)
Soya beans (dry)
b_t
b_t
b_t
b_t
v_f
v_f
v_f
v_f
ofd
ofd
ofd
ofd
ofd
ofd
v_f
v_f
ofd
26
26
26
26
4
4
4
4
25
25
25
25
25
25
4
4
25
Figure A2. Scatterplot of item-wise price increases and expenditure shares
Note: The markers in the scatterplot correspond to the 12 items modelled in the demand system.
53
Robustness Checks for demand system estimation
We conduct two robustness checks of the demand system. First, we expand the original model
specification by controlling for sub-regional heterogeneity, and use instrumental variables for
commodity prices and sub-region dummy variables. Given the importance of transportation in
the supply chain for food items and energy sources (MEMD, 2015), we construct instrumental
variables for prices to mitigate the influence of transportation costs in final commodity prices,
which may result in simultaneous increases of several commodity prices, leading to a bad
control problem (Angrist and Pischke 2008).
We first conduct regressions of the pass-through of transport to commodity prices (all energy22
and food items), constructing a district-level panel dataset of commodity prices from the UNPS
for 2009-2020. We draw on monthly international diesel prices to proxy for local transportation
costs. All coefficients of the log of commodity prices on the log of diesel prices are positive
and statistically significant, reflecting a sizeable effect of transport costs on commodity prices.
We then construct instruments for food and energy prices as follows:
ln 𝑝𝑖𝑣 = ln 𝑝𝑖 − 𝛽𝑖 ln 𝑑
where 𝛽𝑖 is the regression coefficient of the price pass-through for commodity 𝑖, and ln 𝑑 is the
international price of diesel (converted to Ugandan shillings) for the respective month. The 12
constructed commodity prices ln 𝑝𝑖𝑣 and the price of diesel are then used as instruments for the
original prices. We additionally introduce sub-region dummy variables as instruments to
capture geographical heterogeneity in demand responses. The estimated compensated price
elasticities of demand are presented in Table A8. While the broad patterns of substitution are
similar to those obtained from the Uganda National Household Survey 2016-17, the model
22
Electricity tariff schedules set by the Electricity Regulatory Authority in Uganda are also influenced by
international fuel prices and the nominal exchange rate.
54
does not meet the criterion for over-identification as Hansen’s J-statistic is large, and we reject the null hypothesis of all instruments being valid.
Table A8: Robustness Check I – Compensated price elasticities of demand
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
E
K
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-4.537***
(0.398)
-1.410***
(0.338)
-1.970***
(0.26)
0.762***
(0.124)
-0.129***
(0.021)
0.298***
(0.062)
0.063***
(0.007)
0.586***
(0.044)
0.171*
(0.092)
0.512***
(0.055)
0.555***
(0.127)
-0.293***
(0.058)
-0.138***
(0.033)
-1.911***
(0.145)
0.090**
(0.043)
0.180***
(0.02)
0.013***
(0.002)
-0.006
(0.005)
-0.001
(0.001)
0.006**
(0.003)
-0.039***
(0.01)
-0.021***
(0.006)
0.132***
(0.011)
-0.034***
(0.004)
-0.630***
(0.083)
0.294**
(0.141)
-3.399***
(0.201)
0.337***
(0.031)
-0.022***
(0.006)
0.077***
(0.018)
0.062***
(0.002)
-0.062***
(0.006)
0.098***
(0.029)
0.146***
(0.018)
0.128***
(0.019)
-0.056***
(0.012)
0.772***
(0.126)
1.859***
(0.203)
1.067***
(0.098)
-1.356***
(0.067)
-0.088***
(0.007)
-0.065***
(0.019)
0.016***
(0.003)
0.002
(0.008)
-0.293***
(0.029)
0.047**
(0.02)
0.044
(0.087)
-0.013
(0.038)
-1.172***
(0.193)
1.202***
(0.185)
-0.629***
(0.163)
-0.790***
(0.06)
-1.480***
(0.033)
0.589***
(0.074)
0.159***
(0.01)
0.050*
(0.03)
-0.019
(0.08)
-0.500***
(0.044)
-2.095***
(0.103)
0.805***
(0.04)
0.452***
(0.095)
-0.097
(0.08)
0.365***
(0.086)
-0.098***
(0.028)
0.099***
(0.012)
-1.936***
(0.048)
-0.047***
(0.004)
0.029**
(0.013)
0.284***
(0.047)
0.165***
(0.023)
0.176***
(0.061)
0.012
(0.025)
1.367***
(0.162)
-0.164
(0.125)
4.214***
(0.166)
0.336***
(0.058)
0.382***
(0.024)
-0.674***
(0.059)
-1.163***
(0.013)
-0.375***
(0.029)
0.211***
(0.071)
-0.509***
(0.035)
0.845***
(0.104)
-0.130***
(0.038)
3.263***
(0.244)
0.364**
(0.183)
-1.073***
(0.099)
0.009
(0.041)
0.031*
(0.019)
0.106**
(0.046)
-0.096***
(0.007)
-1.163***
(0.028)
0.493***
(0.074)
-0.300***
(0.03)
0.163**
(0.081)
-0.150***
(0.037)
0.241*
(0.13)
-0.564***
(0.143)
0.431***
(0.127)
-0.406***
(0.04)
-0.003
(0.012)
0.263***
(0.044)
0.014***
(0.005)
0.125***
(0.019)
-2.782***
(0.095)
0.232***
(0.033)
0.002
(0.06)
0.143***
(0.026)
0.549***
(0.059)
-0.225***
(0.064)
0.489***
(0.059)
0.050**
(0.021)
-0.059***
(0.005)
0.117***
(0.017)
-0.025***
(0.002)
-0.058***
(0.006)
0.177***
(0.025)
-1.103***
(0.017)
0.052**
(0.021)
0.040***
(0.01)
0.846***
(0.193)
2.059***
(0.167)
0.613***
(0.092)
0.066
(0.131)
-0.352***
(0.017)
0.176***
(0.062)
0.059***
(0.007)
0.045**
(0.022)
0.002
(0.066)
0.074**
(0.03)
-2.345***
(0.093)
0.298***
(0.038)
-2.013***
(0.398)
-2.406***
(0.303)
-1.197***
(0.252)
-0.09
(0.259)
0.610***
(0.03)
0.056
(0.115)
-0.041***
(0.012)
-0.185***
(0.045)
0.697***
(0.129)
0.255***
(0.067)
1.343***
(0.169)
-1.622***
(0.109)
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me), Milk & Eggs (Mi), Cheese,
Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Robust standard errors in parentheses. The Hansen J chi-sq. statistic for over-identification is
5693.11 (p = 0.0000). N = 13,582. *** p < 0.01, ** p < 0.05, * p < 0.1.
55
Second, we estimate elasticities using an alternative pooled cross-sectional dataset – the
Uganda National Panel Survey, over the 2009-10 to 2019-20 period, for both expenditure and
price data. After excluding outliers and observations with missing price and expenditure data,
we obtain a total pooled cross-sectional sample of 15,950 households. We additionally control
for time effects, using year dummy variables as instruments. The estimated compensated price
elasticities of demand are presented in Table A9. Results are broadly similar to those obtained
using the Uganda National Household Survey 2016-17. In particular, the patterns of
substitution between electricity and firewood, as well as between kerosene, charcoal and
firewood, corroborate those obtained from the cross-sectional UNHS dataset.
56
Table A9: Robustness Check II – Compensated price elasticities of demand
Electricity
Kerosene
Charcoal
Firewood
Cereals
Fruits
Vegetables
Meat & Fish
Milk & Eggs
Cheese, Oils & Fats
Alcohol & Tobacco
Other Food & Drink
E
K
C
Fw
Ce
Fr
V
Me
Mi
Ch
Al
O
-4.344***
(0.242)
-0.762***
(0.044)
-0.437***
(0.032)
0.422***
(0.034)
0.338***
(0.03)
0.182***
(0.024)
-0.177***
(0.031)
-0.276***
(0.033)
-0.018
(0.036)
0.109***
(0.012)
0.246***
(0.035)
0.401***
(0.077)
-0.280***
(0.016)
-0.939***
(0.022)
0.009***
(0.002)
-0.010*
(0.006)
0.022***
(0.004)
0.008***
(0.003)
0.003
(0.004)
-0.049***
(0.006)
0.009*
(0.005)
0.006***
(0.001)
0.008**
(0.004)
0.052***
(0.005)
-0.585***
(0.043)
0.032***
(0.007)
-0.931***
(0.008)
-0.086***
(0.007)
0.018***
(0.006)
-0.030***
(0.005)
0.012*
(0.007)
0.032***
(0.007)
0.007
(0.008)
0.004
(0.002)
0.009
(0.007)
0.273***
(0.018)
0.628***
(0.051)
-0.042*
(0.025)
-0.096***
(0.007)
-0.910***
(0.023)
-0.143***
(0.012)
-0.022***
(0.008)
0.028**
(0.013)
0.064***
(0.016)
-0.082***
(0.015)
0.007*
(0.003)
-0.024*
(0.013)
-0.031**
(0.015)
1.109***
(0.099)
0.200***
(0.033)
0.045***
(0.014)
-0.316***
(0.026)
-1.195***
(0.029)
0.129***
(0.017)
0.139***
(0.026)
-0.230***
(0.027)
0.157***
(0.027)
-0.014*
(0.008)
-0.060**
(0.026)
-0.293***
(0.035)
0.656***
(0.087)
0.074***
(0.028)
-0.081***
(0.013)
-0.054***
(0.02)
0.141***
(0.019)
-1.178***
(0.021)
0.080***
(0.023)
-0.176***
(0.021)
-0.243***
(0.023)
0.021***
(0.007)
-0.117***
(0.021)
-0.006
(0.034)
-0.251***
(0.045)
0.011
(0.015)
0.013*
(0.008)
0.027**
(0.012)
0.060***
(0.011)
0.032***
(0.009)
-1.022**
(0.019)
0.204***
(0.013)
0.030**
(0.014)
-0.045***
(0.004)
0.030**
(0.014)
0.045**
(0.019)
-0.156***
(0.019)
-0.076***
(0.009)
0.013***
(0.003)
0.024***
(0.006)
-0.040***
(0.005)
-0.028***
(0.003)
0.081***
(0.005)
-0.965***
(0.008)
-0.025***
(0.006)
0.011***
(0.001)
-0.001
(0.005)
0.078***
(0.007)
-0.021
(0.041)
0.027*
(0.015)
0.006
(0.007)
-0.062***
(0.011)
0.054***
(0.009)
-0.076***
(0.007)
0.024**
(0.011)
-0.051***
(0.011)
-0.981***
(0.017)
-0.017***
(0.003)
-0.004
(0.01)
0.157***
(0.015)
0.946***
(0.104)
0.138***
(0.026)
0.024
(0.016)
0.040*
(0.02)
-0.036*
(0.02)
0.050***
(0.016)
-0.273***
(0.024)
0.164***
(0.022)
-0.133***
(0.024)
-1.033***
(0.011)
0.208***
(0.025)
-0.394***
(0.043)
0.467***
(0.066)
0.041**
(0.021)
0.013
(0.01)
-0.030*
(0.017)
-0.035**
(0.015)
-0.062***
(0.011)
0.040**
(0.019)
-0.003
(0.016)
-0.007
(0.017)
0.046***
(0.005)
-1.035***
(0.022)
-0.238***
(0.028)
0.831***
(0.16)
0.294***
(0.027)
0.422***
(0.028)
-0.043**
(0.021)
-0.185**
(0.022)
-0.003
(0.02)
0.066**
(0.028)
0.287***
(0.024)
0.287***
(0.028)
-0.094***
(0.01)
-0.260***
(0.031)
-1.043***
(0.086)
Note: The items are - Electricity (E), Kerosene (K), Charcoal (C), Firewood (Fw), Cereals (Ce), Fruits (Fr), Vegetables (V), Meat & Fish (Me), Milk & Eggs (Mi), Cheese,
Oils & Fats (Ch), Alcohol & Tobacco (Al) and Other food and drink (O). Robust standard errors in parentheses. As the model includes year dummy variables as instruments,
we compute Hansen’s J chi-sq statistic to test for overidentifying restrictions. The Hansen J chi-sq. statistic for over-identification is 3020.99 (p = 0.0000). The model does
not meet the criterion for over-identification as Hansen’s J-statistic is large. N = 15,949. *** p < 0.01, ** p < 0.05, * p < 0.1.
57