## Appendix I. Utility Pricing and Regulation

## Source details

**Canonical URL:** [Appendix I. Utility Pricing and Regulation](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023118-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023118-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023118-print-pdf.pdf.json)

---

### Rule for Awarding Lump-sum Discounts to Households across Locality Welfare Deciles
- Mechanism:
  - Discounts are lump-sum transfers independent of electricity consumption; eligibility based on locality median per capita income.
  - Localities are ranked by median per capita income and divided into locality welfare deciles.
  - Discount generosity by locality welfare decile:
    - Decile 1: B
    - Decile 2: 0.9*B
    - Decile 3: 0.8*B
    - Decile 4: 0.7*B
    - Decile 5: 0.4*B
    - Decile 6: 0.2*B
    - Decile >= Decile 7: B = 0
- Key parameters and financing:
  - Reference benefit amount B = $51.1/month per household (corresponds to consumption of 260kWh/month).
  - For Decile 1 households with median household size of 7, B = $51.1/month corresponds to 37 kWh/person.
  - Flat tariff to finance discounts: $0.2/kWh (27 percent above the cost-recovery tariff of $0.16/kWh).
  - Total discount budget in the illustration: 0.27 percent of GDP.
  - Contextual baseline numbers:
    - Baseline IBT average tariff across household consumers: $0.105 per kWh (66 percent of the cost-recovery tariff of $0.16 per kWh).
    - Baseline fiscal cost of the IBT to the budget: 0.7 percent of GDP.
    - Total subsidies under the baseline IBT example: about 0.45 percent of GDP.
- Intended effects and empirical illustration results:
  - Caps maximum tariff faced by high-consumption households at $0.2/kWh, lower than scaled-up IBT top-block tariffs ($0.38 and $0.60 per kWh in Reform 1).
  - Low-welfare deciles: average prices fall for households in the lowest welfare deciles receiving larger discounts; a large majority of households in the bottom four welfare deciles are better off under geographic targeting than under the scaled-up IBT.
  - Progressivity: financing burden is more concentrated on higher-welfare households and distribution of benefits is more progressive than scaled-up IBT.
  - Losses among the lowest welfare households are reduced relative to scaled-up IBT, though horizontal inequities remain (vulnerable households in higher-welfare localities can be excluded).
  - Compared to baseline IBT, the geography-targeted discount budget (0.27 percent of GDP) is considerably smaller than baseline IBT subsidies (about 0.45 percent of GDP) while allowing more generous lump-sum benefits for included households (e.g., most generous lump-sum discount $50/month vs $16 under the baseline IBT in the example).
  - Reform yields some winners among low-welfare households despite an overall increase in the average tariff.
- Implementation considerations and data requirements:
  - Depends on quality and granularity of locality-level welfare proxies (e.g., poverty maps, housing prices).
  - Homogeneity of household welfare within localities matters; greater heterogeneity increases exclusion errors.
  - Utility-available information (household location) is sufficient to implement geographic targeting; no household-level PMT data required for this reform.
  - Financing via a higher flat tariff requires acceptance of charging above cost recovery while providing explicit bill discounts to targeted households.
- Trade-offs and policy insights:
  - Geographic lump-sum discounts balance fiscal consolidation and protection better than scaling up IBT top-block tariffs:
    - Reduce extreme tariff spikes that can incentivize going off-grid by capping tariffs at $0.2/kWh versus $0.38 and $0.60/kWh under Reform 1.
    - Improve progressivity and reduce magnitude of losses among many low-welfare households while using a smaller subsidy budget (0.27 percent of GDP vs 0.45 percent of GDP).
  - Limitations:
    - Geographic targeting is relatively crude and can exclude vulnerable households in higher-welfare localities (horizontal inequity).
    - Better targeting (e.g., household-level proxy-means test or per capita allocation) requires more detailed household data but can reduce exclusion errors and improve progressivity.
  - Policy mitigations:
    - Strengthen locality welfare data and granularity.
    - Consider per capita rather than household-level benefits to better protect larger, lower-welfare households.
    - Integrate discount schemes with social safety net databases where feasible.

### Box 2 — Cliffhanger Effects Under Volume Differentiated Tariffs (VDT)
- Concept and mechanism:
  - VDT applies a single flat tariff rate to all consumption of a user, with the tariff rate increasing as consumption increases.
  - The higher tariff applies to all consumption once a higher consumption tier is reached, creating "cliff edge" effects where small increases in consumption can lead to large bill increases.
- Comparison with increasing-block tariffs (IBT) — key findings:
  - VDT yields a higher average tariff at higher consumption levels because the higher tariff applies to all consumption rather than only the increment above a block threshold.
  - Households that cross a block threshold can see large increases in final bills; example: households with consumption just above the maximum in the third consumption block could see their electricity bill increase by 65 percent.
  - Cliff-edge effects disproportionately harm lower welfare households.
- Distributional implications:
  - Within-decile dispersion of budget shares is substantially larger under VDT than under IBT because households in every welfare decile have consumption that falls into every block.
  - Larger within-decile dispersion reflects exacerbated horizontal inequities; lower welfare households with different consumption levels (e.g., different household size) are treated very differently.
- Figure notes and data:
  - Panel A: dotted line = final bill under baseline IBT; solid lines = VDT.
  - VDT final bills were calculated using the same blocks and prices as the baseline IBT.
  - Source: Authors’ calculations using household survey data.

### Regression-based Proxy-means Test (PMT) Design and Targeting Performance
- Regression methodology:
  - A linear regression of per capita consumption on household characteristics:
    - a. Household size
    - b. Urban or Rural
    - c. Province
    - d. Age of household members: number below 5 years, between 6 and 10 years-old, and older than 66
    - e. Type of dwelling
    - f. Area of dwelling
    - g. Wall material
    - h. Rent value
    - i. Total number of rooms
    - j. Assets: eleven variables (number of refrigerators, TVs, computers, private cars, washers, freezers, ovens, microwaves, dishwashers, vacuum cleaners, and tablets)
- Prediction and targeting procedure:
  - Predicted per capita consumption estimated for each household from regression results.
  - Households categorized into deciles based on predicted per capita consumption.
  - Discounts awarded according to the rule in Table 1 (as applied in the PMT scheme).
- Targeting accuracy (comparative metrics Geographic vs PMT):
  - "% of HH" by welfare decile matched to targeted deciles:
    - Decile 1: 68% Geographic, 90% PMT
    - Decile 2: 55% Geographic, 72% PMT
    - Decile 3: 52% Geographic, 55% PMT
    - Decile 4: 40% Geographic, 37% PMT
    - Decile 5: 22% Geographic, 22% PMT
    - Decile 6: 18% Geographic, 14% PMT
    - Decile 7: 17% Geographic, 6% PMT
    - Decile 8: 12% Geographic, 3% PMT
    - Decile 9: 12% Geographic, 1% PMT
    - Decile 10: 6% Geographic, 0% PMT
  - Paired numeric entries (selected per-decile examples presented in appendix tables):
    - Decile 1: 6.8 and 6.87; 23% and 10%; 6.53 and 5.79; 9% and 0%; 6.98 and 5.45
    - Decile 2: 5.94 and 6.29; 28% and 26%; 6.19 and 5.9; 17% and 2%; 6.73 and 4.33
    - Decile 3: 5.5 and 5.76; 29% and 40%; 5.7 and 5.55; 19% and 5%; 5.85 and 4.76
    - Decile 4: 5.06 and 5.61; 40% and 51%; 5.32 and 5.35; 20% and 12%; 5.9 and 4.45
    - Decile 5: 4.75 and 5.28; 53% and 57%; 5.01 and 5.24; 25% and 21%; 5.33 and 4.24
    - Decile 6: 4.36 and 4.74; 42% and 49%; 4.6 and 5.1; 41% and 37%; 4.85 and 4.04
    - Decile 7: 4.18 and 4.67; 28% and 36%; 4.16 and 4.98; 55% and 57%; 4.52 and 3.94
    - Decile 8: 3.35 and 3.99; 26% and 20%; 3.45 and 4.68; 62% and 77%; 4.2 and 3.7
    - Decile 9: 3.28 and 3.44; 20% and 10%; 3.29 and 4.59; 68% and 89%; 3.71 and 3.46
    - Decile 10: 2.6 and 1.56; 11% and 2%; 2.81 and 4.67; 83% and 98%; 3.09 and 3
- Performance summary:
  - PMT yields higher coverage of the poorest deciles (e.g., Decile 1: 90% under PMT vs 68% under Geographic).
  - Geographic targeting performs worse at reaching the poorest deciles but may be easier to implement using utility-held location data.
- Distributional impacts visualization notes:
  - Appendix Figure 2 shows change in household budget (change in electricity bill) by welfare decile; dots = average change; dot sizes = percentage of losers and winners within each decile.
  - Horizontal axis scale shown from -15 to 5 (marks at -15, -10, -5, 0, 5) labeled "Change in Household Budget".
  - Figure panels include: Geographic PC, PMT, PMT PC, Baseline IBT, Scaled-up IBT, Geographic, with welfare deciles 1 through 10 on vertical axis and percent quantiles 25, 50, 75 for "Change in Household Budget by Welfare Decile".

*Source — IMF Working Paper section “Rule for Awarding Lump-sum Discounts to Households across Locality Welfare Deciles”; Box 2 “Cliffhanger Effects Under Volume Differentiated Tariffs”; regression and PMT appendix material (excerpt).*

### Appendix I. Utility Pricing and Regulation .............................................................................

### Appendix I. Utility Pricing and Regulation

### Contents overview
- Appears on page "27"
- Listed as part of the PDF: "wpiea2023118-print-pdf - Appendix I. Utility Pricing and Regulation ...................................................................................................... 27"

### Boxes referenced
- "1 Cost Recovery Components and Efficiency ....................................................................................................... 7"
- "2. Cliffhanger Effects Under Volume Differentiated Tariffs ................................................................................. 26"

### Figures referenced
- "1. Electricity Cost-recovery Rates in Industrial and Residential Sectors ............................................................. 10"
- "2. Fiscal Burden of Electricity Subsidies ............................................................................................................. 11"
- "3. Baseline Tariff Schedule ................................................................................................................................. 13"
- "4. Scaled-up IBT ................................................................................................................................................. 16"
- "5. Geographic Household Level Discount Scheme ............................................................................................. 19"
- "6. Geographic Per Capita Discount Scheme by Welfare Decile ......................................................................... 20"
- "7. Proxy-means Test Discount Scheme .............................................................................................................. 22"
- "8. PMT Per Capita Discount Scheme ................................................................................................................. 23"
- "10. Volume Differentiated Tariffs ........................................................................... Error! Bookmark not defined."

### Tables referenced
- "1. Summary of Reforms ...................................................................................................................................... 15"

*Source: wpiea2023118-print-pdf - Appendix I. Utility Pricing and Regulation (wpiea2023118-print-pdf.pdf)*

### 2. Rule for Awarding Lump-sum Discounts to Households across Locality Welfare Deciles .............................. 18

### 2. Rule for Awarding Lump-sum Discounts to Households across Locality Welfare Deciles .............................. 18

### Rule description and implementation mechanics
- Discounts are awarded as lump-sum transfers that do not depend on electricity consumption levels but rather on proxies for household welfare (i.e., locality median per capita income).  
- Localities (groupings of census tracts) are ranked by median per capita income and divided into locality welfare deciles.  
- Discount generosity decreases with locality welfare decile following the rule specified in the table:
  - Decile 1: B
  - Decile 2: 0.9*B
  - Decile 3: 0.8*B
  - Decile 4: 0.7*B
  - Decile 5: 0.4*B
  - Decile 6: 0.2*B
  - Decile >= Decile 7: B = 0

### Key parameters, reference amounts, and fiscal financing
- Reference benefit amount B is set to $51.1/month per household (corresponds to consumption of 260kWh/month).  
- For Decile 1 households with median household size of 7, B = $51.1/month corresponds to 37 kWh/person.  
- The flat tariff charged to finance the discounts is $0.2/kWh (27 percent above the cost-recovery tariff of $0.16/kWh).  
- The total discount budget used in the illustration is equivalent to 0.27 percent of GDP.  
- Contextual baseline and related numbers used in the broader analysis:
  - Baseline IBT average tariff across household consumers: $0.105 per kWh (66 percent of the cost-recovery tariff of $0.16 per kWh).  
  - Baseline fiscal cost of the IBT to the budget: 0.7 percent of GDP.  
  - Total subsidies under the baseline IBT example: about 0.45 percent of GDP (comparison reference).

### Intended effects and empirical findings (illustration results)
- The flat above-cost-recovery tariff ($0.2/kWh) combined with locality-based lump-sum benefits:
  - Caps the maximum tariff faced by high-consumption households at $0.2/kWh, which is lower than the top-block tariffs under a scaled-up IBT (top two blocks become $0.38 and $0.60 per kWh in Reform 1).  
  - Brings average prices down for households in the lowest welfare deciles receiving larger discounts; a large majority of households in the bottom four welfare deciles are better off under this geographic targeting than under the scaled-up IBT.  
  - Improves overall progressivity relative to scaled-up IBT: the financing burden is more concentrated on higher-welfare households and the distribution of benefits is more progressive.  
  - Reduces the magnitude of losses among the lowest welfare households compared with scaled-up IBT, though horizontal inequities remain (vulnerable households located in higher-welfare localities can be excluded and thus hit hard).  
- Compared to the baseline IBT:
  - The geography-targeted discount budget (0.27 percent of GDP) is considerably smaller than baseline IBT subsidies (about 0.45 percent of GDP), yet allows more generous lump-sum benefits for those included (e.g., the most generous lump-sum discount at $50/month vs $16 under the baseline IBT in the example).  
  - The reform yields some winners among low-welfare households despite an overall increase in the average tariff.

### Implementation considerations and data requirements
- The effectiveness of locality-based targeting depends on:
  - Quality and granularity of locality-level welfare proxies (e.g., poverty maps, housing prices).  
  - Degree of homogeneity of household welfare within localities; more heterogeneity raises exclusion errors for vulnerable households.  
- Information typically available to utilities (household location) is sufficient to implement geographic targeting; no household-level proxy-means information is required for Reform 2.  
- Financing the discounts via a higher flat tariff requires that utilities or policymakers accept charging a flat tariff above cost recovery while providing explicit, visible bill discounts to targeted households.

### Trade-offs and policy insights
- Geographic lump-sum discounts (Reform 2) balance fiscal consolidation and protection better than scaling up IBT top-block tariffs:
  - They reduce extreme tariff spikes that can incentivize going off-grid (by capping tariffs at $0.2/kWh vs top-block rates $0.38 and $0.60/kWh under Reform 1).  
  - They improve progressivity and reduce the magnitude of losses among many low-welfare households while using a smaller subsidy budget (0.27 percent of GDP vs 0.45 percent of GDP).  
- Remaining limitations:
  - Geographic targeting is a relatively crude instrument and can exclude vulnerable households living in higher-welfare localities (horizontal inequity).  
  - Better targeting (e.g., proxy-means test at household level or per capita allocation) requires more detailed household data but can further reduce exclusion errors and improve progressivity.  
- Policy design options to mitigate exclusion:
  - Strengthen locality welfare data and granularity.  
  - Consider per capita rather than household-level benefits to better protect larger, lower-welfare households.  
  - Integrate discount schemes with broader social safety net databases where feasible to improve targeting accuracy.

*Italic: Source — IMF Working Paper section “Rule for Awarding Lump-sum Discounts to Households across Locality Welfare Deciles” (excerpt).*

### Box 2. Cliffhanger Effects Under Volume Differentiated Tariffs

### Box 2. Cliffhanger Effects Under Volume Differentiated Tariffs

### Concept and mechanism of Volume Differentiated Tariffs (VDT)
- Volume differentiated tariffs (VDT) apply a single flat tariff rate to all consumption of a user, with the tariff rate increasing as consumption increases.
- Under VDT the higher tariff is applied to all consumption (not just the increment above a block threshold), creating potential "cliff edge" effects where small increases in consumption can lead to large increases in the final bill.
- VDT schemes are often observed in practice or discussed as policy options.

### Comparison with increasing-block tariffs (IBT) — key findings
- The VDT results in a higher average tariff at higher consumption levels because the higher tariff applies to all consumption rather than only to consumption within that tariff block.
- This mechanism increases household electricity bills, particularly for households whose consumption crosses a block threshold.
- Example finding: households with consumption just above the maximum in the third consumption block could see their electricity bill increase by 65 percent.
- Such “cliff edge” effects can be particularly problematic for lower welfare households.

### Distributional implications (Panel B)
- Households in every welfare decile have consumption that falls into every block; as a result, the range of budget shares within any given welfare decile is substantially larger under VDT than under IBT.
- Panel B shows the distribution of budget shares by welfare decile for both tariff schedules:
  - Dots represent the average budget share corresponding to the electricity bill among households within each welfare decile.
  - Bars represent the 5th and 95th percentiles of the distribution of budget shares within each welfare decile.
- The larger within-decile dispersion under VDT reflects exacerbated horizontal inequities where lower welfare households with different consumption levels (e.g., reflecting different household size) are treated very differently.

### Figure notes and data provenance
- Panel A: dotted line represents final bill amount as a function of quantity consumed under the baseline IBT; solid lines correspond to the VDT.
- Final bills corresponding to the VDT were calculated based on the same blocks and prices as the baseline IBT.
- Source: Authors’ calculations using household survey data.

*Source: Box 2. Cliffhanger Effects Under Volume Differentiated Tariffs, IMF Working Paper content provided.*

### 1. A linear regression was estimated of per capita consumption on the following household characteristics:

### 1. A linear regression was estimated of per capita consumption on the following household characteristics:

### Methodology and Variables
- A linear regression model was estimated of per capita consumption on the following household characteristics:
  - a. Household size
  - b. Urban or Rural
  - c. Province
  - d. Age of household members. Three variables were used: number of household members below 5 years of age, between 6 and 10 years-old, and older than 66.
  - e. Type of dwelling
  - f. Area of dwelling
  - g. Wall material
  - h. Rent value
  - i. Total number of rooms
  - j. Assets. Eleven variables: number of refrigerators, TVs, computers, private cars, washers, freezers, ovens, microwaves, dishwashers, vacuum cleaners, and tablets;

### Prediction and Targeting Procedure
- Based on the regression model results, a predicted value of per capita consumption was estimated for each household;
- Households were categorized into deciles based on the predicted value of per capita consumption;
- Discounts were awarded according to the rule in Table 1.

### Performance of the Proxy-means Test (PMT)
- Appendix Figure 1 notes:
  - Figure displays the transition matrix comparing actual per capita household income with predicted per capita income, with households allocated into welfare deciles according to each criterion.
  - Values represent percentage of households falling in each cell.
  - Welfare is based on household food and non-food consumption.
- The figure indicates percent match on a 0.0 to 0.6 Percent Match scale (labels shown: 0.0, 0.2, 0.4, 0.6).

### Targeting Accuracy (Appendix Table 2)
- Note: For “% of HH”, each cell in the table displays the percentage of households belonging to both the corresponding welfare decile (rows) and the targeted deciles (columns). Targeted deciles in Geographic targeting are based on the estimated welfare of the areas where households are located. Targeted deciles in the PMT targeting are based on the household-level predicted values of a proxy-means-test, as explained in the text and in Annex II.
- "% of HH" by welfare decile: Geographic vs PMT
  - Decile 1: 68% Geographic, 90% PMT
  - Decile 2: 55% Geographic, 72% PMT
  - Decile 3: 52% Geographic, 55% PMT
  - Decile 4: 40% Geographic, 37% PMT
  - Decile 5: 22% Geographic, 22% PMT
  - Decile 6: 18% Geographic, 14% PMT
  - Decile 7: 17% Geographic, 6% PMT
  - Decile 8: 12% Geographic, 3% PMT
  - Decile 9: 12% Geographic, 1% PMT
  - Decile 10: 6% Geographic, 0% PMT
- Paired numeric entries (table presents additional paired statistics per decile; examples as shown in the table):
  - Decile 1: 6.8 and 6.87; 23% and 10%; 6.53 and 5.79; 9% and 0%; 6.98 and 5.45
  - Decile 2: 5.94 and 6.29; 28% and 26%; 6.19 and 5.9; 17% and 2%; 6.73 and 4.33
  - Decile 3: 5.5 and 5.76; 29% and 40%; 5.7 and 5.55; 19% and 5%; 5.85 and 4.76
  - Decile 4: 5.06 and 5.61; 40% and 51%; 5.32 and 5.35; 20% and 12%; 5.9 and 4.45
  - Decile 5: 4.75 and 5.28; 53% and 57%; 5.01 and 5.24; 25% and 21%; 5.33 and 4.24
  - Decile 6: 4.36 and 4.74; 42% and 49%; 4.6 and 5.1; 41% and 37%; 4.85 and 4.04
  - Decile 7: 4.18 and 4.67; 28% and 36%; 4.16 and 4.98; 55% and 57%; 4.52 and 3.94
  - Decile 8: 3.35 and 3.99; 26% and 20%; 3.45 and 4.68; 62% and 77%; 4.2 and 3.7
  - Decile 9: 3.28 and 3.44; 20% and 10%; 3.29 and 4.59; 68% and 89%; 3.71 and 3.46
  - Decile 10: 2.6 and 1.56; 11% and 2%; 2.81 and 4.67; 83% and 98%; 3.09 and 3

### Distributional Impacts: Winners and Losers (Appendix Figure 2)
- Notes on Appendix Figure 2:
  - Change in household budget reflects the change in electricity bill resulting from the reform.
  - Dots represent average change in household budget.
  - Dot sizes represent percentage of losers and winners within each welfare decile.
  - Further details are provided in Annex II.
- Figure panels and labels shown include: Geographic PC, PMT, PMT PC, Baseline IBT, Scaled-up IBT, Geographic.
- Horizontal axis scale shown from -15 to 5 (marks at -15, -10, -5, 0, 5) labeled "Change in Household Budget".
- Vertical axis shows Welfare Decile 1 through 10 and percent quantiles 25, 50, 75 for "Change in Household Budget by Welfare Decile".

*IMF Working Papers — The Distributional and Fiscal Implications of Public Utility Pricing, Working Paper No. WP/2023/118*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023118-print-pdf.pdf_