measuring impacts of hiv/aids on african rural economies

Michigan State University, Dept. of Agricultural Economics

Measuring Impacts of HIV/AIDS

on African Rural Economies

T.S. JayneMichigan State University

Informal presentation at HSRC, JohannesburgJuly 13, 2004


Effective Response to HIV/AIDS Requires Knowledge of How Households Respond to the

Disease• Three-pronged attack:

– Prevention– Treatment– Mitigation

• All depend on solid information on how individuals and households adapt and respond

• 20+ years after the onset of the disease, the empirical foundation for the design of programs is still weak


Objectives

• To understand how affected households respond/adapt to prime-age mortality

• To measure impacts on– Family size and composition– Crop production– Non-farm income– Asset levels

• To consider implications for policy


Characteristics of the national samples

Country Sample size Year(s) of surveys

Panel or cross-sectional

Kenya n=1422 n=1266

1997, 2000, 2002

Panel

Malawi n=420n=372

1990, 2002

Panel

Mozambique

n=4908 2002 Cross-section

Rwanda n=1395 2002 Cross-section

Zambia n=6922 2000 Cross-section


4

6

8

10

12

14

8 12 16 20 24 28

HIV+ prevalence rate, 1999

Adu

lt m

orta

lity

rate

199

6-20

01

CentralCopperbelt

Eastern

Luapula

Lusaka

Northern

Northwestern

Southern

Western


What have we learned so far about measuring impacts?


Importance of panel (longitudinal) analysis


0

5

10

15

20

25

30

35

40

45

Pre-Mortality Post-Mortality

Afflicted hhsUnafflicted hhs

Importance of panel (longitudinal) analysis

Household income


Finding #2

Afflicted households/individuals are not random• Early 1990s: positively correlated

with income, wealth, education, mobility

• More recent evidence: increasingly concentrated among the poor


Per Capita Income Status of Afflicted Households (ex ante) -

Kenya

Deceased prime-age males

Deceased prime-age females

Poorest 25% 50.0 34.1

2nd quartile 19.6 14.6

3rd quartile 15.2 34.1

Wealthiest 25% 15.2 17.2


Per Capita Income Status of Afflicted Households - Zambia

Deceased prime-age males

Deceased prime-age females

Poorest 25% 18.9 19.6

2nd quartile 23.6 24.9

3rd quartile 20.5 26.5

Wealthiest 25% 37.0 29.0


Finding #3: Certain factors affect the magnitude of impacts

on households

• Strong evidence that impacts depend on:

– Initial level of household vulnerability (assets, wealth)

– Sex of the deceased– Position in household of deceased– Ability of household to attract new members– Characteristics of adults remaining in household

(e.g, skills, education level)


Gender Effects of Mortality on Crop Cultivation

• In Kenya:– Death of male head - 0.9 acre to cash

crops (e.g., sugarcane, horticulture)– Death of female head - 1.8 acre to

cereals, tubers


Effects of Death on Farm Production Sensitive to Gender, Position in HH

• Death of Male hh-head 68% reduction in value of crop output

• Death of Female head/spouse less dramatic but still negative effects

• Why Effects of Male Prime Age Mortality are Greater?– Loss of female ag. labor to caregiving– Loss of higher-return crops

• Death of other hh member – insignificant effects on ag.


Finding #4

• In some countries, the majority of prime-age mortality is among older sons / daughters, not heads / spouses


Position in household of deceased p.a. individuals

1585

2476

5842

5050

Zambia: head/spouse other

4456

5644

5347

4753

Rwanda: head/spouse other

1387

4060

6832

6040

Mozambique: head/spouse other

5248

5446

5545

3862

Malawi: head/spouse other

2773

5941

3467

2476

Kenya: head/spouse other

AfflictedM F

Non-afflictedM F


Finding 5: Effects More Severe on the Poor

• Very few significant effects detected among households in top half of asset distribution

• Effects on ag production and non-farm income were larger and more highly significant among the poor


Implications - I

• Not clear that afflicted households need or should be urged to use:– labor-saving crop technologies

• Why?– Afflicted hhs, on average, have as much labor

and land/labor ratios as non-affected hhs– crops / techniques that reduce labor input per

acre may sacrifice income and food produced per acre

– Must take into account population density and extent of under-employed labor


Population Size, 2000 vs. 2025 (projected)

Seven Most Highly Afflicted Countries

02468

101214161820

Males < 20

Males 20-59

Females < 20

Females 20-59

20002025

millions


Implication - II

• Not clear that afflicted households should be urged to grow:– “more nutritious” foods

• Why?– Crops that maximize nutrition / kg produced ≠ – maximize nutrition / acre or income / acre-- need to take account of which crops provide

greatest return to land / labor in a given area


1,2502505Crop Y

1,00010010Crop X

Nutritional units per acre

Kgs produced per acre

Nutritional units / kg produced


Summary

• Adult mortality’s greatest effects are:– On the relatively poor– When male head dies – When death is other than the hh

head/spouse, the household is better able to draw back other members to help the hh adjust


What are we learning about “community effects”:

• What determines community “resilience”?• Local institutions/traditions influence

resilience– Example of sugarcane outgrower programs in

Kenya– Land tenure / land inheritance

• H0: resilience is influenced by• Initial level of poverty in community • Rules governing women’s rights and access to resources

– e.g. can widows retain land after husband’s death?

• Matrilineal vs. patrilineal effects


Need for appropriate balance between:

• Investing in long-term productivity growth (education, infrastructure, markets)

vs• Targeted assistance to affected HHs

• Pro-poor development is important to mitigate economic effects of HIV/AIDS


“Difference-in-Difference” Approach

y - x

y y2 y1

x x2 x1

difference

Afflicted households

Households not afflicted

Difference

20001997


Adult Mortality Rates - Women

0

5

10

15

20

25

30

35

40

Women 15-24 Women 25-34 Women 35-44

ExpectedKenyaKisumu / Siaya


Adult Mortality Rates - Men

0

10

20

30

40

50

60

70

80

90

Men 15-24 Men 25-34 Men 35-44

ExpectedKenyaKisumu/Siaya


Country

Non-Affected

HHs

HH with Male Death

HH with Female Death

Kenya (1997 ex ante) 23.4% 28.3% 21.0%

Kenya (2000 ex post) 22.0% 34.5% 16.2%

Mozambique (2002 ex post)* 28.0% 32.0% 26.0%

Rwanda (2002 ex post) 32.0% 32.0% 28.0%

Zambia (2000 ex post)* 35.0% 26.0% 32.0%

Notes:

Table 10. Percentage of Area Cultivated to Roots & Tubers by Country for Households With and With a PA Death

* for Moz and Zambia, this is % of cultivated area in cassava, only including households which grow cassava

---- mean value ----

measuring impacts of hiv/aids on african rural economies

Documents

weakmichigan state university

poormichigan state university

agricultural economicsfinding

agricultural economicsposition

agricultural economicswhat

impacts of hivaids

affected households

household e