does market access mitigate the impact of seasonality on child growth?panel data evidence from...

20
Does market access mitigate the impact of seasonality on child growth? Panel data evidence from northern Ethiopia Kibrewossen Abay Kalle Hirvonen International Food Policy Research Institute CSAE Conference – March 21 2016

Upload: essp2

Post on 12-Apr-2017

154 views

Category:

Data & Analytics


0 download

TRANSCRIPT

Page 1: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Does market access mitigate the impact of seasonality on child growth?

Panel data evidence from northern Ethiopia

Kibrewossen AbayKalle Hirvonen

International Food Policy Research Institute

CSAE Conference – March 21 2016

Page 2: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Motivation #1• Child undernutrition still a major problem • Sub-Saharan Africa: 40% of stunted and 22% underweight children

• Lack of irrigation means that agricultural production takes places in seasonal cycles• Implications for access to and prices of foods (Hirvonen et al 2016, Kaminski

et al 2014, Gilbert et al 2015).

• In the nutrition literature seasonal energy stress considered a major contributor to child undernutrition (e.g. Vaitla et al 2009)• Even temporary low energy intakes can have serious nutritional

implications, especially for children

Page 3: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Motivation #2• Previous studies documented households closer to market • Smooth their consumption (Handa and Mlay 2006)• More likely to participate in non-farm activities (Jacoby and Minten 2009)• Better price for their produce (Minten 2009)• Enjoy better diets (Stifel & Minten, 2015)• Consumption less dependent on own agricultural production (Hirvonen &

Hoddinott 2014).

• Could better market access insulate children from seasonal weight fluctuations?• We want to test this hypothesis

Page 4: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Data #1• The study focuses on Tigray, northernmost region of Ethiopia• The data used in this paper come from a household panel survey from

South-Eastern part of Tigray• The purpose of the survey was to evaluate the impact of Social Cash

Transfer Pilot Program• 7 rounds of data between May 2012 and May 2014, with

anthropometric measures of children• Ideal for studying seasonality

Page 5: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Data #2• The survey was fielded in two districts- one urban and one rural• We restrict our study in the rural district

• Covers 8 sub-districts and 27 villages• 2,387 households selected in the final sample

• There are four markets accessible to these 8 sub-districts• The Euclidian distance to the nearest market ranges from 0.5 km to 9 km• About one fourth of the villages are located within 3.5km distance from

food market• We categorize the interview dates into lean and non-lean seasons

• 3 lean season rounds and 4 non-lean season months

Page 6: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia
Page 7: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Methods• We model anthropometric outcome for child i at time t residing in village v as:

=1 if child measured during lean season distance to the nearest food market (linear, kms) household level controls: head age sex and literacy, hh wealth and demographics child level controls: age (spline function), sex linear time trend

Page 8: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Methods• To account for possibility of non-linear relationship, we also estimate:

=1 if food sufficient season and close to food market=1 if food sufficient season and not close to food market=1 if lean season and village is close to marketReference: lean season and far from market

Expected

Page 9: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Results

Page 10: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Results• Seasonality exerts considerable fluctuations on children’s weights• Average child (34 months of age) in the sample is 0.34 kilograms lighter in the

lean season

• Children located farther away from food markets have lower WAZ and WHZ• coefficient on the interaction term appears insignificant • seasonal fluctuations do not differ between the more or less remote

households

• These results are consistent in alternative specifications (village and household fixed effects model)

Page 11: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Results

Page 12: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Results• WHZ scores during lean season and close to market are on average

0.18 units higher than lean season and far from market• WAZ scores during lean season and close to market are on average

0.19 units higher than lean season and far from market• Sufficient season WAZ scores are 0.31 units higher in villages with

good market access compared to villages with poorer market access• For an average child 0.31 translates in to 0.53 kilograms

• Children’s weights fluctuate considerably across the two seasons regardless of distance from food market (next figure)

Page 13: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Predictive margins for WAZ

Page 14: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Potential pathways• What could explain higher WAZ and WHZ among children located

closer to markets?

• Children located in villages with better market access enjoy more diverse diets, both in lean and non-lean seasons

• The content of this diet however changes across the seasons for these children near markets• Animal source food consumption increases considerably during the good season

• We find little evidence that WHZ and WAZ are driven by diarrhea outbreaks

Page 15: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia
Page 16: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Summary of the findings• Children located closer to markets are healthier and better nourished• But: their (age or height adjusted) weights are subjected to

considerable seasonal fluctuations• Magnitude is similar to the children residing in more remote areas

• Further analysis suggests that these patterns are driven by diets• Children located closer to markets consume more diverse diets• But the content of this diets vary: children near markets are much more likely

to consume milk or milk products during the good season

• Limitation: no data on food quantities

Page 17: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Implications for policy• Seasonality remains an important problem that has been neglected in

the policy arenas (Kaminski et al, 2014)• More emphasis needed in nutrition programming

• Market access does not insulate children from the impacts of seasonality• Other policies are needed:

• Better storage technologies (~ but mainly for less perishable staple food crops)• Irrigation – so that food can be produced outside the main cropping season• Broader integration of the markets (considerable within country variation in the

timing of the cropping seasons)

Page 18: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Appendix

Page 19: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Data• We use these rich data to study seasonality in child health outcomes• Seven rounds of panel data between May 2012 and May 2014

Survey round Start date Finish date

Round 1 (base-line)* 6 May 2012 26 Jun 2012Round 2 6 Oct 2012 27 Oct 2012Round 3 9 Mar 2013 29 Mar 2013Round 4* 20 Jul 2013 11 Aug 2013Round 5 6 Nov 2013 26 Nov 2013Round 6 3 Mar 2014 25 Mar 2014Round 7 (end-line)* 16 May 2014 17 Jul 2014

* Indicates lean season (May to September)- the classification is based on sources of consumption (here)

Page 20: Does market access mitigate the impact of seasonality on child growth?Panel data evidence from northern Ethiopia

Results