## 3. Determinants of Gas Demand

## Source details

**Canonical URL:** [3. Determinants of Gas Demand](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12247.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12247.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12247.pdf.json)

---

### A. Introduction
- Implicit subsidies to households through under-priced household gas and heating tariffs are increasingly costly: almost 5 percent of GDP per annum.
- Political resistance has deferred full tariff adjustment to marginal cost (import parity) to the medium-term; authorities favor a multiyear approach beginning with energy-saving infrastructure upgrades and access to low-cost loans for home renovation.
- Untargeted under-pricing:
  - favors high consumers (typically wealthier households),
  - drains government finances,
  - sustains energy over-consumption,
  - dampens investment in delivery systems,
  - undermines incentives for domestic production expansion.
- IFI advice summarized:
  - gradual but immediate tariff increases,
  - reforms to social assistance to target the most vulnerable,
  - credible schedule of regular tariff increases to incentivize energy savings,
  - increased building-level metering to improve billing transparency and willingness to pay,
  - unified, means-tested social assistance and targeted explicit subsidies to redirect assistance to those most in need.
- Quantitative illustrative findings:
  - A 20 percent tariff increase → households reduce consumption by around 5 percent for gas and 3 percent for heating, with reductions increasing with income level.
  - Absolute consumer surplus loss for wealthier households is 1½ times larger than for lower income households; as a percent of income, lower income households experience a greater welfare loss.
  - Uniform tariff increases can be regressive, but targeted social assistance can fully defray costs to poorer households; government savings from tariff increases can more than offset welfare losses and allow higher growth-enhancing expenditures.

### B. Stylized Facts: Economic Implications of Gas Sector Subsidies
- Sector structure and distortions:
  - State-owned enterprises dominate gas and heating; Naftogaz (NG) is a state-owned holding covering exploration, production, imports, storage, transit, and domestic distribution.
  - Heating utilities (largely municipal) purchase gas from NG; gas is about 90 percent of the energy input in heating generation.
  - Household prices are at one fifth of import costs; heating utilities at one third of import costs; heating utilities typically embed only half their gas cost into household heating tariffs.
  - Despite tariff increases in 2010–2011 (50 percent increase in gas-component in August 2010; 15 percent increase in heating tariffs in January 2011), household gas and heating tariffs remain among the lowest in Europe and well below import parity.
- NG and heating utility financials and fiscal cost:
  - NG’s losses in 2011: 1½ percent of GDP.
  - Naftogaz snapshot (as presented):
    - Revenues 1148.7 (in bln UAH)
    - Gas Sales 866.6
      - Households 90.7
      - Heating Utilities 80.6
      - Industry 564.3
      - Other 131.0
    - Gas Transit Fees 251.9
    - Other 30.2
    - Expenditures13510.3 (presented in source text)
    - Gas Imports 745.6
    - Domestic Gas 60.5
    - Other 554.2
    - Cash Balance -21 -1.6
  - Heating utilities’ deficits reflect under-priced tariffs, inefficient distribution, and household payment arrears; meter coverage under 40 percent.
- Social assistance and distributional patterns:
  - Benefits compensate special categories; explicit subsidies are given when utility expenditure exceeds 15 percent of income (10 percent for pensioners/disabled).
  - Benefits are more than ¾ of the government’s social assistance budget for gas and heating (0.2 percent of GDP).
  - Only 20 percent of benefits go to the poorest two income quintiles; 30 percent go to the wealthiest quintile.
  - About 85 percent of people below the poverty line do not receive any social assistance for household utilities.
  - Only 5–12 cents per dollar of social assistance in this area contributes to reducing the poverty gap.
- Aggregate fiscal and economic impacts:
  - Overall natural gas-related transfers from government to households total 6 percent of GDP:
    - below-cost gas and heating tariff pricing: 5 percent of GDP,
    - social assistance (explicit subsidies and benefits): ¼ percent of GDP,
    - heating utility subsidies: ¼ percent of GDP,
    - debt arrears: ½ percent of GDP.
  - Underinvestment has reduced domestic production from around 20 to just over 15 billion cubic meters over the past decade.
  - Estimated investment needs: annual investments of US$1.5 billion in unexplored gas endowments could halve Russian import volume within a decade (World Bank staff estimates).
- Ukraine gas consumption, 2011 (total = 49 billion cubic meters):
  - Households 38%
  - Utilities 21%
  - Budget 2%
  - Industries 39%
- Impact of below-cost tariffs on Naftogaz (applying latest 2012 data):
  - Households: Volume V = 16.65 (bcm); Price paid to NG P = 4.44 (USD/tcm); Import price Pm = 15 (USD/tcm); NG loss = -6.0 (US$ billions) / -3.2 (in percent GDP).
  - Heating Utilities: V = 8.81; P = 18.54; Pm = 15; NG loss = -2.6 / -1.4.
  - Total: V = 25.4; Total NG loss = -8.6 / -4.7.

### C. Theory: Implications of Higher Gas and Heating Tariffs
- Household trade-offs:
  - Reduce gas/heating consumption, cut other expenditures or savings; response varies by income.
- Welfare measurement:
  - Short-run expenditure dynamics illustrated by areas on demand curve: initial expenditure rise (areas C+D+E), then reduced net expenditure increase (C-B) after behavioral response.
  - Consumer surplus falls; part of surplus loss (area C) is transferred to NG/utilities as higher revenues.
- Heterogeneous impacts across income:
  - Wealthier households: more elastic demand, larger nominal expenditure increases and larger consumer surplus losses; greater ability to invest in insulation and energy-saving.
  - Poorer households: more inelastic demand, less scope to adjust consumption, greater welfare loss as percent of income.
- Macroeconomic effects:
  - Short run: higher tariffs raise energy bills, crowd out non-energy spending, and increase precautionary savings.
  - Countervailing effects: higher demand for energy-saving materials and renovations that reduce consumption over time.
- Social assistance trade-offs and reforms:
  - Tariff increases raise cost of benefits proportionally if benefits are untargeted.
  - Reformed social assistance could:
    - eliminate benefits to the wealthy,
    - provide targeted explicit subsidies to poor households (compensation when utility expenditure > 15 percent of income),
    - provide cash subsidies equivalent to consumer surplus loss for the poor.
  - Government savings from closing implicit subsidies can be reallocated to social assistance and growth-promoting investment.
- Extreme illustrative case:
  - Raising tariff to import price Pm = P2 eliminates implicit subsidies (areas C+D+E for poor; C’+D’+E’ for wealthy).
  - Transferring implicit subsidy savings from wealthy to poor could maintain poor households’ welfare while leaving net savings for infrastructure because (C’+D’+E’) > (C+D+E).

### D. Model (estimation strategy)
- Objective: estimate household gas and heating demand as functions of price, income, housing and household characteristics to assess tariff increase impacts on welfare, expenditures, consumer surplus, and social assistance needs.
- Specification highlights:
  - Log-linear demand for annual heat: ln(heat_i) = f(effective price of heat, income quintile categorical variables, household income, dwelling floor area, household characteristics vector including number of people, number of rooms, year of major renovation, apartment vs house, rural dummy, pre-1980 building dummy, ability to control heating, non-government supplied heating dummy).
  - Effective price = flat rate price net of heating-related social assistance for each household.
  - Gas demand modeled analogously, with dummies for gas water heater and ease of gas availability.
  - Heterogeneous price elasticities by income quintile:
    - Quintiles 1–2 share same elasticity (basic-need consumers).
    - Quintiles 3, 4, 5 have separate elasticities β + β_j (j=3,4,5).
- Robustness checks:
  - Homogeneous price elasticity across quintiles,
  - Model with household income as continuous interaction with price.
- Data:
  - 2010 household survey (quarterly) by State Statistics Survey of Ukraine: 10,428 households representing ~17 million households (with weights).
  - Variables: socio-economic status, living standards, income, consumption expenditures (including gas and heating).
  - Household tariffs from Naftogaz; social assistance data from Ministry of Labor.
  - 2012 projections derived by adjusting 2010 incomes for IMF inflation and wage growth projections and applying actual 2012 gas/heating prices to the model.

### E. Results (summary to the point provided)
- Demand elasticities (short run):
  - Heating demand is price inelastic.
  - Heating: falls by 1.82 percent when tariffs increase by 10 percent.
  - Gas demand is price inelastic; gas: falls by 2.63 to 2.83 percent in response to a 10 percent tariff hike.
  - Homogeneous price elasticity and continuous price-income interaction elasticity evaluated at average income fall within the range of heterogeneous price elasticities.
- Heterogeneity:
  - Wealthier households’ gas and heating demands are more elastic than poorer households.
  - The wealthiest income quintile’s heating and gas price elasticities are, respectively, 10 and 8 percent higher than those of the lowest two income quintiles.
  - No difference between elasticities of the first two quintiles and the third quintile; very little difference with the fourth quintile.
  - Continuous income-based estimates confirm increasing price elasticity across the third and fourth income quintiles.
- Income elasticities:
  - A 10 percent increase in annual household income causes heating consumption to increase by about 1.1 percent.
  - A 10 percent increase in annual household income causes gas consumption to increase by about 1.7 percent.
- Determinants and selected coefficients (OLS and variants):
  - Heating: LN(price) = -0.170 ***(Std Error 0.027) in OLS; LN(price) = -0.177 *** in homogeneous model.
  - Heating: LN(income) ≈ 0.057 to 0.111 depending on specification.
  - Gas: LN(price) = -0.264 ***(Std Error 0.026) in OLS; LN(price) = -0.270 *** in homogeneous model.
  - Gas: LN(income) ≈ 0.100 to 0.172 depending on specification.
  - Gas: Apartment coefficient = -1.298 ***(Std Error 0.022).
  - Ease of gas availability coefficient for gas = 0.731 ***(Std Error 0.022).
  - Heating regressions: R-squared = 0.327; Observations = 3,376.
  - Gas regressions: R-squared ≈ 0.558–0.559; Observations = 8,957.

### Impact of a 20 percent tariff hike in 2012 — household welfare and expenditures
- Comparative sensitivity:
  - Household expenditures on heating are almost twice as sensitive as gas to tariff increases.
  - Heating price elasticity is a third smaller than that of gas; households lower heating consumption less and thus spend more on heating relative to gas after tariff increases.
- Average household impacts:
  - Impact on the average household is well below one percent of income.
  - Households in the third and fourth quintiles (together 40 percent of households) unable to adjust consumption volumes: utilities’ expenditures rise by 2/3 to 3/4 of 1 percent of income.
  - With some flexibility to reduce consumption, these households can limit expenditure increase to 1/2 to 2/3 of one percent of income.
  - Consumer surplus losses for the average household are very small.
- Wealthy households (top 20 percent):
  - If initially unable to adjust volumes, utilities’ expenditures rise by less than 1/2 of one percent of income.
  - Near term consumption reduction by wealthy is nearly 1½ times that of the poor, bringing expenditures down to 1/3 of one percent of their income.
  - Consumer surplus loss absolute value is greater for wealthy households but smaller in proportion to income.
- Poorest households (bottom 40 percent):
  - Initially unable to adjust volumes, expenditures rise by just over one percent of income.
  - After reducing volumes, utilities expenditures near one percent of income.
  - Poorest households suffer the greatest consumer surplus loss in proportion to income across all income groups.

### Fiscal and social assistance implications (20 percent tariff hike)
- Current social assistance costs: 0.2 percent of GDP.
- After 20 percent tariff hike:
  - Social assistance costs rise by 0.02 percent of GDP (to 0.22 percent of GDP).
  - Direct subsidies increase from 0.02 to 0.03 percent of GDP.
  - Three quarters of the increase goes to households already receiving social assistance for utilities expenditures above 15 percent of income; remainder to newly applying households.
  - Additional cash compensation for lost consumer surplus of all social assistance recipients would cost 0.01 percent of GDP.
- Utility revenues and government savings:
  - NG and heating utilities’ revenues will rise by 0.3 percent of GDP after taking consumption adjustments into account.
  - Elimination of benefits to middle class and wealthy households would free almost 0.2 percent of GDP.
  - Additional targeted social assistance costs (0.02 percent of GDP) are minimal relative to these savings.
- Policy options for redeploying savings:
  - Fully compensate all but the wealthy for their additional costs: cost 0.6 percent of GDP.
  - Compensate only the poorest households: cost 0.2 percent of GDP.
  - Use savings to finance energy-saving infrastructure investment and domestic gas exploration and development, and growth-enhancing activities.

### Policy conclusions and recommendations
- Necessity and effects of tariff increases:
  - Gas and heating tariff increases are necessary to address economic inefficiencies and distributional distortions created by below-cost pricing.
  - Tariff increases reduce implicit subsidies favoring larger (wealthy) consumers, lower overconsumption, and incentivize adoption of energy-saving technologies.
  - Over time, industrial energy efficiency would increase with less illegal siphoning of gas from households to industries as the price differential shrinks.
- Phasing and design:
  - A schedule of gradual tariff increases will provide incentives for reduced energy consumption while ensuring affordability and giving time to adjust.
  - A single measured tariff increase (example: 20 percent) has little impact on the welfare of the wealthy; a medium-term credible schedule is needed to motivate lower consumption by wealthy.
- Social assistance reform:
  - Reformed, better-targeted social assistance can partially compensate the poor and middle class and reduce distributional distortions.
  - Eliminating untargeted benefits received by wealthy households would free almost 0.2 percent of GDP for targeted assistance.
  - A reformed system could ensure no household in the first four quintiles pays more than 15 percent of income for utilities expenditures.
  - Cash compensation for consumer surplus losses can enhance public support for subsidy phase-out.
- Use of government savings:
  - Under-pricing gas and heating provides a 5 percent of GDP subsidy to households.
  - Even with just a 20 percent tariff increase the government will gain almost 0.3 percent of GDP.
  - Savings should first finance energy-saving infrastructure investment to reduce the burden of tariff increases on households.
  - Additional savings from gradual tariff increases should be invested in domestic gas exploration and development and growth-enhancing activities to produce greater long-term gains.

*Source: _wp12247 (IMF working paper excerpt provided).*

### Bibliography ...........................................................................................................

### Bibliography ............................................................................................................................22

### Tables
- 1. Summary Statistics...............................................................................................................19
- 2. Determinants of Heating Demand .......................................................................................20

*Source: _wp12247 - Bibliography (PDF).*

### 3. Determinants of Gas Demand ..........................................................................................

### 3. Determinants of Gas Demand

### A. Introduction
- Implicit subsidies to households through under-priced household gas and heating tariffs have been a prevailing policy in Ukraine and are increasingly costly (almost 5 percent of GDP per annum).
- Political resistance has deferred full tariff adjustment to marginal cost (import parity) to the medium-term; authorities favor a multiyear approach beginning with energy-saving infrastructure upgrades and access to low-cost loans for home renovation.
- Untargeted under-pricing:
  - is a poor instrument for distributional goals (favors high consumers, typically wealthier households),
  - drains government finances,
  - sustains energy over-consumption,
  - dampens investment in delivery systems,
  - undermines incentives for domestic production expansion.
- IFI advice summarized:
  - gradual but immediate tariff increases,
  - reforms to social assistance to target the most vulnerable,
  - credible schedule of regular tariff increases to incentivize energy savings,
  - increased building-level metering to improve billing transparency and willingness to pay,
  - unified, means-tested social assistance and targeted explicit subsidies to redirect assistance to those most in need.
- Quantitative illustrative findings:
  - A 20 percent tariff increase → households reduce consumption by around 5 percent for gas and 3 percent for heating, with reductions increasing with income level.
  - Absolute consumer surplus loss for wealthier households is 1½ times larger than for lower income households; as a percent of income, lower income households experience a greater welfare loss.
  - Uniform tariff increases can be regressive, but targeted social assistance can fully defray costs to poorer households; government savings from tariff increases can more than offset welfare losses and allow higher growth-enhancing expenditures.

### B. Stylized Facts: Economic Implications of Gas Sector Subsidies
- Sector structure and current distortions:
  - State-owned enterprises dominate gas and heating; Naftogaz (NG) is a state-owned holding covering exploration, production, imports, storage, transit, and domestic distribution.
  - Heating utilities (largely municipal) purchase gas from NG; gas is about 90 percent of the energy input in heating generation.
  - Household and heating tariffs remain heavily regulated: household prices are at one fifth of import costs, heating utilities at one third of import costs; heating utilities typically embed only half their gas cost into household heating tariffs.
  - Despite tariff increases in 2010–2011 (50 percent increase in gas-component in August 2010; 15 percent increase in heating tariffs in January 2011), household gas and heating tariffs remain among the lowest in Europe and well below import parity.
- NG and heating utility financials and fiscal cost:
  - NG’s losses in 2011: 1½ percent of GDP.
  - Naftogaz financial snapshot (as presented):
    - Revenues 1148.7 (in bln UAH)
    - Gas Sales 866.6
      - Households 90.7
      - Heating Utilities 80.6
      - Industry 564.3
      - Other 131.0
    - Gas Transit Fees 251.9
    - Other 30.2
    - Expenditures 1351 0.3 (?) — (note: presented in source as "Expenditures13510.3")
    - Gas Imports 745.6
    - Domestic Gas 60.5
    - Other 554.2
    - Cash Balance -21 -1.6
  - Heating utilities’ deficits due to under-priced tariffs, inefficient distribution, and household payment arrears; meter coverage under 40 percent.
- Social assistance and distributional patterns:
  - Benefits vs. explicit subsidies:
    - Benefits compensate special categories (socially important professions, war veterans, Chernobyl victims); explicit subsidies are given when utility expenditure exceeds 15 percent of income (10 percent for pensioners/disabled).
    - Benefits are more than ¾ of the government’s social assistance budget for gas and heating (0.2 percent of GDP).
    - Only 20 percent of benefits go to the poorest two income quintiles; 30 percent go to the wealthiest quintile.
    - About 85 percent of people below the poverty line do not receive any social assistance for household utilities.
    - Only 5–12 cents per dollar of social assistance in this area contributes to reducing the poverty gap.
- Aggregate fiscal and economic impacts:
  - Overall natural gas-related transfers from government to households total 6 percent of GDP.
    - Bulk reflects below-cost gas and heating tariff pricing: 5 percent of GDP.
    - Social assistance (explicit subsidies and benefits): ¼ percent of GDP.
    - Heating utility subsidies: ¼ percent of GDP.
    - Debt arrears: ½ percent of GDP.
  - Underinvestment has reduced domestic production from around 20 to just over 15 billion cubic meters over the past decade, raising import dependence; outdated transport and delivery systems cause large technical gas loss.
  - Estimated investment needs: annual investments of US$1.5 billion in unexplored gas endowments could halve Russian import volume within a decade, eventually permitting self-sufficiency and exports (World Bank staff estimates).
- Ukraine gas consumption, 2011 (total = 49 billion cubic meters):
  - Households 38%
  - Utilities 21%
  - Budget 2%
  - Industries 39%
- Impact of below-cost tariffs on Naftogaz (presented figures):
  - Households: Volume consumed, V = 16.65 (bcm); Price paid to NG, P = 4.44 (USD/tcm); Import price, Pm = 15 (USD/tcm); NG loss = -6.0 (US$ billions) / -3.2 (in percent GDP).
  - Heating Utilities: V = 8.81; P = 18.54; Pm = 15; NG loss = -2.6 / -1.4.
  - Total: V = 25.4; Total NG loss = -8.6 / -4.7.
  - (Table noted as "Applying latest 2012 data"; sources: Authorities, Staff estimates.)

### C. Theory: Implications of Higher Gas and Heating Tariffs
- Households face trade-offs: reduce gas/heating consumption, cut other expenditures or savings; response varies by income.
- Welfare effects measured via changes in expenditures and consumer surplus (Hausman (1981); Vartia (1983)):
  - Short-run expenditure dynamics:
    - Immediate tariff increase yields maximum expenditure rise at near-inelastic consumption (households move from point 1 to 2); expenditure increases by areas C+D+E (from A+B to A+B+C+D+E).
    - Short-term behavioral response reduces consumption at the new tariff (move from point 2 to 3); net expenditure increase becomes C-B (from A+B to A+C).
  - Consumer surplus falls: part of surplus loss (area C) is transferred to NG/utilities as higher revenues.
- Heterogeneous impacts across income:
  - Wealthier households: more elastic demand, larger nominal expenditure increases (C’-B’ > C-B) and larger consumer surplus losses (C’+D’ > C+D), but greater ability to invest in insulation and energy-saving.
  - Poorer households: more inelastic demand, less scope to adjust consumption, greater welfare loss as percent of income.
- Macroeconomic effects:
  - Short run: higher tariffs (and inelastic demand) raise energy bills, crowd out non-energy spending, and increase precautionary savings.
  - Countervailing effects include higher demand for energy-saving materials, renovations, and later construction of energy-efficient homes that reduce consumption and reverse initial crowding out.
- Social assistance trade-offs:
  - Tariff increases raise cost of benefits proportionally since many benefit recipients are wealthy consumers.
  - Reformed social assistance could:
    - eliminate benefits to the wealthy,
    - provide targeted explicit subsidies to poor households (compensation when utility expenditure > 15 percent of income),
    - provide cash subsidies equivalent to consumer surplus loss for the poor.
  - Government savings from closing implicit subsidies can be reallocated to social assistance and growth-promoting investment.
- Extreme illustrative case:
  - Raising tariff to import price Pm = P2 eliminates implicit subsidies (areas C+D+E for poor; C’+D’+E’ for wealthy).
  - Welfare of the poor could be maintained by transferring implicit subsidy savings from wealthy to poor (or via cash transfers).
  - There would remain savings for infrastructure and growth investment because (C’+D’+E’) > (C+D+E).

### D. Model (estimation strategy)
- Objective: estimate household gas and heating demand as functions of price, income, housing and household characteristics to assess tariff increase impacts on welfare, expenditures, consumer surplus, and social assistance needs.
- Key modeling features:
  - Log-linear demand specification for annual heat consumption: ln(heat_i) = f(effective price of heat, income quintile categorical variables, household income, dwelling floor area, household characteristics vector including number of people, number of rooms, year of major renovation, apartment vs house, rural dummy, pre-1980 building dummy, ability to control heating, non-government supplied heating dummy).
  - Effective price = flat rate price net of heating-related social assistance (benefits and subsidies) for each household.
  - Gas demand modeled analogously, with dummy variables for presence of gas water heater and ease of gas availability substituting for some heating dummies.
  - Heterogeneous price elasticities by income quintile:
    - For first two income quintiles: same price elasticity (basic-need consumers).
    - For quintiles 3, 4, 5: separate elasticities β + β_j (j=3,4,5).
  - Robustness checks:
    - Homogeneous price elasticity across quintiles,
    - Model with household income as continuous interaction with price.
- Data:
  - 2010 household survey (quarterly) by State Statistics Survey of Ukraine: 10,428 households representing ~17 million households (with weights).
  - Variables: socio-economic status, living standards, income, consumption expenditures (including gas and heating).
  - Household heating and gas tariffs from Naftogaz; social assistance data from Ministry of Labor.
  - Effective household prices (net of social assistance) used in demand estimation.
  - 2012 projections derived by adjusting 2010 incomes for IMF inflation and wage growth projections and applying actual 2012 gas/heating prices to the model.

### E. Results (summary to the point provided)
- Demand elasticities:
  - Both gas and heating demand are price inelastic in the short run.
  - Heating consumption will fall by 1.65 to ... (results continue in source).
  - (Full elasticity estimates and additional quantitative welfare, expenditure, and social assistance impacts follow in the source beyond the excerpt provided.)

*Source: _wp12247 - 3. Determinants of Gas Demand (IMF PDF chapter).*

### 1.82 percent when tariffs increase by 10 percent. Gas demand is also inelastic, though more

### _wp12247 - 1.82 percent when tariffs increase by 10 percent. Gas demand is also inelastic, though more

### Elasticity estimates and heterogeneity
- Heating price elasticity: falls by 1.82 percent when tariffs increase by 10 percent (implied heterogeneous estimates across income).
- Gas price elasticity: falls by 2.63 to 2.83 percent in response to a 10 percent tariff hike.
- Homogeneous price elasticity and continuous price-income interaction elasticity evaluated at average income fall within the range of heterogeneous price elasticities.
- Wealthier households’ gas and heating demands are more elastic than poorer households:
  - The Wald test rejects the hypothesis that the fifth quintile elasticity equals that of the first two quintiles (test for ߚହ = 0 rejected for both gas and heating).
  - The wealthiest income quintile’s heating and gas price elasticities are, respectively, 10 and 8 percent higher than those of the lowest two income quintiles.
  - No difference between elasticities of the first two quintiles and the third quintile; very little difference with the fourth quintile.
- Continuous income-based estimates confirm increasing price elasticity across the third and fourth income quintiles.
- Income elasticities:
  - A 10 percent increase in annual household income causes heating consumption to increase by about 1.1 percent.
  - A 10 percent increase in annual household income causes gas consumption to increase by about 1.7 percent.

### Determinants of heating and gas consumption (coefficients and patterns)
- General findings (all with expected sign and statistical significance where noted):
  - Larger apartment/house and more people → greater gas and heating demand.
    - Heating demand more sensitive to physical size (LN(Area) coefficient positive and large).
    - Gas demand more sensitive to household size.
  - Heating demand higher for apartment buildings over individual homes and for urban over rural households:
    - Rural households have alternative forms of heating (including wood burning).
    - Apartment buildings tend to have poorer insulation quality and receive piped hot water from a central source.
  - Gas demand lower in apartments and urban areas (these households less likely to use gas for heating water).
  - Better renovated apartments and individual homes consume less heating and gas (Year of major renovation negative coefficient).
  - Centralized heating increases consumption relative to independent household-controlled heating (Central heating coefficient positive).
  - Gas consumption rises when gas is used to heat water (Gas water heater coefficient positive) and when gas availability is easier (Ease of gas availability coefficient positive).
- Selected coefficient magnitudes from OLS (continuous interaction models where applicable):
  - Heating: LN(price) = -0.170 ***(Std Error 0.027) in OLS; LN(price) = -0.177 *** (price elasticity evaluated at average income) in homogeneous model.
  - Heating: LN(income) ≈ 0.057 to 0.111 depending on specification.
  - Gas: LN(price) = -0.264 ***(Std Error 0.026) in OLS; LN(price) = -0.270 *** in homogeneous model.
  - Gas: LN(income) ≈ 0.100 to 0.172 depending on specification.
  - Gas: Apartment coefficient = -1.298 *** (Std Error 0.022).
  - Ease of gas availability coefficient for gas = 0.731 *** (Std Error 0.022).
- Model fit and samples:
  - Heating regressions: R-squared = 0.327; Observations = 3,376.
  - Gas regressions: R-squared ≈ 0.558–0.559; Observations = 8,957.

### Impact of a 20 percent tariff hike in 2012 — household welfare and expenditures
- Comparative sensitivity:
  - Household expenditures on heating are almost twice as sensitive as gas to tariff increases.
  - Heating price elasticity is a third smaller than that of gas; households lower heating consumption less and thus spend more on heating relative to gas after tariff increases.
- Average household impact:
  - The impact on the average household is well below one percent of income.
  - For households in the third and fourth quintiles (together 40 percent of households) unable to adjust consumption volumes, utilities’ expenditures rise by 2/3 to 3/4 of 1 percent of income.
  - With some flexibility to reduce consumption, these households can limit their expenditure increase to 1/2 to 2/3 of one percent of income.
  - Consumer surplus losses for the average household are very small (consumer surplus loss measured as triangle D in the demand curve graphs).
- Wealthy households (top 20 percent):
  - If initially unable to adjust volumes, utilities’ expenditures rise by less than 1/2 of one percent of income.
  - In the near term they reduce consumption nearly 1½ times that of the poor, bringing expenditures down to 1/3 of one percent of their income.
  - Consumer surplus loss absolute value is greater for wealthy households but smaller in proportion to income.
- Poorest households (bottom 40 percent):
  - Initially unable to adjust volumes, expenditures rise by just over one percent of income.
  - After reducing volumes, utilities expenditures are nearing one percent of their income.
  - Poorest households suffer the greatest consumer surplus loss in proportion to income across all income groups.
- Distributional graphs and quintile-specific consumption changes are reported (household gas consumption change with 20 percent price increase shown in thousands of cubic meters by income quintile).

### Fiscal and social assistance implications
- Current social assistance costs: 0.2 percent of GDP.
- After 20 percent tariff hike:
  - Social assistance costs rise by 0.02 percent of GDP (to 0.22 percent of GDP).
  - Direct subsidies increase from 0.02 to 0.03 percent of GDP.
  - Three quarters of the increase goes to households already receiving social assistance for utilities expenditures above 15 percent of income; remainder to newly applying households.
  - Additional cash compensation for lost consumer surplus of all social assistance recipients would cost 0.01 percent of GDP.
- Utility revenues and government savings:
  - NG and heating utilities’ revenues will rise by 0.3 percent of GDP after taking consumption adjustments into account.
  - Elimination of benefits to middle class and wealthy households would free almost 0.2 percent of GDP.
  - Additional targeted social assistance costs (0.02 percent of GDP) are minimal relative to these savings.
- Policy options for redeploying savings:
  - Fully compensate all but the wealthy for their additional costs: cost 0.6 percent of GDP.
  - Compensate only the poorest households: cost 0.2 percent of GDP.
  - Use savings to finance energy-saving infrastructure investment and domestic gas exploration and development, and growth-enhancing activities.

### Policy conclusions and recommendations
- Necessity and effects of tariff increases:
  - Gas and heating tariff increases are necessary to address economic inefficiencies and distributional distortions created by below-cost pricing.
  - Tariff increases reduce implicit subsidies favoring larger (wealthy) consumers, lower overconsumption, and incentivize adoption of energy-saving technologies.
  - Over time, industrial energy efficiency would increase with less illegal siphoning of gas from households to industries as the price differential shrinks.
- Phasing and design:
  - A schedule of gradual tariff increases will provide incentives for reduced energy consumption while ensuring affordability, mitigating impact on poor and middle class, and giving time to adjust.
  - A single measured tariff increase (example: 20 percent) has little impact on the welfare of the wealthy; medium-term credible schedule needed to motivate lower consumption by wealthy.
- Social assistance reform:
  - Reformed, better-targeted social assistance can partially compensate the poor and middle class and reduce distributional distortions.
  - Eliminating untargeted benefits received by wealthy households would free almost 0.2 percent of GDP for targeted assistance.
  - A reformed system could ensure no household in the first four quintiles pays more than 15 percent of income for utilities expenditures.
  - Cash compensation for consumer surplus losses can enhance public support for subsidy phase-out.
- Use of government savings:
  - Under-pricing gas and heating provides a 5 percent of GDP subsidy to households.
  - Even with just a 20 percent tariff increase the government will gain almost 0.3 percent of GDP.
  - Savings should first finance energy-saving infrastructure investment to reduce the burden of tariff increases on households.
  - Additional savings from gradual tariff increases should be invested in domestic gas exploration and development and growth-enhancing activities to produce greater long-term gains.

*Source: IMF working paper content provided in the supplied PDF excerpt.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp12247.pdf_