International Food Policy Research Institute (IFPRI) and Ethiopian Development Research Institute (EDRI) in collaboration with Ethiopian Economics Association (EEA). Eleventh International Conference on Ethiopian Economy. July 18-20, 2013
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Effect of Climate Shock on Cognitive Development of Children in Ethiopia
1. Effect of Climate Shock on Cognitive Development of
Children in Ethiopia
Tilahun Asmare, WVE
and
Guush Berhane, ESSP, IFPRI
Addis Ababa, Ethiopia
1
2. PRESENTATION OUTLINE
Introduction
Problem statement
Objective
Research Methodology
Data Source
Conceptual framework
Model Specification
Results and Discussions
Descriptive Analysis
Econometric Analysis
Conclusion and recommendation
Conclusion
recommendation 2
3. INTRODUCTION
• Climate-related shock, drought and flood threaten the
livelihood of nations specially developing countries that
have low coping mechanism and resilience capacity.
• Ethiopia faced by adverse climate shock (e.g drought) that
leads to a loss of household income, a reduction in
consumption, and/or a loss of productive asset (Darcon,
2005).
• Drought often defined as a continuous interval of time during
which actual moisture supply at a given place is consistently less
than normally expected (Roy and Hirway, 2007).
• Drought is covariant and slow onset shock
3
4. INTRODUCTION
The short term impact of drought
loss of food production and non-food production,
loss of employment;
shortage of water, fodder and fuel wood;
indebtedness, migration etc
long-term impact
Low agriculture growth
stunting,
low cognitive development of children that it turn
negatively affects the development of the country as it
lost its children intelligence.
4
5. INTRODUCTION
Cognitive development is the construction of thought
processes, including remembering, problem solving, and
decision making, from childhood through adolescent and
adulthood (cengage 2005).
Many researchers used Peabody picture vocabulary test
(PPVT) as measurement of cognitive of children (desai
(1989), Bau (1999,McCullah &Joshi (2002), Duc (2009),
Nair (2009).
5
6. STATEMENT OF THE PROBLEM
‘notwithstanding major investments in improving the
numbers and the qualifications of teachers and the
availability of equipment, student achievement has not
sufficiently improved. (MOE, 2010).
The Evolution of scores obtained in NLA shows the
composite score of grade 4 decreased to 40.9% in
2007/08 from its result 2000/01 that was 47.9% and the
composite score of grade 8 decreased from 41.1% of
2007/08 to 35.6% in 2000/01 (NOE, 2008).
6
7. STATEMENT OF THE PROBLEM:
Though drought is the most frequent challenge of the country, the
national ESDP do not include it as a factor that determinants of
children academic achievement.
Most researches undertake on cognitive development of children are
mainly focused on students/individual factors, household/ family
factors and school factors and give little or no attention to weather
related shocks.
Most research is focused on the impact of idiosyncratic shocks (such
as child labor) on children educational outcomes (test scores)
(Woldehanna, 2012.
As far as my knowledge there is no any related researches
undertaken in the country that can shows the impact of weather
related shock such as drought on PPVT test score
7
8. OBJECTIVES OF THE STUDY
The General objective of this study is to examine the effect of
climate related drought shocks on children cognitive
development and school outcomes. The specific objectives are:
To investigate the impact of climate related shock such as
drought exposure on cognitive ability of children.
To examine the impact of drought on other academic
achievement measurements such as reading and writing skills.
To measure the impact of productive safety net (PSNP)
interventions on cognitive development of children aged 12
and 15.
8
9. RESEARCH METHODOLOGY
Data source
Secondary data (Young Lives longitudinal dataset
Second and third round old cohort
Used purposive sampling strategy in selecting sentinel sites and a
random strategy used in the household selection within each sentinel
site (Woldehanna , 2011).
The sampling was undertaken in five regions (namely Amhara, Addis
Ababa, Tigray, Oromia and SNNP
3 to 5 woredas were selected in each region, gives total 20 woredas.
50 old cohort children (of age 7.5 to 8.5 years) were selected randomly
in each sentinel site.
9
10. RESEARCH METHODOLOGY
Children test score broadly depend on their initial skill endowment,
income and Time (Millett & Shah, 2012) .
Initial skill endowments include biological or initial characteristics of
children such as sex, birth order, race, birth weight etc.
HH income has direct effect, determine the amount of investment on
cognitive development children goods and services such as books that
increases students test results up on consumption.
To increase household income the household needs many inputs such labor
that can be own or hired agricultural labor.
The decision to use either of the two sources depends on the wage rate.
When the wage is high households may use own labor including child labor that
in turn reduce children test score. 10
11. RESEARCH METHODOLOGY
Children time consumption: hours spent on study or school
time positively affects children test score.
The biological effect of drought on cognitive development is
very straightforward.
Drought exposed children (either prenatal or postnatal exposure) are
simply lower ability, so conditional on the same level of schooling, they
still score lower on achievement tests because they attend less
efficiently (Bryce & Manisha , 2012).
Drought also negatively affects the income of household.
Hence any food or cash support to the drought affected
households helps to reduce the negative impact of drought.
11
13. RESEARCH METHODOLOGY
We used Panel data that have following advantages over pooled
data (Baltagi 2004)
(1) account for heterogeneity across individual units which is assumed
away in pooled data
(2) deal with time-invariant omitted variables as we can and in pooled data
(3) are less likely to have problems with autocorrelation and
multicollinearity as time series data
There are basically two types of panel models, the fixed effects
and the random effects model.
They differ by their assumptions how the heterogeneity is
captured and estimation techniques.
13
14. RESEARCH METHODOLOGY
The fixed effect model assumes that individual
heterogeneity is captured by the intercept term.
The random effects model assume in some sense that the
individual effects are captured by the intercept and a
random component.
This random component is not associated with the
regresses on the right hand side and part of the error term.
14
15. RESEARCH METHODOLOGY
The critical difference between Fixed Effect and Random Effect is that FE
allowed for correlation between the unobserved individual specific effect
and the explanatory variables whereas RE requires these to be
uncorrelated.
Having this, we run a basic fixed effect regression and random effect
regression of independent variables cognitive achievement outcome of
interest (standard score in PPVT test).
To choose from FE and RE we used Hausman
In this case, Random effects (RE) is preferred under the null hypothesis due
to higher efficiency, while under the alternative fixed effects (FE) is at least
consistent and thus preferred.
Using the test of Hausman test we rejected the null hypothesis and
accepted the alternative hypothesis that is fixed effect model preferred.
15
16. DATA AND DESCRIPTIVE EXPLANATIONS
Variable Obs Mean Std dv Min Max
Standard PPVT test core 1648 91.17658 26.38066 40 160
Dummy for drought affected 1663 .3968731 .4893964 0 1
Dummy for PSNP participation 1663 .439567 .4964837 0 1
Rural PSNP participating 1663 .3505713 .4772924 0 1
rural drought affected PSNP beneficiaries 1662 .2178099 .4128817 0 1
dummy for Child health worse 1660 .086747 .2815488 0 1
Hours spent on paid work 1659 .2932188 1.339566 0 11
Dummy for missed more than a week 1511 .1250827 .3309222 0 1
Dummy for land ownership 1664 .9206731 .2703296 0 1
Hours spent on study 1659 1.794039 1.158055 0 8
Grade completed 1504 4.746676 1.917542 0 11
School distance in minute 1505 25.45116 19.5518 0 130
dummy for credit access 1663 .2958509 .4565616 0 1
Dummy household condition poor 1664 .6790865 .4669679 0 1
16
17. RESULTS AND DISCUSSION
Drought shock and PPVT test score by region and year
Source: Author’s calculations based on Young Lives data
Standard PPVT score Amhara Oromia SNNP Tigray Total
not affected
round 2 83 96 88 96 90
round 3 96 99 104 115 101
Change rate (%) 15.66 3.13 18.18 19.79 12.2
affected
round 2 76 76 76 88 80
round 3 79 82 84 98 88
Change rate (%) 3.95 7.89 10.53 11.36 10
Gap between affected and
non-affected
round 2 7 20 12 8 10
round 3 17 17 20 17 13
17
19. RESULT AND DISCUSSION
Drought enrolment and grade completed
Source: Author’s calculations based on Young Lives data
outcomes
Full
sample
(R2)
Non
Affected
(R2)
Affected
(R2)
Full
sample
(R3)
non
Affected
(R3)
Affected
(R3)
Enrollment (%) 97.37 97.30 97.50 89.60 92.16 85
grade completed
(progress)
4.25 4.3 3.54 5.74 5.7 5.15
school Attendance 5.32 5.24 5.52 10.64 9.53 14.09
Taken grade 8 21% 27% 11%
19
20. RESULT AND DISCUSSION
old cohort children time use on typical day by drought shock
Source: Author’s calculations based on Young Lives data
on a typical days
hours spent on
Round 2 Round 3
Full
sample
non
Affected
Affected
Full
sample
non
Affected
Affected
caring for others 0.65 0.65 0.66 0.7 0.75 0.74
Domestic work 2.3 2.34 2.2 2.7 2.67 2.75
Farming activity 1.69 1.42 2.17 1.57 1.31 1.88
Paying work 0.15 0.11 0.25 0.42 0.36 0.51
School 5.15 5.27 4.93 5.52 5.31 5.05
Study outside
school
1.72 1.81 1.56 1.86 2 1.7
Sleep 9.04 8.91 9.26 8.65 8.56 8.77
Leisure 2.71 2.91 2.33 2.81 3.02 2.52
child labor 4.79 4.52 5.28 5.39 5.09 5.88
20
21. RESULT AND DISCUSSION
old cohort children reading and writing skill by drought incidence
Source: Author’s calculations based on Young Lives data
9% 12% 12%
67%
9%
29%
62%
14%
19% 19%
47%
16%
39%
45%
0%
10%
20%
30%
40%
50%
60%
70%
80%
can't read reads
letters
reads
words
reads
sentence
No with
difficulty
without
difficult
Reading level Writing skills
not Affected
by drought
Affected by
drought
21
22. RESULT AND DISCUSSION
Households affected by drought and PSNP participation
Source: Author’s calculations based on Young Lives data
Region
Affected
PSNP
participate
d R2)
affected
but
supported
(R2)
cohort
mean
PPVT
(R2)
Affected
PSNP
participat
ed (R3)
affected
but
supporte
d (R3)
cohort
mean
PPVT
(R3)
response
capacity
growth
Amhara 77% 78% 80.57 20% 72% 88.31 -7.60%
Oromia 70% 73% 88.29 53% 97% 99.33 32.60%
SNNP 61% 73% 85.69 7% 86% 95.54 18.70%
Tigray 93% 93% 91.82 69% 97% 102.95 5.00%
Total 78% 81% 91.29 40% 89% 102.95 8.90%
22
23. RESULT AND DISCUSSION
During round two more than 78 percent of the
households was PSNP participants but reduced to 40
percent during round three that partly may be due
graduation.
drought responding capacity almost fifty percent
reduction.
other food/cash support from small projects (may be
as emergency relief programs) increased as a result
percentage of drought affected but supported
households increased in round three.
23
24. RESULT AND DISCUSSION
Economic Estimation
PPVT test estimation
Dependent variable standard PPVT test score Fe model Re model
Dummy for drought affected -6.732429*** -8.048916***
-2.92 -4.93
dummy for Child health worse -4.658618 -5.031659**
-1.53 -2.31
Hours spent on paid work -2.010522* -2.463492***
-1.75 -3.26
rural drought affected PSNP beneficiaries 6.428558** 3.934902**
2.28 2.02
Dummy for missed more than a week -3.907601 -2.166414
-1.59 -1.21
Dummy for landownership 3.887701 -11.25698***
0.88 -4.88
Hours spent on study per day 2.248315 2.562031***
2.74 4.36
Grade completed 3.503539*** 4.636148***
3.91 12.86
School distance in minute -0.0535196*** -0.0553434*
-1.04 -1.74
24
25. RESULT AND DISCUSSION
Economic Estimation
PPVT test estimation
Dependent variable standard PPVT test score Fe model Re model
dummy for credit access 2.29007 0.1541255
1.31 0.12
Dummy household condition poor -4.268674*** -3.525394***
-2.22 -2.76
R-sq: within 0.137 0.1193
between 0.2468 0.3153
overall 0.1974 0.2425
Number of obs 1484 1484
Number of groups 812 812
F test that all u_i=0 1.71
sigma_u 19.439141 10.434154
sigma_e 19.718549 19.718549
Rho 0.49286489 0.21875258
Hausman test of FE vs. RE; 2(17)=28.81, p=0.0024
Robust z statistics and t value in parentheses for RE and Fe respectively
* significant at 10%; ** significant at 5%; *** significant at 1% 25
26. RESULTS AND DISCUSSION
We found that drought shock affected cognitive
development of children negatively. (Similar to the findings
of Aderman, 2006 and Mill and Shah (2012)
The PPVT test score of old cohort children affected by
drought shock reduced by almost 7 points than that of old
cohort children not affected by drought shock.
26
27. RESULTS AND DISCUSSION
We found that child health worse has negative effect on PPVT
score similar to the findings of Spernak 2006
hours spent on paid work, land ownership,
hours spent on study and grade completion have positive effect
test score similar to the findings of Ray and Lancaster, 2003
Meanwhile old cohort children from poor household registered
lower results than that of non-drought affected old cohort children.
The impact of drought dummy for missed class more than a week
with PPVT test score is negative thought it is not statistically
significant.
27
28. RESULTS AND DISCUSSION
We found that drought affected rural children whose household
participated in the productive safety nets program scores 6 points higher
than of non participant drought affected rural children .
We found that child health hours spent on paid work, land ownership,
hours spent on study, grade completion have a significant and positive
relationship with standard PPVT test score.
Meanwhile old cohort children from poor household registered lower
results than that of non-drought affected old cohort children.
The impact of drought dummy for missed class more than a week with
PPVT test score is negative thought it is not statistically significant.
dummy credit access has positive relationship children.
28
29. CONCLUSIONS AND RECOMMENDATIONS
Conclusions
Drought shock has a negative impact on children PPVT test score. Old
cohort children affected by drought scores PPVT test score 7 point lower
than those not affected.
Household condition poor and school distance in minutes negatively
affect PPVT score
Drought also negatively affects school enrollment , grade completion
school attendance, hours spent on study outside school.
Children who are exposed to drought shock score significantly worse in
reading and writing skills.
29
30. CONCLUSIONS AND RECOMMENDATIONS
Conclusions
productive safety net programme participation has a positive and
significant impact on drought affected children cognitive development.
Moreover productive safety improves children cognitive development by
improving household economic condition. So drought mitigation
measures such as food aid reverses help to improve cognitive
development of children affected by the drought.
30
31. CONCLUSIONS AND RECOMMENDATIONS
Recommendations
Strengthening prevention and mitigation measures (DRR)
the identification of efficient agricultural management practices, and
provision of timely advice to farmers
improving drought responding capacity/ emergency response (especially
Amhara region)
as strengthen their early warning systems such as Short message
system (SMS) throughout all woredas of the region.
Moreover microfinance institution can also create access for credit
service to the drought affected households.
National Learning Assessment has to include as factor that affects
children academic achievement.
31
Methrological – based on the degree of dryness and duration of dry periodHydrological – when the surface and ground water levels falls below the averageAgricultural drought- shortage of water for crop growth – departure from the normalEcological- occurs when primary productivity of natural ecosystem declines.Socio economic –agregate of all the abovedroughts when
The poorest carry the heaviest burden of the effects of disasters across different determinants and outcomes of human capital. Finally, although the occurrence of natural hazards is mostly out of control of authorities, there still is a significant room for policy action to minimize their impacts on the accumulation of human capital (Baez J. F., 2010).
The Education Sector Development Program IV, an action plan of Ethiopian government in education sector development for 2010/2011 – 2014/2015 academic years.It clearly identified poor academic performance of children as one the major challenges education sector development in Ethiopia. The government of Ethiopia in collaboration with its stakeholders undertakes National Learning Assessments (NLA) with the main purpose to find out the extent to which learning takes place and determine the main factors that influence the learning outcomes of students.
It is an innovative collaborative study investigating the changing nature of childhood poverty over 15 years in four countries (namely: Ethiopia, India, Peru and Vietnam) and seeks to contribute to broader poverty reduction policy knowledge and practice. However, this paper is based on Ethiopia data set a project led by the University of on childhood poverty, the Ethiopia dataset. Young Lives study is tracking the overall development of 12,000 children using longitudinal quantitative and qualitative research over a 15-year period in the four countries, 3, 000 children in each country. Similar to others countries in Ethiopia, the research has been following two groups of children as old cohort 2,000 children aged between 6 and 18 months and 1,000 children between 7.5 and 8.5 years old at the start of the research, 2002. The paper uses the second and third round longitudinal surveys undertaken in Ethiopia. The survey collected both young cohort and old cohort children data. However the paper limits itself to uses old cohort children datasets. During the two consecutive round surveys the cohort age was 12 and 15 respectively.
The descriptive statistics shows that, the average Standard Peabody Picture vocabulary test score was nighty one (See Table 6). The table also shows that largest and smallest standard Peabody Picture vocabulary test score of the total respondents was 40 and 160 respectively. The table also shows that forty percent of households reported that they encountered with incidence of drought shock. On average in the two rounds, PSNP participation reached forty four in general and 35 percent households in rural site. This shows PSNP participation is in favor of urban households.
PPVT test score of children affected by drought is lower than those not affected by drought shock in all regions and all rounds.The mean standard test score of old cohort children was 87 (R2) and 96 (R3) In round two PPVT test score of drought affected children in Amhara, Oromia and SNNP was 76 while 88 in Tigray. Meanwhile the PPVT of children not affected during the same round was higher in Tigray and Oromia followed by SNNP and Amhara. During the third round, the PPVT test score of drought affected children of Amhara, Oromia and SNNP was almost similar but significantly higher in Tigray. The test score gap between those affected and non-affected R2, highest in Oromia followed by SNNP, Tigray and Amhara. R3, similar except higher in SNNP. The growth rate of PPVT test score of non affected is higher than those children affect by drought shock except in Oromia region.To sum up, the standard PPVT test score of the cohort grows between the two rounds with higher growth rate for drought non-affected than those affected by the shock. The standard test score gab between drought affected and non-affected children of the cohort widens during round three except in the case of Oromia regional statesTo sum up, the standard PPVT test score of the cohort grows between the two rounds with higher growth rate for drought non-affected than those affected by the shock. The standard test score gab between drought affected and non-affected children of the cohort widens during round three except in the case of Oromia regional states
PPVT test score of children affected by drought is lower than those not affected by drought shock in all regions and all rounds.The mean standard test score of old cohort children was 87 (R2) and 96 (R3) In round two PPVT test score of drought affected children in Amhara, Oromia and SNNP was 76 while 88 in Tigray. Meanwhile the PPVT of children not affected during the same round was higher in Tigray and Oromia followed by SNNP and Amhara. During the third round, the PPVT test score of drought affected children of Amhara, Oromia and SNNP was almost similar but significantly higher in Tigray. The test score gap between those affected and non-affected R2, highest in Oromia followed by SNNP, Tigray and Amhara. R3, similar except higher in SNNP. The growth rate of PPVT test score of non affected is higher than those children affect by drought shock except in Oromia region.To sum up, the standard PPVT test score of the cohort grows between the two rounds with higher growth rate for drought non-affected than those affected by the shock. The standard test score gab between drought affected and non-affected children of the cohort widens during round three except in the case of Oromia regional statesTo sum up, the standard PPVT test score of the cohort grows between the two rounds with higher growth rate for drought non-affected than those affected by the shock. The standard test score gab between drought affected and non-affected children of the cohort widens during round three except in the case of Oromia regional states
grade completion was 4.25 (R2) and 5.74 (R3), though it is expected to be 5 and 8 respectively. The results reveals that there was good completion rate at early grades and highly reduce when the children age and grade level increases.The grade completion difference/gap between drought affected and non affected during second round was wider than in round two that implies drought affected students’ dropout rate and/or grade repetition rate is very high during the second round that is because with increase in age of the cohort their importance for paid and unpaid work increased. grade completion was 4.25 (R2) and 5.74 (R3), though it is expected to be 5 and 8 respectively. The results reveals that there was good completion rate at early grades and highly reduce when the children age and grade level increases.The grade completion difference/gap between drought affected and non affected during second round was wider than in round two that implies drought affected students’ dropout rate and/or grade repetition rate is very high during the second round that is because with increase in age of the cohort their importance for paid and unpaid work increased.
Child labor (hours spent on caring for others, doing domestic activity, and paid work, family farm /cattle herding) was 4.79 hours per day during round two and 5.39 during round three. During the third round survey, it was found that all child labor components except hours spent on family farm /cattle herding increased. child labor of affected was 5.29 and non affected was only 4.11 hours per day during the second round.In 2009 child labor of affected was 5.88 and non affected 4.55 hours per day. The gap widens between rounds, 1.18 hour per day in the second and 1.33 hours per day during the third round. It also found that child labor is high in rural Ethiopia compared with the urban one during both rounds and the increase in hours on child labor in rural is higher than that of the urban children. It is also observed that even the rural urban child labor gap widens as the age of children increase, i.e. 1.91 hours per day at age of 12 and 2.31 hours per day when their age reached 15 years old. In general child labor is high in Ethiopia and which is not in compliance with child right convention that says child labor is child work if it is more than two hours. In general child labor is high in Ethiopia and which is not in compliance with child right convention that says child labor is child work if it is more than two hours. The table helps to see the effect of drought on time use by old cohort children. During the second round, old cohort children from drought shock affected spent higher hours on child labor components except on domestic work. Similarly, hours spent on school and study and leisure is lower among drought affected children. Hours spent on schools and study outside schools have strong effected on children academic achievement. However in round three all child labor components are higher among drought affected children. Similar to the second round, hours spent on the schools and study outside school of children affected by drought is lower than those of not affected. In general, it is observed that drought shock forced children to spend more hours on child labor and less hours on school and study outside school that in turn, as a result lowers children cognitive development or students’ academic performance.