## Annex 1. Supplementary Tables

## Source details

**Canonical URL:** [Annex 1. Supplementary Tables](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15250.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15250.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15250.pdf.json)

---

### Introduction and purpose
- Fuel subsidies exacerbate income inequalities and are not a cost-effective approach to protecting the poor.
- Two channels by which subsidies worsen inequality:
  - Benefits distributed in proportion to household energy consumption; higher-income households typically have more energy-intensive consumption baskets.
  - Public expenditures on energy subsidies can crowd out more redistributive public spending or require financing through regressive taxation.
- Governments reluctant to remove subsidies because higher fuel prices reduce household real incomes, especially for lower-income households.
- Paper updates cross-country evidence on welfare impact of fuel price increases by extending Arze del Granado, Coady, and Gillingham (2012) from 20 countries (2005–09) to 32 countries up to 2014.
- Objective: quantify magnitude and distribution of income losses from fuel price increases to inform design of more cost-effective protection for vulnerable groups.

### Trends in international oil prices and domestic pass-through
- International oil prices rose steadily after the mid-2000s and surged in 2007–first half 2008; collapsed within six months after mid-2008 peak; rebounded by April 2011; declined in late 2014–early 2015 offering a policy window.
- Incomplete pass-through raised subsidies substantially; combined cost for diesel, gasoline, and kerosene exceeded 3 percent of GDP on an annualized basis in more than half of developing countries (when these fuels alone are considered).
- Summary statistics (median pass-through Percent for diesel by period):
  - Developing Economies {118}: End-2004–Mid-2008 = 66; Mid-2008–End-2008 = 22; End-2008–End-2012 = 49; End-2013–End-2014 = 11
  - Developing Asia {28}: 65; 12; 54; 40
  - Latin America and the Caribbean {32}: 51; 28; 83; 12
  - Middle East and North Africa {19}: 13; 1; 10; 0
  - Sub-Saharan Africa {39}: 85; 38; 35; 0
- Median fiscal cost (Percent of GDP) for gasoline, diesel, and kerosene:
  - Developing Economies {118}: End-2004–Mid-2008 = 1.92; Mid-2008–End-2008 = −3.14; End-2008–End-2012 = 1.74; End-2013–End-2014 = −0.88
  - Developing Asia {28}: 1.43; −4.68; 2.24; −0.89
  - Latin America and the Caribbean {32}: 2.50; −2.98; 1.27; −0.80
  - Middle East and North Africa {19}: 1.93; −3.44; 3.20; −1.89
  - Sub-Saharan Africa {39}: 2.15; −3.42; 1.50; −0.92
- Median net-tax (US$/liter) by period (End-2004, Mid-2008, End-2008, End-2012, End-2014):
  - Developing Economies {118}: 0.10; −0.08; 0.33; 0.22; 0.37
  - Developing Asia {28}: −0.11; −0.37; 0.17; 0.05; 0.16
  - Latin America and the Caribbean {32}: 0.09; −0.08; 0.29; 0.22; 0.29
  - Middle East and North Africa {19}: −0.14; −0.78; −0.12; −0.36; 0.07
  - Sub-Saharan Africa {39}: 0.27; 0.14; 0.54; 0.32; 0.47
- After mid-2008 collapse many countries passed through little of the decline (pass-through fell below 20 percent in more than half of developing countries; below 1 percent in most of the Middle East and North Africa), increasing net taxes and reducing subsidies temporarily.

### Methodology (estimating welfare impact)
- Total household welfare impact = direct impact + indirect impact.
- Direct impact:
  - Higher prices for fuels consumed for cooking, heating, lighting, and personal transport.
  - Computed from household survey data on expenditures as budget shares; budget share provides estimate of welfare impact of a doubling of fuel price absent demand response (short-term impact or upper bound).
- Indirect impact:
  - Higher prices for other goods and services as fuel cost increases feed into production costs and consumer prices.
  - Estimated using price-shifting models (based on Coady and Newhouse (2006)) assuming increases in fuel production costs are fully passed forward to domestic prices.
  - Requires input-output tables or production-structure data; outdated input-output data can cause under- or overestimation.
  - Key modeling assumptions: goods are nontraded; domestic production technologies exhibit constant returns to scale; demand is completely price inelastic — implying short-term impact or upper bound.
  - Indirect impacts computed by estimating price increases for consumption categories and multiplying by household budget shares; aggregated to household indirect welfare impact.
- Distribution:
  - Percentage impacts averaged across household consumption quintiles (based on consumption per capita or per adult equivalent).
  - Distribution classified as progressive if percentage welfare loss increases with household consumption; regressive if it decreases.

### Welfare impact of a uniform $0.25 per liter price increase (magnitude, composition, and distribution)
- Method: country studies recalculated to a common price increase of $0.25 per liter by scaling welfare impacts proportionally; electricity tariff changes not included in new recalculations due to lack of data.
- Key aggregated results:
  - Average total (direct plus indirect) impact of a $0.25 per liter increase in fuel prices: 5.5 percent decline in household real incomes (authors’ calculations based on reviewed studies).
  - Regional range of total impact:
    - South and Central America: 3.5 percent decline in household real incomes.
    - Middle East: 7.0 percent decline in household real incomes.
    - All Regions average: 5.5 percent.
  - Indirect impacts account for a substantial share of total impact (approximately half on average).
- Drivers of regional differences:
  - Higher simulated impacts in Middle East and Central Asia linked to comparatively low retail fuel prices — a given $0.25 per liter increase represents a larger percentage price increase where retail prices are low.

### Indirect versus direct impacts (Ratio of Fuel Consumption to GDP)
- On average, the indirect impact accounts for 55 percent of the total impact.
- Regional shares of the indirect effect:
  - sub-Saharan Africa: 65 percent of the total impact
  - South and Central America: 60 percent of the total impact
  - Asia and Pacific: less than 50 percent of the total impact
  - Middle East and Central Asia: less than 50 percent of the total impact
- Composition of indirect impact (budget-share weighted averages across consumption categories):
  - Budget share (Percent of household total consumption): Food 49.2; Transport 4.1; Others 43.9
  - Price effect (Percentage increase): Food 0.03; Transport 0.10; Others 0.04
  - Indirect impact (Percent of household total consumption): Food 1.4; Transport 0.4; Others 1.6
- Interpretation:
  - “Others” has a slightly lower budget share than food but a greater fuel intensity, producing a higher price effect; on average “Others” account for slightly less than 50 percent of the indirect impact and food about 40 percent.
  - Transport absorbs only 4.1 percent of household budgets on average but accounts for about 12 percent of the indirect impact because of its relatively high energy intensity.

### Direct impacts and fuel-specific patterns
- Fuel composition differs across regions:
  - Kerosene relatively important in Africa due to low household access to electricity.
  - Kerosene much less important in the Middle East and in South and Central America where access to electricity is more extensive.
- Access caveat: if only half of households have access to a fuel (e.g., electricity or LPG), the impact on users will be double the average impact reported for all households—estimates can substantially underestimate impacts on users.

### Distributional patterns of welfare impacts
- Overall distribution:
  - Total, direct, and indirect welfare impacts are approximately distributionally neutral: percentage decrease in welfare is very similar across income groups.
  - A $0.25 per liter increase in fuel prices decreases household real incomes by, on average, 5.5 percent.
  - Approximately half of this impact comes through the indirect effect on prices of other goods and services.
- Fuel-specific distribution:
  - Gasoline and electricity impacts are strongly progressive.
  - Kerosene impact is strongly regressive.
  - LPG impact varies by region: slightly progressive on average but very regressive in the Middle East and Central Asia.
- Quintile-level indicators shown in figures indicate totals around 5.4–5.7 percent by quintile in example charts.

### Subsidy leakages and cost-effectiveness
- Universal fuel subsidies are badly targeted; higher total consumption of well-off households causes substantial leakage of benefits to higher-income groups.
- On average:
  - The top income quintile receives more than six times more in total subsidies than the bottom quintile.
  - For gasoline and LPG, top quintile receives:
    - Gasoline: 27 times the bottom quintile
    - LPG: 12 times the bottom quintile
- Aggregate concentration:
  - The richest 20 percent capture, on average, more than six times more in fuel subsidies than the poorest 20 percent.
  - More than 83 percent of gasoline subsidy benefits accrue to the richest 40 percent of households.
- Cost of providing $1 to the poorest 40 percent via fuel subsidies:
  - Using gasoline subsidies, the cost to the budget of transferring one dollar to the poorest 40 percent is about 14 dollars (nearly 93 out of every 100 dollars of gasoline subsidy “leaks” to the top three quintiles).
  - For kerosene the cost-benefit ratio is about 3 dollars.
- Comparison with a safety-net program:
  - If 15 out of every 100 dollars allocated to a safety net program is absorbed by administrative costs and 80 percent of the remaining 85 dollars in beneficiary transfers reaches the poor (i.e., 68 percent of the total budget), then the cost-benefit ratio for such a program is 1.5 dollars — substantially lower than via fuel subsidies.

### Implications for electricity pricing
- Nonlinear electricity pricing (block tariffs with a lower lifeline tariff) can mitigate impacts but often perform worse than believed:
  - Many poor households do not benefit from lifeline tariffs because they lack access to electricity.
  - Many poor households with larger family sizes consume above lifeline thresholds and thus do not benefit.
  - Financing lifeline subsidies via higher tariffs on larger consumers can make large-consuming poor households worse off.
  - Conclusion: nonlinear pricing is a very crude protection mechanism with important negative side effects.

### Protecting low-income households — policy recommendations
- Key objective: implement well-targeted measures to mitigate impact of energy price increases on the poor to build public support for subsidy reform.
- Preferred measures:
  - Targeted cash transfers or near-cash transfers:
    - Provide beneficiary flexibility and remove costly, abuse-prone government distribution of subsidized energy.
    - Examples cited: Armenia’s targeted cash transfer in electricity reform; Indonesia’s nonconditional cash transfer covering 35 percent of the population.
  - Conditional cash transfers:
    - Link benefits to investments in education and health to protect incomes and build human capital.
- If cash transfers infeasible:
  - Expand existing programs that can be scaled quickly with improved targeting: school meals, public works, reductions in education and health user fees, subsidized mass urban transport, subsidies for water and electricity connection costs.
  - Consider temporarily delaying price increases for products more important to the poor (for example kerosene).
- Electricity-specific measures:
  - Maintain lower lifeline tariffs but recognize their limits; subsidize connection costs instead of consumption where appropriate (example: Kenya).
- Design and sequencing of reform:
  - Develop a comprehensive reform plan with clear objectives, measures, and timelines.
  - Develop an effective communication strategy with transparency.
  - Appropriately phase and sequence price increases (weigh gradualism versus short-term savings and distortion risks).
- Improve efficiency of energy state-owned enterprises (SOEs) to reduce fiscal burdens and signal reforms are not protecting inefficiency.
- Depoliticize energy pricing:
  - Consider automatic pricing formulas linking domestic prices to international prices with built-in price smoothing to share international volatility between consumers and the budget and avoid sudden large domestic price jumps.
  - Price-smoothing rules trade off lower price volatility against higher fiscal volatility.

### Data coverage, adjustment factors, and selected table highlights
- Data coverage (Annex Table 1.1) includes Year of Report; Year of Household Survey; Year of Input-Output Matrix; Author Institution. Selected entries:
  - Angola — Year of Report: 2014; Year of Household Survey: 2008/2009; Year of Input-Output Matrix: 2014; Author Institution: IMF
  - Cameroon — Year of Report: 2007; Year of Household Survey: 2001; Year of Input-Output Matrix: 2001; Author Institution: IMF
  - Ghana — Year of Report: 2005; Year of Household Survey: 1999; Year of Input-Output Matrix: 1993; Author Institution: IMF
  - India — Year of Report: 2013; Year of Household Survey: 2010; Year of Input-Output Matrix: 2004; Author Institution: IMF
  - Indonesia — Year of Report: 2014; Year of Household Survey: 2012; Year of Input-Output Matrix: 2011; Author Institution: World Bank
- Adjustment factors by country and fuel product (Annex Table 1.2) — selected exact values:
  - Angola — Gasoline: 0.32; Kerosene: 0.69; Diesel: 0.48; LPG: 0.54
  - Bangladesh — Gasoline: 0.54; Kerosene: 0.86; Diesel: 0.86; LPG: 0.70
  - Ghana — Gasoline: 0.96; Kerosene: 1.42; Diesel: 1.02; LPG: 0.99
  - India (two studies) — Study 1: Gasoline: 0.28; Kerosene: 1.23; Diesel: 0.46; LPG: 0.37; Study 2: Gasoline: 0.20; Kerosene: 0.92; Diesel: 0.29; LPG: 0.03
  - Jordan (two studies) — Study 1: Gasoline: 0.53; Kerosene: 1.25; Diesel: 1.29; LPG: 0.91; Study 2: Gasoline: 0.25; Kerosene: 0.34; Diesel: 0.34; LPG: 0.33
  - Peru — Gasoline: 0.27; Kerosene: 0.27; Diesel: 0.27; LPG: 0.27
- Direct and indirect welfare impacts (Annex Table 1.3) — regional averages (Percent of household total consumption):
  - Africa (average) — Gasoline: 0.3; Kerosene: 0.9; LPG: 0.2; Electricity: 0.4; Indirect: 1.7; Total: 5.0
  - South and Central America (average) — Gasoline: 0.2; Kerosene: 0.2; LPG: 0.3; Electricity: 0.8; Indirect: 1.4; Total: 3.5
  - Asia and Pacific (average) — Gasoline: 0.6; Kerosene: 1.0; LPG: 0.4; Electricity: 1.3; Indirect: 3.2; Total: 5.9
  - Middle East and Central Asia (average) — Gasoline: 0.9; Kerosene: 0.4; LPG: 0.9; Electricity: 1.4; Indirect: 3.5; Total: 7.0
  - All regions (average) — Gasoline: 0.4; Kerosene: 0.8; LPG: 0.4; Electricity: 0.9; Indirect: 2.5; Total: 5.5
- Selected country-specific total impacts (Percent of household total consumption):
  - Cameroon — Total: 3.6
  - Gabon — Total: 5.6
  - Central African Republic — Total: 3.5
  - Senegal — Total: 2.1
  - Ghana — Total: 17.3
  - Congo, Republic of — Total: 9.0
  - Angola — Total: 4.2
  - South Africa — Total: 0.9
  - Bolivia — Total: 3.6
  - Peru — Total: 1.7
  - Honduras — Total: 5.0
  - Bangladesh — Total: 3.2 (one study) and 0.3 (another study)
  - India — Total: 3.6 (one study) and 2.5 (another study)
  - Indonesia — Total: 8.8
  - Malaysia — Total: 8.5
  - Jordan — Totals: 12.1 and 5.8 (two studies)
  - Lebanon — Total: 7.1
  - Yemen — Total: 4.4

### Annex Table 1.4 — Composition of Indirect Impact (selected country examples)
- Indirect impact = budget share × price effect for a $0.25 per liter increase; values reported as "Percent of household total consumption."
- Selected country rows:
  - Bangladesh — Budget Share: Food 65.7; Transport 2.7; Others 29.9. Price Effect: Food 0.02; Transport 0.07; Others 0.01. Indirect Impact: Food 1.00; Transport 0.19; Others 0.31
  - Cameroon — Budget Share: Food 44.5; Transport 5.9; Others 47.3. Price Effect: Food 0.01; Transport 0.06; Others 0.01. Indirect Impact: Food 0.35; Transport 0.34; Others 0.63
  - Ghana — Budget Share: Food 45.2; Transport 3.2; Others 46.0. Price Effect: Food 0.12; Transport 0.47; Others 0.10. Indirect Impact: Food 5.25; Transport 1.50; Others 4.93
  - Jordan2/ — Budget Share: Food 41.0; Transport 0.0; Others 53.3. Price Effect: Food 0.06; Transport N.A.; Others 0.06. Indirect Impact: Food 2.58; Transport N.A.; Others 3.77
  - Malaysia — Budget Share: Food 20.7; Transport 15.9; Others 53.5. Price Effect: Food 0.06; Transport 0.09; Others 0.09. Indirect Impact: Food 1.29; Transport 1.45; Others 2.74
  - Yemen — Budget Share: Food 69.5; Transport 0.9; Others 26.0. Price Effect: Food 0.03; Transport 0.03; Others 0.02. Indirect Impact: Food 2.20; Transport 0.03; Others 0.51
- Aggregate all countries (average):
  - Budget Share: Food 49.2; Transport 4.1; Others 43.9
  - Price Effect: Food 0.03; Transport 0.10; Others 0.04
  - Indirect Impact: Food 1.4; Transport 0.4; Others 1.6

### Conclusions (empirical findings and policy conclusions)
- Main quantitative findings:
  - A $0.25 per liter increase in fuel prices reduces household real incomes by, on average, 5.5 percent.
  - Approximately 50 percent of this impact is indirect (through higher prices of other goods and services).
- Policy conclusions:
  - Fuel subsidies are badly targeted, fiscally costly, and primarily benefit higher-income groups.
  - Withdrawal of subsidies can significantly reduce real incomes of low-income households; hence well-targeted mitigation measures are critical.
  - Recommended measures include well-targeted cash transfers, temporary retention of subsidies on goods important to the poor, expansion of existing safety nets, increased pro-poor spending in health/education/infrastructure, and adoption of automatic pricing mechanisms with price smoothing for transitional protection.

*Source: _wp15250 - Annex 1. Supplementary Tables*

### Annex 1. Supplementary Tables ..........................................................................................

### Annex 1. Supplementary Tables

### Introduction and purpose
- Fuel subsidies exacerbate income inequalities and are not a cost-effective approach to protecting the poor.
- Two channels by which subsidies worsen inequality:
  - Benefits distributed in proportion to household energy consumption; higher-income households typically have more energy-intensive consumption baskets.
  - Public expenditures on energy subsidies can crowd out more redistributive public spending or require financing through regressive taxation.
- Governments remain reluctant to remove subsidies because higher fuel prices reduce household real incomes, especially for lower-income households.
- This paper updates cross-country evidence on the welfare impact of fuel price increases by extending the survey in Arze del Granado, Coady, and Gillingham (2012) from 20 countries (2005–09) to 32 countries up to 2014.
- Objective: quantify magnitude and distribution of income losses from fuel price increases to inform design of more cost-effective protection for vulnerable groups.

### Trends in international oil prices and domestic pass-through
- International oil prices rose steadily after the mid-2000s and surged in 2007–first half 2008.
- Many developing countries did not fully pass international price increases to domestic consumers:
  - More than half of developing countries passed on less than two-thirds of the increase in international diesel prices during the mid-2000s to mid-2008 period.
  - Pass-through was especially low in the Middle East and North Africa: half of the countries passed through less than 13 percent of international price increases.
- Incomplete pass-through raised subsidies substantially; combined cost for diesel, gasoline, and kerosene exceeded 3 percent of GDP on an annualized basis in more than half of developing countries (when these fuels alone are considered).
- Price dynamics and pass-through patterns (selected summary statistics from Table 1):
  - Median pass-through (Percent) for diesel:
    - Developing Economies {118}: End-2004–Mid-2008 = 66; Mid-2008–End-2008 = 22; End-2008–End-2012 = 49; End-2013–End-2014 = 11
    - Developing Asia {28}: 65; 12; 54; 40
    - Latin America and the Caribbean {32}: 51; 28; 83; 12
    - Middle East and North Africa {19}: 13; 1; 10; 0
    - Sub-Saharan Africa {39}: 85; 38; 35; 0
  - Median fiscal cost (Percent of GDP) for gasoline, diesel, and kerosene:
    - Developing Economies {118}: End-2004–Mid-2008 = 1.92; Mid-2008–End-2008 = −3.14; End-2008–End-2012 = 1.74; End-2013–End-2014 = −0.88
    - Developing Asia {28}: 1.43; −4.68; 2.24; −0.89
    - Latin America and the Caribbean {32}: 2.50; −2.98; 1.27; −0.80
    - Middle East and North Africa {19}: 1.93; −3.44; 3.20; −1.89
    - Sub-Saharan Africa {39}: 2.15; −3.42; 1.50; −0.92
  - Median net-tax (US$/liter) by period (End-2004, Mid-2008, End-2008, End-2012, End-2014):
    - Developing Economies {118}: 0.10; −0.08; 0.33; 0.22; 0.37
    - Developing Asia {28}: −0.11; −0.37; 0.17; 0.05; 0.16
    - Latin America and the Caribbean {32}: 0.09; −0.08; 0.29; 0.22; 0.29
    - Middle East and North Africa {19}: −0.14; −0.78; −0.12; −0.36; 0.07
    - Sub-Saharan Africa {39}: 0.27; 0.14; 0.54; 0.32; 0.47
- After mid-2008 peak, international prices collapsed within six months; many countries passed through little of the decline to consumers (pass-through fell below 20 percent in more than half of developing countries; below 1 percent in most of the Middle East and North Africa), which increased net taxes and reduced subsidies temporarily.
- International prices rebounded by April 2011, reversing many temporary fiscal gains and increasing fiscal costs of controlled prices again.
- The decline in international prices in late 2014–early 2015 presented another policy window to adopt durable subsidy reforms; knowledge of magnitude and distribution of household impacts is crucial for designing mitigating measures.

### Methodology (estimating welfare impact)
- Total household welfare impact of higher domestic fuel prices composed of two channels:
  - Direct impact: higher prices for fuels consumed for cooking, heating, lighting, and personal transport.
    - Requires household survey data on expenditures to compute budget share for each fuel product.
    - Budget share provides estimate of welfare impact of a doubling of fuel price absent demand response (interpreted as short-term impact or upper bound).
  - Indirect impact: higher prices for other goods and services as fuel cost increases feed into production costs and consumer prices.
    - Estimated using price-shifting models based on Coady and Newhouse (2006), assuming increases in fuel production costs are fully passed forward to domestic prices.
    - Requires input-output tables or production structure data; outdated input-output data can cause underestimation (if fuel cost shares have risen since table) or overestimation (if they have fallen).
    - Key modeling assumptions: goods are nontraded; domestic production technologies exhibit constant returns to scale; demand is completely price inelastic. These imply estimates are short-term impacts or upper bounds.
    - Indirect impacts are computed by estimating price increases for consumption categories and multiplying by household budget shares for those categories from surveys; aggregated to household indirect welfare impact.
- Total impact: sum of direct and indirect impacts.
  - Distribution across income groups estimated by averaging impacts across household consumption quintiles (based on consumption per capita or per adult equivalent).
  - Classification: distribution is progressive if percentage welfare loss increases with household consumption; regressive if it decreases.
  - Effectiveness of subsidized fuel prices for protecting the poor depends on share of total subsidy benefits accruing to lower-income households.

### Welfare impact of a uniform $0.25 per liter price increase (magnitude, composition, and distribution)
- Review approach:
  - Country studies (mainly IMF and World Bank staff) recalculated to a common price increase of $0.25 per liter by scaling welfare impacts proportionally to reflect differing initial retail price levels.
  - Annex Table 1.1 lists case studies and data; Annex Table 1.2 contains adjustment factors; Annex Table 1.3 presents country- and product-specific results.
  - Electricity tariff changes not included in new recalculations due to lack of recent data in reviewed reports; original article results on electricity are reported for reference.
- Key aggregated results:
  - Average total (direct plus indirect) impact of a $0.25 per liter increase in fuel prices: 5.5 percent decline in household real incomes (authors’ calculations based on reviewed studies).
  - Regional range of total impact:
    - South and Central America: 3.5 percent decline in household real incomes.
    - Middle East: 7.0 percent decline in household real incomes.
    - All Regions average: 5.5 percent.
  - Indirect impacts account for a substantial share of total impact (see Figure 2 and Figure 3 panels referenced).
- Drivers of regional differences:
  - Higher simulated impacts in Middle East and Central Asia linked to comparatively low retail fuel prices (a given $0.25 per liter increase represents a larger percentage price increase where retail prices are low), implying a volume effect via higher consumption shares where retail prices are low.
  - Inverse relationship observed between retail prices of diesel and gasoline and consumption of these products as a share of national income: higher retail prices in advanced and emerging European economies associated with lower consumption; Middle East and Central Asia exhibit lower prices and higher consumption.
  - These patterns are consistent with higher energy prices providing stronger incentives for improving energy efficiency.

### Implementation considerations and policy implications (as discussed)
- Estimates should be interpreted as short-term impacts or upper bounds because of:
  - Assumption of no (or fully inelastic) demand response in many components.
  - Potential outdatedness of input-output data used to compute indirect effects.
- Knowledge of magnitude and distribution of household impacts is crucial for:
  - Designing safety net measures to protect the poor and vulnerable when reforms are implemented.
  - Making subsidy reforms politically feasible and durable by targeting compensation effectively.
- The evidence confirms that a very large share of benefits from fuel price subsidies accrues to high-income households, reinforcing income inequalities and suggesting scope for more cost-effective targeted social protection.

*Source: _wp15250 - Annex 1. Supplementary Tables*

### 2. Ratio of Fuel Consumption to GDP

### 2. Ratio of Fuel Consumption to GDP

### Indirect versus direct impacts
- On average, the indirect impact accounts for 55 percent of the total impact.
- Regional shares of the indirect effect:
  - sub-Saharan Africa: 65 percent of the total impact
  - South and Central America: 60 percent of the total impact
  - Asia and Pacific: less than 50 percent of the total impact
  - Middle East and Central Asia: less than 50 percent of the total impact
- Rationale: a high proportion of total fuel consumption is for intermediate use, so indirect effects (price pass-through to other goods and services) are a sizable component of total household welfare impact.
- Composition of indirect impact (budget-share weighted averages across consumption categories):
  - Budget share (Percent of household total consumption)
    - Food: 49.2
    - Transport: 4.1
    - Others: 43.9
  - Price effect (Percentage increase)
    - Food: 0.03
    - Transport: 0.10
    - Others: 0.04
  - Indirect impact (Percent of household total consumption)
    - Food: 1.4
    - Transport: 0.4
    - Others: 1.6
- Interpretation:
  - “Others” has a slightly lower budget share than food but a greater fuel intensity, producing a higher price effect; on average “Others” account for slightly less than 50 percent of the indirect impact and food about 40 percent.
  - Transport absorbs only 4.1 percent of household budgets on average but accounts for about 12 percent of the indirect impact because of its relatively high energy intensity.

### Direct impacts and fuel-specific patterns
- Fuel composition differs across regions:
  - Kerosene is relatively important in Africa due to low household access to electricity.
  - Kerosene is much less important in the Middle East and in South and Central America where access to electricity is more extensive.
- Access caveat: if only half of households have access to a fuel (for example electricity or LPG), the impact on users will be double the average impact reported for all households in the table—estimates can substantially underestimate impacts on users.

### Distributional patterns of welfare impacts
- Overall distribution:
  - Total, direct, and indirect welfare impacts are approximately distributionally neutral: the percentage decrease in welfare is very similar across income groups.
  - A $0.25 per liter increase in fuel prices decreases household real incomes by, on average, 5.5 percent.
  - Approximately half of this impact comes through the indirect effect on prices of other goods and services.
- Fuel-specific distribution:
  - Gasoline and electricity impacts are strongly progressive.
  - Kerosene impact is strongly regressive.
  - LPG impact varies by region: slightly progressive on average but very regressive in the Middle East and Central Asia.
- Quintile-level example (from figures):
  - Indicators shown include Direct impact, Indirect impact, Total impact by income quintile (Bottom, 2, 3, 4, Top) with example percentage-point values displayed in figures (e.g., values around 2.3, 2.4, 2.6, 2.5 for indirect/direct components and totals around 5.4–5.7 percent).

### Subsidy leakages and cost-effectiveness
- Universal fuel subsidies are badly targeted; higher total consumption of well-off households causes substantial leakage of benefits to higher-income groups.
- On average:
  - The top income quintile receives more than six times more in total subsidies than the bottom quintile.
  - For gasoline and LPG, top quintile receives:
    - Gasoline: 27 times the bottom quintile
    - LPG: 12 times the bottom quintile
- Aggregate concentration:
  - The richest 20 percent capture, on average, more than six times more in fuel subsidies than the poorest 20 percent.
  - More than 83 percent of gasoline subsidy benefits accrue to the richest 40 percent of households.
- Cost of providing $1 to the poorest 40 percent via fuel subsidies:
  - Using gasoline subsidies, the cost to the budget of transferring one dollar to the poorest 40 percent is about 14 dollars (nearly 93 out of every 100 dollars of gasoline subsidy “leaks” to the top three quintiles).
  - For kerosene the cost-benefit ratio is about 3 dollars.
- Example comparison with a safety-net program:
  - If 15 out of every 100 dollars allocated to a safety net program is absorbed by administrative costs and 80 percent of the remaining 85 dollars in beneficiary transfers reaches the poor (i.e., 68 percent of the total budget), then the cost-benefit ratio for such a program is 1.5 dollars—substantially lower than via fuel subsidies.

### Implications for electricity pricing
- Nonlinear electricity pricing (block tariffs with a lower lifeline tariff) can mitigate impacts but often perform worse than believed:
  - Many poor households do not benefit from lifeline tariffs because they lack access to electricity.
  - Many poor households with larger family sizes consume above lifeline thresholds and thus do not benefit.
  - Financing lifeline subsidies via higher tariffs on larger consumers can make large-consuming poor households worse off.
  - Conclusion: nonlinear pricing is a very crude protection mechanism with important negative side effects.

### Protecting low-income households — policy recommendations
- Key objective: implement well-targeted measures to mitigate the impact of energy price increases on the poor to build public support for subsidy reform.
- Preferred measures:
  - Targeted cash transfers or near-cash transfers:
    - Provide beneficiary flexibility and remove costly, abuse-prone government distribution of subsidized energy.
    - Examples: Armenia’s targeted cash transfer in electricity reform; Indonesia’s nonconditional cash transfer covering 35 percent of the population.
  - Conditional cash transfers:
    - Link benefits to investments in education and health to protect incomes and build human capital.
  - If cash transfers are infeasible:
    - Expand existing programs that can be scaled quickly with improved targeting: school meals, public works, reductions in education and health user fees, subsidized mass urban transport, subsidies for water and electricity connection costs.
    - Consider temporarily delaying price increases for products more important to the poor (for example kerosene).
  - Electricity-specific measures:
    - Maintain lower lifeline tariffs but recognize their limits; subsidize connection costs instead of consumption where appropriate (example: Kenya).
  - Design and sequencing of reform:
    - Develop a comprehensive reform plan with clear objectives, measures, and timelines.
    - Develop an effective communication strategy with transparency.
    - Appropriately phase and sequence price increases (weigh gradualism versus short-term savings and distortion risks).
  - Improve efficiency of energy state-owned enterprises (SOEs) to reduce fiscal burdens and signal reforms are not protecting inefficiency.
  - Depoliticize energy pricing:
    - Consider automatic pricing formulas linking domestic prices to international prices with built-in price smoothing to share international volatility between consumers and the budget and avoid sudden large domestic price jumps.
    - Price-smoothing rules trade off lower price volatility against higher fiscal volatility.
- International examples and experience:
  - Countries that used targeted cash transfers and program expansion: Armenia, Indonesia, Gabon, Ghana, Niger, Nigeria, Mozambique, Morocco, Philippines.
  - Examples of automatic pricing mechanisms and price smoothing in practice: Chile, Peru, Mauritius, South Africa, the Philippines, Turkey.
  - India liberalized gasoline prices in 2010 and diesel prices in 2014.

### Conclusions (summary of empirical findings)
- The paper revisits distribution of benefits from fuel subsidies and updates previous review findings.
- Main quantitative findings:
  - A $0.25 per liter increase in fuel prices reduces household real incomes by, on average, 5.5 percent.
  - Approximately 50 percent of this impact is indirect (through higher prices of other goods and services).
- Policy conclusions:
  - Fuel subsidies are badly targeted, fiscally costly, and primarily benefit higher-income groups.
  - Withdrawal of subsidies can significantly reduce real incomes of low-income households; hence well-targeted mitigation measures are critical.
  - Recommended measures include well-targeted cash transfers, temporary retention of subsidies on goods important to the poor, expansion of existing safety nets, increased pro-poor spending in health/education/infrastructure, and adoption of automatic pricing mechanisms with price smoothing for transitional protection.

*Source: _wp15250 - 2. Ratio of Fuel Consumption to GDP*

### ANNEX 1. SUPPLEMENTARY TABLES

### ANNEX 1. SUPPLEMENTARY TABLES

### Data coverage and study metadata (Annex Table 1.1)
- Table contents: Year of Report; Year of Household Survey; Year of Input-Output Matrix; Author Institution.
- Geographic coverage: Africa, South and Central America, Asia and Pacific, Middle East and Central Asia.
- Selected country entries (preserving exact years and institutions):
  - Angola — Year of Report: 2014; Year of Household Survey: 2008/2009; Year of Input-Output Matrix: 2014; Author Institution: IMF
  - Cameroon — Year of Report: 2007; Year of Household Survey: 2001; Year of Input-Output Matrix: 2001; Author Institution: IMF
  - Ghana — Year of Report: 2005; Year of Household Survey: 1999; Year of Input-Output Matrix: 1993; Author Institution: IMF
  - Kenya — Year of Report: 2010; Year of Household Survey: 2006; Year of Input-Output Matrix: n.a.; Author Institution: World Bank
  - India — Year of Report: 2013; Year of Household Survey: 2010; Year of Input-Output Matrix: 2004; Author Institution: IMF
  - Indonesia — Year of Report: 2014; Year of Household Survey: 2012; Year of Input-Output Matrix: 2011; Author Institution: World Bank
  - Jordan — Year of Report: 2011; Year of Household Survey: 2008/2009; Year of Input-Output Matrix: 2006; Author Institution: IMF
  - Yemen — Year of Report: 2013; Year of Household Survey: 2006; Year of Input-Output Matrix: 2009; Author Institution: IMF
- Note included in table: n.a. = not available. New studies in bold (format preserved in original table).

### Adjustment factors by country and fuel product (Annex Table 1.2)
- Table provides adjustment factors for four fuel products: Gasoline, Kerosene, Diesel, LPG.
- Selected country-factor entries (exact values preserved):
  - Angola — Gasoline: 0.32; Kerosene: 0.69; Diesel: 0.48; LPG: 0.54
  - Bangladesh — Gasoline: 0.54; Kerosene: 0.86; Diesel: 0.86; LPG: 0.70
  - Bolivia — Gasoline: 0.54; Kerosene: 0.71; Diesel: 0.52; LPG: 0.53
  - Cameroon — Gasoline: 0.40; Kerosene: 0.92; Diesel: 0.48; LPG: 0.44
  - Ghana — Gasoline: 0.96; Kerosene: 1.42; Diesel: 1.02; LPG: 0.99
  - India (two studies reported) — Study 1: Gasoline: 0.28; Kerosene: 1.23; Diesel: 0.46; LPG: 0.37; Study 2: Gasoline: 0.20; Kerosene: 0.92; Diesel: 0.29; LPG: 0.03
  - Indonesia — Gasoline: 0.53; Kerosene: 0.94; Diesel: n.a.; LPG: n.a.
  - Jordan (two studies reported) — Study 1: Gasoline: 0.53; Kerosene: 1.25; Diesel: 1.29; LPG: 0.91; Study 2: Gasoline: 0.25; Kerosene: 0.34; Diesel: 0.34; LPG: 0.33
  - Malaysia — Kerosene: 0.41; Diesel: 0.45; LPG: n.a. (Author: IMF/Universiti Utara Malaysia)
  - Peru — Gasoline: 0.27; Kerosene: 0.27; Diesel: 0.27; LPG: 0.27
  - Philippines — Kerosene: 0.66; Diesel: 0.81; LPG: 0.82; Electricity: 1.04 (table layout preserves these placements)
- Notes from table: LPG = liquefied petroleum gas; n.a. = not available. For countries with more than one study available, results are reported in chronological order; refer to Table 1.1 for year of report/underlying data.

### Direct and indirect welfare impacts of fuel price increases (Annex Table 1.3)
- Table structure: Direct impacts by product (Gasoline, Kerosene, LPG, Electricity); Indirect impacts; Total impact. All values reported as "Percent of household total consumption."
- Regional averages (exact values preserved):
  - Africa (average) — Gasoline: 0.3; Kerosene: 0.9; LPG: 0.2; Electricity: 0.4; Indirect: 1.7; (other indirect columns) 3.2; Total: 5.0
  - South and Central America (average) — Gasoline: 0.2; Kerosene: 0.2; LPG: 0.3; Electricity: 0.8; Indirect: 1.4; (other indirect columns) 2.0; Total: 3.5
  - Asia and Pacific (average) — Gasoline: 0.6; Kerosene: 1.0; LPG: 0.4; Electricity: 1.3; Indirect: 3.2; (other indirect columns) 2.7; Total: 5.9
  - Middle East and Central Asia (average) — Gasoline: 0.9; Kerosene: 0.4; LPG: 0.9; Electricity: 1.4; Indirect: 3.5; (other indirect columns) 3.4; Total: 7.0
  - All regions (average) — Gasoline: 0.4; Kerosene: 0.8; LPG: 0.4; Electricity: 0.9; Indirect: 2.5; (other indirect columns) 3.0; Total: 5.5
  - Only regions with indirect effect — Gasoline: 0.5; Kerosene: 0.8; LPG: 0.3; Electricity: 0.7; Indirect: 2.4; (other indirect columns) 3.0; Total: 5.4
- Selected country-specific total impacts (exact values preserved where available):
  - Cameroon — Total: 3.6 (Direct and Indirect components shown in table)
  - Gabon — Total: 5.6
  - Central African Republic — Total: 3.5
  - Senegal — Total: 2.1
  - Ghana — Total: 17.3 (components: Direct by product and Indirect contributions reported in table)
  - Congo, Republic of — Total: 9.0
  - Burkina Faso — Total: 2.0
  - Angola — Total: 4.2
  - South Africa — Total: 0.9
  - Bolivia — Total: 3.6
  - Peru — Total: 1.7
  - El Salvador — Total: 1.7
  - Honduras — Total: 5.0
  - Paraguay — Total: 1.3
  - Bangladesh — Total: 3.2 (one study) and 0.3 (another study)
  - Sri Lanka — Total: 5.3
  - India — Total: 3.6 (one study) and 2.5 (another study)
  - Indonesia — Total: 8.8
  - Malaysia — Total: 8.5
  - Philippines — Total: 1.9
  - Thailand — Total: 1.1
  - Vietnam — Total: 0.8
  - Jordan — Totals: 12.1 (study with large electricity indirect effect) and 5.8 (other study)
  - Lebanon — Total: 7.1
  - Sudan — Total: 2.8
  - Yemen — Total: 4.4
- Notes from table: LPG = liquefied petroleum gas; n.a. = not available. Values are percent of household total consumption. Table reports direct impacts by product and indirect (economy-wide) impacts, summed to total welfare impact.

*Source: Authors’ calculations based on results from reviewed studies.*

### Annex Table 1.4. Composition of Indirect Impact

### Annex Table 1.4. Composition of Indirect Impact

### Summary of table purpose and structure
- The table reports composition of indirect impact expressed as "Percent of household total consumption".
- Columns and groupings in the table:
  - Budget Share: Food, Transport, Others1/
  - Price Effect: Food, Transport, Others1/
  - Indirect Impact: Food, Transport, Others1/
- The indirect impact is defined as the product of budget shares and the price impact for a $0.25 per liter increase in fuel prices.
- [N.A.] indicates information not available in the country study.
- Footnote 1/: Category "others" excludes fuel products directly consumed by households for cooking, lighting, and transport. Hence, budget shares do not add to 100, but to 100 minus household's share of consumption in fuel products.
- Footnote 2/: Jordan’s report did not disaggregate transport from other nonfood expenditures.

### Country-level composition (selected country rows exactly as reported)
- Bangladesh
  - Budget Share: Food 65.7, Transport 2.7, Others 29.9
  - Price Effect: Food 0.02, Transport 0.07, Others 0.01
  - Indirect Impact: Food 1.00, Transport 0.19, Others 0.31
- Burkina Faso
  - Budget Share: Food 57.0, Transport 0.7, Others 40.9
  - Price Effect: Food 0.01, Transport 0.07, Others 0.01
  - Indirect Impact: Food 0.31, Transport 0.05, Others 0.30
- Cameroon
  - Budget Share: Food 44.5, Transport 5.9, Others 47.3
  - Price Effect: Food 0.01, Transport 0.06, Others 0.01
  - Indirect Impact: Food 0.35, Transport 0.34, Others 0.63
- Central African Republic
  - Budget Share: Food 47.1, Transport 0.9, Others 51.3
  - Price Effect: Food 0.02, Transport 0.14, Others 0.03
  - Indirect Impact: Food 1.14, Transport 0.13, Others 1.57
- Congo, Republic of
  - Budget Share: Food 41.4, Transport 10.9, Others 46.4
  - Price Effect: Food 0.02, Transport 0.09, Others 0.12
  - Indirect Impact: Food 0.98, Transport 1.02, Others 5.71
- Gabon
  - Budget Share: Food 48.8, Transport 3.5, Others 45.5
  - Price Effect: Food 0.04, Transport 0.03, Others 0.03
  - Indirect Impact: Food 1.83, Transport 0.11, Others 1.48
- Ghana
  - Budget Share: Food 45.2, Transport 3.2, Others 46.0
  - Price Effect: Food 0.12, Transport 0.47, Others 0.10
  - Indirect Impact: Food 5.25, Transport 1.50, Others 4.93
- Jordan2/
  - Budget Share: Food 41.0, Transport 0.0, Others 53.3
  - Price Effect: Food 0.06, Transport N.A., Others 0.06
  - Indirect Impact: Food 2.58, Transport N.A., Others 3.77
- Malaysia
  - Budget Share: Food 20.7, Transport 15.9, Others 53.5
  - Price Effect: Food 0.06, Transport 0.09, Others 0.09
  - Indirect Impact: Food 1.29, Transport 1.45, Others 2.74
- Mali
  - Budget Share: Food 45.6, Transport 0.9, Others 52.0
  - Price Effect: Food 0.01, Transport 0.12, Others 0.01
  - Indirect Impact: Food 0.59, Transport 0.11, Others 0.72
- Paraguay
  - Budget Share: Food 40.8, Transport 8.6, Others 48.5
  - Price Effect: Food 0.01, Transport 0.05, Others 0.00
  - Indirect Impact: Food 0.38, Transport 0.44, Others 0.25
- Peru
  - Budget Share: Food 51.2, Transport 1.8, Others 46.2
  - Price Effect: Food 0.01, Transport 0.03, Others 0.01
  - Indirect Impact: Food 0.42, Transport 0.05, Others 0.27
- Senegal
  - Budget Share: Food 49.3, Transport 3.0, Others 46.7
  - Price Effect: Food 0.00, Transport 0.01, Others 0.02
  - Indirect Impact: Food 0.25, Transport 0.02, Others 0.81
- Sri Lanka
  - Budget Share: Food 56.1, Transport 2.5, Others 38.7
  - Price Effect: Food 0.03, Transport 0.10, Others 0.02
  - Indirect Impact: Food 1.80, Transport 0.25, Others 0.58
- Sudan
  - Budget Share: Food 63.0, Transport 4.9, Others 29.7
  - Price Effect: Food 0.03, Transport 0.10, Others N.A.
  - Indirect Impact: Food 1.94, Transport 0.50, Others N.A.
- Yemen
  - Budget Share: Food 69.5, Transport 0.9, Others 26.0
  - Price Effect: Food 0.03, Transport 0.03, Others 0.02
  - Indirect Impact: Food 2.20, Transport 0.03, Others 0.51

### Aggregate statistics and shares
- All countries (average)
  - Budget Share: Food 49.2, Transport 4.1, Others 43.9
  - Price Effect: Food 0.03, Transport 0.10, Others 0.04
  - Indirect Impact: Food 1.4, Transport 0.4, Others 1.6
- Share in total impact
  - Gasoline/other split listed as: 18.43 59.37 22.20  40 12 48

### Notes and definitions reiterated from the table
- The price effect is the proportionate increase in prices resulting from a $0.25 per liter increase in fuel prices.
- The indirect impact equals budget share × price effect.
- Category "others" excludes fuel products directly consumed by households for cooking, lighting, and transport; therefore budget shares sum to less than 100 by the household share of fuel consumption.
- Jordan’s report did not disaggregate transport from other nonfood expenditures.
- [N.A.] indicates information not available in the country study.

*Source: Annex Table 1.4. Composition of Indirect Impact (from the supplied IMF content unit).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15250.pdf_