AIDS, Reversal of the Demographic Transition and Economic Development: Evidence from Africa

1
AIDS, Reversal of the Demographic Transition and
Economic Development Evidence from Africa

• Sebnem Kalemli-Ozcan
• University of Houston and NBER

2
Big Question

• How important are disease burdens in explaining
• LR growth and income differences?
• Fogel Health and nutrition are important source
  of growth in the West
• Sachs Malaria is the single most important
  factor in Africas underdevelopment
• Bloom and Canning Better health, more rapid
  growth, higher development
• Acemoglu and Johnson No effect

3
Is there a Big LR Impact?

• Still debatedReverse causality hard to solve
• If there is, must come from
• Human Capital intergenerational effects
• Fertility intra hh resource allocation
• Trade and Investment

4
Why study AIDS?

• 40 million people living with HIV/AIDS
• ? 2/3 in Africa
• Economic impact greater than the other diseases
• always fatal but kills slowly
• works through numerous channels
• affects prime-aged adults
• widespread in urban areas
• educated and upper class individuals

5
Changes in Female Life Expectancy in African
Countries with high HIV/AIDS Prevalence, 1950-2000
6
Changes in Female Adult Mortality in African
Countries with high HIV/AIDS Prevalence, 1950-2000
7
Macro Empirical Literature Mixed Results

• Negative effect
• Over (1992) Bonnel (2000) Lorentzen, McMillan,
  and Wacziarg (2005)
• Papageorgiou, Stoytcheva (2004)
• Bell, Devarajan, and Gersbach (2003) Corrigan,
  Gloom, and Mendez (2005) Cuddington (1993)
• No effect
• Bloom and Mahal (1997) Werker, Ahuja, and
  Wendell (2006)
• Positive effect Young (2005, 2006)

8
Problems with the Macro Empirical Literature

• Reverse causality
• Numerous channels
• - vary by time and/or by country
• - hard to control in aggregate growth framework
• Savings and Productivity
• Human Capital
• Foreign Investment (Haacker (2002) Alsan, Bloom,
  and Canning (2004))
• Demographics, Fertility, Population growth

9
This paper Fertility/HK Channel

• Consensus on less HK accumulation
• Supply of education orphans, teachers die,
• Demand for education lower rate of return,
  quality-quantity trade-off
• Lower population growth (assuming no fertility
  response) UNAIDS (2004)
• Even lower population growth with fertility
  responding negatively Young (2005, 2006)

10
I ask

• What is the effect of HIV/AIDS on fertility
• and HK accumulation in the aggregate data?

11
I show

• Panel regressions 44 African countries,
  1985-2000
• Results
• Positive effect of AIDS on fertility
• ? 2 more children
• Negative effect of AIDS on HK
• ? 38 percentage points less primary schooling

12
Implications

• Disastrous effects on economic development
• Reversal of fertility transition coming decades
• Support for models emphasizing demand for
  children and quality-quantity trade-off in the
  face of high mortality
• (Ehrlich and Lui (1991) Meltzer (1992) Boldrin
  and Jones (2002) Doepke (2004) Chakraborty
  (2004) Soares (2005) Sah (1991) Kalemli-Ozcan
  (2003) Tamura (2004))

13
Why do my results differ than Youngs?

• The results are consistent with Young (2005)
• South Africa negative relation between HIV/AIDS
  and TFR
• The results differ from Young (2006)
• Assumptions on behavioral fertility response
• - unprotected sexual activity declines
• - returns to labor increase
• Data
• Empirics

14
Behavioral Response of Fertility to the Epidemic

• People unaware of the epidemic no change
• Infected people who know their status
• ? decrease their fertility
• Uninfected people aware of high mortality
• ? decrease their fertility (Youngs assumptions)
• ? increase their fertility (precautionary demand)
  
• Infected people who do not know their status
• ? behave like uninfected people

15
Knowledge/Risk Perception in Africa Not good

• Jacob Zuma, South Africas former deputy
  president took a shower to minimize the risk of
  infection after having sex with an HIV-positive
  woman (Economist, 2006)
• Data from DHS surveys
• Micro empirical research
• ? Serious informational issues

16
Table 1 HIV Testing Statistics, 1988-2004
Tested during ANC visit
Counseled during ANC visit
Receiving test results
Receiving a test
Requesting and receiving a test, receiving
results
9
Botswana
2
Burundi
9
36
5
21
18
Cameron
5
Cote dIvoire
6
Gambia
7
15
4
Guinea-Bissau
3
29
8.5
Kenya
9
Lesotho
1
51
2
4
3
Mozambique
1
Niger
1
24
3
7
4
Nigeria
3
Senegal
2
Sierra Leone
3
Togo
22
43
8
Zambia
17
Risk Perception, Selected African Countries,
1994-99
18
Knowledge/Risk Perception Empirics

• Oster (2005) knowledge of own status low
• Setel (1995) 30 of tested learn results
• Thornton (2006) 20 tested, half learn
• Thornton (2006), Higgins (1991), Green (1994),
  Temmerman et al. (1990) no effect of counseling
• Allen et al. (1993) desire to have children
  among HIV- positive is 45.

19
Behavioral Response of Fertility to the Epidemic
Empirics

• Oster (2005 2006)
• very small decrease in sexual activity behavior
  
• change only for rich people with higher LE
• Mwaluko et al. (2003), Williams et al. (2003),
  Bloom et al (2000), Stoneburner and Low Beer
  (2004), Lagarde et al. (1997), Ngwshemi et al.
  (1996), Caldwell et al. (1999)
• ? no change in sexual behavior
• Dupas (2006)
• Luke and Munshi (2004)
• Francis (2006)

20
A disease of young adults HIV/AIDS Prevalence by
Age in Botswana
21
Probability of a Zimbabwean child aged 15 dying
before age 50, 1980-1997
22
Data

• 44 African countries for 1985-2000
• (1985, 1987, 1990, 1992, 1995, 1997, 2000)
• Dependent Variables
• Total fertility rates WB, WDI and DHS
• School enrollment rates WB, WDI

23
WB Data Total Fertility Rate in African
Countries, 1985-2000
24
DHS Data Total Fertility Rate in African
Countries, 1985-2000
25
A puzzle Increase in TFR

• Most recent DHS surveys show an uptick in the
• total fertility rate
• Kenya study by Westcoff and Cross (2006)
• The increase in the proportion of women who want
  more children is puzzlingit seems reasonable to
  conclude that the increase in child and youth
  mortality because if AIDS has a role in the
  changes in reproductive intentions

26
Gross Primary School Enrollment in African
Countries, 1985-2000
27
Data

• Independent Variables
• AIDS prevalence UNAIDS epidemiological fact
  sheets - reported AIDS cases
• HIV prevalence U.S. Census Bureau HIV
  Surveillance Database rates among pregnant
  women
• ? Both data have problems

28
Caveats of HIV/AIDS Data

• AIDS
• Serious underreporting
• Captures mortality better
• HIV
• Problematic time variation within a country,
  measure different places in different years
• Captures recent infections better

29
Cross-Section Regressions AIDS and TFR
30
Table 4 Fertility in a Cross- Section of
CountriesDependent Variable is Average Total
Fertility Rate, 1985-2000
Base Sample (4)
Base Sample (3)
Wide Spread (2)
Base Sample (1)
0.20 (0.07)
- -
0.21 (0.08)
0.18 (0.06)
Log Avg. AIDS Incid. 19852000
- -
0.18 (0.07)
- -
- -
Log Avg. HIV Prev. 1985-2000
- -
-0.02 (0.008)
-0.02 (0.008)
-0.02 (0.008)
Avg. Female Sec. School, 1985-2000
-0.01 (0.01)
- -
- -
- -
Avg. Female Prim. School, 1985-2000
-0.34 (0.12)
-0.07 (0.14)
-0.15 (0.14)
-0.17 (0.13)
Log Avg. GDP p.c., 1985-2000
0.02 (0.004)
0.01 (0.003)
0.01 (0.003)
0.01 (0.003)
Avg. Infant Mort., 1985-2000
0.75 41
0.75 41
0.77 38
0.77 41
R2 N
31
Table 4-Robustness
32
Table 5 Fertility in a Cross-Section of
Countries Perceptions
(4)
(3)
(2)
(1)
0.02 (0.007)
- -
- -
- -
Heard of HIV/AIDS, 19882000
- -
0.02 (0.007)
0.02 (0.004)
0.02 (0.0008)
Know someone died of AIDS,1993-2000
-0.02 (0.008)
- -
- -
- -
Avg. Female Sec. School, 1985-2000
-0.04 (0.17)
- -
-0.70 (0.13)
- -
Log Avg. GDP p.c., 1985-2000
0.01 (0.004)
0.03 (0.005)
- -
- -
Avg. Infant Mort., 1985-2000
0.75 30
0.78 12
0.82 12
0.26 12
R2 N
33
Table 6 Fertility in a Cross-Section of
Countries AIDS in 1985Dependent Variable is
Total Fertility Rate in 2000
Available Sample (4)
Base Sample (3)
Base Sample (2)
Base Sample (1)
0.11 (0.04)
0.10 (0.05)
- -
0.10 (0.05)
Log AIDS Incidence in 1985
- -
- -
0.18 (0.06)
- -
Log HIV Prevalence in 1985
-0.01 (0.10)
- -
-0.004 (0.01)
-0.002 (0.01)
Female Secondary School in 1985
- -
0.01 (0.05)
- -
- -
Female Primary School in 1985
-0.23 (0.22)
-0.27 (0.11)
-0.10 (0.14)
-0.23 (0.14)
Log GDP p.c. in 1985
0.01 (0.002)
0.02 (0.004)
0.02 (0.004)
0.02 (0.004)
Infant mortality in 1985
-0.03 (0.013)
- -
- -
- -
Contraceptive Use in 2000
0.81 27
0.68 41
0.71 41
0.68 41
R2 N
34
Panel Regressions AIDS and TFR
35
Table 7a Fertility in a Panel of Countries
-Dependent Var. Total Fertility Rate-
Base Sample (4)
Base Sample (3)
Base Sample (2)
Base Sample (1)
OLS
OLS
OLS
OLS
Yes
Yes
No
No
Fixed Effects
-0.08 (0.01)
-0.10 (0.01)
-0.07 (0.01)
-0.09 (0.01)
Time Trend
- -
0.03 (0.01)
- -
0.07 (0.02)
Log AIDS Incid.
-0.03 (0.03)
- -
0.08 (0.02)
- -
Log HIV Prev.
-0.001 (0.003)
-0.001 (0.003)
-0.02 (0.003)
-0.02 (0.003)
Female Sec. School,
-0.18 (0.06)
-0.21 (0.07)
-0.09 (0.05)
-0.15 (0.05)
Log GDP p.c.
0.002 (0.003)
0.001 (0.003)
0.01 (0.001)
0.01 (0.001)
Infant Mort.
0.78 228
0.79 228
0.74 228
0.73 228
R2 N
36
Table 7b Fertility in a Panel of Countries
-Dependent Var. Total Fertility Rate-
Base Sample (4)
Base Sample (3)
Base Sample (2)
Base Sample (1)
WLS
WLS
WLS
WLS
Yes
Yes
No
No
Fixed Effects
-0.08 (0.003)
-0.09 (0.003)
-0.08 (0.004)
-0.10 (0.005)
Time Trend
- -
0.02 (0.006)
- -
0.07 (0.01)
Log AIDS Incid.
-0.02 (0.02)
- -
0.06 (0.01)
- -
Log HIV Prev.
-0.003 (0.001)
-0.004 (0.001)
-0.02 (0.002)
-0.02 (0.002)
Female Sec. School,
-0.21 (0.04)
-0.23 (0.05)
-0.16 (0.03)
-0.16 (0.03)
Log GDP p.c.
0.001 (0.001)
0.001 (0.001)
0.01 (0.001)
0.01 (0.001)
Infant Mort.
0.79 228
0.78 228
0.73 228
0.73 228
R2 N
37
Table 8 Fertility in a Panel of Countries
Robustness-Dependent var. Total Fertility Rate-
No South Africa (5)
Whole Africa (4)
No Rich (3)
Base Sample (2)
Base Sample (1)
WLS
WLS
WLS
WLS
WLS
Yes No
Yes No
Yes No
Yes No
Yes Yes
Country Fix. Eff. Time Fix. Eff.
-0.09 (0.002)
-0.09 (0.002)
-0.09 (0.002)
-0.09 (0.002)
- -
Time Trend
- -
- -
- -
-0.001 (0.001)
- -
Time Trend2
0.02 (0.005)
0.02 (0.005)
0.02 (0.006)
0.02 (0.006)
0.01 (0.004)
Log AIDS Incidence
- -
- -
-0.002 (0.003)
-0.004 (0.001)
-0.004 (0.002)
Female Sec. School,
-0.16 (0.04)
-0.16 (0.04)
-0.08 (0.007)
-0.21 (0.04)
-0.20 (0.04)
Log GDP per capita
0.78 273
0.78 280
0.78 198
0.78 228
0.30 228
R2 N
38
Regression of Total Fertility Rate on AIDS
Prevalence controlling for other regressors
39
Panel Regressions AIDS and HK
40
Table 9 Human Capital Invest. in a Panel of
Countries-Dependent Var. Gross Primary Sch.
Enroll.-
Base Sample (5)
Base Sample (4)
Base Sample (3)
Base Sample (2)
Base Sample (1)
WLS
WLS
WLS
OLS
OLS
Yes
Yes
Yes
Yes
Yes
Fixed Effects
2.18 (0.24)
0.96 (0.22)
2.18 (0.26)
1.17 (0.47)
2.50 (0.56)
Time Trend
-3.93 (0.58)
- -
-4.25 (0.63)
- -
-5.04 (1.16)
Log AIDS Incidence
- -
-2.11 (1.04)
- -
-3.83 (2.09)
- -
Log HIV Prevalence
25.90 (5.36)
19.64 (5.16)
26.79 (5.56)
20.52 (8.54)
24.64 (8.00)
Log GDP per capita
- -
-0.71 (0.10)
-0.92 (0.13)
-1.09 (0.25)
-1.19 (0.25)
Infant Mortality
-0.48 (0.06)
- -
- -
- -
- -
Mortality Under Age 5
0.50 228
0.28 228
0.32 228
0.26 228
0.32 228
R2 N
41
Regression of Human Capital Investment on AIDS
Prevalence after controlling for other regressors
42
Table 10 Human Capital Investment in a panel of
Countries Robustness
Sec. Sch. Enroll.
Prim. Sch. Enroll.
Prim. Sch. Enroll.
Prim. Sch. Enroll.
Dependent Variable
Base Sample (4)
No Rich (3)
Base Sample (2)
Base Sample (1)
WLS
WLS
WLS
WLS
Yes No
Yes No
Yes No
Yes Yes
Country Fix. Eff. Time Fix. Eff.
0.74 (0.17)
2.46 (0.30)
2.21 (0.26)
- -
Time Trend
- -
- -
0.04 (0.02)
- -
Time Trend2
-0.80 (0.40)
-2.43 (0.61)
-4.22 (0.61)
-2.25 (1.05)
Log AIDS Incidence
20.37 (2.65)
5.89 (6.28)
26.99 (5.36)
19.55 (5.10)
Log GDP p.c.
-0.03 (0.11)
-0.11 (0.24)
-0.93 (0.10)
-1.10 (0.09)
Infant Mortality
0.25 192
0.59 198
0.60 228
0.50 228
R2 N
43
Table 11a Instrumental Variable Regressions
Average Fertility Rate 1985-2000
Average Fertility Rate 1985-2000
Dependent variable
(2)
(1)
0.60 (0.31)
0.13 (0.08)
Log Avg. AIDS Incid, 19852000
-0.03 (0.02)
-0.02 (0.01)
Avg. Female Sec. School, 1985-2000
-0.04 (0.3)
-0.12 (0.12)
Log Average GDP p.c., 1985-2000
0.02 (0.006)
0.01 (0.003)
Average Infant Mortality, 1985-2000
0.34
0.70
R2
17 1 17 STD, premarital
40 1 40 circumcision
Countries Time N Instrument
44
Table 11b Instrumental Variable Regressions
Gross Primary School Enroll.
Fertility Rate
Dependent variable
(2)
(1)
9.01 (2.91)
-0.14 (0.03)
Time Trend
-10.25 (5.93)
0.09 (0.05)
Log AIDS Incidence
- -
-0.001 (0.001)
Female Sec. School
20.74 (4.67)
0.24 (0.26)
Log GDP per capita
-1.25 (0.56)
-0.01 (0.01)
Infant Mortality
0.10
0.82
R2
21 3 63 premarital
21 3 63 premarital
Countries Time N Instrument
45
Table 12 Fertility in a Cross-Section of
RegionsDependent variable is Total Fertility
Rate in 19982004
Pooled OLS (Weighted) (4)
Pooled OLS (3)
Pooled OLS (Weighted) (2)
Pooled (OLS) (1)
0.27 (0.08)
0.28 (0.08)
0.40 (0.11)
Log HIV Prevalence, 19851990
0.43 (0.11)
Yes
Yes
No
No
Country Dummies
0.69 41
0.57 41
0.26 41
0.17 41
R2 N
46
Why do my results differ than Youngs?

• The results are consistent with Young (2005)
• South Africa negative relation between HIV/AIDS
  and TFR
• The results differ from Young (2006)
• Assumptions on behavioral fertility response
• - unprotected sexual activity declines
• - returns to labor increase
• Data
• Empirics

47
HIV/AIDS in South Africa
48
Fertility Enrollment in South Africa
49
HIV/AIDS and TFR in South Africa
50
HIV/AIDS and TFR in Botswana
51
HIV/AIDS and TFR in Gabon
52
HIV/AIDS and TFR in Uganda
53
Why do my results differ than Youngs?

• Is South Africa representative?
• Cohort-Specific trends
• Spillovers between cohorts
• Omitted variables Education, Urbanization
• AIDS vs HIV Time variation
• Sample
• Fertility measure

54
Conclusions

• AIDS epidemic has a statistically and
  economically significant () effect on fertility
  and (-) effect on human capital accumulation
• Reversal in the fertility transition in coming
  decades and significant reductions in aggregate
  human capital investment
• ? already started for Kenya and Uganda
• Lower growth and welfare for current and future
  African generations

55
Additional Slides

• Descriptive Statistics
• Robustness
• Figure 2 HIV/AIDS over time
• Figure 5a Survival time
• HIV/AIDS statistics-world
• Forecast error results

56
Table 2 Descriptive Statistics
Min
Max
Std.Dev
Mean
2.2
7.6
1.1
5.8
Average Total Fertility Rate, 19852000
58.8
281.6
56.8
160.1
Average Primary School Enrollment, 19852000 ()
10.0
169.5
36.9
49.5
Average Secondary School Enrollment, 19852000 ()
0.02
298.0
47.4
25.6
Average AIDS Incidence, 19852000 (per 100, 000)
0.09
25.7
6.3
6.9
Average HIV Prevalence, 19852000 ()
101.8
6246.4
1317.4
860.8
Average GDP per capita, 19852000 (PPP 1995 s)
5.5
70.9
14.2
31.5
Average Urban Population, 19852000 ()
2.2
6.0
0.7
3.2
Average Population over 65, 19852000 ()
15.2
184.0
37.3
103.2
Average Infant Mortality, 19852000 (per 1000)
18.7
321.4
65.2
165.1
Average Mortality Under 5, 1985-2000 (per 1000)
236.7
579.7
78.8
465
Average Adult Mortality, 19852000 (per 1000)
22.1
140.7
31.6
73.4
Average Primary School for Female,19852000 ()
3.2
90.8
20.1
21.6
Average Secondary School for Female, 19852000 ()
5.7
78.8
17.5
28
Average Secondary School for Male, 19852000 ()
57
Table 3 Correlation Matrix
Male Sec. Sch.
Female Sec. Sch.
Female Prim. Sch.
Adult Mort.
Age 5 Mort.
Infant Mort.
Pop. Over 65
Urban Pop.
Log GDP
Log HIV
Log Aids
1.00
Log Aids
1.00
0.76
Log HIV
1.00
-0.04
0.16
Log GDP
1.00
0.59
-0.15
0.01
Urban Pop.
1.00
0.47
0.50
-0.22
-0.07
Pop. Over 65
1.00
-0.34
-0.30
-0.63
0.04
-0.18
Infant Mort.
1.00
0.98
-0.34
-0.33
-0.66
0.04
-0.18
Age 5 Mort.
1.00
0.44
0.42
-0.38
-0.43
-0.25
0.59
0.43
Adult Mort.
1.00
-0.06
-0.68
-0.66
0.32
0.34
0.69
0.14
0.31
Fem. Prim Sch.
1.00
0.78
-0.24
-.0.75
-0.71
0.38
0.49
0.85
0.07
0.19
Fem. Sec. Sch.
1.00
0.93
0.73
-0.35
-0.73
-0.70
0.39
0.62
0.83
0.05
0.22
Male Sec. Sch.
58
Table 8-Robustness
59
Table 10-Robustness
60
HIV/AIDS Prevalence in Selected African
Countries, 1985-2000
61
Survival Rates in Low/Mid Income Countries
62
Regional HIV/AIDS Statistics,end of 2003

63
Regional HIV/AIDS Statistics,end of 2003
64
Forecast Error in Fertility and AIDS
prevalence(2000)Dependent Variable Forecast
Error in Fertility