This document summarizes a study examining the drivers of fertility decline in Ethiopia between 2005 and 2011. The study uses survey data to analyze changes in factors like education levels, occupations, wealth, child mortality, and access to family planning services. Regression and decomposition analyses show that in rural areas, where most of the decline occurred, increased female education and labor force participation along with expanded access to health extension workers were the primary factors explaining lower fertility rates. While progress has been made, further investment in education and family planning is still needed to continue reducing Ethiopia's high fertility.
1. ETHIOPIAN DEVELOPMENT
RESEARCH INSTITUTE
Drivers of the Demographic
Transition in Ethiopia
Authors: Fanaye Tadesse
Derek Headey
IFPRI ESSP-II
Ethiopian Economic Association
Conference
July 19-21, 2012
Addis Ababa
1
2. 1.Introduction
• After years of neglect, there is renewed interest in
reducing fertility rates, largely via family planning &
female education
• Recent papers in Copenhagen Consensus and Lancet
suggest high returns through several channels:
• Better maternal/child health & nutrition outcomes
• High savings & education>>faster economic growth
• Also strong correlations between HH size and poverty
2
3. 1.Introduction
• Rapid population growth long been a concern in
Ethiopia
– High population density relative to agric. resources
often linked to rural poverty and famines
– Population-induced deforestation and soil degradation
– Very poor health and nutrition outcomes
• Driver of population growth in Ethiopia – High Fertility
• Hence Ethiopian government has long sought to
reduce fertility
• Fertility declined from 7.1 in 1990 to 5.4 in 2005 and
then to 4.8 in 2011
4. 1. Introduction
• A very rapid decline in fertility rates over 2005-
2011, solely based on reduction in rural areas: from 6
children in 2005 to 5.5 in 2011
• Last 5-6 years has seen additional instrument of health
extension workers (HEWs), largely in rural areas
• Many other changes: big increases in education, rapid
economic growth, infrastructure, etc
• So in this paper we try to understand the reason behind
the fertility decline between the years 2005 and 2010
5. 2. Theories of fertility
• Demographers tend to emphasize supply side explanations
• Economic theories tend to emphasize demand-side
determinants, or fertility as a conscious choice (Becker)
• Children are both consumption & investment goods
• Substitution from quantity to quality as family income
increases (Becker and Lewis 1973).
• Opportunity costs matter: children have both explicit costs
and implicit costs (e.g. women’s time/wages)
6. 2. Economic theories of fertility
• Women’s education has other possible links:
– Increases age at marriage (and vice versa)
– Empowers women
– Raising the knowledge of contraceptives & benefits of
fewer children and larger birth intervals
• These different theories have long produced skepticism
among economists about the relevance of family
planning: “Development is the best contraceptive!”
• But a recent review of family planning found substantive
evidence that it often works (Kohle 2012)
7. 3. Data
• Data used for this study is the EDHS rounds 2005 and
2011.
• Nationally representative survey of women between
15 and 49 in both rural and urban Ethiopia.
– 14,070 in 2005
– 16,515 in 2011
• Contains both individual level and household level
variables
• Topics include fertility, family planning, child mortality,
child health, nutrition and knowledge of HIV/AIDS.
8. 4. Methods
• We use the Blinder-Oaxaca decomposition to look at
major drivers of fertility decline between 2005 and 2011
• Pseudo-panel technique that disaggregates predicted
change into changes in levels of explanatory
variables, and changes in coefficients (“returns”)
• Dependent variable is number of children-ever-born
Endowment effect Coefficient Effect Interaction Effect
9. 4. Methods
• We also disaggregate by rural and urban because of
substantial differences in
– Initial fertility, education, occupations, access, etc.
– Dynamics (all the recent change was rural fertility)
– Different trends in explanatory variables
– Quality of services: schools, health centers
• Disaggregation by age group might also pick up
parameter heterogeneity
• We also test whether HEWs influence different
groups differently (e.g. less education – Portner et al)
10. 4. Methods
Construction of some variables
1. DHS does not have information on consumption or
income variables, but measures a wealth index
constructed from the information on asset holdings
– The way indices are constructed is the same for both
2005 and 2011
– Separate indices were constructed for urban and rural
areas and these were mapped to the national index
using weights generated by regressing the national index
on the separate indices (Rutstein, 2008)
11. 4. Methods
2. Possible endogeneity of Child Mortality
– Rather than including the child mortality of the household, we
took the average of the child mortality in the cluster.
– A locally formed expectation rather than actual experience of
the household.
3. Endogeneity of family planning interventions
– HEWs captured by the percentage of women visited by the
HEW; so measures activity of different HEWs
– Admittedly, HEW activity may not be random (Portner et al)
– Similarly, contraceptive knowledge was constructed at village
level. Helps address endogeneity, and capture network effects
12. 5. Descriptive Results:
Children ever born by age category
Urban
Rural
All women 1.65
1.64 3.46
All women
3.27
45-49 5.58
5.53 45-49 7.54
7.66
40-44 4.57
4.34 40-44 7
6.92
35-39 3.36
3.39 35-39 6.16
5.93
2.42
30-34 4.79
2.4 30-34
4.6
25-29 1.35
1.66 25-29 3.13
2.87
20-24 0.5
0.5 20-24 1.31
1.24
15-19 0.07
0.04 15-19 0.27
0.16
2005 2011
Difference significant for 15-
Difference significant for 25-29 19, 25-29 and 34-39 age groups
age group and for all women
15. 5. Regression & decomposition results
• The age and area-specific results are a bit too
detailed to report, so here’s a summary
Urban areas:
• Most robust coefficients across age brackets & years
were wealth, women’s education, child mortality
• Not much effect of HEWs
Rural areas:
• Most robust effects were education, women’s
occupation, and HEWs
• Wealth & mortality effects not very robust
16. 5. Regression & decomposition results
• We report decomposition for rural areas only, since
urban decline was not significant
• Consistent with survey results, only 3 age groups had
significant predicted differences across the two
rounds: 15-19 years, 26-29 years, 35-39 years.
• Decline in fertility is mainly explained by endowment
differences between the two periods.
• Coefficient and interaction effects are not significant
(i.e. fairly highly degree of parameter stability)
17. Decomposition by age group
0.05
Predicted decline in children ever born
0.02
0
-0.01
-0.08
-0.05 -0.11 Health extension
Husband's educ.
-0.1 -0.06 Women's occup.
Women's educ.
-0.02 -0.06
-0.15 Demographics
-0.2
15-19 25-29 34-39
Mother's age brackets
18. Educational levels and occupations driving the fertility in
rural areas
0.000
-0.005
-0.010
-0.020 -0.025 -0.024
-0.026 -0.025
-0.030 -0.032
-0.040
-0.045
-0.050 -0.054
-0.056
-0.060
primary Secondary primary Secondary Higher Clerical Other Clerical Agriculture
Educ Educ. Educ Educ. Educ occupation
15-19 25-29 35-39
19. Conclusion
• Ethiopia has recently witnessed a rapid reduction in
fertility, entirely driven by trends in rural areas
• The largest driver of that reduction was increased female
labor force participation, followed by equal contributions
from women’s education and HEWs
• This is good news for Ethiopia, good news for
economists, and good news for demographers!
• However, there is still a long way to go.
• Just 3% of women have secondary education or above.
• Only 17% of women report HEW visits
• Given their high returns, both interventions warrant
further investment in the future
20. Future work on
• Analyses of desired number of children (preferred
fertility level – since unmet need for contraceptive is
still high at 27 percent).
• Possible regional differences that need to be
explored more