1. Resilience, Poverty and Malnutrition in Mali
Rebecca Pietrelli
Economist
ESA Division-FAO
Marco d’Errico Economist
ESA Division-FAO
Francesca Grazioli
Economist
ESA Division-FAO
#sdafrica2015
26-27 November 2015, Dakar

2. Outline Contribution Resilience measurement Data
Identification strategy
Conclusions
Data
Findings
Outline

3. (1) To estimate the resilience capacity index (RCI) by using the FAO - Resilience Index Measurement and Analysis (RIMA) model in Mali (2009/10).
(2) Is resilience capacity a determinant of household expenditure?
(3) Does resilience capacity affect child malnutrition?
Contribution

4. Resilience measurement
The RIMA model adopts a 2-step procedure:
The pillars (Assets – AST; Access to basic Services – ABS; Adaptive Capacity – AC; Sensitivity - S) are estimated through Factor Analysis from observed variables.
A Structural Equation Model (SEM) predicts the Resilience index by identifying the relation between the pillars:
Resilience measurement
Figure 1. Resilience index and pillars

5. Multiple Indicator Cluster Survey - Enquête Légère Intégrée aprés des Ménages (MICS-ELIM) -> 8,660 Households interviewed in 2009/10.
Data
2. Communes survey: 703 communes surveyed in 2008.

6. Data
Figure 2. HH expenditure by low, medium and high resilience

7. Data
. Following WHO (2006), is has been specifically considered:
stunting, a measure of chronic nutritional deficiency, based on a child’s height and age;
2) wasting, a measure of short-term nutritional deficiency, based on a child’s weight and height;
3) underweight, a composite measure of chronic and acute nutritional deficiency, based on a child’s weight and age.
1) A child is considered stunted if his Height-for-Age Z-score (HAZ) falls below 2 standard deviations below the normal height for his age-gender group.
2) A child is considered wasted if his Weight-for-Height Z-score (WHZ) falls below 2 standard deviations below the normal weight for his age-gender group.
3) A child is considered underweight if his Weight-for-Age Z-score (WAZ) falls below 2 standard deviations below the normal height for his age-gender group.
Figure 3. Percentage of households with malnourished children by low, medium and high resilience

8. Identification strategy
Two-stage least square regressions (2SLS):
𝑌 ℎ,𝑡 =𝛽 𝑅𝑒𝑠 ℎ,𝑡 +𝛾 𝑿 ℎ,𝑡 +𝛿 𝑫 ℎ,𝑡 + 𝜀 ℎ,𝑡 (1)
household expenditure per capita;
three dummies for having at least one stunted, wasted or underweight child;
three continuous variables expressing the number of stunted, wasted and under-weight children in the household.
𝑅𝑒𝑠 ℎ,𝑡 = 𝛽 𝑇𝑒𝑐ℎ 𝑆𝑒𝑟𝑣𝑖𝑐𝑒𝑠 𝑐,𝑡−1 +𝛾 𝑿 ℎ,𝑡 +𝛿 𝑫 ℎ,𝑡 + 𝜀 ℎ,𝑡 (2)
Identification strategy
In the case of (ii) we use IVProbit since the outcome variable is a dummy.
There are several reasons why resilience might be endogenous.
First, there is reverse causality as expenditure or malnutrition might affect resilience. The richer household may spend more on, for example, education of household members, a key aspect of household resilience. Children’s poor health might also affect family resilience, if additional resources have to be spent on childcare.
Second, there may be omitted factors that affect both resilience, on one hand, and expenditure or child malnutrition, on the other. Some example may be household ability or networks.
Third, there may be a measurement error of resilience itself.

9. Identification strategy
Validity of the instrumental variable:
1. The instrument must be exogenous
𝐶𝑜𝑣 𝑇𝑒𝑐ℎ 𝑆𝑒𝑟𝑣𝑖𝑐𝑒, 𝑂𝑢𝑡𝑐𝑜𝑚𝑒𝑠 =0
2. The instrument must be correlated with the endogenous explanatory variable
𝐶𝑜𝑣 𝑇𝑒𝑐ℎ 𝑆𝑒𝑟𝑣𝑖𝑐𝑒, 𝑅𝑒𝑠 ≠0
Identification strategy
The instrument need to be:
Valid – it has to influence the outcome only by the resilience index. Any direct effect of the presence of state on the outcomes can not be reported. We use a time lag.
Strong – There must be a strong correlation between the instrumental variable and resilience. We can test the correlation by OLS.

10. Table 1. First-stage: Instrumenting regression results for Resilience
Resilience
Number of technical services of the State (per 100 inhabitants)
3.102***
(0.337)
HH size
( )
Squared HH size
2.31e-05
(5.52e-05)
Age of HH head
***
( )
Male of HH head
(0.0274)
Constant
1.474***
(0.0409)
Observations
8,548
R-squared
0.351
Regional dummies are included.
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Findings

11. Findings Table 2 OLS and 2SLS models of household expenditure (a) (b)
(a)
(b)
OLS
2SLS
Resilience
0.272***
0.504***
( )
(0.0650)
HH size
***
***
( )
( )
Squared HH size
***
***
(3.05e-05)
(3.31e-05)
Age of HH head
***
( )
( )
Male of HH head
0.0764***
0.0865***
(0.0151)
(0.0167)
Kayes *
0.271***
(0.0198)
(0.0874)
Koulikoro
0.0163
0.318***
(0.0196)
(0.0867)
Sikasso
-0.436***
-0.143*
(0.0189)
(0.0841)
Segou
-0.189***
0.126
(0.0904)
Mopti
-0.116***
0.230**
(0.0199)
(0.0988)
Tomboctou
***
0.289***
(0.0215)
(0.105)
Gao
***
0.299***
(0.0224)
(0.103)
Kidal
0.0548**
0.489***
(0.0241)
(0.124)
Constant
13.16***
12.81***
(0.0242)
(0.100)
Observations
8,548
R-squared
0.602
0.531
Cragg-Donald Wald F statistic
84.75
Angrist-Pischke F test
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Findings

12. Table 3 Probit and IV Probit models of having malnourished children
Stunting
Wasting
Underweight
(a)
(b)
Probit
IVProbit
Resilience
-0.367***
-0.868***
-0.284***
-0.817***
-0.140***
-0.785***
(0.0216)
(0.137)
(0.0224)
(0.144)
(0.0240)
(0.147)
Controlling for HH characteristics and regional dummies.
Findings
Table 4 Probit and IV Probit models of the N. of malnourished children
Stunting
Wasting
Underweight
(a)
(b)
OLS
2SLS
Resilience
-0.178***
-0.392***
-0.101***
-0.322***
-0.031***
-0.211***
(0.0101)
(0.104)
( )
(0.0877)
( )
(0.0618)
Controlling for HH characteristics and regional dummies.

13. Conclusions Household resilience capacity has the potential to:
increase household expenditure
decrease the probability of having malnourished children
and decrease the number of malnourished children.
-> More evidence from other countries and different HH profiles is needed.
Conclusions

14. THANK YOU! Marco.Derrico@fao.org Rebecca.Pietrelli@fao.org
Contact us: