## _wp06126

## Source details

**Canonical URL:** [_wp06126](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06126.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06126.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06126.pdf.json)

---

### Introduction and scope
- HIV/AIDS identified as one of the largest obstacles to development in many countries, particularly in Africa.
- Life expectancy estimated to have declined by more than 20 years in some worst-affected countries (United Nations Populations Division, 2005).
- Average life expectancy in Africa has been declining as a consequence of HIV/AIDS since 1990 (by 2½ years overall).
- Focus: impact of HIV/AIDS on poverty and implications for attaining MDG 1 (“to eradicate extreme poverty and hunger”), with recognition of implications for MDGs 2–5 and MDG 3.

### Channels through which HIV/AIDS affects poverty and growth
- Direct health effects: increased mortality and morbidity among working-age population.
- Labor-force channel: deaths among working-age segments reduce the workforce and unambiguously lower GDP growth rates.
- GDP per capita channel: evidence mixed; models treating mortality as exogenous can imply mitigating effects where unemployed workers replace dying workers.
- Longer-run channels: higher domestic production costs, reduced attractiveness for investment, and adverse effects on human capital accumulation.

### Data description and key country indicators
- Countries analyzed: Ghana, Kenya, Swaziland, Zambia.
- HIV prevalence, end-2003 (percent of population, ages 15-49): Ghana 3.1; Kenya 6.7; Swaziland 38.8; Zambia 16.5.
- Crude death rate, 2003 (percent): Ghana 1.1; Kenya 1.6; Swaziland 2.7; Zambia 2.3.
  - Of which: AIDS (percent): Ghana 0.1; Kenya 0.5; Swaziland 1.6; Zambia 0.8.
- Life expectancy at birth, 1985-1990 (years): Ghana 56; Kenya 59; Swaziland 57; Zambia 49.
- Life expectancy at birth, 2000-2005 (years): Ghana 57; Kenya 47; Swaziland 33; Zambia 37.
- Total orphans, 2003 (percent of population, ages 0-17): Ghana 10.0; Kenya 11.3; Swaziland 16.7; Zambia 18.3.
  - Of which: AIDS orphans (percent): Ghana 1.7; Kenya 4.3; Swaziland 10.8; Zambia 10.5.
- Poverty and inequality (selected):
  - Average daily per capita income (US$): Ghana 2.05; Kenya 2.43; Swaziland 2.06; Zambia 1.53.
  - Population below $1 per day (in percent): Ghana 44.7; Kenya 27.8; Swaziland 39.7; Zambia 56.9.
  - Population below $2 per day (in percent): Ghana 73.7; Kenya 63.6; Swaziland 74.4; Zambia 83.0.
  - Gini coefficient: Ghana 0.54; Kenya 0.48; Swaziland 0.51; Zambia 0.48.
  - Income dependency ratio: Ghana 3.06; Kenya 2.68; Swaziland 3.47; Zambia 2.81.
- Distributional patterns:
  - Prevalence by sex (In percent): Ghana Male 1.5 / Female 2.7; Kenya Male 4.6 / Female 8.7; Swaziland Male 30.9 / Female 38.6; Zambia Male 12.9 / Female 17.8.
  - Urban vs Rural prevalence (In percent): Ghana Urban 2.4 / Rural 2.0; Kenya Urban 9.7 / Rural 5.3; Swaziland Urban 36.5 / Rural 32.3; Zambia Urban 23.8 / Rural 10.2.
  - Prevalence by income quintile examples (In percent): Ghana Poorest 1.4 / Second 2.2 / Middle 3.1 / Fourth 2.3 / Richest 2.1; Kenya Poorest 3.3 / Second 6.0 / Middle 4.7 / Fourth 7.3 / Richest 9.7.
  - Prevalence tends to be higher in urban areas and, where available by wealth quintile, appears lower in the poorest segments; prevalence often high just above the poverty line.

### Methodology
- Two-step simulation analysis:
  1. Randomly assign HIV status to individuals in expenditure surveys based on socioeconomic characteristics and prevalence by subgroup.
  2. Simulate economic impact using assumptions about morbidity and mortality costs, income effects, labor market replacement, and social insurance mechanisms; estimate effects on income per capita, income distribution, and poverty rates over a 10-year horizon.
- Implementation details:
  - Simulations executed in Stata/SE 8.0; procedure reiterated fifty times per scenario; final estimates are the average of the fifty outputs.
  - Prevalence imposition rules vary by country (age, sex, region, urban/rural, expenditure quintile or sentinel site).
  - If subgroup sample small, assign average prevalence; if outside tested age range, assign half the nearest sampled age-group prevalence.
  - Assume male-to-female prevalence ratio ~82 percent (UNAIDS, 2004).
  - Infection timing: HIV-infected individuals assumed to live an average of ten years with the infection; year of infection drawn uniformly; new infections simulated annually similarly.
- Economic impact assumptions (exact parameters preserved):
  - HIV health-related expenditures assumed equivalent to 25 percent of the income of a household worker in urban areas, and to 50 percent in rural areas.
  - Funeral expenditures for every HIV casualty equivalent to four months of household expenditures.
  - Reduction of 15 percent in the income of any worker in the household who is HIV-infected and in the last two years of infection.
  - Labor market: job vacated by death filled by a previously unemployed individual within half a year (0.5 year); replacements drawn from same income decile with perfect geographic, gender, and age mobility.
  - Social insurance baseline: household left without any income earner receives an allowance of 25 percent of its initial income.
- Important modeling simplifications:
  - HIV prevalence rates held constant at survey rates.
  - Population does not age; household structure remains constant except for mortality-induced changes.
  - Infection transmission between individuals and household composition dynamics (beyond simple financial support) are not modeled.

### Impact of HIV/AIDS on poverty and inequality — Base Case Scenario (selected exact indicators)
- Base Case assumptions: jobs replaced after 0.5 year; households left with no income earner receive a 25 percent allowance.
- Table 5: Base Case, Year 0 → Year 10 → Change (%)
  - Ghana:
    - Income per capita (US$): Year 0 2.49; Year 10 2.47; Change (%) -0.55
    - Poverty headcount ($2/day,%) Year 0 66.38; Year 10 66.38; Change (%) 0.003
    - Poverty gap ($2/day,%) Year 0 34.21; Year 10 34.99; Change (%) 2.3
    - Poverty headcount ($1/day,%) Year 0 34.15; Year 10 35.74; Change (%) 4.6
    - Income dependency ratio: Year 0 2.44; Year 10 2.44; Change (%) -0.4
    - Gini coefficient: Year 0 53.85; Year 10 54.49; Change (%) 1.2
  - Kenya:
    - Income per capita (US$): Year 0 2.80; Year 10 2.78; Change (%) -0.90
    - Poverty headcount ($2/day,%) Year 0 58.92; Year 10 60.53; Change (%) 2.7
    - Poverty gap ($2/day,%) Year 0 25.17; Year 10 27.06; Change (%) 7.5
    - Poverty headcount ($1/day,%) Year 0 23.46; Year 10 25.80; Change (%) 10.0
    - Income dependency ratio: Year 0 2.68; Year 10 2.64; Change (%) -2.0
    - Gini coefficient: Year 0 48.43; Year 10 50.41; Change (%) 4.1
  - Swaziland:
    - Income per capita (US$): Year 0 2.87; Year 10 2.65; Change (%) -7.51
    - Poverty headcount ($2/day,%) Year 0 59.90; Year 10 64.00; Change (%) 6.9
    - Poverty gap ($2/day,%) Year 0 26.00; Year 10 33.36; Change (%) 28.3
    - Poverty headcount ($1/day,%) Year 0 23.28; Year 10 33.00; Change (%) 41.8
    - Income dependency ratio: Year 0 3.47; Year 10 3.00; Change (%) -14.3
    - Gini coefficient: Year 0 50.63; Year 10 55.74; Change (%) 10.1
  - Zambia:
    - Income per capita (US$): Year 0 1.75; Year 10 1.57; Change (%) -9.95
    - Poverty headcount ($2/day,%) Year 0 79.17; Year 10 81.48; Change (%) 2.9
    - Poverty gap ($2/day,%) Year 0 41.67; Year 10 48.59; Change (%) 16.6
    - Poverty headcount ($1/day,%) Year 0 51.04; Year 10 57.52; Change (%) 12.7
    - Income dependency ratio: Year 0 2.81; Year 10 2.67; Change (%) -4.7
    - Gini coefficient: Year 0 47.77; Year 10 53.23; Change (%) 11.4
- Key base-case findings:
  - Fall in average income is significant in high-prevalence countries (Swaziland and Zambia); in single-digit prevalence countries (Ghana and Kenya), income falls below 1 percent over 10 years.
  - Relation between income impact and prevalence not strictly positive: average income falls by a larger share in Zambia (10 percent) than in Swaziland (7.5 percent) despite higher prevalence in Swaziland, reflecting distributional characteristics.
  - Poverty and inequality effects heterogeneous: dramatic increases in $1/day poverty and Gini coefficient in high-prevalence countries (Swaziland: $1/day headcount up 41.8 percent; Gini up 10.1 percent).

### Worker-replacement scenarios and impacts (selected exact results)
- Scenarios modeled:
  - Baseline: vacant jobs filled after 0.5 year.
  - Immediate worker replacement: unemployed fill jobs immediately.
  - No worker replacement: deceased workers are not replaced.
- Baseline (selected repeats of Table 5 shown above).
- Immediate worker replacement — selected changes:
  - Income per capita (US$): Ghana 2.49 → 2.48 → -0.18; Kenya 2.80 → 2.80 → -0.26; Swaziland 2.87 → 2.79 → -4.12; Zambia 1.75 → 1.62 → -6.25.
  - Poverty headcount ($2/day,%): Ghana 66.38 → 66.38 → 0.001; Kenya 58.92 → 59.38 → 0.8; Swaziland 59.90 → 62.14 → 3.8; Zambia 79.17 → 80.62 → 1.8.
  - Gini coefficient: Ghana 53.85 → 54.06 → 0.4; Kenya 48.43 → 49.00 → 1.2; Swaziland 50.63 → 53.43 → 5.5; Zambia 47.77 → 51.20 → 7.2.
- No worker replacement — selected changes:
  - Income per capita (US$): Ghana 2.49 → 2.46 → -1.29; Kenya 2.80 → 2.68 → -4.60; Swaziland 2.87 → 2.46 → -14.38; Zambia 1.75 → 1.51 → -13.65.
  - Poverty headcount ($1/day,%): Ghana 34.15 → 35.90 → 5.1; Kenya 23.46 → 26.53 → 13.1; Swaziland 23.28 → 35.53 → 52.6; Zambia 51.04 → 58.66 → 14.9.
  - Gini coefficient: Ghana 53.85 → 54.45 → 1.1; Kenya 48.43 → 49.63 → 2.4; Swaziland 50.63 → 56.35 → 11.3; Zambia 47.77 → 53.48 → 11.9.
- Interpretation:
  - Immediate replacement minimizes impacts; no replacement maximizes negative effects.
  - Falls in income per capita largest where HIV prevalence exceeds unemployment (Swaziland, Zambia).
  - Under no-worker-replacement, $1-a-day poverty in Swaziland increases by more than 50 percent.

### Social insurance mechanisms: modeled variants and effects (selected exact results)
- Horizontal-transfer scenarios within same income decile:
  - "Free fall" (no support).
  - Baseline ("25% allowance"): transfer to households whose income would otherwise fall to zero of 25 percent of initial income.
  - "75% allowance": transfer of 75 percent of initial income.
- Selected impacts:
  - "Free fall" poverty headcount ($2/day,%): Ghana 66.38 → 66.57 → 0.3; Kenya 58.92 → 61.04 → 3.6; Swaziland 59.90 → 65.33 → 9.1; Zambia 79.17 → 82.53 → 4.3.
  - "Baseline (25% allowance)" poverty gap ($2/day,%): Ghana 34.21 → 34.99 → 2.3; Kenya 25.17 → 27.06 → 7.5; Swaziland 26.00 → 33.36 → 28.3; Zambia 41.67 → 48.59 → 16.6.
  - "75% allowance" Gini coefficient changes: Ghana 53.85 → 54.34 → 0.9; Kenya 48.43 → 49.61 → 2.4; Swaziland 50.63 → 53.34 → 5.3; Zambia 47.77 → 49.68 → 4.0.
- Interpretation:
  - Horizontal transfers do not change average income per capita but mitigate depth of poverty increases.
  - Even a 25 percent allowance reduces increases in headcount poverty and the poverty gap.
  - Horizontal support reduces depth but imposes burdens on supporting households; vertical transfers (government programs) not modeled.

### Determinants of income and poverty impacts; normalized results
- Main determinants:
  - HIV prevalence (scale of epidemic).
  - Distributional factors: prevalence concentrated among higher-income groups raises aggregate income impacts; prevalence concentrated among those just above the poverty line raises poverty impacts.
  - Size of the population at risk (proportion just above the poverty line).
  - Age profile and geographic distribution (urban vs. rural).
- Normalized impact assuming 5 percent HIV prevalence (Table 8, Baseline scenario, exact values):
  - Income per capita (US$): Ghana -1.30; Kenya -0.67; Swaziland -1.20; Zambia -3.19.
  - Poverty headcount ($2/day, in percent): Ghana 0.01; Kenya 2.05; Swaziland 1.09; Zambia 0.94.
  - Poverty gap ($2/day, in percent): Ghana 5.40; Kenya 5.59; Swaziland 4.51; Zambia 5.33.
  - Poverty headcount ($1/day, in percent): Ghana 11.05; Kenya 7.43; Swaziland 6.65; Zambia 4.07.
  - Income dependency ratio: Ghana -0.85; Kenya -1.50; Swaziland -2.27; Zambia -1.52.
  - Gini coefficient: Ghana 2.79; Kenya 3.05; Swaziland 1.61; Zambia 3.66.
- Income-per-decile patterns (selected exact decile-change vectors):
  - Ghana deciles 1–10 (% change): -0.8, -0.9, -0.8, -0.8, -3.1, -3.0, -0.1, -0.9, 0.6, -0.5.
  - Kenya deciles 1–10 (% change): -0.2, -0.2, -1.0, -0.6, -1.4, -0.1, -1.1, -0.3, 2.3, -0.2.
  - Swaziland deciles 1–10 (% change): -10.3, -12.0, -11.3, -9.1, -10.0, -8.8, -8.2, -4.8, -1.1, -7.5.
  - Zambia deciles 1–10 (% change): -5.4, -6.5, -5.6, -6.5, -8.1, -8.8, -8.9, -7.5, -5.5, -9.5.
- Location of the $1-a-day line by country (note 16): between deciles 3 and 4 in Ghana; between deciles 2 and 3 in Kenya and Swaziland; between deciles 4 and 5 in Zambia.
- Interpretation:
  - When prevalence higher among higher-income groups (Zambia), aggregate income per capita falls more.
  - When prevalence concentrated just above the poverty line (Kenya, Ghana, Swaziland), poverty impacts are amplified even if average income per capita does not fall much.
  - Size of population just above the poverty line explains high poverty impact in Swaziland.

### Conclusions and policy implications
- Simulation results:
  - Impact of HIV/AIDS on poverty indicators often exceeds impact on average incomes; poverty incidence and poverty gap increase more than expected from declines in income per capita in three sample countries.
  - Key determinants of poverty impact: size of population at risk of falling into poverty and HIV prevalence among that group.
  - Results robust to labor market assumptions and modeled horizontal social insurance functioning.
- Policy implications (preserved as stated):
  - Even modest or negligible changes in average income per capita can conceal substantial increases in poverty; poverty reduction efforts can be frustrated by the epidemic.
  - HIV/AIDS threatens attainment of MDG 1 (“eradicate extreme poverty and hunger”) and other MDGs affected by increased poverty (for example, in education).
  - Informal horizontal support mechanisms have limited capacity to mitigate the impact on poverty; an effective social safety net would likely require vertical transfers (e.g., government programs).
- Methodological implications:
  - Assessments of welfare implications of HIV/AIDS cannot be limited to effects on income per capita.
  - Studies confined to households directly affected can be misleading; broader, distribution-sensitive analysis is needed.
  - Future research will benefit from newer household surveys and Demographic and Health Surveys (DHSs).

*Source: _wp06126 (Authors’ calculations based on HIV-prevalence and income/expenditure surveys listed in Table 2).*

### References..............................................................................................................

### _wp06126 - References

### Introduction and scope
- HIV/AIDS is identified as one of the largest obstacles to development in many countries, particularly in Africa.
- In some of the worst-affected countries, life expectancy is estimated to have declined by more than 20 years (United Nations Populations Division, 2005).
- Average life expectancy in Africa has been declining as a consequence of HIV/AIDS since 1990 (by 2½ years overall).
- The paper focuses on the impact of HIV/AIDS on poverty and the implications for attaining MDG 1 (“to eradicate extreme poverty and hunger”), while recognizing implications for MDGs 2–5 and MDG 3.

### Channels through which HIV/AIDS affects poverty and growth
- Direct health effects: increased mortality and morbidity among working-age population.
- Labor-force channel: deaths among working-age segments reduce the workforce and unambiguously lower GDP growth rates.
- GDP per capita channel: evidence is mixed on whether HIV/AIDS lowers GDP per capita; initial models treating mortality as an exogenous shock suggest mitigating effects where unemployed workers replace dying workers.
- Longer-run channels: higher domestic production costs and reduced attractiveness for investment (Haacker, 2002) and adverse effects on human capital accumulation (Bell and others, 2004) can lead to more negative long-run outcomes.

### Prior empirical and simulation evidence
- Some studies predict an increase in GDP per capita if a decreasing population shares existing stocks of capital and natural resources.
- General equilibrium model simulations have implied that income per capita would be largely unaffected by the HIV/AIDS epidemic (ING Baring, 2000; and Kambou and others, 1993).
- Longer-horizon analyses typically arrive at more negative assessments because of investment and human-capital channels (Haacker, 2002; Bell and others, 2004).

### Household-level impacts and distributional patterns
- Morbidity costs include a fall in income of infected working members and higher HIV-related expenditures for sick individuals; mortality leads to income loss and funeral expenses.
- The disease fuels poverty beyond per capita income effects through differential ability to cope: poorer households have lower access to financial mechanisms and are less able to manage sudden higher expenditures and income reductions (Whiteside, 2002).
- Evidence shows that home-based assistance can fail to reach the poorest segments of society (Steinberg and others, 2002).
- Costs of the disease relative to household income are twice larger in rural than in urban areas (Steinberg and others, 2002).
- In South Africa, incidence, depth, and severity of poverty are higher among households affected by HIV/AIDS; affected households are more likely to experience chronic poverty and income variation (Booysen, 2003).
- Much of the existing evidence is limited to unrepresentative population groups and short time horizons.

### This study: methodology, country choice, and analytical advances
- The study uses simulation analysis to evaluate the impact of HIV/AIDS on poverty more thoroughly, incorporating both morbidity and mortality costs within country-specific income distributions and demographic patterns.
- Four sub-Saharan African countries are analyzed: Ghana, Kenya, Swaziland, and Zambia. These countries were chosen because they represent very different settings in terms of size, level and distribution of HIV/AIDS prevalence, demographic structure, geographical location, and type of data sources.
- The analysis incorporates:
  - Costs and effects of increased mortality.
  - Costs and effects of rising morbidity (reduced labor income of infected individuals and higher HIV-related household expenditures).
  - Funeral expenses triggered by deaths.
  - Mechanisms through which employment vacated by AIDS victims is filled (labor market replacement assumptions).
  - Social insurance mechanisms to support destitute households.
- The cross-country approach allows relating the impact of HIV/AIDS to country-specific demographic and socioeconomic characteristics and the distribution of HIV prevalence across population groups.

### Comparative context and link to earlier work
- Earlier simulations for Botswana (Greener and others, 2000; Greener, 2004) suggested the headcount poverty index rises substantially and that income losses of the poorest quartile (in relative terms) are twice those of the average household; those simulations primarily treated the disease as a one-time mortality shock and did not fully account for morbidity costs.
- The present study extends that work by including morbidity costs, covering several countries (including lower-prevalence cases), and exploiting cross-country differences to identify factors driving heterogeneous outcomes.

*Source: _wp06126 - References (excerpt).*

### Section II discusses the data and projections used in the main simulations; Section III

### _wp06126 - Section II discusses the data and projections used in the main simulations; Section III

### II. DATA DESCRIPTION
- Sample spans countries with very different demographic profiles and HIV prevalence rates.
- Table 1 highlights selected indicators:
  - HIV prevalence, end-2003 (percent of population, ages 15-49): Ghana 3.1; Kenya 6.7; Swaziland 38.8; Zambia 16.5.
  - Crude death rate, 2003 (percent): Ghana 1.1; Kenya 1.6; Swaziland 2.7; Zambia 2.3.
    - Of which: AIDS (percent): Ghana 0.1; Kenya 0.5; Swaziland 1.6; Zambia 0.8.
  - Life expectancy at birth, 1985-1990 (years): Ghana 56; Kenya 59; Swaziland 57; Zambia 49.
  - Life expectancy at birth, 2000-2005 (years): Ghana 57; Kenya 47; Swaziland 33; Zambia 37.
  - Total orphans, 2003 (percent of population, ages 0-17): Ghana 10.0; Kenya 11.3; Swaziland 16.7; Zambia 18.3.
    - Of which: AIDS orphans (percent): Ghana 1.7; Kenya 4.3; Swaziland 10.8; Zambia 10.5.
- Key notes:
  - Ghana: relatively low HIV prevalence (3 percent) though about 10 percent of all deaths can be attributed to HIV/AIDS.
  - Swaziland: currently estimated as highest HIV prevalence worldwide (39 percent noted in narrative; Table 1 value 38.8 percent).
  - High-prevalence countries show substantial increases in mortality, sharp drops in life expectancy, and increased orphan rates.
- Data sources and sample specifics (Table 2):
  - HIV prevalence data sources and years:
    - Ghana: Demographic and health survey; Year 2003; Sample size 9,760 individuals; Overall HIV prevalence 2.1 (In percent).
    - Kenya: Demographic and health survey; Year 2003; Sample size 8,486 individuals; Overall HIV prevalence 6.7.
    - Swaziland: 8th sentinel report; Year 2002; Sample size 2,787 pregnant women; Overall HIV prevalence 31.4.
    - Zambia: Demographic and health survey; Year 2001/02; Sample size 3,950 individuals; Overall HIV prevalence 15.6.
  - Prevalence rates coverage by characteristic: region, urban vs rural, sex, age, wealth (varies by country; see table).
  - Income/expenditure data sources and years:
    - Ghana: Living standards measurement survey, 1998/99; Sample size 50,713 individuals; 10,874 households.
    - Kenya: Welfare monitoring survey, 1997.
    - Swaziland: Household income and expenditure survey, 1995.
    - Zambia: Living conditions monitoring survey, 1998.
  - Income data less available and less reliable than consumption data; household expenditure used as proxy for household income.
- Poverty and inequality key indicators (Table 3):
  - Average daily per capita income (US$): Ghana 2.05; Kenya 2.43; Swaziland 2.06; Zambia 1.53.
  - Population below $1 per day (in percent): Ghana 44.7; Kenya 27.8; Swaziland 39.7; Zambia 56.9.
  - Population below $2 per day (in percent): Ghana 73.7; Kenya 63.6; Swaziland 74.4; Zambia 83.0.
  - Poverty gap ($2 per day; in percent): Ghana 41.1; Kenya 28.6; Swaziland 38.1; Zambia 46.0.
  - Gini coefficient: Ghana 0.54; Kenya 0.48; Swaziland 0.51; Zambia 0.48.
  - Income dependency ratio: Ghana 3.06; Kenya 2.68; Swaziland 3.47; Zambia 2.81.
- Distributional observations (from Table 4 and text):
  - Prevalence by sex (In percent): Ghana Male 1.5 / Female 2.7; Kenya Male 4.6 / Female 8.7; Swaziland Male 30.9 / Female 38.6; Zambia Male 12.9 / Female 17.8 (note: Zambia female/male based also on UNAIDS estimates).
  - Urban vs Rural prevalence (In percent): Ghana Urban 2.4 / Rural 2.0; Kenya Urban 9.7 / Rural 5.3; Swaziland Urban 36.5 / Rural 32.3; Zambia Urban 23.8 / Rural 10.2.
  - Age-specific prevalence (In percent) shown for five-year bands from 15-19 up to 55-60 with exact values provided in Table 4.
  - Prevalence by income quintile (In percent): Ghana Poorest 1.4 / Second 2.2 / Middle 3.1 / Fourth 2.3 / Richest 2.1; Kenya Poorest 3.3 / Second 6.0 / Middle 4.7 / Fourth 7.3 / Richest 9.7.
  - Prevalence tends to be higher in urban areas; where available by wealth quintile, prevalence appears lower in the poorest segments.
  - Prevalence rates tend to be high just above the poverty line (upper end of second quintile in Ghana; lower end in Kenya).

### III. METHODOLOGY
- Two-step analysis:
  1. Randomly assign an HIV status to individuals in the expenditure survey sample based on individual socioeconomic characteristics and HIV prevalence information by category (summarized in Table 4).
  2. Simulate impact using assumptions about costs to affected households, income effects, and existing social insurance mechanisms to estimate effects on income per capita, income distribution, and poverty rates.
- Implementation details:
  - Simulations carried out in Stata/SE 8.0; procedure reiterated fifty times for each scenario; final estimates are the average of the fifty outputs.
  - DHS and Income and Expenditure household surveys available in Stata format for sample countries.
- Procedure for imposing prevalence on expenditure surveys:
  - For Ghana and Kenya: assign prevalence rates by age group, sex, province/region, urban/rural residence, and expenditure quintile (assumed related to wealth quintiles of source database).
  - For Zambia: transpose prevalence rates by age group, sex, province/region, urban/rural residence.
  - For Swaziland: transfer prevalence according to sentinel site and age group.
  - If subgroup has no HIV-positive individuals (small sample), assign average prevalence rate of entire population; if individual falls outside age range tested, assign half the prevalence rate of the nearest sampled age group.
  - Assume ratio between male and female HIV prevalence about 82 percent (based on UNAIDS, 2004).
  - Random assignment allows clustering but tends to overestimate number of households with at least one infection and underestimate households with multiple infections.
- Infection dynamics and timing:
  - For initial infected group, assume HIV-infected individuals live an average of ten years with the infection and randomly determine year of infection from a uniform distribution.
  - Each year determine newly infected individuals through similar random selection.
- Economic impact assumptions:
  - HIV-related health expenditures added to household minimum expenditure (poverty line):
    - HIV health-related expenditures assumed equivalent to 25 percent of the income of a household worker in urban areas, and to 50 percent in rural areas.
    - Funeral expenditures for every HIV casualty equivalent to four months of household expenditures.
  - Income effects:
    - If household member dies, household loses that member's income; per-capita income changes accordingly.
    - Reduction of 15 percent in the income of any worker in the household who is HIV-infected and in the last two years of infection.
  - Labor market model (simple, to capture general equilibrium effects):
    - A job vacated by death is filled by a previously unemployed individual within half a year (0.5 year).
    - Vacated jobs assigned to unemployed individuals from households in the same income decile (restricts mobility by economic status) with perfect geographic, gender, and age mobility assumed.
  - Social insurance support:
    - Baseline scenario: household left without any income earner receives an allowance of 25 percent of its initial income (captures formal schemes for upper deciles and informal community support for lower deciles).
- Important modeling simplifications and limitations:
  - HIV prevalence rates assumed constant at survey rates (no evolving epidemic dynamics modeled).
  - Population does not age; household structure remains constant apart from changes induced by increased mortality.
  - Does not model infection dynamics between individuals explicitly (notably cohabitating couples).
  - Household composition dynamics (e.g., movement of orphans, formation of new relationships) are not modeled beyond simple financial support; may distort age structure of people living with HIV/AIDS.

### IV. IMPACT OF HIV/AIDS ON POVERTY AND INEQUALITY
- Measurement approach:
  - Focus on $1-a-day poverty line (MDG-adopted) for poverty headcount index and poverty gap.
  - Also present headcount index based on $2-per-day line.
  - Simulations are dynamic: estimate year-by-year effects of mortality and morbidity over a 10-year horizon under various scenarios.
- Base Case Scenario assumptions:
  - Jobs replaced after half a year following death.
  - Households left without any income earner receive a 25 percent allowance.
- Table 5: Impact on Poverty and Inequality Indicators — Base Case Scenario (values provided exactly)
  - Ghana:
    - Income per capita (US$): Year 0 2.49; Year 10 2.47; Change (%) -0.55
    - Poverty headcount ($2/day,%) Year 0 66.38; Year 10 66.38; Change (%) 0.003
    - Poverty gap ($2/day,%) Year 0 34.21; Year 10 34.99; Change (%) 2.3
    - Poverty headcount ($1/day,%) Year 0 34.15; Year 10 35.74; Change (%) 4.6
    - Income dependency ratio: Year 0 2.44; Year 10 2.44; Change (%) -0.4
    - Gini coefficient: Year 0 53.85; Year 10 54.49; Change (%) 1.2
  - Kenya:
    - Income per capita (US$): Year 0 2.80; Year 10 2.78; Change (%) -0.90
    - Poverty headcount ($2/day,%) Year 0 58.92; Year 10 60.53; Change (%) 2.7
    - Poverty gap ($2/day,%) Year 0 25.17; Year 10 27.06; Change (%) 7.5
    - Poverty headcount ($1/day,%) Year 0 23.46; Year 10 25.80; Change (%) 10.0
    - Income dependency ratio: Year 0 2.68; Year 10 2.64; Change (%) -2.0
    - Gini coefficient: Year 0 48.43; Year 10 50.41; Change (%) 4.1
  - Swaziland:
    - Income per capita (US$): Year 0 2.87; Year 10 2.65; Change (%) -7.51
    - Poverty headcount ($2/day,%) Year 0 59.90; Year 10 64.00; Change (%) 6.9
    - Poverty gap ($2/day,%) Year 0 26.00; Year 10 33.36; Change (%) 28.3
    - Poverty headcount ($1/day,%) Year 0 23.28; Year 10 33.00; Change (%) 41.8
    - Income dependency ratio: Year 0 3.47; Year 10 3.00; Change (%) -14.3
    - Gini coefficient: Year 0 50.63; Year 10 55.74; Change (%) 10.1
  - Zambia:
    - Income per capita (US$): Year 0 1.75; Year 10 1.57; Change (%) -9.95
    - Poverty headcount ($2/day,%) Year 0 79.17; Year 10 81.48; Change (%) 2.9
    - Poverty gap ($2/day,%) Year 0 41.67; Year 10 48.59; Change (%) 16.6
    - Poverty headcount ($1/day,%) Year 0 51.04; Year 10 57.52; Change (%) 12.7
    - Income dependency ratio: Year 0 2.81; Year 10 2.67; Change (%) -4.7
    - Gini coefficient: Year 0 47.77; Year 10 53.23; Change (%) 11.4
- Key findings under base case assumptions:
  - Fall in average income is significant in high-prevalence countries (Swaziland and Zambia); in single-digit prevalence countries (Ghana and Kenya), HIV causes a fall in income below 1 percent over 10 years.
  - Relation between income impact and prevalence is not strictly positive:
    - Average income falls by a considerably larger share in Zambia (10 percent) than in Swaziland (7.5 percent), despite much higher prevalence in Swaziland. This is largely due to distributional characteristics of the disease across the population.
  - Poverty and inequality effects are heterogeneous:
    - Large increases in $1/day poverty headcount and Gini coefficient in high-prevalence countries (e.g., Swaziland: $1/day headcount up 41.8 percent; Gini up 10.1 percent).
    - Poverty gap increases markedly in Swaziland (28.3 percent) and Zambia (16.6 percent) over 10 years.
- Scenario analysis (text indicates additional scenarios are presented in subsequent subsections):
  - Variations consider replacement timing of deceased workers and extent of social insurance mechanisms (details of alternative scenarios are described in later subsections beyond the provided excerpt).

*Source: Authors' calculations and survey data as presented in the document.*

### Section V.

### Section V.

### Key findings on income, poverty, and distributional effects
- The increases in headcount poverty and in poverty gaps are sizable in most sample countries; the share of the population living with incomes below $1 per day increases closely in line with HIV prevalence rates.
- Poverty increases most dramatically in Swaziland, by 10 percentage points, and only slightly in Ghana, by 1.5 percentage points.
- Effects on average income and poverty measures are not perfectly correlated; distributional factors matter.
- Cross-country evidence (Kraay, 2004) suggests a 1 percent decrease in income per capita is on average associated with a 1.11 percent increase in poverty incidence ($1/day). The simulations here indicate much stronger impacts on poverty for given income changes:
  - If average income were to decline by 1 percent, the simulated increase in $1/day poverty prevalence would be: Ghana 8 percent, Kenya 11 percent, Swaziland 6 percent, Zambia 1.3 percent.
- Kraay (2004) finds that a 1 percent decrease in average income leads to a 1.44 percent increase in the poverty gap; the paper’s results show poverty gaps increase by more than 4 percent with every 1 percent increase in average income in all sample countries except Zambia.

### Worker replacement: scenarios and impacts
- Labor market scenarios modeled:
  - Baseline: vacant jobs filled after half a year.
  - Immediate worker replacement: unemployed fill jobs immediately.
  - No worker replacement: deceased workers are not replaced.
- Baseline scenario ("Worker replacement after half a year") — selected indicators (Without AIDS → With AIDS (Year 10) → Change (In per-cent)):
  - Income per capita (US$): Ghana 2.49 → 2.47 → -0.55; Kenya 2.80 → 2.78 → -0.90; Swaziland 2.87 → 2.65 → -7.51; Zambia 1.75 → 1.57 → -9.95.
  - Poverty headcount ($2/day,%): Ghana 66.38 → 66.38 → 0.0; Kenya 58.92 → 60.53 → 2.7; Swaziland 59.90 → 64.00 → 6.9; Zambia 79.17 → 81.48 → 2.9.
  - Poverty gap ($2/day,%): Ghana 34.21 → 34.99 → 2.3; Kenya 25.17 → 27.06 → 7.5; Swaziland 26.00 → 33.36 → 28.3; Zambia 41.67 → 48.59 → 16.6.
  - Poverty headcount ($1/day,%): Ghana 34.15 → 35.74 → 4.6; Kenya 23.46 → 25.80 → 10.0; Swaziland 23.28 → 33.00 → 41.8; Zambia 51.04 → 57.52 → 12.7.
  - Income dependency Ratio: Ghana 2.44 → 2.44 → -0.4; Kenya 2.68 → 2.64 → -2.0; Swaziland 3.47 → 3.00 → -14.3; Zambia 2.81 → 2.67 → -4.7.
  - Gini coefficient: Ghana 53.85 → 54.49 → 1.2; Kenya 48.43 → 50.41 → 4.1; Swaziland 50.63 → 55.74 → 10.1; Zambia 47.77 → 53.23 → 11.4.
- "Immediate worker replacement" — selected changes (Without AIDS → With AIDS (Year 10) → Change (In per-cent)):
  - Income per capita (US$): Ghana 2.49 → 2.48 → -0.18; Kenya 2.80 → 2.80 → -0.26; Swaziland 2.87 → 2.79 → -4.12; Zambia 1.75 → 1.62 → -6.25.
  - Poverty headcount ($2/day,%): Ghana 66.38 → 66.38 → 0.001; Kenya 58.92 → 59.38 → 0.8; Swaziland 59.90 → 62.14 → 3.8; Zambia 79.17 → 80.62 → 1.8.
  - Income dependency ratio: Ghana 2.44 → 2.42 → -1.0; Kenya 2.68 → 2.60 → -2.9; Swaziland 3.47 → 2.59 → -25.4; Zambia 2.81 → 2.44 → -13.0.
  - Gini coefficient: Ghana 53.85 → 54.06 → 0.4; Kenya 48.43 → 49.00 → 1.2; Swaziland 50.63 → 53.43 → 5.5; Zambia 47.77 → 51.20 → 7.2.
- "No worker replacement" — selected indicators:
  - Income per capita (US$): Ghana 2.49 → 2.46 → -1.29; Kenya 2.80 → 2.68 → -4.60; Swaziland 2.87 → 2.46 → -14.38; Zambia 1.75 → 1.51 → -13.65.
  - Poverty headcount ($1/day,%): Ghana 34.15 → 35.90 → 5.1; Kenya 23.46 → 26.53 → 13.1; Swaziland 23.28 → 35.53 → 52.6; Zambia 51.04 → 58.66 → 14.9.
  - Poverty gap ($2/day,%): Ghana 34.21 → 35.21 → 2.9; Kenya 25.17 → 27.70 → 10.0; Swaziland 26.00 → 35.92 → 38.1; Zambia 41.67 → 49.91 → 19.7.
  - Income dependency Ratio: Ghana 2.44 → 2.47 → 0.9; Kenya 2.68 → 2.75 → 2.4; Swaziland 3.47 → 3.75 → 8.0; Zambia 2.81 → 2.91 → 3.5.
  - Gini coefficient: Ghana 53.85 → 54.45 → 1.1; Kenya 48.43 → 49.63 → 2.4; Swaziland 50.63 → 56.35 → 11.3; Zambia 47.77 → 53.48 → 11.9.
- Interpretation:
  - Immediate replacement minimizes the impact on all indicators; no replacement maximizes negative effects.
  - Falls in income per capita are largest where HIV prevalence exceeds unemployment (Swaziland, Zambia).
  - Under no-worker-replacement, $1-a-day poverty in Swaziland increases by more than 50 percent.
  - Income dependency ratios fall under the baseline and immediate-replacement scenarios but rise under no-replacement in some countries.
  - Changes in income inequality are similar across scenarios except in Kenya.

### Social insurance mechanisms: modeling and effects
- Social insurance modeled as transfers between households within the same income decile (horizontal transfers), representing formal insurance in the formal sector and informal community support.
- Three scenarios presented (Table 7):
  - "Free fall" (no support).
  - Baseline ("25% allowance"): transfer to households whose income would otherwise fall to zero of 25 percent of initial income, financed by households in the same decile.
  - "75% allowance": transfer of 75 percent of initial income, financed similarly.
- Selected impacts (Without AIDS → With AIDS (Year 10) → Change (In per-cent)):
  - "Free fall" poverty headcount ($2/day,%): Ghana 66.38 → 66.57 → 0.3; Kenya 58.92 → 61.04 → 3.6; Swaziland 59.90 → 65.33 → 9.1; Zambia 79.17 → 82.53 → 4.3.
  - "Baseline (25% allowance)" poverty gap ($2/day,%): Ghana 34.21 → 34.99 → 2.3; Kenya 25.17 → 27.06 → 7.5; Swaziland 26.00 → 33.36 → 28.3; Zambia 41.67 → 48.59 → 16.6.
  - "75% allowance" Gini coefficient changes: Ghana 53.85 → 54.34 → 0.9; Kenya 48.43 → 49.61 → 2.4; Swaziland 50.63 → 53.34 → 5.3; Zambia 47.77 → 49.68 → 4.0.
- Interpretation:
  - Modeled social insurance mechanisms do not change average income per capita (transfers are redistributive within the country) but mitigate the depth of poverty increases.
  - Even the "25 percent" allowance notably reduces increases in headcount poverty and the poverty gap.
  - Horizontal support reduces depth but places burdens on supporting households; vertical transfers (e.g., government programs for orphans and vulnerable children) are not modeled here and would require a more elaborate model.

### Determinants of the impact on income and poverty
- Main determinants identified:
  - HIV prevalence (scale of epidemic).
  - Distributional factors: prevalence concentrated among higher-income groups raises aggregate income impacts; prevalence concentrated among those just above the poverty line raises poverty impacts.
  - Size of the population at risk (proportion just above the poverty line).
  - Age profile and geographic distribution (urban vs. rural) of prevalence.
- Normalized impact assuming a 5 percent HIV prevalence rate (Table 8, Baseline scenario):
  - Income per capita (US$): Ghana -1.30; Kenya -0.67; Swaziland -1.20; Zambia -3.19.
  - Poverty headcount ($2/day, in percent): Ghana 0.01; Kenya 2.05; Swaziland 1.09; Zambia 0.94.
  - Poverty gap ($2/day, in percent): Ghana 5.40; Kenya 5.59; Swaziland 4.51; Zambia 5.33.
  - Poverty headcount ($1/day, in percent): Ghana 11.05; Kenya 7.43; Swaziland 6.65; Zambia 4.07.
  - Income dependency ratio: Ghana -0.85; Kenya -1.50; Swaziland -2.27; Zambia -1.52.
  - Gini coefficient: Ghana 2.79; Kenya 3.05; Swaziland 1.61; Zambia 3.66.
- Income-per-decile impacts (Table 9): HIV/AIDS reduces income more in higher-income deciles in Zambia and in lower-income deciles in Ghana, Kenya, and Swaziland. Example decile change (%) highlights:
  - Ghana change a) to b) (%): deciles 1–10 = -0.8, -0.9, -0.8, -0.8, -3.1, -3.0, -0.1, -0.9, 0.6, -0.5.
  - Kenya change a) to b) (%): deciles 1–10 = -0.2, -0.2, -1.0, -0.6, -1.4, -0.1, -1.1, -0.3, 2.3, -0.2.
  - Swaziland change a) to b) (%): deciles 1–10 = -10.3, -12.0, -11.3, -9.1, -10.0, -8.8, -8.2, -4.8, -1.1, -7.5.
  - Zambia change a) to b) (%): deciles 1–10 = -5.4, -6.5, -5.6, -6.5, -8.1, -8.8, -8.9, -7.5, -5.5, -9.5.
- The $1-a-day line location by country (note 16): between deciles 3 and 4 in Ghana; between deciles 2 and 3 in Kenya and Swaziland; between deciles 4 and 5 in Zambia.
- Interpretation:
  - When HIV prevalence is higher among higher-income groups (Zambia), aggregate income per capita falls more.
  - When prevalence is concentrated just above the poverty line (Kenya, Ghana, Swaziland), poverty impacts are amplified even if average income per capita does not fall much.
  - The size of the population just above the poverty line matters: Swaziland has a much higher percentage just above the poverty line (explaining higher poverty impact relative to income), Kenya somewhat higher than Ghana.

### Conclusions and policy implications
- Simulation analysis confirms that the impact of HIV/AIDS on poverty indicators often exceeds its impact on average incomes; in three sample countries poverty incidence and the poverty gap increase more than expected from declines in income per capita.
- Key determinants of the impact on poverty are:
  - The size of the population at risk of falling into poverty.
  - HIV prevalence among that population group.
- Results are robust to labor market assumptions and to modeled social insurance functioning (horizontal transfers).
- Policy implications:
  - Even modest or negligible changes in average income per capita can conceal substantial increases in poverty; poverty reduction efforts can be frustrated by the epidemic.
  - HIV/AIDS threatens attainment of MDG 1 (“eradicate extreme poverty and hunger”) and other MDGs affected by increased poverty (for example, in education).
  - Informal horizontal support mechanisms have limited capacity to mitigate the impact on poverty; an effective social safety net would likely require vertical transfers (e.g., government programs).
- Methodological implications:
  - Assessments of welfare implications of HIV/AIDS cannot be limited to effects on income per capita.
  - Studies confined to households directly affected can be misleading; broader, distribution-sensitive analysis is needed.
  - Future research will benefit from newer household surveys and Demographic and Health Surveys (DHSs) for a better understanding of the socioeconomic dimension of the epidemic.

*Source: Authors’ calculations based on HIV-prevalence and income/expenditure surveys listed in Table 2.*

### REFERENCES

### REFERENCES

### Works on the economic consequences of HIV/AIDS and related macroeconomic analysis
- Bell, Clive, Shantayanan Devarajan, and Hans Gersbach, 2003, “The Long-Run Economic Costs of AIDS: Theory and an Application to South Africa,” Policy Research Working Paper No. 3152 (Washington:World Bank).
- ———, 2004, “Thinking About the Long-Run Economic Costs of AIDS,” in Haacker (2004).
- Haacker, Markus, 2002, “The Economic Consequences of HIV/AIDS in Southern Africa.” Working Paper WP/02/38 (Washington DC: International Monetary Fund).
- Haacker, Markus (ed.), 2004, The Macroeconomics of HIV/AIDS (Washington DC: International Monetary Fund).
- Haacker, Markus, 2006, “Global Trends in Life Expectancy and Income per Capita,” unpublished draft (Washington DC: International Monetary Fund).
- ING Barings, South African Research, 2000, “Economic Impact of AIDS in South Africa. A Dark Cloud on the Horizon” (Johannesburg, South Africa).
- Kambou, Gérard, Shantayanan Devarajan, and Mead Over, 1992, “The Economic Impact of AIDS in an African Country: Simulations with a Computable General Equilibrium Model of Cameroon,” Journal of African Economies, Vol. 1 No. 1, pp. 109–30.
- Bell, Clive; Devarajan, Shantayanan; Gersbach, Hans — listed above (see 2003 and 2004 entries).

### Studies on poverty, inequality, and household impacts related to HIV/AIDS
- Greener, Robert, 2004, “Impact of HIV/AIDS on Poverty and Inequality,” in Haacker (2004).
- Greener, Robert, Keith Jefferis, and Happy Siphambe, 2000, “The Impact of HIV/AIDS on Poverty and Inequality in Botswana,” South African Journal of Economics, Vol. 68, No. 5, pp. 888–915.
- Booysen, Frederick le Roux, 2003, “Poverty Dynamics and HIV/AIDS Related Morbidity and Mortality in South Africa,” paper presented at an international conference on “Empirical Evidence for the Demographic and Socio-Economic Impact of AIDS,” Health Economics and HIV/AIDS Research Division, University of KwaZulu-Natal, South Africa, March 26–28.
- Steinberg, M., Johnson, S., Schierhout, G., and Ndegwa, D., 2002. “Hitting Home: How Households Cope with the Impact of the HIV/AIDS Epidemic.” The Henry Kaiser Family Foundation.
- Whiteside, Alan, 2002, “Poverty and HIV/AIDS in Africa,” Third World Quarterly, Vol 23, No 2, pp. 313-332, 2002.

### Country demographic and health surveys; sentinel and household surveys
- Ghana, 2004, “Ghana Demographic and Health Survey 2003,” Ghana Statistical Service (GSS), 2004, Noguchi Memorial Institute for Medical Research (NMIMR), and ORC Macro, Calverton, Maryland: GSS, NMIMR, and ORC Macro.
- Kenya, 2003, “Kenya Demographic and Health Survey 2003,” Central Bureau of Statistics (CBS) [Kenya], Ministry of Health (MOH) [Kenya], and ORC Macro. 2004. Calverton, Maryland: CBS, MOH, and ORC Macro.
- Zambia, 2003, “Zambia Demographic and Health Survey 2001-2002”, Central Statistical Office [Zambia], Central Board of Health [Zambia], and ORC Macro. 2003, Calverton, Maryland, USA: Central Statistical Office, Central Board of Health, and ORC Macro.
- Shisana, O., 2002. “Nelson Mandela/HSRC Study of HIV/AIDS: Household Survey 2002.” Human Sciences Research Council Publishers.
- Swaziland Ministry of Health and Social Welfare, “8th HIV Sentinel Serosurveillance Report Year 2002.” Swaziland National AIDS/STDs Programme, Mbabane, Swaziland.
- Swaziland, 2002, “8th HIV Sentinel Serosurveillance Report Year 2002,” Swaziland National AIDS/STDs Programme, Ministry of Health and Social Welfare, 2002, Mbabane, Swaziland.
- Joint United Nations Programme on HIV/AIDS (UNAIDS), 2004, 2004 Report on the Global AIDS Epidemic (Geneva: UNAIDS).
- Joint United Nations Programme on HIV/AIDS (UNAIDS), United Nations Children’s Fund (UNICEF), and U.S. Agency for International Development (USAID), 2004, Children on the Brink 2004 (Geneva, New York, and Washington).

### Pensions, social security, microcredit, and coverage of the informal sector
- Barbone, Luca, and Luis-Alvaro Sanchez B., 1999, “Pensions and Social Security in Sub-Saharan Africa: Issues and Options,” World Bank Africa Region Working Paper No. 4 (Washington:World Bank).
- Baruti, Elias, 2003, “Extension of Social Security Coverage to the Poor in the Informal Urban Sector in Tanzania Through Micro Credit,” paper presented at the Inaugural Tanzanian Biennial Development Forum.

### Cross-country growth, indicators, and population datasets
- Kraay, Art, 2004. “When is Growth Pro-Poor? Cross-Country Evidence,” World Bank Policy Research Working Paper 3225 (Washington DC: World Bank).
- World Bank, 2005, “World Development Indicators 2004”, World Bank World Development Indicators, Washington DC.
- United Nations Population Division, 2005, World Population Prospects: The 2004 Revision, CD-ROM Edition - Comprehensive Dataset (New York: United Nations).

### Other referenced items
- Kambou, Gérard; Devarajan, Shantayanan; Over, Mead — see Kambou et al., 1992 entry above.
- UNAIDS – see Joint United Nations Programme on HIV/AIDS.
- Ghana, Kenya, Zambia survey citations — see entries under Demographic and Health Surveys above.

### Appendix: Initial Expenditure Distribution in the Four Countries Covered
- Charts labeled for: Ghana, Kenya, Swaziland, Zambia.
- Axis labels shown: "Expenditure (PPP U.S. dollars per day)" and "Percent of Population".
- Source: Household Income and Expenditure Surveys cited in Table 2.

*Source: _wp06126 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06126.pdf_